diff options
488 files changed, 15167 insertions, 13552 deletions
@@ -1,4 +1,34 @@ -# Names should be added to this file like so: -# Name or Organization <email address> +# This file is automatically generated from the git commit history +# by tools/gen_authors.sh. +Adrian Grange <agrange@google.com> +Alex Converse <alex.converse@gmail.com> +Andres Mejia <mcitadel@gmail.com> +Fabio Pedretti <fabio.ped@libero.it> +Frank Galligan <fgalligan@google.com> +Fredrik Söderquist <fs@opera.com> +Fritz Koenig <frkoenig@google.com> +Giuseppe Scrivano <gscrivano@gnu.org> +Guillermo Ballester Valor <gbvalor@gmail.com> +James Zern <jzern@google.com> +Jan Kratochvil <jan.kratochvil@redhat.com> +Jeff Muizelaar <jmuizelaar@mozilla.com> +Jim Bankoski <jimbankoski@google.com> +Johann Koenig <johannkoenig@google.com> +John Koleszar <jkoleszar@google.com> +Justin Clift <justin@salasaga.org> +Justin Lebar <justin.lebar@gmail.com> +Luca Barbato <lu_zero@gentoo.org> +Makoto Kato <makoto.kt@gmail.com> +Michael Kohler <michaelkohler@live.com> +Paul Wilkins <paulwilkins@google.com> +Pavol Rusnak <stick@gk2.sk> +Philip Jägenstedt <philipj@opera.com> +Scott LaVarnway <slavarnway@google.com> +Timothy B. Terriberry <tterribe@xiph.org> +Tom Finegan <tomfinegan@google.com> +Yaowu Xu <yaowu@google.com> +Yunqing Wang <yunqingwang@google.com> Google Inc. +The Mozilla Foundation +The Xiph.Org Foundation @@ -2,18 +2,15 @@ LOCAL_PATH := $(call my-dir) include $(CLEAR_VARS) LOCAL_SRC_FILES = \ - vpx_codec/src/vpx_codec.c \ - vpx_codec/src/vpx_decoder.c \ - vpx_codec/src/vpx_decoder_compat.c \ - vpx_codec/src/vpx_image.c \ - vpx_codec/src/vpx_encoder.c \ + vpx/src/vpx_codec.c \ + vpx/src/vpx_decoder.c \ + vpx/src/vpx_image.c \ vpx_mem/vpx_mem.c \ vpx_scale/generic/vpxscale.c \ vpx_scale/generic/yv12config.c \ vpx_scale/generic/yv12extend.c \ vpx_scale/generic/scalesystemdependant.c \ vpx_scale/generic/gen_scalers.c \ - vp8/common/segmentation_common.c \ vp8/common/alloccommon.c \ vp8/common/blockd.c \ vp8/common/debugmodes.c \ @@ -47,13 +44,14 @@ LOCAL_SRC_FILES = \ vp8/decoder/dboolhuff.c \ vp8/decoder/decodemv.c \ vp8/decoder/decodframe.c \ - vp8/decoder/demode.c \ vp8/decoder/dequantize.c \ vp8/decoder/detokenize.c \ vp8/decoder/generic/dsystemdependent.c \ vp8/decoder/onyxd_if.c \ vp8/decoder/threading.c \ + vp8/vp8_dx_iface.c \ vpx_config.c \ + vp8/decoder/arm/neon/idct_blk_neon.c \ LOCAL_CFLAGS := \ -DHAVE_CONFIG_H=vpx_config.h @@ -102,9 +100,10 @@ ASM_FILES = \ vp8/common/arm/neon/sixtappredict4x4_neon.s \ vp8/common/arm/neon/sixtappredict8x4_neon.s \ vp8/common/arm/neon/sixtappredict8x8_neon.s \ - vp8/decoder/arm/neon/dequantdcidct_neon.s \ - vp8/decoder/arm/neon/dequantidct_neon.s \ - vp8/decoder/arm/neon/dequantizeb_neon.s + vp8/common/arm/neon/dc_only_idct_add_neon.s \ + vp8/decoder/arm/neon/dequantizeb_neon.s \ + vp8/decoder/arm/neon/dequant_dc_idct_neon.s \ + vp8/decoder/arm/neon/dequant_idct_neon.s \ # All the assembly sources must be converted from ADS to GAS compatible format VPX_GEN := $(addprefix $(intermediates)/, $(ASM_FILES)) @@ -1,3 +1,68 @@ +2010-09-02 v0.9.2 + - Enhancements: + Disable frame dropping by default + Improved multithreaded performance + Improved Force Key Frame Behaviour + Increased rate control buffer level precision + Fix bug in 1st pass motion compensation + ivfenc: correct fixed kf interval, --disable-kf + - Speed: + Changed above and left context data layout + Rework idct calling structure. + Removed unnecessary MB_MODE_INFO copies + x86: SSSE3 sixtap prediction + Reworked IDCT to include reconstruction (add) step + Swap alt/gold/new/last frame buffer ptrs instead of copying. + Improve SSE2 loopfilter functions + Change bitreader to use a larger window. + Avoid loopfilter reinitialization when possible + - Quality: + Normalize quantizer's zero bin and rounding factors + Add trellis quantization. + Make the quantizer exact. + Updates to ARNR filtering algorithm + Fix breakout thresh computation for golden & AltRef frames + Redo the forward 4x4 dct + Improve the accuracy of forward walsh-hadamard transform + Further adjustment of RD behaviour with Q and Zbin. + - Build System: + Allow linking of libs built with MinGW to MSVC + Fix target auto-detection on mingw32 + Allow --cpu= to work for x86. + configure: pass original arguments through to make dist + Fix builds without runtime CPU detection + msvs: fix install of codec sources + msvs: Change devenv.com command line for better msys support + msvs: Add vs9 targets. + Add x86_64-linux-icc target + - Bugs: + Potential crashes on older MinGW builds + Fix two-pass framrate for Y4M input. + Fixed simple loop filter, other crashes on ARM v6 + arm: fix missing dependency with --enable-shared + configure: support directories containing .o + Replace pinsrw (SSE) with MMX instructions + apple: include proper mach primatives + Fixed rate control bug with long key frame interval. + Fix DSO link errors on x86-64 when not using a version script + Fixed buffer selection for UV in AltRef filtering + + +2010-06-17 v0.9.1 + - Enhancements: + * ivfenc/ivfdec now support YUV4MPEG2 input and pipe I/O + * Speed optimizations + - Bugfixes: + * Rate control + * Prevent out-of-bounds accesses on invalid data + - Build system updates: + * Detect toolchain to be used automatically for native builds + * Support building shared libraries + * Better autotools emulation (--prefix, --libdir, DESTDIR) + - Updated LICENSE + * http://webmproject.blogspot.com/2010/06/changes-to-webm-open-source-license.html + + 2010-05-18 v0.9.0 - Initial open source release. Welcome to WebM and VP8! diff --git a/CleanSpec.mk b/CleanSpec.mk deleted file mode 100644 index b84e1b6..0000000 --- a/CleanSpec.mk +++ /dev/null @@ -1,49 +0,0 @@ -# Copyright (C) 2007 The Android Open Source Project -# -# Licensed under the Apache License, Version 2.0 (the "License"); -# you may not use this file except in compliance with the License. -# You may obtain a copy of the License at -# -# http://www.apache.org/licenses/LICENSE-2.0 -# -# Unless required by applicable law or agreed to in writing, software -# distributed under the License is distributed on an "AS IS" BASIS, -# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -# See the License for the specific language governing permissions and -# limitations under the License. -# - -# If you don't need to do a full clean build but would like to touch -# a file or delete some intermediate files, add a clean step to the end -# of the list. These steps will only be run once, if they haven't been -# run before. -# -# E.g.: -# $(call add-clean-step, touch -c external/sqlite/sqlite3.h) -# $(call add-clean-step, rm -rf $(PRODUCT_OUT)/obj/STATIC_LIBRARIES/libz_intermediates) -# -# Always use "touch -c" and "rm -f" or "rm -rf" to gracefully deal with -# files that are missing or have been moved. -# -# Use $(PRODUCT_OUT) to get to the "out/target/product/blah/" directory. -# Use $(OUT_DIR) to refer to the "out" directory. -# -# If you need to re-do something that's already mentioned, just copy -# the command and add it to the bottom of the list. E.g., if a change -# that you made last week required touching a file and a change you -# made today requires touching the same file, just copy the old -# touch step and add it to the end of the list. -# -# ************************************************ -# NEWER CLEAN STEPS MUST BE AT THE END OF THE LIST -# ************************************************ - -# For example: -#$(call add-clean-step, rm -rf $(OUT_DIR)/target/common/obj/APPS/AndroidTests_intermediates) -#$(call add-clean-step, rm -rf $(OUT_DIR)/target/common/obj/JAVA_LIBRARIES/core_intermediates) -#$(call add-clean-step, find $(OUT_DIR) -type f -name "IGTalkSession*" -print0 | xargs -0 rm -f) -#$(call add-clean-step, rm -rf $(PRODUCT_OUT)/data/*) - -# ************************************************ -# NEWER CLEAN STEPS MUST BE AT THE END OF THE LIST -# ************************************************ @@ -1,22 +1,20 @@ -Copyright (c) 2010, Google, Inc. - -All rights reserved. +Copyright (c) 2010, Google Inc. All rights reserved. Redistribution and use in source and binary forms, with or without -modification, are permitted provided that the following conditions -are met: +modification, are permitted provided that the following conditions are +met: -- Redistributions of source code must retain the above copyright - notice, this list of conditions and the following disclaimer. + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. -- Redistributions in binary form must reproduce the above - copyright notice, this list of conditions and the following - disclaimer in the documentation and/or other materials provided - with the distribution. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in + the documentation and/or other materials provided with the + distribution. -- Neither the name of Google nor the names of its contributors may - be used to endorse or promote products derived from this software - without specific prior written permission. + * Neither the name of Google nor the names of its contributors may + be used to endorse or promote products derived from this software + without specific prior written permission. THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT @@ -30,19 +28,3 @@ THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. -Subject to the terms and conditions of the above License, Google -hereby grants to You a perpetual, worldwide, non-exclusive, -no-charge, royalty-free, irrevocable (except as stated in this -section) patent license to make, have made, use, offer to sell, sell, -import, and otherwise transfer this implementation of VP8, where such -license applies only to those patent claims, both currently owned by -Google and acquired in the future, licensable by Google that are -necessarily infringed by this implementation of VP8. If You or your -agent or exclusive licensee institute or order or agree to the -institution of patent litigation against any entity (including a -cross-claim or counterclaim in a lawsuit) alleging that this -implementation of VP8 or any code incorporated within this -implementation of VP8 constitutes direct or contributory patent -infringement, or inducement of patent infringement, then any rights -granted to You under this License for this implementation of VP8 -shall terminate as of the date such litigation is filed. @@ -0,0 +1,22 @@ +Additional IP Rights Grant (Patents) + +"This implementation" means the copyrightable works distributed by +Google as part of the WebM Project. + +Google hereby grants to you a perpetual, worldwide, non-exclusive, +no-charge, royalty-free, irrevocable (except as stated in this section) +patent license to make, have made, use, offer to sell, sell, import, +transfer, and otherwise run, modify and propagate the contents of this +implementation of VP8, where such license applies only to those patent +claims, both currently owned by Google and acquired in the future, +licensable by Google that are necessarily infringed by this +implementation of VP8. This grant does not include claims that would be +infringed only as a consequence of further modification of this +implementation. If you or your agent or exclusive licensee institute or +order or agree to the institution of patent litigation against any +entity (including a cross-claim or counterclaim in a lawsuit) alleging +that this implementation of VP8 or any code incorporated within this +implementation of VP8 constitutes direct or contributory patent +infringement, or inducement of patent infringement, then any patent +rights granted to you under this License for this implementation of VP8 +shall terminate as of the date such litigation is filed. @@ -9,18 +9,18 @@ COMPILING THE APPLICATIONS/LIBRARIES: the application. 1. Prerequisites - + * All x86 targets require the Yasm[1] assembler be installed. * All Windows builds require that Cygwin[2] be installed. * Building the documentation requires PHP[3] and Doxygen[4]. If you do not have these packages, you must pass --disable-install-docs to the configure script. - + [1]: http://www.tortall.net/projects/yasm [2]: http://www.cygwin.com [3]: http://php.net [4]: http://www.doxygen.org - + 2. Out-of-tree builds Out of tree builds are a supported method of building the application. For an out of tree build, the source tree is kept separate from the object @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/build/.gitattributes b/build/.gitattributes new file mode 100644 index 0000000..03db79b --- /dev/null +++ b/build/.gitattributes @@ -0,0 +1,2 @@ +*-vs8/*.rules -crlf +*-msvs/*.rules -crlf diff --git a/build/arm-wince-vs8/.gitattributes b/build/arm-wince-vs8/.gitattributes deleted file mode 100644 index be1eeb9..0000000 --- a/build/arm-wince-vs8/.gitattributes +++ /dev/null @@ -1 +0,0 @@ -*.rules -crlf diff --git a/build/arm-wince-vs8/obj_int_extract.bat b/build/arm-wince-vs8/obj_int_extract.bat index e58bdd6..a361fc3 100644 --- a/build/arm-wince-vs8/obj_int_extract.bat +++ b/build/arm-wince-vs8/obj_int_extract.bat @@ -1,12 +1,13 @@ @echo off -REM Copyright (c) 2010 The VP8 project authors. All Rights Reserved. -REM -REM Use of this source code is governed by a BSD-style license and patent -REM grant that can be found in the LICENSE file in the root of the source -REM tree. All contributing project authors may be found in the AUTHORS -REM file in the root of the source tree. +REM Copyright (c) 2010 The WebM project authors. All Rights Reserved. +REM +REM Use of this source code is governed by a BSD-style license +REM that can be found in the LICENSE file in the root of the source +REM tree. An additional intellectual property rights grant can be found +REM in the file PATENTS. All contributing project authors may +REM be found in the AUTHORS file in the root of the source tree. echo on -cl /I ".\\" /I "..\vp6_decoder_sdk" /I "..\vp6_decoder_sdk\vpx_ports" /D "NDEBUG" /D "_WIN32_WCE=0x420" /D "UNDER_CE" /D "WIN32_PLATFORM_PSPC" /D "WINCE" /D "_LIB" /D "ARM" /D "_ARM_" /D "_UNICODE" /D "UNICODE" /D "HAVE_CONFIG_H" /FD /EHsc /MT /GS- /fp:fast /GR- /Fo"Pocket_PC_2003__ARMV4_\%1/" /Fd"Pocket_PC_2003__ARMV4_\%1/vc80.pdb" /W3 /nologo /c /TC ..\vp6_decoder_sdk\vp6_decoder\algo\common\arm\dec_asm_offsets_arm.c
+cl /I ".\\" /I "..\vp6_decoder_sdk" /I "..\vp6_decoder_sdk\vpx_ports" /D "NDEBUG" /D "_WIN32_WCE=0x420" /D "UNDER_CE" /D "WIN32_PLATFORM_PSPC" /D "WINCE" /D "_LIB" /D "ARM" /D "_ARM_" /D "_UNICODE" /D "UNICODE" /FD /EHsc /MT /GS- /fp:fast /GR- /Fo"Pocket_PC_2003__ARMV4_\%1/" /Fd"Pocket_PC_2003__ARMV4_\%1/vc80.pdb" /W3 /nologo /c /TC ..\vp6_decoder_sdk\vp6_decoder\algo\common\arm\dec_asm_offsets_arm.c
obj_int_extract.exe rvds "Pocket_PC_2003__ARMV4_\%1/dec_asm_offsets_arm.obj"
diff --git a/build/arm-wince-vs8/vpx_decoder.sln b/build/arm-wince-vs8/vpx.sln index 2262057..3e49929 100644 --- a/build/arm-wince-vs8/vpx_decoder.sln +++ b/build/arm-wince-vs8/vpx.sln @@ -8,7 +8,7 @@ Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "example", "example.vcproj", EndProject Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "obj_int_extract", "obj_int_extract.vcproj", "{E1360C65-D375-4335-8057-7ED99CC3F9B2}" EndProject -Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "vpx_decoder", "vpx_decoder.vcproj", "{DCE19DAF-69AC-46DB-B14A-39F0FAA5DB74}" +Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "vpx", "vpx.vcproj", "{DCE19DAF-69AC-46DB-B14A-39F0FAA5DB74}" ProjectSection(ProjectDependencies) = postProject {E1360C65-D375-4335-8057-7ED99CC3F9B2} = {E1360C65-D375-4335-8057-7ED99CC3F9B2} EndProjectSection diff --git a/build/make/Makefile b/build/make/Makefile index 412629e..1ca747a 100755 --- a/build/make/Makefile +++ b/build/make/Makefile @@ -1,10 +1,11 @@ ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## @@ -18,6 +19,7 @@ ifeq ($(target),) done all: .DEFAULT clean:: .DEFAULT +install:: .DEFAULT # Note: md5sum is not installed on OS X, but openssl is. Openssl may not be @@ -29,7 +31,7 @@ md5sum := $(firstword $(wildcard \ md5sum := $(if $(filter %openssl,$(md5sum)),$(md5sum) dgst -md5,$(md5sum)) TGT_CC:=$(word 3, $(subst -, ,$(TOOLCHAIN))) -install: +dist: @for t in $(ALL_TARGETS); do \ $(MAKE) --no-print-directory target=$$t $(MAKECMDGOALS) || exit $$?;\ done @@ -37,13 +39,8 @@ install: @if [ -d "$(DIST_DIR)/src" ]; then \ mkdir -p "$(DIST_DIR)/build"; \ cd "$(DIST_DIR)/build"; \ - if [ $(TGT_CC) = "rvct" ] ; then \ - echo "../src/configure --target=$(TOOLCHAIN) --libc=$(ALT_LIBC)"; \ - ../src/configure --target=$(TOOLCHAIN) --libc=$(ALT_LIBC); \ - else \ - echo "../src/configure --target=$(TOOLCHAIN)"; \ - ../src/configure --target=$(TOOLCHAIN); \ - fi; \ + echo "Rerunning configure $(CONFIGURE_ARGS)"; \ + ../src/configure $(CONFIGURE_ARGS); \ $(if $(filter vs%,$(TGT_CC)),make NO_LAUNCH_DEVENV=1;) \ fi @if [ -d "$(DIST_DIR)" ]; then \ @@ -55,7 +52,6 @@ install: fi -svnstat: ALL_TARGETS:=$(firstword $(ALL_TARGETS)) endif ifneq ($(target),) @@ -94,10 +90,10 @@ clean:: rm -f $(OBJS-yes) $(OBJS-yes:.o=.d) $(OBJS-yes:.asm.s.o=.asm.s) rm -f $(CLEAN-OBJS) +.PHONY: dist +dist: .PHONY: install -install: -.PHONY: install-helper -install-helper: +install:: $(BUILD_PFX)%.c.d: %.c $(if $(quiet),@echo " [DEP] $@") @@ -218,6 +214,20 @@ $(1): $(qexec)$$(AR) $$(ARFLAGS) $$@ $$? endef +define so_template +# Not using a pattern rule here because we don't want to generate empty +# archives when they are listed as a dependency in files not responsible +# for creating them. +# +# This needs further abstraction for dealing with non-GNU linkers. +$(1): + $(if $(quiet),@echo " [LD] $$@") + $(qexec)$$(LD) -shared $$(LDFLAGS) \ + -Wl,--no-undefined -Wl,-soname,$$(SONAME) \ + -Wl,--version-script,$$(SO_VERSION_SCRIPT) -o $$@ \ + $$(filter %.o,$$?) $$(extralibs) +endef + define lipo_lib_template $(1): $(addsuffix /$(1),$(FAT_ARCHS)) $(if $(quiet),@echo " [LIPO] $$@") @@ -254,6 +264,9 @@ DOCS=$(call cond_enabled,CONFIG_INSTALL_DOCS,DOCS) @touch $@ INSTALL-DOCS=$(call cond_enabled,CONFIG_INSTALL_DOCS,INSTALL-DOCS) +ifeq ($(MAKECMDGOALS),dist) +INSTALL-DOCS+=$(call cond_enabled,CONFIG_INSTALL_DOCS,DIST-DOCS) +endif .install-docs: .docs $(addprefix $(DIST_DIR)/,$(INSTALL-DOCS)) @touch $@ @@ -265,6 +278,9 @@ BINS=$(call enabled,BINS) @touch $@ INSTALL-BINS=$(call cond_enabled,CONFIG_INSTALL_BINS,INSTALL-BINS) +ifeq ($(MAKECMDGOALS),dist) +INSTALL-BINS+=$(call cond_enabled,CONFIG_INSTALL_BINS,DIST-BINS) +endif .install-bins: .bins $(addprefix $(DIST_DIR)/,$(INSTALL-BINS)) @touch $@ @@ -275,8 +291,12 @@ LIBS=$(call enabled,LIBS) .libs: $(LIBS) @touch $@ $(foreach lib,$(filter %_g.a,$(LIBS)),$(eval $(call archive_template,$(lib)))) +$(foreach lib,$(filter %so.$(VERSION_MAJOR).$(VERSION_MINOR).$(VERSION_PATCH),$(LIBS)),$(eval $(call so_template,$(lib)))) INSTALL-LIBS=$(call cond_enabled,CONFIG_INSTALL_LIBS,INSTALL-LIBS) +ifeq ($(MAKECMDGOALS),dist) +INSTALL-LIBS+=$(call cond_enabled,CONFIG_INSTALL_LIBS,DIST-LIBS) +endif .install-libs: .libs $(addprefix $(DIST_DIR)/,$(INSTALL-LIBS)) @touch $@ @@ -289,6 +309,9 @@ PROJECTS=$(call enabled,PROJECTS) @touch $@ INSTALL-PROJECTS=$(call cond_enabled,CONFIG_INSTALL_PROJECTS,INSTALL-PROJECTS) +ifeq ($(MAKECMDGOALS),dist) +INSTALL-PROJECTS+=$(call cond_enabled,CONFIG_INSTALL_PROJECTS,DIST-PROJECTS) +endif .install-projects: .projects $(addprefix $(DIST_DIR)/,$(INSTALL-PROJECTS)) @touch $@ @@ -296,26 +319,30 @@ clean:: rm -f .projects .install-projects $(PROJECTS) endif -# If there are any source files to be installed, then include the build +# If there are any source files to be distributed, then include the build # system too. -ifneq ($(call enabled,INSTALL-SRCS),) - INSTALL-SRCS-yes += configure - INSTALL-SRCS-yes += build/make/configure.sh - INSTALL-SRCS-yes += build/make/gen_asm_deps.sh - INSTALL-SRCS-yes += build/make/Makefile - INSTALL-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_def.sh - INSTALL-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_proj.sh - INSTALL-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_sln.sh - INSTALL-SRCS-$(CONFIG_RVCT) += build/make/armlink_adapter.sh +ifneq ($(call enabled,DIST-SRCS),) + DIST-SRCS-yes += configure + DIST-SRCS-yes += build/make/configure.sh + DIST-SRCS-yes += build/make/gen_asm_deps.sh + DIST-SRCS-yes += build/make/Makefile + DIST-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_def.sh + DIST-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_proj.sh + DIST-SRCS-$(CONFIG_MSVS) += build/make/gen_msvs_sln.sh + DIST-SRCS-$(CONFIG_MSVS) += build/x86-msvs/yasm.rules + DIST-SRCS-$(CONFIG_RVCT) += build/make/armlink_adapter.sh # # This isn't really ARCH_ARM dependent, it's dependant on whether we're # using assembly code or not (CONFIG_OPTIMIZATIONS maybe). Just use # this for now. - INSTALL-SRCS-$(ARCH_ARM) += build/make/obj_int_extract.c - INSTALL-SRCS-$(ARCH_ARM) += build/make/ads2gas.pl - INSTALL-SRCS-yes += $(target:-$(TOOLCHAIN)=).mk + DIST-SRCS-$(ARCH_ARM) += build/make/obj_int_extract.c + DIST-SRCS-$(ARCH_ARM) += build/make/ads2gas.pl + DIST-SRCS-yes += $(target:-$(TOOLCHAIN)=).mk endif INSTALL-SRCS := $(call cond_enabled,CONFIG_INSTALL_SRCS,INSTALL-SRCS) +ifeq ($(MAKECMDGOALS),dist) +INSTALL-SRCS += $(call cond_enabled,CONFIG_INSTALL_SRCS,DIST-SRCS) +endif .install-srcs: $(addprefix $(DIST_DIR)/src/,$(INSTALL-SRCS)) @touch $@ @@ -329,13 +356,5 @@ endif BUILD_TARGETS += .docs .libs .bins INSTALL_TARGETS += .install-docs .install-srcs .install-libs .install-bins all-$(target): $(BUILD_TARGETS) -install: $(INSTALL_TARGETS) - -# -# Development helper targets -# -ifneq ($(SRC_PATH_BARE),) -.PHONY: svnstat -svnstat: - svn stat $(SRC_PATH_BARE) -endif +install:: $(INSTALL_TARGETS) +dist: $(INSTALL_TARGETS) diff --git a/build/make/ads2gas.pl b/build/make/ads2gas.pl index 6fcba84..3dff048 100755 --- a/build/make/ads2gas.pl +++ b/build/make/ads2gas.pl @@ -1,11 +1,12 @@ #!/usr/bin/perl ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## diff --git a/build/make/ads2gas_apple.pl b/build/make/ads2gas_apple.pl index 569c3e7..5014c61 100755 --- a/build/make/ads2gas_apple.pl +++ b/build/make/ads2gas_apple.pl @@ -1,11 +1,12 @@ #!/usr/bin/env perl ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## diff --git a/build/make/armlink_adapter.sh b/build/make/armlink_adapter.sh index dcaa82c..571e46e 100755 --- a/build/make/armlink_adapter.sh +++ b/build/make/armlink_adapter.sh @@ -1,11 +1,12 @@ #!/bin/bash ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## diff --git a/build/make/configure.sh b/build/make/configure.sh index ed3a34f..e9f8a2b 100755 --- a/build/make/configure.sh +++ b/build/make/configure.sh @@ -1,13 +1,13 @@ #!/bin/bash ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## configure.sh ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## This script is sourced by the main configure script and contains +## utility functions and other common bits that aren't strictly libvpx +## related. +## +## This build system is based in part on the FFmpeg configure script. ## - # @@ -120,8 +120,8 @@ EOF show_targets() { while [ -n "$*" ]; do - if [ "${1%%-*}" == "${2%%-*}" ]; then - if [ "${2%%-*}" == "${3%%-*}" ]; then + if [ "${1%%-*}" = "${2%%-*}" ]; then + if [ "${2%%-*}" = "${3%%-*}" ]; then printf " %-24s %-24s %-24s\n" "$1" "$2" "$3" shift; shift; shift else @@ -348,7 +348,6 @@ true } write_common_target_config_mk() { - [ -n "$2" ] && local have_config_h="-DHAVE_CONFIG_H=\"${2##*/}\"" local CC=${CC} enabled ccache && CC="ccache ${CC}" @@ -367,7 +366,7 @@ AS=${AS} STRIP=${STRIP} NM=${NM} -CFLAGS = ${CFLAGS} ${have_config_h} +CFLAGS = ${CFLAGS} ARFLAGS = -rus\$(if \$(quiet),c,v) LDFLAGS = ${LDFLAGS} ASFLAGS = ${ASFLAGS} @@ -379,7 +378,7 @@ EOF fmt_deps = sed -e 's;^__image.axf;\$(dir \$@)\$(notdir \$<).o \$@;' #hide EOF else cat >> $1 << EOF -fmt_deps = sed -e 's;^\(.*\)\.o;\$(dir \$@)\1\$(suffix \$<).o \$@;' #hide +fmt_deps = sed -e 's;^\([a-zA-Z0-9_]*\)\.o;\$(dir \$@)\1\$(suffix \$<).o \$@;' EOF fi @@ -396,8 +395,6 @@ EOF write_common_target_config_h() { cat > ${TMP_H} << EOF /* This file automatically generated by configure. Do not edit! */ -#define INLINE ${INLINE} -#define FORCEINLINE ${FORCEINLINE:-${INLINE}} #define RESTRICT ${RESTRICT} EOF print_config_h ARCH "${TMP_H}" ${ARCH_LIST} @@ -443,7 +440,13 @@ process_common_cmdline() { disable builtin_libc alt_libc="${optval}" ;; - --libc) + --prefix=*) + prefix="${optval}" + ;; + --libdir=*) + libdir="${optval}" + ;; + --libc|--prefix|--libdir) die "Option ${opt} requires argument" ;; --help|-h) show_help @@ -464,6 +467,18 @@ process_cmdline() { done } + +post_process_common_cmdline() { + prefix="${prefix:-/usr/local}" + prefix="${prefix%/}" + libdir="${libdir:-${prefix}/lib}" + libdir="${libdir%/}" + if [ "${libdir#${prefix}}" = "${libdir}" ]; then + die "Libdir ${libdir} must be a subdirectory of ${prefix}" + fi +} + + post_process_cmdline() { true; } @@ -479,6 +494,43 @@ setup_gnu_toolchain() { } process_common_toolchain() { + if [ -z "$toolchain" ]; then + gcctarget="$(gcc -dumpmachine 2> /dev/null)" + + # detect tgt_isa + case "$gcctarget" in + *x86_64*|*amd64*) + tgt_isa=x86_64 + ;; + *i[3456]86*) + tgt_isa=x86 + ;; + esac + + # detect tgt_os + case "$gcctarget" in + *darwin8*) + tgt_isa=universal + tgt_os=darwin8 + ;; + *darwin9*) + tgt_isa=universal + tgt_os=darwin9 + ;; + *mingw32*|*cygwin*) + [ -z "$tgt_isa" ] && tgt_isa=x86 + tgt_os=win32 + ;; + *linux*|*bsd*) + tgt_os=linux + ;; + esac + + if [ -n "$tgt_isa" ] && [ -n "$tgt_os" ]; then + toolchain=${tgt_isa}-${tgt_os}-gcc + fi + fi + toolchain=${toolchain:-generic-gnu} is_in ${toolchain} ${all_platforms} || enabled force_toolchain \ @@ -730,6 +782,9 @@ process_common_toolchain() { soft_enable ssse3 case ${tgt_os} in + win*) + enabled gcc && add_cflags -fno-common + ;; solaris*) CC=${CC:-${CROSS}gcc} LD=${LD:-${CROSS}gcc} @@ -744,11 +799,23 @@ process_common_toolchain() { setup_gnu_toolchain add_cflags -use-msasm -use-asm add_ldflags -i-static + enabled x86_64 && add_cflags -ipo -no-prec-div -static -xSSE3 -axSSE3 + enabled x86_64 && AR=xiar + case ${tune_cpu} in + atom*) + tune_cflags="-x" + tune_cpu="SSE3_ATOM" + ;; + *) + tune_cflags="-march=" + ;; + esac ;; gcc*) add_cflags -m${bits} add_ldflags -m${bits} link_with_cc=gcc + tune_cflags="-march=" setup_gnu_toolchain ;; esac @@ -781,6 +848,7 @@ process_common_toolchain() { ;; universal*|*-gcc|generic-gnu) link_with_cc=gcc + enable gcc setup_gnu_toolchain ;; esac @@ -834,8 +902,8 @@ EOF # glibc needs these if enabled linux; then - add_cflags -D_LARGEFILE_SOURCE - add_cflags -D_FILE_OFFSET_BITS=64 + add_cflags -D_LARGEFILE_SOURCE + add_cflags -D_FILE_OFFSET_BITS=64 fi } @@ -844,8 +912,8 @@ process_toolchain() { } print_config_mk() { - prefix=$1 - makefile=$2 + local prefix=$1 + local makefile=$2 shift 2 for cfg; do upname="`toupper $cfg`" @@ -856,8 +924,8 @@ print_config_mk() { } print_config_h() { - prefix=$1 - header=$2 + local prefix=$1 + local header=$2 shift 2 for cfg; do upname="`toupper $cfg`" @@ -888,6 +956,7 @@ process() { else echo "# ${self} $@" > ${logfile} fi + post_process_common_cmdline post_process_cmdline process_toolchain process_detect diff --git a/build/make/gen_asm_deps.sh b/build/make/gen_asm_deps.sh index c1118e1..7c6c5d5 100755 --- a/build/make/gen_asm_deps.sh +++ b/build/make/gen_asm_deps.sh @@ -1,11 +1,12 @@ #!/bin/bash ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## diff --git a/build/make/gen_msvs_def.sh b/build/make/gen_msvs_def.sh index 68b2406..4defcc2 100755 --- a/build/make/gen_msvs_def.sh +++ b/build/make/gen_msvs_def.sh @@ -1,11 +1,12 @@ #!/bin/bash ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## diff --git a/build/make/gen_msvs_proj.sh b/build/make/gen_msvs_proj.sh index eb482a4..584477f 100755 --- a/build/make/gen_msvs_proj.sh +++ b/build/make/gen_msvs_proj.sh @@ -1,16 +1,18 @@ #!/bin/bash ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## self=$0 self_basename=${self##*/} +self_dirname=$(dirname "$0") EOL=$'\n' show_help() { @@ -204,7 +206,7 @@ for opt in "$@"; do ;; --ver=*) vs_ver="$optval" case $optval in - [78]) + [789]) ;; *) die Unrecognized Visual Studio Version in $opt ;; @@ -246,6 +248,8 @@ case "${vs_ver:-8}" in ;; 8) vs_ver_id="8.00" ;; + 9) vs_ver_id="9.00" + ;; esac [ -n "$name" ] || die "Project name (--name) must be specified!" @@ -291,8 +295,8 @@ case "$target" in x86*) platforms[0]="Win32" # these are only used by vs7 - asm_Debug_cmdline="yasm -Xvc -g cv8 -f \$(PlatformName) ${yasmincs} \$(InputPath)" - asm_Release_cmdline="yasm -Xvc -f \$(PlatformName) ${yasmincs} \$(InputPath)" + asm_Debug_cmdline="yasm -Xvc -g cv8 -f \$(PlatformName) ${yasmincs} "\$(InputPath)"" + asm_Release_cmdline="yasm -Xvc -f \$(PlatformName) ${yasmincs} "\$(InputPath)"" ;; arm*|iwmmx*) case "${name}" in @@ -342,19 +346,19 @@ generate_vcproj() { open_tag ToolFiles case "$target" in - x86*) $uses_asm && tag DefaultToolFile FileName="yasm.rules" + x86*) $uses_asm && tag ToolFile RelativePath="$self_dirname/../x86-msvs/yasm.rules" ;; arm*|iwmmx*) - if [ "$name" == "vpx_decoder" ];then + if [ "$name" == "vpx" ];then case "$target" in armv5*) - tag DefaultToolFile FileName="armasmv5.rules" + tag ToolFile RelativePath="$self_dirname/../arm-wince-vs8/armasmv5.rules" ;; armv6*) - tag DefaultToolFile FileName="armasmv6.rules" + tag ToolFile RelativePath="$self_dirname/../arm-wince-vs8/armasmv6.rules" ;; iwmmxt*) - tag DefaultToolFile FileName="armasmxscale.rules" + tag ToolFile RelativePath="$self_dirname/../arm-wince-vs8/armasmxscale.rules" ;; esac fi @@ -374,7 +378,7 @@ generate_vcproj() { if [ "$target" == "armv6-wince-vs8" ] || [ "$target" == "armv5te-wince-vs8" ] || [ "$target" == "iwmmxt-wince-vs8" ] || [ "$target" == "iwmmxt2-wince-vs8" ];then case "$name" in - vpx_decoder) tag Tool \ + vpx) tag Tool \ Name="VCPreBuildEventTool" \ CommandLine="call obj_int_extract.bat \$(ConfigurationName)" tag Tool \ @@ -385,7 +389,7 @@ generate_vcproj() { ExecutionBucket="7" \ Optimization="0" \ AdditionalIncludeDirectories="$incs" \ - PreprocessorDefinitions="_DEBUG;_WIN32_WCE=\$(CEVER);UNDER_CE;\$(PLATFORMDEFINES);WINCE;DEBUG;_LIB;\$(ARCHFAM);\$(_ARCHFAM_);_UNICODE;UNICODE;HAVE_CONFIG_H" \ + PreprocessorDefinitions="_DEBUG;_WIN32_WCE=\$(CEVER);UNDER_CE;\$(PLATFORMDEFINES);WINCE;DEBUG;_LIB;\$(ARCHFAM);\$(_ARCHFAM_);_UNICODE;UNICODE;" \ MinimalRebuild="true" \ RuntimeLibrary="1" \ BufferSecurityCheck="false" \ @@ -404,7 +408,7 @@ generate_vcproj() { ExecutionBucket="7" \ Optimization="0" \ AdditionalIncludeDirectories="$incs" \ - PreprocessorDefinitions="_DEBUG;_WIN32_WCE=\$(CEVER);UNDER_CE;\$(PLATFORMDEFINES);WINCE;DEBUG;_CONSOLE;\$(ARCHFAM);\$(_ARCHFAM_);_UNICODE;UNICODE;HAVE_CONFIG_H" \ + PreprocessorDefinitions="_DEBUG;_WIN32_WCE=\$(CEVER);UNDER_CE;\$(PLATFORMDEFINES);WINCE;DEBUG;_CONSOLE;\$(ARCHFAM);\$(_ARCHFAM_);_UNICODE;UNICODE;" \ MinimalRebuild="true" \ RuntimeLibrary="1" \ BufferSecurityCheck="false" \ @@ -435,7 +439,7 @@ generate_vcproj() { Name="VCCLCompilerTool" \ Optimization="0" \ AdditionalIncludeDirectories="$incs" \ - PreprocessorDefinitions="WIN32;_DEBUG;_CRT_SECURE_NO_WARNINGS;$defines" \ + PreprocessorDefinitions="WIN32;_DEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \ RuntimeLibrary="$debug_runtime" \ UsePrecompiledHeader="0" \ WarningLevel="3" \ @@ -508,7 +512,7 @@ generate_vcproj() { if [ "$target" == "armv6-wince-vs8" ] || [ "$target" == "armv5te-wince-vs8" ] || [ "$target" == "iwmmxt-wince-vs8" ] || [ "$target" == "iwmmxt2-wince-vs8" ];then case "$name" in - vpx_decoder) tag DeploymentTool \ + vpx) tag DeploymentTool \ ForceDirty="-1" \ RegisterOutput="0" ;; @@ -532,7 +536,7 @@ generate_vcproj() { if [ "$target" == "armv6-wince-vs8" ] || [ "$target" == "armv5te-wince-vs8" ] || [ "$target" == "iwmmxt-wince-vs8" ] || [ "$target" == "iwmmxt2-wince-vs8" ];then case "$name" in - vpx_decoder) tag Tool \ + vpx) tag Tool \ Name="VCPreBuildEventTool" \ CommandLine="call obj_int_extract.bat \$(ConfigurationName)" tag Tool \ @@ -544,7 +548,7 @@ generate_vcproj() { Optimization="2" \ FavorSizeOrSpeed="1" \ AdditionalIncludeDirectories="$incs" \ - PreprocessorDefinitions="NDEBUG;_WIN32_WCE=\$(CEVER);UNDER_CE;\$(PLATFORMDEFINES);WINCE;_LIB;\$(ARCHFAM);\$(_ARCHFAM_);_UNICODE;UNICODE;HAVE_CONFIG_H" \ + PreprocessorDefinitions="NDEBUG;_WIN32_WCE=\$(CEVER);UNDER_CE;\$(PLATFORMDEFINES);WINCE;_LIB;\$(ARCHFAM);\$(_ARCHFAM_);_UNICODE;UNICODE;" \ RuntimeLibrary="0" \ BufferSecurityCheck="false" \ UsePrecompiledHeader="0" \ @@ -563,7 +567,7 @@ generate_vcproj() { Optimization="2" \ FavorSizeOrSpeed="1" \ AdditionalIncludeDirectories="$incs" \ - PreprocessorDefinitions="NDEBUG;_WIN32_WCE=\$(CEVER);UNDER_CE;\$(PLATFORMDEFINES);WINCE;_CONSOLE;\$(ARCHFAM);\$(_ARCHFAM_);_UNICODE;UNICODE;HAVE_CONFIG_H" \ + PreprocessorDefinitions="NDEBUG;_WIN32_WCE=\$(CEVER);UNDER_CE;\$(PLATFORMDEFINES);WINCE;_CONSOLE;\$(ARCHFAM);\$(_ARCHFAM_);_UNICODE;UNICODE;" \ RuntimeLibrary="0" \ BufferSecurityCheck="false" \ UsePrecompiledHeader="0" \ @@ -593,7 +597,7 @@ generate_vcproj() { x86*) tag Tool \ Name="VCCLCompilerTool" \ AdditionalIncludeDirectories="$incs" \ - PreprocessorDefinitions="WIN32;NDEBUG;_CRT_SECURE_NO_WARNINGS;$defines" \ + PreprocessorDefinitions="WIN32;NDEBUG;_CRT_SECURE_NO_WARNINGS;_CRT_SECURE_NO_DEPRECATE;$defines" \ RuntimeLibrary="$release_runtime" \ UsePrecompiledHeader="0" \ WarningLevel="3" \ @@ -670,7 +674,7 @@ generate_vcproj() { if [ "$target" == "armv6-wince-vs8" ] || [ "$target" == "armv5te-wince-vs8" ] || [ "$target" == "iwmmxt-wince-vs8" ] || [ "$target" == "iwmmxt2-wince-vs8" ];then case "$name" in - vpx_decoder) tag DeploymentTool \ + vpx) tag DeploymentTool \ ForceDirty="-1" \ RegisterOutput="0" ;; diff --git a/build/make/gen_msvs_sln.sh b/build/make/gen_msvs_sln.sh index b670ec5..9cf0900 100755 --- a/build/make/gen_msvs_sln.sh +++ b/build/make/gen_msvs_sln.sh @@ -1,11 +1,12 @@ #!/bin/bash ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## @@ -24,7 +25,7 @@ files. Options: --help Print this message --out=outfile Redirect output to a file - --ver=version Version (7,8) of visual studio to generate for + --ver=version Version (7,8,9) of visual studio to generate for --target=isa-os-cc Target specifier EOF exit 1 @@ -192,11 +193,11 @@ ${TAB}rm -rf "$platform"/"$config" ifneq (\$(found_devenv),) ifeq (\$(CONFIG_VS_VERSION),7) $nows_sln_config: $outfile -${TAB}devenv.com $outfile /build "$config" +${TAB}devenv.com $outfile -build "$config" else $nows_sln_config: $outfile -${TAB}devenv.com $outfile /build "$sln_config" +${TAB}devenv.com $outfile -build "$sln_config" endif else @@ -223,7 +224,7 @@ for opt in "$@"; do ;; --ver=*) vs_ver="$optval" case $optval in - [78]) + [789]) ;; *) die Unrecognized Visual Studio Version in $opt ;; @@ -234,7 +235,7 @@ for opt in "$@"; do 7) sln_vers="8.00" sln_vers_str="Visual Studio .NET 2003" ;; - 8) + [89]) ;; *) die "Unrecognized Visual Studio Version '$optval' in $opt" ;; @@ -256,6 +257,9 @@ case "${vs_ver:-8}" in 8) sln_vers="9.00" sln_vers_str="Visual Studio 2005" ;; + 9) sln_vers="10.00" + sln_vers_str="Visual Studio 2008" + ;; esac for f in "${file_list[@]}"; do diff --git a/build/make/obj_int_extract.c b/build/make/obj_int_extract.c index cef14e7..e01870f 100644 --- a/build/make/obj_int_extract.c +++ b/build/make/obj_int_extract.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -16,7 +17,7 @@ #if defined(_MSC_VER) #include <io.h> #include <share.h> -#include "vpx_ports/vpx_integer.h" +#include "vpx/vpx_integer.h" #else #include <stdint.h> #include <unistd.h> diff --git a/build/make/version.sh b/build/make/version.sh index 2bda701..3efb956 100755 --- a/build/make/version.sh +++ b/build/make/version.sh @@ -1,11 +1,12 @@ #!/bin/bash ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## diff --git a/build/x86-msvs/yasm.rules b/build/x86-msvs/yasm.rules new file mode 100644 index 0000000..ee1fefb --- /dev/null +++ b/build/x86-msvs/yasm.rules @@ -0,0 +1,115 @@ +<?xml version="1.0" encoding="utf-8"?>
+<VisualStudioToolFile
+ Name="Yasm"
+ Version="8.00"
+ >
+ <Rules>
+ <CustomBuildRule
+ Name="YASM"
+ DisplayName="Yasm Assembler"
+ CommandLine="yasm -Xvc -f $(PlatformName) [AllOptions] [AdditionalOptions] [Inputs]"
+ Outputs="[$ObjectFileName]"
+ FileExtensions="*.asm"
+ ExecutionDescription="Assembling $(InputFileName)"
+ ShowOnlyRuleProperties="false"
+ >
+ <Properties>
+ <StringProperty
+ Name="Defines"
+ DisplayName="Definitions"
+ Category="Pre-Defined Symbols"
+ Description="Specify pre-defined symbols ('symbol' or 'symbol = value') "
+ Switch="-D [value]"
+ Delimited="true"
+ Inheritable="true"
+ />
+ <StringProperty
+ Name="IncludePaths"
+ DisplayName="Include Paths"
+ Category="Configuration"
+ Description="Set the paths for any additional include files"
+ Switch="-I [value]"
+ Delimited="true"
+ Inheritable="true"
+ />
+ <StringProperty
+ Name="UnDefines"
+ DisplayName="Remove Definitions"
+ Category="Pre-Defined Symbols"
+ Description="Remove pre-defined symbols "
+ Switch="-U [value]"
+ Delimited="true"
+ Inheritable="true"
+ />
+ <StringProperty
+ Name="ObjectFileName"
+ DisplayName="Object File Name"
+ Category="Output"
+ Description="Select the output file name"
+ Switch="-o [value]"
+ DefaultValue="$(IntDir)\$(InputName).obj"
+ />
+ <StringProperty
+ Name="ListFileName"
+ DisplayName="List File Name"
+ Category="Output"
+ Description="Select an output listing by setting its file name"
+ Switch="-l [value]"
+ />
+ <StringProperty
+ Name="PreIncludeFile"
+ DisplayName="Pre Include File"
+ Category="Configuration"
+ Description="Select a pre-included file by setting its name"
+ Switch="-P [value]"
+ />
+ <BooleanProperty
+ Name="Debug"
+ DisplayName="Debug Information"
+ Category="Output"
+ Description="Generate debugging information"
+ Switch="-g cv8"
+ />
+ <EnumProperty
+ Name="PreProc"
+ DisplayName="Pre-Processor"
+ Category="Configuration"
+ Description="Select the pre-processor ('nasm' or 'raw')"
+ >
+ <Values>
+ <EnumValue
+ Value="0"
+ Switch="-rnasm"
+ DisplayName="Nasm "
+ />
+ <EnumValue
+ Value="1"
+ Switch="-rraw"
+ DisplayName="Raw"
+ />
+ </Values>
+ </EnumProperty>
+ <EnumProperty
+ Name="Parser"
+ DisplayName="Parser"
+ Category="Configuration"
+ Description="Select the parser for Intel ('nasm') or AT&T ( 'gas') syntax"
+ >
+ <Values>
+ <EnumValue
+ Value="0"
+ Switch="-pnasm"
+ DisplayName="Nasm"
+ />
+ <EnumValue
+ Value="1"
+ Switch="-pgas"
+ DisplayName="Gas"
+ />
+ </Values>
+ </EnumProperty>
+ </Properties>
+ </CustomBuildRule>
+ </Rules>
+</VisualStudioToolFile>
+
@@ -1,13 +1,16 @@ #!/bin/bash ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## configure ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## This script is the front-end to the build system. It provides a similar +## interface to standard configure scripts with some extra bits for dealing +## with toolchains that differ from the standard POSIX interface and +## for extracting subsets of the source tree. In theory, reusable parts +## of this script were intended to live in build/make/configure.sh, +## but in practice, the line is pretty blurry. +## +## This build system is based in part on the FFmpeg configure script. ## - #source_path="`dirname \"$0\"`" source_path=${0%/*} @@ -24,18 +27,18 @@ Advanced options: supported by hardware [auto] ${toggle_codec_srcs} in/exclude codec library source code ${toggle_debug_libs} in/exclude debug version of libraries - ${toggle_eval_limit} enable limited evaluation build ${toggle_md5} support for output of checksum data ${toggle_static_msvcrt} use static MSVCRT (VS builds only) ${toggle_vp8} VP8 codec support ${toggle_psnr} output of PSNR data, if supported (encoders) ${toggle_mem_tracker} track memory usage - ${toggle_eval_limit} decoder limitted to 500 frames ${toggle_postproc} postprocessing ${toggle_multithread} multithreaded encoding and decoding. ${toggle_spatial_resampling} spatial sampling (scaling) support ${toggle_realtime_only} enable this option while building for real-time encoding ${toggle_runtime_cpu_detect} runtime cpu detection + ${toggle_shared} shared library support + ${toggle_arm_asm_detok} assembly version of the detokenizer (ARM platforms only) Codecs: Codecs can be selectively enabled or disabled individually, or by family: @@ -105,10 +108,13 @@ all_platforms="${all_platforms} x86-solaris-gcc" all_platforms="${all_platforms} x86-win32-gcc" all_platforms="${all_platforms} x86-win32-vs7" all_platforms="${all_platforms} x86-win32-vs8" +all_platforms="${all_platforms} x86-win32-vs9" all_platforms="${all_platforms} x86_64-darwin9-gcc" all_platforms="${all_platforms} x86_64-linux-gcc" +all_platforms="${all_platforms} x86_64-linux-icc" all_platforms="${all_platforms} x86_64-solaris-gcc" all_platforms="${all_platforms} x86_64-win64-vs8" +all_platforms="${all_platforms} x86_64-win64-vs9" all_platforms="${all_platforms} universal-darwin8-gcc" all_platforms="${all_platforms} universal-darwin9-gcc" all_platforms="${all_platforms} generic-gnu" @@ -135,11 +141,12 @@ if [ ${doxy_major:-0} -ge 1 ]; then [ $doxy_minor -eq 5 ] && [ $doxy_patch -ge 3 ] && enable doxygen fi -# install everything, by default +# install everything except the sources, by default. sources will have +# to be enabled when doing dist builds, since that's no longer a common +# case. enabled doxygen && php -v >/dev/null 2>&1 && enable install_docs enable install_bins enable install_libs -enable install_srcs enable optimizations enable fast_unaligned #allow unaligned accesses, if supported by hw @@ -157,8 +164,8 @@ if ! enabled alt_tree_layout; then [ -d ${source_path}/vp8 ] && CODECS="${CODECS} vp8_encoder vp8_decoder" else # customer environment -[ -f ${source_path}/../include/vp8cx.h ] && CODECS="${CODECS} vp8_encoder" -[ -f ${source_path}/../include/vp8dx.h ] && CODECS="${CODECS} vp8_decoder" +[ -f ${source_path}/../include/vpx/vp8cx.h ] && CODECS="${CODECS} vp8_encoder" +[ -f ${source_path}/../include/vpx/vp8dx.h ] && CODECS="${CODECS} vp8_decoder" [ -f ${source_path}/../lib/*/*mt.lib ] && soft_enable static_msvcrt fi @@ -224,11 +231,8 @@ CONFIG_LIST=" dequant_tokens dc_recon - new_tokens - eval_limit runtime_cpu_detect postproc - postproc_generic multithread psnr ${CODECS} @@ -238,6 +242,8 @@ CONFIG_LIST=" static_msvcrt spatial_resampling realtime_only + shared + arm_asm_detok " CMDLINE_SELECT=" extra_warnings @@ -264,10 +270,7 @@ CMDLINE_SELECT=" dequant_tokens dc_recon - new_tokens - eval_limit postproc - postproc_generic multithread psnr ${CODECS} @@ -276,6 +279,8 @@ CMDLINE_SELECT=" mem_tracker spatial_resampling realtime_only + shared + arm_asm_detok " process_cmdline() { @@ -354,7 +359,6 @@ process_targets() { enabled codec_srcs && DIST_DIR="${DIST_DIR}-src" ! enabled postproc && DIST_DIR="${DIST_DIR}-nopost" ! enabled multithread && DIST_DIR="${DIST_DIR}-nomt" - enabled eval_limit && DIST_DIR="${DIST_DIR}-eval" ! enabled install_docs && DIST_DIR="${DIST_DIR}-nodocs" DIST_DIR="${DIST_DIR}-${tgt_isa}-${tgt_os}" case "${tgt_os}" in @@ -365,8 +369,27 @@ process_targets() { if [ -f "${source_path}/build/make/version.sh" ]; then local ver=`"$source_path/build/make/version.sh" --bare $source_path` DIST_DIR="${DIST_DIR}-${ver}" + ver=${ver%%-*} + VERSION_PATCH=${ver##*.} + ver=${ver%.*} + VERSION_MINOR=${ver##*.} + ver=${ver#v} + VERSION_MAJOR=${ver%.*} fi - enabled child || echo "DIST_DIR?=${DIST_DIR}" >> config.mk + enabled child || cat <<EOF >> config.mk +ifeq (\$(MAKECMDGOALS),dist) +DIST_DIR?=${DIST_DIR} +else +DIST_DIR?=\$(DESTDIR)${prefix} +endif +LIBSUBDIR=${libdir##${prefix}/} + +VERSION_MAJOR=${VERSION_MAJOR} +VERSION_MINOR=${VERSION_MINOR} +VERSION_PATCH=${VERSION_PATCH} + +CONFIGURE_ARGS=${CONFIGURE_ARGS} +EOF enabled child || echo "CONFIGURE_ARGS?=${CONFIGURE_ARGS}" >> config.mk # @@ -385,6 +408,12 @@ process_targets() { } process_detect() { + if enabled shared; then + # Can only build shared libs on a subset of platforms. Doing this check + # here rather than at option parse time because the target auto-detect + # magic happens after the command line has been parsed. + enabled linux || die "--enable-shared only supported on ELF for now" + fi if [ -z "$CC" ]; then echo "Bypassing toolchain for environment detection." enable external_build @@ -419,7 +448,7 @@ EOF check_header pthread.h check_header sys/mman.h - check_header vpx_ports/vpx_integer.h -I${source_path} && enable vpx_ports + check_header vpx/vpx_integer.h -I${source_path} && enable vpx_ports } process_toolchain() { @@ -485,7 +514,7 @@ process_toolchain() { enabled gcc || soft_disable ccache if enabled mips; then enable dequant_tokens - enable dc_recon + enable dc_recon fi # Enable the postbuild target if building for visual studio. @@ -494,8 +523,6 @@ process_toolchain() { enable solution vs_version=${tgt_cc##vs} all_targets="${all_targets} solution" - INLINE=__inline - FORCEINLINE=__forceinline ;; esac @@ -1,10 +1,11 @@ ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## @@ -47,8 +48,8 @@ docs/html/index.html: doxyfile $(CODEC_DOX) $(TXT_DOX) @doxygen $< DOCS-yes += docs/html/index.html -INSTALL-DOCS-yes = $(wildcard docs/html/*) -INSTALL-DOCS-$(CONFIG_CODEC_SRCS) += $(addprefix src/,$(CODEC_DOX)) -INSTALL-DOCS-$(CONFIG_CODEC_SRCS) += src/libs.doxy_template -INSTALL-DOCS-yes += CHANGELOG -INSTALL-DOCS-yes += README +DIST-DOCS-yes = $(wildcard docs/html/*) +DIST-DOCS-$(CONFIG_CODEC_SRCS) += $(addprefix src/,$(CODEC_DOX)) +DIST-DOCS-$(CONFIG_CODEC_SRCS) += src/libs.doxy_template +DIST-DOCS-yes += CHANGELOG +DIST-DOCS-yes += README diff --git a/example_xma.c b/example_xma.c index cc4e591..72eb470 100644 --- a/example_xma.c +++ b/example_xma.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -15,10 +16,10 @@ #include <string.h> #define VPX_CODEC_DISABLE_COMPAT 1 #include "vpx_config.h" -#include "vpx_decoder.h" -#include "vpx_integer.h" +#include "vpx/vpx_decoder.h" +#include "vpx/vpx_integer.h" #if CONFIG_VP8_DECODER -#include "vp8dx.h" +#include "vpx/vp8dx.h" #endif static char *exec_name; diff --git a/examples.mk b/examples.mk index 036a423..00ffc70 100644 --- a/examples.mk +++ b/examples.mk @@ -1,10 +1,11 @@ ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## @@ -14,18 +15,21 @@ UTILS-$(CONFIG_DECODERS) += ivfdec.c ivfdec.SRCS += md5_utils.c md5_utils.h ivfdec.SRCS += vpx_ports/vpx_timer.h -ivfdec.SRCS += vpx_ports/vpx_integer.h +ivfdec.SRCS += vpx/vpx_integer.h ivfdec.SRCS += args.c args.h vpx_ports/config.h ivfdec.GUID = BA5FE66F-38DD-E034-F542-B1578C5FB950 ivfdec.DESCRIPTION = Full featured decoder UTILS-$(CONFIG_ENCODERS) += ivfenc.c -ivfenc.SRCS += args.c args.h vpx_ports/config.h -ivfenc.SRCS += vpx_ports/mem_ops.h vpx_ports/mem_ops_aligned.h +ivfenc.SRCS += args.c args.h y4minput.c y4minput.h +ivfenc.SRCS += vpx_ports/config.h vpx_ports/mem_ops.h +ivfenc.SRCS += vpx_ports/mem_ops_aligned.h ivfenc.GUID = 548DEC74-7A15-4B2B-AFC3-AA102E7C25C1 ivfenc.DESCRIPTION = Full featured encoder -UTILS-$(CONFIG_DECODERS) += example_xma.c -example_xma.GUID = A955FC4A-73F1-44F7-135E-30D84D32F022 -example_xma.DESCRIPTION = External Memory Allocation mode usage + +# XMA example disabled for now, not used in VP8 +#UTILS-$(CONFIG_DECODERS) += example_xma.c +#example_xma.GUID = A955FC4A-73F1-44F7-135E-30D84D32F022 +#example_xma.DESCRIPTION = External Memory Allocation mode usage GEN_EXAMPLES-$(CONFIG_DECODERS) += simple_decoder.c simple_decoder.GUID = D3BBF1E9-2427-450D-BBFF-B2843C1D44CC @@ -84,8 +88,6 @@ ifeq ($(HAVE_ALT_TREE_LAYOUT),yes) INC_PATH := $(SRC_PATH_BARE)/../include else LIB_PATH-yes += $(if $(BUILD_PFX),$(BUILD_PFX),.) - INC_PATH-yes += $(SRC_PATH_BARE)/vpx_codec - INC_PATH-yes += $(SRC_PATH_BARE)/vpx_ports INC_PATH-$(CONFIG_VP8_DECODER) += $(SRC_PATH_BARE)/vp8 INC_PATH-$(CONFIG_VP8_ENCODER) += $(SRC_PATH_BARE)/vp8 LIB_PATH := $(call enabled,LIB_PATH) @@ -99,6 +101,7 @@ LDFLAGS += $(addprefix -L,$(LIB_PATH)) UTILS = $(call enabled,UTILS) GEN_EXAMPLES = $(call enabled,GEN_EXAMPLES) ALL_EXAMPLES = $(UTILS) $(GEN_EXAMPLES) +UTIL_SRCS = $(foreach ex,$(UTILS),$($(ex:.c=).SRCS)) ALL_SRCS = $(foreach ex,$(ALL_EXAMPLES),$($(ex:.c=).SRCS)) CODEC_EXTRA_LIBS=$(sort $(call enabled,CODEC_EXTRA_LIBS)) @@ -120,8 +123,10 @@ $(eval $(if $(filter universal%,$(TOOLCHAIN)),LIPO_OBJS,BUILD_OBJS):=yes) # Create build/install dependencies for all examples. The common case # is handled here. The MSVS case is handled below. NOT_MSVS = $(if $(CONFIG_MSVS),,yes) -INSTALL-BINS-$(NOT_MSVS) += $(addprefix bin/,$(ALL_EXAMPLES:.c=)) -INSTALL-SRCS-yes += $(ALL_SRCS) +DIST-BINS-$(NOT_MSVS) += $(addprefix bin/,$(ALL_EXAMPLES:.c=)) +INSTALL-BINS-$(NOT_MSVS) += $(addprefix bin/,$(UTILS:.c=)) +DIST-SRCS-yes += $(ALL_SRCS) +INSTALL-SRCS-yes += $(UTIL_SRCS) OBJS-$(NOT_MSVS) += $(if $(BUILD_OBJS),$(call objs,$(ALL_SRCS))) BINS-$(NOT_MSVS) += $(addprefix $(BUILD_PFX),$(ALL_EXAMPLES:.c=)) diff --git a/examples/decode_to_md5.txt b/examples/decode_to_md5.txt index 0599b13..b3dd568 100644 --- a/examples/decode_to_md5.txt +++ b/examples/decode_to_md5.txt @@ -26,21 +26,21 @@ is processed, then U, then V. It is important to honor the image's `stride` values. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_DX unsigned char md5_sum[16]; -md5_ctx_t md5; +MD5Context md5; int i; -md5_init(&md5); +MD5Init(&md5); for(plane=0; plane < 3; plane++) { unsigned char *buf =img->planes[plane]; for(y=0; y<img->d_h >> (plane?1:0); y++) { - md5_update(&md5, buf, img->d_w >> (plane?1:0)); + MD5Update(&md5, buf, img->d_w >> (plane?1:0)); buf += img->stride[plane]; } } -md5_finalize(&md5, md5_sum); +MD5Final(md5_sum, &md5); for(i=0; i<16; i++) fprintf(outfile, "%02x",md5_sum[i]); fprintf(outfile, " img-%dx%d-%04d.i420\n", img->d_w, img->d_h, diff --git a/examples/decoder_tmpl.c b/examples/decoder_tmpl.c index deea449..ba3ac98 100644 --- a/examples/decoder_tmpl.c +++ b/examples/decoder_tmpl.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -16,11 +17,9 @@ #include <stdarg.h> #include <string.h> #define VPX_CODEC_DISABLE_COMPAT 1 -#include "vpx_decoder.h" -#if CONFIG_VP8_DECODER && !defined(interface) -#include "vp8dx.h" +#include "vpx/vpx_decoder.h" +#include "vpx/vp8dx.h" #define interface (&vpx_codec_vp8_dx_algo) -#endif @EXTRA_INCLUDES diff --git a/examples/decoder_tmpl.txt b/examples/decoder_tmpl.txt index 3287d50..6da38c2 100644 --- a/examples/decoder_tmpl.txt +++ b/examples/decoder_tmpl.txt @@ -1,7 +1,7 @@ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INCLUDES #define VPX_CODEC_DISABLE_COMPAT 1 -#include "vpx_decoder.h" -#include "vp8dx.h" +#include "vpx/vpx_decoder.h" +#include "vpx/vp8dx.h" #define interface (&vpx_codec_vp8_dx_algo) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ DEC_INCLUDES diff --git a/examples/encoder_tmpl.c b/examples/encoder_tmpl.c index 2df893a..fdfc3af 100644 --- a/examples/encoder_tmpl.c +++ b/examples/encoder_tmpl.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -16,12 +17,10 @@ #include <stdarg.h> #include <string.h> #define VPX_CODEC_DISABLE_COMPAT 1 -#include "vpx_encoder.h" -#if CONFIG_VP8_ENCODER && !defined(interface) -#include "vp8cx.h" +#include "vpx/vpx_encoder.h" +#include "vpx/vp8cx.h" #define interface (&vpx_codec_vp8_cx_algo) #define fourcc 0x30385056 -#endif @EXTRA_INCLUDES #define IVF_FILE_HDR_SZ (32) @@ -129,7 +128,7 @@ int main(int argc, char **argv) { height = strtol(argv[2], NULL, 0); if(width < 16 || width%2 || height <16 || height%2) die("Invalid resolution: %ldx%ld", width, height); - if(!vpx_img_alloc(&raw, IMG_FMT_YV12, width, height, 1)) + if(!vpx_img_alloc(&raw, VPX_IMG_FMT_I420, width, height, 1)) die("Faile to allocate image", width, height); if(!(outfile = fopen(argv[4], "wb"))) die("Failed to open %s for writing", argv[4]); diff --git a/examples/encoder_tmpl.txt b/examples/encoder_tmpl.txt index db1ea77..87055ca 100644 --- a/examples/encoder_tmpl.txt +++ b/examples/encoder_tmpl.txt @@ -1,7 +1,7 @@ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_INCLUDES #define VPX_CODEC_DISABLE_COMPAT 1 -#include "vpx_encoder.h" -#include "vp8cx.h" +#include "vpx/vpx_encoder.h" +#include "vpx/vp8cx.h" #define interface (&vpx_codec_vp8_cx_algo) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ENC_INCLUDES diff --git a/examples/gen_example_code.sh b/examples/gen_example_code.sh index f2e45c5..1133b49 100755 --- a/examples/gen_example_code.sh +++ b/examples/gen_example_code.sh @@ -1,11 +1,12 @@ #!/bin/bash ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## diff --git a/examples/gen_example_doxy.php b/examples/gen_example_doxy.php index 08beade..701bbd3 100755 --- a/examples/gen_example_doxy.php +++ b/examples/gen_example_doxy.php @@ -1,11 +1,12 @@ #!/usr/bin/env php /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/examples/gen_example_text.sh b/examples/gen_example_text.sh index 0e1f796..9a8703d 100755 --- a/examples/gen_example_text.sh +++ b/examples/gen_example_text.sh @@ -1,11 +1,12 @@ #!/bin/bash ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## diff --git a/examples/includes/PHP-Markdown-Extra-1.2.3/markdown.php.orig b/examples/includes/PHP-Markdown-Extra-1.2.3/markdown.php.orig deleted file mode 100644 index b0ed7c9..0000000 --- a/examples/includes/PHP-Markdown-Extra-1.2.3/markdown.php.orig +++ /dev/null @@ -1,2909 +0,0 @@ -<?php -# -# Markdown Extra - A text-to-HTML conversion tool for web writers -# -# PHP Markdown & Extra -# Copyright (c) 2004-2008 Michel Fortin -# <http://www.michelf.com/projects/php-markdown/> -# -# Original Markdown -# Copyright (c) 2004-2006 John Gruber -# <http://daringfireball.net/projects/markdown/> -# - - -define( 'MARKDOWN_VERSION', "1.0.1m" ); # Sat 21 Jun 2008 -define( 'MARKDOWNEXTRA_VERSION', "1.2.3" ); # Wed 31 Dec 2008 - - -# -# Global default settings: -# - -# Change to ">" for HTML output -@define( 'MARKDOWN_EMPTY_ELEMENT_SUFFIX', " />"); - -# Define the width of a tab for code blocks. -@define( 'MARKDOWN_TAB_WIDTH', 4 ); - -# Optional title attribute for footnote links and backlinks. -@define( 'MARKDOWN_FN_LINK_TITLE', "" ); -@define( 'MARKDOWN_FN_BACKLINK_TITLE', "" ); - -# Optional class attribute for footnote links and backlinks. -@define( 'MARKDOWN_FN_LINK_CLASS', "" ); -@define( 'MARKDOWN_FN_BACKLINK_CLASS', "" ); - - -# -# WordPress settings: -# - -# Change to false to remove Markdown from posts and/or comments. -@define( 'MARKDOWN_WP_POSTS', true ); -@define( 'MARKDOWN_WP_COMMENTS', true ); - - - -### Standard Function Interface ### - -@define( 'MARKDOWN_PARSER_CLASS', 'MarkdownExtra_Parser' ); - -function Markdown($text) { -# -# Initialize the parser and return the result of its transform method. -# - # Setup static parser variable. - static $parser; - if (!isset($parser)) { - $parser_class = MARKDOWN_PARSER_CLASS; - $parser = new $parser_class; - } - - # Transform text using parser. - return $parser->transform($text); -} - - -### WordPress Plugin Interface ### - -/* -Plugin Name: Markdown Extra -Plugin URI: http://www.michelf.com/projects/php-markdown/ -Description: <a href="http://daringfireball.net/projects/markdown/syntax">Markdown syntax</a> allows you to write using an easy-to-read, easy-to-write plain text format. Based on the original Perl version by <a href="http://daringfireball.net/">John Gruber</a>. <a href="http://www.michelf.com/projects/php-markdown/">More...</a> -Version: 1.2.2 -Author: Michel Fortin -Author URI: http://www.michelf.com/ -*/ - -if (isset($wp_version)) { - # More details about how it works here: - # <http://www.michelf.com/weblog/2005/wordpress-text-flow-vs-markdown/> - - # Post content and excerpts - # - Remove WordPress paragraph generator. - # - Run Markdown on excerpt, then remove all tags. - # - Add paragraph tag around the excerpt, but remove it for the excerpt rss. - if (MARKDOWN_WP_POSTS) { - remove_filter('the_content', 'wpautop'); - remove_filter('the_content_rss', 'wpautop'); - remove_filter('the_excerpt', 'wpautop'); - add_filter('the_content', 'mdwp_MarkdownPost', 6); - add_filter('the_content_rss', 'mdwp_MarkdownPost', 6); - add_filter('get_the_excerpt', 'mdwp_MarkdownPost', 6); - add_filter('get_the_excerpt', 'trim', 7); - add_filter('the_excerpt', 'mdwp_add_p'); - add_filter('the_excerpt_rss', 'mdwp_strip_p'); - - remove_filter('content_save_pre', 'balanceTags', 50); - remove_filter('excerpt_save_pre', 'balanceTags', 50); - add_filter('the_content', 'balanceTags', 50); - add_filter('get_the_excerpt', 'balanceTags', 9); - } - - # Add a footnote id prefix to posts when inside a loop. - function mdwp_MarkdownPost($text) { - static $parser; - if (!$parser) { - $parser_class = MARKDOWN_PARSER_CLASS; - $parser = new $parser_class; - } - if (is_single() || is_page() || is_feed()) { - $parser->fn_id_prefix = ""; - } else { - $parser->fn_id_prefix = get_the_ID() . "."; - } - return $parser->transform($text); - } - - # Comments - # - Remove WordPress paragraph generator. - # - Remove WordPress auto-link generator. - # - Scramble important tags before passing them to the kses filter. - # - Run Markdown on excerpt then remove paragraph tags. - if (MARKDOWN_WP_COMMENTS) { - remove_filter('comment_text', 'wpautop', 30); - remove_filter('comment_text', 'make_clickable'); - add_filter('pre_comment_content', 'Markdown', 6); - add_filter('pre_comment_content', 'mdwp_hide_tags', 8); - add_filter('pre_comment_content', 'mdwp_show_tags', 12); - add_filter('get_comment_text', 'Markdown', 6); - add_filter('get_comment_excerpt', 'Markdown', 6); - add_filter('get_comment_excerpt', 'mdwp_strip_p', 7); - - global $mdwp_hidden_tags, $mdwp_placeholders; - $mdwp_hidden_tags = explode(' ', - '<p> </p> <pre> </pre> <ol> </ol> <ul> </ul> <li> </li>'); - $mdwp_placeholders = explode(' ', str_rot13( - 'pEj07ZbbBZ U1kqgh4w4p pre2zmeN6K QTi31t9pre ol0MP1jzJR '. - 'ML5IjmbRol ulANi1NsGY J7zRLJqPul liA8ctl16T K9nhooUHli')); - } - - function mdwp_add_p($text) { - if (!preg_match('{^$|^<(p|ul|ol|dl|pre|blockquote)>}i', $text)) { - $text = '<p>'.$text.'</p>'; - $text = preg_replace('{\n{2,}}', "</p>\n\n<p>", $text); - } - return $text; - } - - function mdwp_strip_p($t) { return preg_replace('{</?p>}i', '', $t); } - - function mdwp_hide_tags($text) { - global $mdwp_hidden_tags, $mdwp_placeholders; - return str_replace($mdwp_hidden_tags, $mdwp_placeholders, $text); - } - function mdwp_show_tags($text) { - global $mdwp_hidden_tags, $mdwp_placeholders; - return str_replace($mdwp_placeholders, $mdwp_hidden_tags, $text); - } -} - - -### bBlog Plugin Info ### - -function identify_modifier_markdown() { - return array( - 'name' => 'markdown', - 'type' => 'modifier', - 'nicename' => 'PHP Markdown Extra', - 'description' => 'A text-to-HTML conversion tool for web writers', - 'authors' => 'Michel Fortin and John Gruber', - 'licence' => 'GPL', - 'version' => MARKDOWNEXTRA_VERSION, - 'help' => '<a href="http://daringfireball.net/projects/markdown/syntax">Markdown syntax</a> allows you to write using an easy-to-read, easy-to-write plain text format. Based on the original Perl version by <a href="http://daringfireball.net/">John Gruber</a>. <a href="http://www.michelf.com/projects/php-markdown/">More...</a>', - ); -} - - -### Smarty Modifier Interface ### - -function smarty_modifier_markdown($text) { - return Markdown($text); -} - - -### Textile Compatibility Mode ### - -# Rename this file to "classTextile.php" and it can replace Textile everywhere. - -if (strcasecmp(substr(__FILE__, -16), "classTextile.php") == 0) { - # Try to include PHP SmartyPants. Should be in the same directory. - @include_once 'smartypants.php'; - # Fake Textile class. It calls Markdown instead. - class Textile { - function TextileThis($text, $lite='', $encode='') { - if ($lite == '' && $encode == '') $text = Markdown($text); - if (function_exists('SmartyPants')) $text = SmartyPants($text); - return $text; - } - # Fake restricted version: restrictions are not supported for now. - function TextileRestricted($text, $lite='', $noimage='') { - return $this->TextileThis($text, $lite); - } - # Workaround to ensure compatibility with TextPattern 4.0.3. - function blockLite($text) { return $text; } - } -} - - - -# -# Markdown Parser Class -# - -class Markdown_Parser { - - # Regex to match balanced [brackets]. - # Needed to insert a maximum bracked depth while converting to PHP. - var $nested_brackets_depth = 6; - var $nested_brackets_re; - - var $nested_url_parenthesis_depth = 4; - var $nested_url_parenthesis_re; - - # Table of hash values for escaped characters: - var $escape_chars = '\`*_{}[]()>#+-.!'; - var $escape_chars_re; - - # Change to ">" for HTML output. - var $empty_element_suffix = MARKDOWN_EMPTY_ELEMENT_SUFFIX; - var $tab_width = MARKDOWN_TAB_WIDTH; - - # Change to `true` to disallow markup or entities. - var $no_markup = false; - var $no_entities = true; - - # Predefined urls and titles for reference links and images. - var $predef_urls = array(); - var $predef_titles = array(); - - - function Markdown_Parser() { - # - # Constructor function. Initialize appropriate member variables. - # - $this->_initDetab(); - $this->prepareItalicsAndBold(); - - $this->nested_brackets_re = - str_repeat('(?>[^\[\]]+|\[', $this->nested_brackets_depth). - str_repeat('\])*', $this->nested_brackets_depth); - - $this->nested_url_parenthesis_re = - str_repeat('(?>[^()\s]+|\(', $this->nested_url_parenthesis_depth). - str_repeat('(?>\)))*', $this->nested_url_parenthesis_depth); - - $this->escape_chars_re = '['.preg_quote($this->escape_chars).']'; - - # Sort document, block, and span gamut in ascendent priority order. - asort($this->document_gamut); - asort($this->block_gamut); - asort($this->span_gamut); - } - - - # Internal hashes used during transformation. - var $urls = array(); - var $titles = array(); - var $html_hashes = array(); - - # Status flag to avoid invalid nesting. - var $in_anchor = false; - - - function setup() { - # - # Called before the transformation process starts to setup parser - # states. - # - # Clear global hashes. - $this->urls = $this->predef_urls; - $this->titles = $this->predef_titles; - $this->html_hashes = array(); - - $in_anchor = false; - } - - function teardown() { - # - # Called after the transformation process to clear any variable - # which may be taking up memory unnecessarly. - # - $this->urls = array(); - $this->titles = array(); - $this->html_hashes = array(); - } - - - function transform($text) { - # - # Main function. Performs some preprocessing on the input text - # and pass it through the document gamut. - # - $this->setup(); - - # Remove UTF-8 BOM and marker character in input, if present. - $text = preg_replace('{^\xEF\xBB\xBF|\x1A}', '', $text); - - # Standardize line endings: - # DOS to Unix and Mac to Unix - $text = preg_replace('{\r\n?}', "\n", $text); - - # Make sure $text ends with a couple of newlines: - $text .= "\n\n"; - - # Convert all tabs to spaces. - $text = $this->detab($text); - - # Turn block-level HTML blocks into hash entries - $text = $this->hashHTMLBlocks($text); - - # Strip any lines consisting only of spaces and tabs. - # This makes subsequent regexen easier to write, because we can - # match consecutive blank lines with /\n+/ instead of something - # contorted like /[ ]*\n+/ . - $text = preg_replace('/^[ ]+$/m', '', $text); - - # Run document gamut methods. - foreach ($this->document_gamut as $method => $priority) { - $text = $this->$method($text); - } - - $this->teardown(); - - return $text . "\n"; - } - - var $document_gamut = array( - # Strip link definitions, store in hashes. - "stripLinkDefinitions" => 20, - - "runBasicBlockGamut" => 30, - ); - - - function stripLinkDefinitions($text) { - # - # Strips link definitions from text, stores the URLs and titles in - # hash references. - # - $less_than_tab = $this->tab_width - 1; - - # Link defs are in the form: ^[id]: url "optional title" - $text = preg_replace_callback('{ - ^[ ]{0,'.$less_than_tab.'}\[(.+)\][ ]?: # id = $1 - [ ]* - \n? # maybe *one* newline - [ ]* - <?(\S+?)>? # url = $2 - [ ]* - \n? # maybe one newline - [ ]* - (?: - (?<=\s) # lookbehind for whitespace - ["(] - (.*?) # title = $3 - [")] - [ ]* - )? # title is optional - (?:\n+|\Z) - }xm', - array(&$this, '_stripLinkDefinitions_callback'), - $text); - return $text; - } - function _stripLinkDefinitions_callback($matches) { - $link_id = strtolower($matches[1]); - $this->urls[$link_id] = $matches[2]; - $this->titles[$link_id] =& $matches[3]; - return ''; # String that will replace the block - } - - - function hashHTMLBlocks($text) { - if ($this->no_markup) return $text; - - $less_than_tab = $this->tab_width - 1; - - # Hashify HTML blocks: - # We only want to do this for block-level HTML tags, such as headers, - # lists, and tables. That's because we still want to wrap <p>s around - # "paragraphs" that are wrapped in non-block-level tags, such as anchors, - # phrase emphasis, and spans. The list of tags we're looking for is - # hard-coded: - # - # * List "a" is made of tags which can be both inline or block-level. - # These will be treated block-level when the start tag is alone on - # its line, otherwise they're not matched here and will be taken as - # inline later. - # * List "b" is made of tags which are always block-level; - # - $block_tags_a_re = 'ins|del'; - $block_tags_b_re = 'p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|address|'. - 'script|noscript|form|fieldset|iframe|math'; - - # Regular expression for the content of a block tag. - $nested_tags_level = 4; - $attr = ' - (?> # optional tag attributes - \s # starts with whitespace - (?> - [^>"/]+ # text outside quotes - | - /+(?!>) # slash not followed by ">" - | - "[^"]*" # text inside double quotes (tolerate ">") - | - \'[^\']*\' # text inside single quotes (tolerate ">") - )* - )? - '; - $content = - str_repeat(' - (?> - [^<]+ # content without tag - | - <\2 # nested opening tag - '.$attr.' # attributes - (?> - /> - | - >', $nested_tags_level). # end of opening tag - '.*?'. # last level nested tag content - str_repeat(' - </\2\s*> # closing nested tag - ) - | - <(?!/\2\s*> # other tags with a different name - ) - )*', - $nested_tags_level); - $content2 = str_replace('\2', '\3', $content); - - # First, look for nested blocks, e.g.: - # <div> - # <div> - # tags for inner block must be indented. - # </div> - # </div> - # - # The outermost tags must start at the left margin for this to match, and - # the inner nested divs must be indented. - # We need to do this before the next, more liberal match, because the next - # match will start at the first `<div>` and stop at the first `</div>`. - $text = preg_replace_callback('{(?> - (?> - (?<=\n\n) # Starting after a blank line - | # or - \A\n? # the beginning of the doc - ) - ( # save in $1 - - # Match from `\n<tag>` to `</tag>\n`, handling nested tags - # in between. - - [ ]{0,'.$less_than_tab.'} - <('.$block_tags_b_re.')# start tag = $2 - '.$attr.'> # attributes followed by > and \n - '.$content.' # content, support nesting - </\2> # the matching end tag - [ ]* # trailing spaces/tabs - (?=\n+|\Z) # followed by a newline or end of document - - | # Special version for tags of group a. - - [ ]{0,'.$less_than_tab.'} - <('.$block_tags_a_re.')# start tag = $3 - '.$attr.'>[ ]*\n # attributes followed by > - '.$content2.' # content, support nesting - </\3> # the matching end tag - [ ]* # trailing spaces/tabs - (?=\n+|\Z) # followed by a newline or end of document - - | # Special case just for <hr />. It was easier to make a special - # case than to make the other regex more complicated. - - [ ]{0,'.$less_than_tab.'} - <(hr) # start tag = $2 - '.$attr.' # attributes - /?> # the matching end tag - [ ]* - (?=\n{2,}|\Z) # followed by a blank line or end of document - - | # Special case for standalone HTML comments: - - [ ]{0,'.$less_than_tab.'} - (?s: - <!-- .*? --> - ) - [ ]* - (?=\n{2,}|\Z) # followed by a blank line or end of document - - | # PHP and ASP-style processor instructions (<? and <%) - - [ ]{0,'.$less_than_tab.'} - (?s: - <([?%]) # $2 - .*? - \2> - ) - [ ]* - (?=\n{2,}|\Z) # followed by a blank line or end of document - - ) - )}Sxmi', - array(&$this, '_hashHTMLBlocks_callback'), - $text); - - return $text; - } - function _hashHTMLBlocks_callback($matches) { - $text = $matches[1]; - $key = $this->hashBlock($text); - return "\n\n$key\n\n"; - } - - - function hashPart($text, $boundary = 'X') { - # - # Called whenever a tag must be hashed when a function insert an atomic - # element in the text stream. Passing $text to through this function gives - # a unique text-token which will be reverted back when calling unhash. - # - # The $boundary argument specify what character should be used to surround - # the token. By convension, "B" is used for block elements that needs not - # to be wrapped into paragraph tags at the end, ":" is used for elements - # that are word separators and "X" is used in the general case. - # - # Swap back any tag hash found in $text so we do not have to `unhash` - # multiple times at the end. - $text = $this->unhash($text); - - # Then hash the block. - static $i = 0; - $key = "$boundary\x1A" . ++$i . $boundary; - $this->html_hashes[$key] = $text; - return $key; # String that will replace the tag. - } - - - function hashBlock($text) { - # - # Shortcut function for hashPart with block-level boundaries. - # - return $this->hashPart($text, 'B'); - } - - - var $block_gamut = array( - # - # These are all the transformations that form block-level - # tags like paragraphs, headers, and list items. - # - "doHeaders" => 10, - "doHorizontalRules" => 20, - - "doLists" => 40, - "doCodeBlocks" => 50, - "doBlockQuotes" => 60, - ); - - function runBlockGamut($text) { - # - # Run block gamut tranformations. - # - # We need to escape raw HTML in Markdown source before doing anything - # else. This need to be done for each block, and not only at the - # begining in the Markdown function since hashed blocks can be part of - # list items and could have been indented. Indented blocks would have - # been seen as a code block in a previous pass of hashHTMLBlocks. - $text = $this->hashHTMLBlocks($text); - - return $this->runBasicBlockGamut($text); - } - - function runBasicBlockGamut($text) { - # - # Run block gamut tranformations, without hashing HTML blocks. This is - # useful when HTML blocks are known to be already hashed, like in the first - # whole-document pass. - # - foreach ($this->block_gamut as $method => $priority) { - $text = $this->$method($text); - } - - # Finally form paragraph and restore hashed blocks. - $text = $this->formParagraphs($text); - - return $text; - } - - - function doHorizontalRules($text) { - # Do Horizontal Rules: - return preg_replace( - '{ - ^[ ]{0,3} # Leading space - ([-*_]) # $1: First marker - (?> # Repeated marker group - [ ]{0,2} # Zero, one, or two spaces. - \1 # Marker character - ){2,} # Group repeated at least twice - [ ]* # Tailing spaces - $ # End of line. - }mx', - "\n".$this->hashBlock("<hr$this->empty_element_suffix")."\n", - $text); - } - - - var $span_gamut = array( - # - # These are all the transformations that occur *within* block-level - # tags like paragraphs, headers, and list items. - # - # Process character escapes, code spans, and inline HTML - # in one shot. - "parseSpan" => -30, - - # Process anchor and image tags. Images must come first, - # because ![foo][f] looks like an anchor. - "doImages" => 10, - "doAnchors" => 20, - - # Make links out of things like `<http://example.com/>` - # Must come after doAnchors, because you can use < and > - # delimiters in inline links like [this](<url>). - "doAutoLinks" => 30, - "encodeAmpsAndAngles" => 40, - - "doItalicsAndBold" => 50, - "doHardBreaks" => 60, - ); - - function runSpanGamut($text) { - # - # Run span gamut tranformations. - # - foreach ($this->span_gamut as $method => $priority) { - $text = $this->$method($text); - } - - return $text; - } - - - function doHardBreaks($text) { - # Do hard breaks: - return preg_replace_callback('/ {2,}\n/', - array(&$this, '_doHardBreaks_callback'), $text); - } - function _doHardBreaks_callback($matches) { - return $this->hashPart("<br$this->empty_element_suffix\n"); - } - - - function doAnchors($text) { - # - # Turn Markdown link shortcuts into XHTML <a> tags. - # - if ($this->in_anchor) return $text; - $this->in_anchor = true; - - # - # First, handle reference-style links: [link text] [id] - # - $text = preg_replace_callback('{ - ( # wrap whole match in $1 - \[ - ('.$this->nested_brackets_re.') # link text = $2 - \] - - [ ]? # one optional space - (?:\n[ ]*)? # one optional newline followed by spaces - - \[ - (.*?) # id = $3 - \] - ) - }xs', - array(&$this, '_doAnchors_reference_callback'), $text); - - # - # Next, inline-style links: [link text](url "optional title") - # - $text = preg_replace_callback('{ - ( # wrap whole match in $1 - \[ - ('.$this->nested_brackets_re.') # link text = $2 - \] - \( # literal paren - [ ]* - (?: - <(\S*)> # href = $3 - | - ('.$this->nested_url_parenthesis_re.') # href = $4 - ) - [ ]* - ( # $5 - ([\'"]) # quote char = $6 - (.*?) # Title = $7 - \6 # matching quote - [ ]* # ignore any spaces/tabs between closing quote and ) - )? # title is optional - \) - ) - }xs', - array(&$this, '_DoAnchors_inline_callback'), $text); - - # - # Last, handle reference-style shortcuts: [link text] - # These must come last in case you've also got [link test][1] - # or [link test](/foo) - # -// $text = preg_replace_callback('{ -// ( # wrap whole match in $1 -// \[ -// ([^\[\]]+) # link text = $2; can\'t contain [ or ] -// \] -// ) -// }xs', -// array(&$this, '_doAnchors_reference_callback'), $text); - - $this->in_anchor = false; - return $text; - } - function _doAnchors_reference_callback($matches) { - $whole_match = $matches[1]; - $link_text = $matches[2]; - $link_id =& $matches[3]; - - if ($link_id == "") { - # for shortcut links like [this][] or [this]. - $link_id = $link_text; - } - - # lower-case and turn embedded newlines into spaces - $link_id = strtolower($link_id); - $link_id = preg_replace('{[ ]?\n}', ' ', $link_id); - - if (isset($this->urls[$link_id])) { - $url = $this->urls[$link_id]; - $url = $this->encodeAttribute($url); - - $result = "<a href=\"$url\""; - if ( isset( $this->titles[$link_id] ) ) { - $title = $this->titles[$link_id]; - $title = $this->encodeAttribute($title); - $result .= " title=\"$title\""; - } - - $link_text = $this->runSpanGamut($link_text); - $result .= ">$link_text</a>"; - $result = $this->hashPart($result); - } - else { - $result = $whole_match; - } - return $result; - } - function _doAnchors_inline_callback($matches) { - $whole_match = $matches[1]; - $link_text = $this->runSpanGamut($matches[2]); - $url = $matches[3] == '' ? $matches[4] : $matches[3]; - $title =& $matches[7]; - - $url = $this->encodeAttribute($url); - - $result = "<a href=\"$url\""; - if (isset($title)) { - $title = $this->encodeAttribute($title); - $result .= " title=\"$title\""; - } - - $link_text = $this->runSpanGamut($link_text); - $result .= ">$link_text</a>"; - - return $this->hashPart($result); - } - - - function doImages($text) { - # - # Turn Markdown image shortcuts into <img> tags. - # - # - # First, handle reference-style labeled images: ![alt text][id] - # - $text = preg_replace_callback('{ - ( # wrap whole match in $1 - !\[ - ('.$this->nested_brackets_re.') # alt text = $2 - \] - - [ ]? # one optional space - (?:\n[ ]*)? # one optional newline followed by spaces - - \[ - (.*?) # id = $3 - \] - - ) - }xs', - array(&$this, '_doImages_reference_callback'), $text); - - # - # Next, handle inline images:  - # Don't forget: encode * and _ - # - $text = preg_replace_callback('{ - ( # wrap whole match in $1 - !\[ - ('.$this->nested_brackets_re.') # alt text = $2 - \] - \s? # One optional whitespace character - \( # literal paren - [ ]* - (?: - <(\S*)> # src url = $3 - | - ('.$this->nested_url_parenthesis_re.') # src url = $4 - ) - [ ]* - ( # $5 - ([\'"]) # quote char = $6 - (.*?) # title = $7 - \6 # matching quote - [ ]* - )? # title is optional - \) - ) - }xs', - array(&$this, '_doImages_inline_callback'), $text); - - return $text; - } - function _doImages_reference_callback($matches) { - $whole_match = $matches[1]; - $alt_text = $matches[2]; - $link_id = strtolower($matches[3]); - - if ($link_id == "") { - $link_id = strtolower($alt_text); # for shortcut links like ![this][]. - } - - $alt_text = $this->encodeAttribute($alt_text); - if (isset($this->urls[$link_id])) { - $url = $this->encodeAttribute($this->urls[$link_id]); - $result = "<img src=\"$url\" alt=\"$alt_text\""; - if (isset($this->titles[$link_id])) { - $title = $this->titles[$link_id]; - $title = $this->encodeAttribute($title); - $result .= " title=\"$title\""; - } - $result .= $this->empty_element_suffix; - $result = $this->hashPart($result); - } - else { - # If there's no such link ID, leave intact: - $result = $whole_match; - } - - return $result; - } - function _doImages_inline_callback($matches) { - $whole_match = $matches[1]; - $alt_text = $matches[2]; - $url = $matches[3] == '' ? $matches[4] : $matches[3]; - $title =& $matches[7]; - - $alt_text = $this->encodeAttribute($alt_text); - $url = $this->encodeAttribute($url); - $result = "<img src=\"$url\" alt=\"$alt_text\""; - if (isset($title)) { - $title = $this->encodeAttribute($title); - $result .= " title=\"$title\""; # $title already quoted - } - $result .= $this->empty_element_suffix; - - return $this->hashPart($result); - } - - - function doHeaders($text) { - # Setext-style headers: - # Header 1 - # ======== - # - # Header 2 - # -------- - # - $text = preg_replace_callback('{ ^(.+?)[ ]*\n(=+|-+)[ ]*\n+ }mx', - array(&$this, '_doHeaders_callback_setext'), $text); - - # atx-style headers: - # # Header 1 - # ## Header 2 - # ## Header 2 with closing hashes ## - # ... - # ###### Header 6 - # - $text = preg_replace_callback('{ - ^(\#{1,6}) # $1 = string of #\'s - [ ]* - (.+?) # $2 = Header text - [ ]* - \#* # optional closing #\'s (not counted) - \n+ - }xm', - array(&$this, '_doHeaders_callback_atx'), $text); - - return $text; - } - function _doHeaders_callback_setext($matches) { - # Terrible hack to check we haven't found an empty list item. - if ($matches[2] == '-' && preg_match('{^-(?: |$)}', $matches[1])) - return $matches[0]; - - $level = $matches[2]{0} == '=' ? 1 : 2; - $block = "<h$level>".$this->runSpanGamut($matches[1])."</h$level>"; - return "\n" . $this->hashBlock($block) . "\n\n"; - } - function _doHeaders_callback_atx($matches) { - $level = strlen($matches[1]); - $block = "<h$level>".$this->runSpanGamut($matches[2])."</h$level>"; - return "\n" . $this->hashBlock($block) . "\n\n"; - } - - - function doLists($text) { - # - # Form HTML ordered (numbered) and unordered (bulleted) lists. - # - $less_than_tab = $this->tab_width - 1; - - # Re-usable patterns to match list item bullets and number markers: - $marker_ul_re = '[*+-]'; - $marker_ol_re = '\d+[.]'; - $marker_any_re = "(?:$marker_ul_re|$marker_ol_re)"; - - $markers_relist = array($marker_ul_re, $marker_ol_re); - - foreach ($markers_relist as $marker_re) { - # Re-usable pattern to match any entirel ul or ol list: - $whole_list_re = ' - ( # $1 = whole list - ( # $2 - [ ]{0,'.$less_than_tab.'} - ('.$marker_re.') # $3 = first list item marker - [ ]+ - ) - (?s:.+?) - ( # $4 - \z - | - \n{2,} - (?=\S) - (?! # Negative lookahead for another list item marker - [ ]* - '.$marker_re.'[ ]+ - ) - ) - ) - '; // mx - - # We use a different prefix before nested lists than top-level lists. - # See extended comment in _ProcessListItems(). - - if ($this->list_level) { - $text = preg_replace_callback('{ - ^ - '.$whole_list_re.' - }mx', - array(&$this, '_doLists_callback'), $text); - } - else { - $text = preg_replace_callback('{ - (?:(?<=\n)\n|\A\n?) # Must eat the newline - '.$whole_list_re.' - }mx', - array(&$this, '_doLists_callback'), $text); - } - } - - return $text; - } - function _doLists_callback($matches) { - # Re-usable patterns to match list item bullets and number markers: - $marker_ul_re = '[*+-]'; - $marker_ol_re = '\d+[.]'; - $marker_any_re = "(?:$marker_ul_re|$marker_ol_re)"; - - $list = $matches[1]; - $list_type = preg_match("/$marker_ul_re/", $matches[3]) ? "ul" : "ol"; - - $marker_any_re = ( $list_type == "ul" ? $marker_ul_re : $marker_ol_re ); - - $list .= "\n"; - $result = $this->processListItems($list, $marker_any_re); - - $result = $this->hashBlock("<$list_type>\n" . $result . "</$list_type>"); - return "\n". $result ."\n\n"; - } - - var $list_level = 0; - - function processListItems($list_str, $marker_any_re) { - # - # Process the contents of a single ordered or unordered list, splitting it - # into individual list items. - # - # The $this->list_level global keeps track of when we're inside a list. - # Each time we enter a list, we increment it; when we leave a list, - # we decrement. If it's zero, we're not in a list anymore. - # - # We do this because when we're not inside a list, we want to treat - # something like this: - # - # I recommend upgrading to version - # 8. Oops, now this line is treated - # as a sub-list. - # - # As a single paragraph, despite the fact that the second line starts - # with a digit-period-space sequence. - # - # Whereas when we're inside a list (or sub-list), that line will be - # treated as the start of a sub-list. What a kludge, huh? This is - # an aspect of Markdown's syntax that's hard to parse perfectly - # without resorting to mind-reading. Perhaps the solution is to - # change the syntax rules such that sub-lists must start with a - # starting cardinal number; e.g. "1." or "a.". - - $this->list_level++; - - # trim trailing blank lines: - $list_str = preg_replace("/\n{2,}\\z/", "\n", $list_str); - - $list_str = preg_replace_callback('{ - (\n)? # leading line = $1 - (^[ ]*) # leading whitespace = $2 - ('.$marker_any_re.' # list marker and space = $3 - (?:[ ]+|(?=\n)) # space only required if item is not empty - ) - ((?s:.*?)) # list item text = $4 - (?:(\n+(?=\n))|\n) # tailing blank line = $5 - (?= \n* (\z | \2 ('.$marker_any_re.') (?:[ ]+|(?=\n)))) - }xm', - array(&$this, '_processListItems_callback'), $list_str); - - $this->list_level--; - return $list_str; - } - function _processListItems_callback($matches) { - $item = $matches[4]; - $leading_line =& $matches[1]; - $leading_space =& $matches[2]; - $marker_space = $matches[3]; - $tailing_blank_line =& $matches[5]; - - if ($leading_line || $tailing_blank_line || - preg_match('/\n{2,}/', $item)) - { - # Replace marker with the appropriate whitespace indentation - $item = $leading_space . str_repeat(' ', strlen($marker_space)) . $item; - $item = $this->runBlockGamut($this->outdent($item)."\n"); - } - else { - # Recursion for sub-lists: - $item = $this->doLists($this->outdent($item)); - $item = preg_replace('/\n+$/', '', $item); - $item = $this->runSpanGamut($item); - } - - return "<li>" . $item . "</li>\n"; - } - - - function doCodeBlocks($text) { - # - # Process Markdown `<pre><code>` blocks. - # - $text = preg_replace_callback('{ - (?:\n\n|\A\n?) - ( # $1 = the code block -- one or more lines, starting with a space/tab - (?> - [ ]{'.$this->tab_width.'} # Lines must start with a tab or a tab-width of spaces - .*\n+ - )+ - ) - ((?=^[ ]{0,'.$this->tab_width.'}\S)|\Z) # Lookahead for non-space at line-start, or end of doc - }xm', - array(&$this, '_doCodeBlocks_callback'), $text); - - return $text; - } - function _doCodeBlocks_callback($matches) { - $codeblock = $matches[1]; - - $codeblock = $this->outdent($codeblock); - $codeblock = htmlspecialchars($codeblock, ENT_NOQUOTES); - - # trim leading newlines and trailing newlines - $codeblock = preg_replace('/\A\n+|\n+\z/', '', $codeblock); - - $codeblock = "<pre><code>$codeblock\n</code></pre>"; - return "\n\n".$this->hashBlock($codeblock)."\n\n"; - } - - - function makeCodeSpan($code) { - # - # Create a code span markup for $code. Called from handleSpanToken. - # - $code = htmlspecialchars(trim($code), ENT_NOQUOTES); - return $this->hashPart("<code>$code</code>"); - } - - - var $em_relist = array( - '' => '(?:(?<!\*)\*(?!\*)|(?<!_)_(?!_))(?=\S)(?![.,:;]\s)', - '*' => '(?<=\S)(?<!\*)\*(?!\*)', - '_' => '(?<=\S)(?<!_)_(?!_)', - ); - var $strong_relist = array( - '' => '(?:(?<!\*)\*\*(?!\*)|(?<!_)__(?!_))(?=\S)(?![.,:;]\s)', - '**' => '(?<=\S)(?<!\*)\*\*(?!\*)', - '__' => '(?<=\S)(?<!_)__(?!_)', - ); - var $em_strong_relist = array( - '' => '(?:(?<!\*)\*\*\*(?!\*)|(?<!_)___(?!_))(?=\S)(?![.,:;]\s)', - '***' => '(?<=\S)(?<!\*)\*\*\*(?!\*)', - '___' => '(?<=\S)(?<!_)___(?!_)', - ); - var $em_strong_prepared_relist; - - function prepareItalicsAndBold() { - # - # Prepare regular expressions for seraching emphasis tokens in any - # context. - # - foreach ($this->em_relist as $em => $em_re) { - foreach ($this->strong_relist as $strong => $strong_re) { - # Construct list of allowed token expressions. - $token_relist = array(); - if (isset($this->em_strong_relist["$em$strong"])) { - $token_relist[] = $this->em_strong_relist["$em$strong"]; - } - $token_relist[] = $em_re; - $token_relist[] = $strong_re; - - # Construct master expression from list. - $token_re = '{('. implode('|', $token_relist) .')}'; - $this->em_strong_prepared_relist["$em$strong"] = $token_re; - } - } - } - - function doItalicsAndBold($text) { - $token_stack = array(''); - $text_stack = array(''); - $em = ''; - $strong = ''; - $tree_char_em = false; - - while (1) { - # - # Get prepared regular expression for seraching emphasis tokens - # in current context. - # - $token_re = $this->em_strong_prepared_relist["$em$strong"]; - - # - # Each loop iteration seach for the next emphasis token. - # Each token is then passed to handleSpanToken. - # - $parts = preg_split($token_re, $text, 2, PREG_SPLIT_DELIM_CAPTURE); - $text_stack[0] .= $parts[0]; - $token =& $parts[1]; - $text =& $parts[2]; - - if (empty($token)) { - # Reached end of text span: empty stack without emitting. - # any more emphasis. - while ($token_stack[0]) { - $text_stack[1] .= array_shift($token_stack); - $text_stack[0] .= array_shift($text_stack); - } - break; - } - - $token_len = strlen($token); - if ($tree_char_em) { - # Reached closing marker while inside a three-char emphasis. - if ($token_len == 3) { - # Three-char closing marker, close em and strong. - array_shift($token_stack); - $span = array_shift($text_stack); - $span = $this->runSpanGamut($span); - $span = "<strong><em>$span</em></strong>"; - $text_stack[0] .= $this->hashPart($span); - $em = ''; - $strong = ''; - } else { - # Other closing marker: close one em or strong and - # change current token state to match the other - $token_stack[0] = str_repeat($token{0}, 3-$token_len); - $tag = $token_len == 2 ? "strong" : "em"; - $span = $text_stack[0]; - $span = $this->runSpanGamut($span); - $span = "<$tag>$span</$tag>"; - $text_stack[0] = $this->hashPart($span); - $$tag = ''; # $$tag stands for $em or $strong - } - $tree_char_em = false; - } else if ($token_len == 3) { - if ($em) { - # Reached closing marker for both em and strong. - # Closing strong marker: - for ($i = 0; $i < 2; ++$i) { - $shifted_token = array_shift($token_stack); - $tag = strlen($shifted_token) == 2 ? "strong" : "em"; - $span = array_shift($text_stack); - $span = $this->runSpanGamut($span); - $span = "<$tag>$span</$tag>"; - $text_stack[0] .= $this->hashPart($span); - $$tag = ''; # $$tag stands for $em or $strong - } - } else { - # Reached opening three-char emphasis marker. Push on token - # stack; will be handled by the special condition above. - $em = $token{0}; - $strong = "$em$em"; - array_unshift($token_stack, $token); - array_unshift($text_stack, ''); - $tree_char_em = true; - } - } else if ($token_len == 2) { - if ($strong) { - # Unwind any dangling emphasis marker: - if (strlen($token_stack[0]) == 1) { - $text_stack[1] .= array_shift($token_stack); - $text_stack[0] .= array_shift($text_stack); - } - # Closing strong marker: - array_shift($token_stack); - $span = array_shift($text_stack); - $span = $this->runSpanGamut($span); - $span = "<strong>$span</strong>"; - $text_stack[0] .= $this->hashPart($span); - $strong = ''; - } else { - array_unshift($token_stack, $token); - array_unshift($text_stack, ''); - $strong = $token; - } - } else { - # Here $token_len == 1 - if ($em) { - if (strlen($token_stack[0]) == 1) { - # Closing emphasis marker: - array_shift($token_stack); - $span = array_shift($text_stack); - $span = $this->runSpanGamut($span); - $span = "<em>$span</em>"; - $text_stack[0] .= $this->hashPart($span); - $em = ''; - } else { - $text_stack[0] .= $token; - } - } else { - array_unshift($token_stack, $token); - array_unshift($text_stack, ''); - $em = $token; - } - } - } - return $text_stack[0]; - } - - - function doBlockQuotes($text) { - $text = preg_replace_callback('/ - ( # Wrap whole match in $1 - (?> - ^[ ]*>[ ]? # ">" at the start of a line - .+\n # rest of the first line - (.+\n)* # subsequent consecutive lines - \n* # blanks - )+ - ) - /xm', - array(&$this, '_doBlockQuotes_callback'), $text); - - return $text; - } - function _doBlockQuotes_callback($matches) { - $bq = $matches[1]; - # trim one level of quoting - trim whitespace-only lines - $bq = preg_replace('/^[ ]*>[ ]?|^[ ]+$/m', '', $bq); - $bq = $this->runBlockGamut($bq); # recurse - - $bq = preg_replace('/^/m', " ", $bq); - # These leading spaces cause problem with <pre> content, - # so we need to fix that: - $bq = preg_replace_callback('{(\s*<pre>.+?</pre>)}sx', - array(&$this, '_DoBlockQuotes_callback2'), $bq); - - return "\n". $this->hashBlock("<blockquote>\n$bq\n</blockquote>")."\n\n"; - } - function _doBlockQuotes_callback2($matches) { - $pre = $matches[1]; - $pre = preg_replace('/^ /m', '', $pre); - return $pre; - } - - - function formParagraphs($text) { - # - # Params: - # $text - string to process with html <p> tags - # - # Strip leading and trailing lines: - $text = preg_replace('/\A\n+|\n+\z/', '', $text); - - $grafs = preg_split('/\n{2,}/', $text, -1, PREG_SPLIT_NO_EMPTY); - - # - # Wrap <p> tags and unhashify HTML blocks - # - foreach ($grafs as $key => $value) { - if (!preg_match('/^B\x1A[0-9]+B$/', $value)) { - # Is a paragraph. - $value = $this->runSpanGamut($value); - $value = preg_replace('/^([ ]*)/', "<p>", $value); - $value .= "</p>"; - $grafs[$key] = $this->unhash($value); - } - else { - # Is a block. - # Modify elements of @grafs in-place... - $graf = $value; - $block = $this->html_hashes[$graf]; - $graf = $block; -// if (preg_match('{ -// \A -// ( # $1 = <div> tag -// <div \s+ -// [^>]* -// \b -// markdown\s*=\s* ([\'"]) # $2 = attr quote char -// 1 -// \2 -// [^>]* -// > -// ) -// ( # $3 = contents -// .* -// ) -// (</div>) # $4 = closing tag -// \z -// }xs', $block, $matches)) -// { -// list(, $div_open, , $div_content, $div_close) = $matches; -// -// # We can't call Markdown(), because that resets the hash; -// # that initialization code should be pulled into its own sub, though. -// $div_content = $this->hashHTMLBlocks($div_content); -// -// # Run document gamut methods on the content. -// foreach ($this->document_gamut as $method => $priority) { -// $div_content = $this->$method($div_content); -// } -// -// $div_open = preg_replace( -// '{\smarkdown\s*=\s*([\'"]).+?\1}', '', $div_open); -// -// $graf = $div_open . "\n" . $div_content . "\n" . $div_close; -// } - $grafs[$key] = $graf; - } - } - - return implode("\n\n", $grafs); - } - - - function encodeAttribute($text) { - # - # Encode text for a double-quoted HTML attribute. This function - # is *not* suitable for attributes enclosed in single quotes. - # - $text = $this->encodeAmpsAndAngles($text); - $text = str_replace('"', '"', $text); - return $text; - } - - - function encodeAmpsAndAngles($text) { - # - # Smart processing for ampersands and angle brackets that need to - # be encoded. Valid character entities are left alone unless the - # no-entities mode is set. - # - if ($this->no_entities) { - $text = str_replace('&', '&', $text); - } else { - # Ampersand-encoding based entirely on Nat Irons's Amputator - # MT plugin: <http://bumppo.net/projects/amputator/> - $text = preg_replace('/&(?!#?[xX]?(?:[0-9a-fA-F]+|\w+);)/', - '&', $text);; - } - # Encode remaining <'s - $text = str_replace('<', '<', $text); - - return $text; - } - - - function doAutoLinks($text) { - $text = preg_replace_callback('{<((https?|ftp|dict):[^\'">\s]+)>}i', - array(&$this, '_doAutoLinks_url_callback'), $text); - - # Email addresses: <address@domain.foo> - $text = preg_replace_callback('{ - < - (?:mailto:)? - ( - [-.\w\x80-\xFF]+ - \@ - [-a-z0-9\x80-\xFF]+(\.[-a-z0-9\x80-\xFF]+)*\.[a-z]+ - ) - > - }xi', - array(&$this, '_doAutoLinks_email_callback'), $text); - - return $text; - } - function _doAutoLinks_url_callback($matches) { - $url = $this->encodeAttribute($matches[1]); - $link = "<a href=\"$url\">$url</a>"; - return $this->hashPart($link); - } - function _doAutoLinks_email_callback($matches) { - $address = $matches[1]; - $link = $this->encodeEmailAddress($address); - return $this->hashPart($link); - } - - - function encodeEmailAddress($addr) { - # - # Input: an email address, e.g. "foo@example.com" - # - # Output: the email address as a mailto link, with each character - # of the address encoded as either a decimal or hex entity, in - # the hopes of foiling most address harvesting spam bots. E.g.: - # - # <p><a href="mailto:foo - # @example.co - # m">foo@exampl - # e.com</a></p> - # - # Based by a filter by Matthew Wickline, posted to BBEdit-Talk. - # With some optimizations by Milian Wolff. - # - $addr = "mailto:" . $addr; - $chars = preg_split('/(?<!^)(?!$)/', $addr); - $seed = (int)abs(crc32($addr) / strlen($addr)); # Deterministic seed. - - foreach ($chars as $key => $char) { - $ord = ord($char); - # Ignore non-ascii chars. - if ($ord < 128) { - $r = ($seed * (1 + $key)) % 100; # Pseudo-random function. - # roughly 10% raw, 45% hex, 45% dec - # '@' *must* be encoded. I insist. - if ($r > 90 && $char != '@') /* do nothing */; - else if ($r < 45) $chars[$key] = '&#x'.dechex($ord).';'; - else $chars[$key] = '&#'.$ord.';'; - } - } - - $addr = implode('', $chars); - $text = implode('', array_slice($chars, 7)); # text without `mailto:` - $addr = "<a href=\"$addr\">$text</a>"; - - return $addr; - } - - - function parseSpan($str) { - # - # Take the string $str and parse it into tokens, hashing embeded HTML, - # escaped characters and handling code spans. - # - $output = ''; - - $span_re = '{ - ( - \\\\'.$this->escape_chars_re.' - | - (?<![`\\\\]) - `+ # code span marker - '.( $this->no_markup ? '' : ' - | - <!-- .*? --> # comment - | - <\?.*?\?> | <%.*?%> # processing instruction - | - <[/!$]?[-a-zA-Z0-9:]+ # regular tags - (?> - \s - (?>[^"\'>]+|"[^"]*"|\'[^\']*\')* - )? - > - ').' - ) - }xs'; - - while (1) { - # - # Each loop iteration seach for either the next tag, the next - # openning code span marker, or the next escaped character. - # Each token is then passed to handleSpanToken. - # - $parts = preg_split($span_re, $str, 2, PREG_SPLIT_DELIM_CAPTURE); - - # Create token from text preceding tag. - if ($parts[0] != "") { - $output .= $parts[0]; - } - - # Check if we reach the end. - if (isset($parts[1])) { - $output .= $this->handleSpanToken($parts[1], $parts[2]); - $str = $parts[2]; - } - else { - break; - } - } - - return $output; - } - - - function handleSpanToken($token, &$str) { - # - # Handle $token provided by parseSpan by determining its nature and - # returning the corresponding value that should replace it. - # - switch ($token{0}) { - case "\\": - return $this->hashPart("&#". ord($token{1}). ";"); - case "`": - # Search for end marker in remaining text. - if (preg_match('/^(.*?[^`])'.preg_quote($token).'(?!`)(.*)$/sm', - $str, $matches)) - { - $str = $matches[2]; - $codespan = $this->makeCodeSpan($matches[1]); - return $this->hashPart($codespan); - } - return $token; // return as text since no ending marker found. - default: - return $this->hashPart($token); - } - } - - - function outdent($text) { - # - # Remove one level of line-leading tabs or spaces - # - return preg_replace('/^(\t|[ ]{1,'.$this->tab_width.'})/m', '', $text); - } - - - # String length function for detab. `_initDetab` will create a function to - # hanlde UTF-8 if the default function does not exist. - var $utf8_strlen = 'mb_strlen'; - - function detab($text) { - # - # Replace tabs with the appropriate amount of space. - # - # For each line we separate the line in blocks delemited by - # tab characters. Then we reconstruct every line by adding the - # appropriate number of space between each blocks. - - $text = preg_replace_callback('/^.*\t.*$/m', - array(&$this, '_detab_callback'), $text); - - return $text; - } - function _detab_callback($matches) { - $line = $matches[0]; - $strlen = $this->utf8_strlen; # strlen function for UTF-8. - - # Split in blocks. - $blocks = explode("\t", $line); - # Add each blocks to the line. - $line = $blocks[0]; - unset($blocks[0]); # Do not add first block twice. - foreach ($blocks as $block) { - # Calculate amount of space, insert spaces, insert block. - $amount = $this->tab_width - - $strlen($line, 'UTF-8') % $this->tab_width; - $line .= str_repeat(" ", $amount) . $block; - } - return $line; - } - function _initDetab() { - # - # Check for the availability of the function in the `utf8_strlen` property - # (initially `mb_strlen`). If the function is not available, create a - # function that will loosely count the number of UTF-8 characters with a - # regular expression. - # - if (function_exists($this->utf8_strlen)) return; - $this->utf8_strlen = create_function('$text', 'return preg_match_all( - "/[\\\\x00-\\\\xBF]|[\\\\xC0-\\\\xFF][\\\\x80-\\\\xBF]*/", - $text, $m);'); - } - - - function unhash($text) { - # - # Swap back in all the tags hashed by _HashHTMLBlocks. - # - return preg_replace_callback('/(.)\x1A[0-9]+\1/', - array(&$this, '_unhash_callback'), $text); - } - function _unhash_callback($matches) { - return $this->html_hashes[$matches[0]]; - } - -} - - -# -# Markdown Extra Parser Class -# - -class MarkdownExtra_Parser extends Markdown_Parser { - - # Prefix for footnote ids. - var $fn_id_prefix = ""; - - # Optional title attribute for footnote links and backlinks. - var $fn_link_title = MARKDOWN_FN_LINK_TITLE; - var $fn_backlink_title = MARKDOWN_FN_BACKLINK_TITLE; - - # Optional class attribute for footnote links and backlinks. - var $fn_link_class = MARKDOWN_FN_LINK_CLASS; - var $fn_backlink_class = MARKDOWN_FN_BACKLINK_CLASS; - - # Predefined abbreviations. - var $predef_abbr = array(); - - - function MarkdownExtra_Parser() { - # - # Constructor function. Initialize the parser object. - # - # Add extra escapable characters before parent constructor - # initialize the table. - $this->escape_chars .= ':|'; - - # Insert extra document, block, and span transformations. - # Parent constructor will do the sorting. - $this->document_gamut += array( - "doFencedCodeBlocks" => 5, - "stripFootnotes" => 15, - "stripAbbreviations" => 25, - "appendFootnotes" => 50, - ); - $this->block_gamut += array( - "doFencedCodeBlocks" => 5, - "doTables" => 15, - "doDefLists" => 45, - ); - $this->span_gamut += array( - "doFootnotes" => 5, - "doAbbreviations" => 70, - ); - - parent::Markdown_Parser(); - } - - - # Extra variables used during extra transformations. - var $footnotes = array(); - var $footnotes_ordered = array(); - var $abbr_desciptions = array(); - var $abbr_word_re = ''; - - # Give the current footnote number. - var $footnote_counter = 1; - - - function setup() { - # - # Setting up Extra-specific variables. - # - parent::setup(); - - $this->footnotes = array(); - $this->footnotes_ordered = array(); - $this->abbr_desciptions = array(); - $this->abbr_word_re = ''; - $this->footnote_counter = 1; - - foreach ($this->predef_abbr as $abbr_word => $abbr_desc) { - if ($this->abbr_word_re) - $this->abbr_word_re .= '|'; - $this->abbr_word_re .= preg_quote($abbr_word); - $this->abbr_desciptions[$abbr_word] = trim($abbr_desc); - } - } - - function teardown() { - # - # Clearing Extra-specific variables. - # - $this->footnotes = array(); - $this->footnotes_ordered = array(); - $this->abbr_desciptions = array(); - $this->abbr_word_re = ''; - - parent::teardown(); - } - - - ### HTML Block Parser ### - - # Tags that are always treated as block tags: - var $block_tags_re = 'p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|address|form|fieldset|iframe|hr|legend'; - - # Tags treated as block tags only if the opening tag is alone on it's line: - var $context_block_tags_re = 'script|noscript|math|ins|del'; - - # Tags where markdown="1" default to span mode: - var $contain_span_tags_re = 'p|h[1-6]|li|dd|dt|td|th|legend|address'; - - # Tags which must not have their contents modified, no matter where - # they appear: - var $clean_tags_re = 'script|math'; - - # Tags that do not need to be closed. - var $auto_close_tags_re = 'hr|img'; - - - function hashHTMLBlocks($text) { - # - # Hashify HTML Blocks and "clean tags". - # - # We only want to do this for block-level HTML tags, such as headers, - # lists, and tables. That's because we still want to wrap <p>s around - # "paragraphs" that are wrapped in non-block-level tags, such as anchors, - # phrase emphasis, and spans. The list of tags we're looking for is - # hard-coded. - # - # This works by calling _HashHTMLBlocks_InMarkdown, which then calls - # _HashHTMLBlocks_InHTML when it encounter block tags. When the markdown="1" - # attribute is found whitin a tag, _HashHTMLBlocks_InHTML calls back - # _HashHTMLBlocks_InMarkdown to handle the Markdown syntax within the tag. - # These two functions are calling each other. It's recursive! - # - # - # Call the HTML-in-Markdown hasher. - # - list($text, ) = $this->_hashHTMLBlocks_inMarkdown($text); - - return $text; - } - function _hashHTMLBlocks_inMarkdown($text, $indent = 0, - $enclosing_tag_re = '', $span = false) - { - # - # Parse markdown text, calling _HashHTMLBlocks_InHTML for block tags. - # - # * $indent is the number of space to be ignored when checking for code - # blocks. This is important because if we don't take the indent into - # account, something like this (which looks right) won't work as expected: - # - # <div> - # <div markdown="1"> - # Hello World. <-- Is this a Markdown code block or text? - # </div> <-- Is this a Markdown code block or a real tag? - # <div> - # - # If you don't like this, just don't indent the tag on which - # you apply the markdown="1" attribute. - # - # * If $enclosing_tag_re is not empty, stops at the first unmatched closing - # tag with that name. Nested tags supported. - # - # * If $span is true, text inside must treated as span. So any double - # newline will be replaced by a single newline so that it does not create - # paragraphs. - # - # Returns an array of that form: ( processed text , remaining text ) - # - if ($text === '') return array('', ''); - - # Regex to check for the presense of newlines around a block tag. - $newline_before_re = '/(?:^\n?|\n\n)*$/'; - $newline_after_re = - '{ - ^ # Start of text following the tag. - (?>[ ]*<!--.*?-->)? # Optional comment. - [ ]*\n # Must be followed by newline. - }xs'; - - # Regex to match any tag. - $block_tag_re = - '{ - ( # $2: Capture hole tag. - </? # Any opening or closing tag. - (?> # Tag name. - '.$this->block_tags_re.' | - '.$this->context_block_tags_re.' | - '.$this->clean_tags_re.' | - (?!\s)'.$enclosing_tag_re.' - ) - (?: - (?=[\s"\'/a-zA-Z0-9]) # Allowed characters after tag name. - (?> - ".*?" | # Double quotes (can contain `>`) - \'.*?\' | # Single quotes (can contain `>`) - .+? # Anything but quotes and `>`. - )*? - )? - > # End of tag. - | - <!-- .*? --> # HTML Comment - | - <\?.*?\?> | <%.*?%> # Processing instruction - | - <!\[CDATA\[.*?\]\]> # CData Block - | - # Code span marker - `+ - '. ( !$span ? ' # If not in span. - | - # Indented code block - (?> ^[ ]*\n? | \n[ ]*\n ) - [ ]{'.($indent+4).'}[^\n]* \n - (?> - (?: [ ]{'.($indent+4).'}[^\n]* | [ ]* ) \n - )* - | - # Fenced code block marker - (?> ^ | \n ) - [ ]{'.($indent).'}~~~+[ ]*\n - ' : '' ). ' # End (if not is span). - ) - }xs'; - - - $depth = 0; # Current depth inside the tag tree. - $parsed = ""; # Parsed text that will be returned. - - # - # Loop through every tag until we find the closing tag of the parent - # or loop until reaching the end of text if no parent tag specified. - # - do { - # - # Split the text using the first $tag_match pattern found. - # Text before pattern will be first in the array, text after - # pattern will be at the end, and between will be any catches made - # by the pattern. - # - $parts = preg_split($block_tag_re, $text, 2, - PREG_SPLIT_DELIM_CAPTURE); - - # If in Markdown span mode, add a empty-string span-level hash - # after each newline to prevent triggering any block element. - if ($span) { - $void = $this->hashPart("", ':'); - $newline = "$void\n"; - $parts[0] = $void . str_replace("\n", $newline, $parts[0]) . $void; - } - - $parsed .= $parts[0]; # Text before current tag. - - # If end of $text has been reached. Stop loop. - if (count($parts) < 3) { - $text = ""; - break; - } - - $tag = $parts[1]; # Tag to handle. - $text = $parts[2]; # Remaining text after current tag. - $tag_re = preg_quote($tag); # For use in a regular expression. - - # - # Check for: Code span marker - # - if ($tag{0} == "`") { - # Find corresponding end marker. - $tag_re = preg_quote($tag); - if (preg_match('{^(?>.+?|\n(?!\n))*?(?<!`)'.$tag_re.'(?!`)}', - $text, $matches)) - { - # End marker found: pass text unchanged until marker. - $parsed .= $tag . $matches[0]; - $text = substr($text, strlen($matches[0])); - } - else { - # Unmatched marker: just skip it. - $parsed .= $tag; - } - } - # - # Check for: Indented code block or fenced code block marker. - # - else if ($tag{0} == "\n" || $tag{0} == "~") { - if ($tag{1} == "\n" || $tag{1} == " ") { - # Indented code block: pass it unchanged, will be handled - # later. - $parsed .= $tag; - } - else { - # Fenced code block marker: find matching end marker. - $tag_re = preg_quote(trim($tag)); - if (preg_match('{^(?>.*\n)+?'.$tag_re.' *\n}', $text, - $matches)) - { - # End marker found: pass text unchanged until marker. - $parsed .= $tag . $matches[0]; - $text = substr($text, strlen($matches[0])); - } - else { - # No end marker: just skip it. - $parsed .= $tag; - } - } - } - # - # Check for: Opening Block level tag or - # Opening Context Block tag (like ins and del) - # used as a block tag (tag is alone on it's line). - # - else if (preg_match('{^<(?:'.$this->block_tags_re.')\b}', $tag) || - ( preg_match('{^<(?:'.$this->context_block_tags_re.')\b}', $tag) && - preg_match($newline_before_re, $parsed) && - preg_match($newline_after_re, $text) ) - ) - { - # Need to parse tag and following text using the HTML parser. - list($block_text, $text) = - $this->_hashHTMLBlocks_inHTML($tag . $text, "hashBlock", true); - - # Make sure it stays outside of any paragraph by adding newlines. - $parsed .= "\n\n$block_text\n\n"; - } - # - # Check for: Clean tag (like script, math) - # HTML Comments, processing instructions. - # - else if (preg_match('{^<(?:'.$this->clean_tags_re.')\b}', $tag) || - $tag{1} == '!' || $tag{1} == '?') - { - # Need to parse tag and following text using the HTML parser. - # (don't check for markdown attribute) - list($block_text, $text) = - $this->_hashHTMLBlocks_inHTML($tag . $text, "hashClean", false); - - $parsed .= $block_text; - } - # - # Check for: Tag with same name as enclosing tag. - # - else if ($enclosing_tag_re !== '' && - # Same name as enclosing tag. - preg_match('{^</?(?:'.$enclosing_tag_re.')\b}', $tag)) - { - # - # Increase/decrease nested tag count. - # - if ($tag{1} == '/') $depth--; - else if ($tag{strlen($tag)-2} != '/') $depth++; - - if ($depth < 0) { - # - # Going out of parent element. Clean up and break so we - # return to the calling function. - # - $text = $tag . $text; - break; - } - - $parsed .= $tag; - } - else { - $parsed .= $tag; - } - } while ($depth >= 0); - - return array($parsed, $text); - } - function _hashHTMLBlocks_inHTML($text, $hash_method, $md_attr) { - # - # Parse HTML, calling _HashHTMLBlocks_InMarkdown for block tags. - # - # * Calls $hash_method to convert any blocks. - # * Stops when the first opening tag closes. - # * $md_attr indicate if the use of the `markdown="1"` attribute is allowed. - # (it is not inside clean tags) - # - # Returns an array of that form: ( processed text , remaining text ) - # - if ($text === '') return array('', ''); - - # Regex to match `markdown` attribute inside of a tag. - $markdown_attr_re = ' - { - \s* # Eat whitespace before the `markdown` attribute - markdown - \s*=\s* - (?> - (["\']) # $1: quote delimiter - (.*?) # $2: attribute value - \1 # matching delimiter - | - ([^\s>]*) # $3: unquoted attribute value - ) - () # $4: make $3 always defined (avoid warnings) - }xs'; - - # Regex to match any tag. - $tag_re = '{ - ( # $2: Capture hole tag. - </? # Any opening or closing tag. - [\w:$]+ # Tag name. - (?: - (?=[\s"\'/a-zA-Z0-9]) # Allowed characters after tag name. - (?> - ".*?" | # Double quotes (can contain `>`) - \'.*?\' | # Single quotes (can contain `>`) - .+? # Anything but quotes and `>`. - )*? - )? - > # End of tag. - | - <!-- .*? --> # HTML Comment - | - <\?.*?\?> | <%.*?%> # Processing instruction - | - <!\[CDATA\[.*?\]\]> # CData Block - ) - }xs'; - - $original_text = $text; # Save original text in case of faliure. - - $depth = 0; # Current depth inside the tag tree. - $block_text = ""; # Temporary text holder for current text. - $parsed = ""; # Parsed text that will be returned. - - # - # Get the name of the starting tag. - # (This pattern makes $base_tag_name_re safe without quoting.) - # - if (preg_match('/^<([\w:$]*)\b/', $text, $matches)) - $base_tag_name_re = $matches[1]; - - # - # Loop through every tag until we find the corresponding closing tag. - # - do { - # - # Split the text using the first $tag_match pattern found. - # Text before pattern will be first in the array, text after - # pattern will be at the end, and between will be any catches made - # by the pattern. - # - $parts = preg_split($tag_re, $text, 2, PREG_SPLIT_DELIM_CAPTURE); - - if (count($parts) < 3) { - # - # End of $text reached with unbalenced tag(s). - # In that case, we return original text unchanged and pass the - # first character as filtered to prevent an infinite loop in the - # parent function. - # - return array($original_text{0}, substr($original_text, 1)); - } - - $block_text .= $parts[0]; # Text before current tag. - $tag = $parts[1]; # Tag to handle. - $text = $parts[2]; # Remaining text after current tag. - - # - # Check for: Auto-close tag (like <hr/>) - # Comments and Processing Instructions. - # - if (preg_match('{^</?(?:'.$this->auto_close_tags_re.')\b}', $tag) || - $tag{1} == '!' || $tag{1} == '?') - { - # Just add the tag to the block as if it was text. - $block_text .= $tag; - } - else { - # - # Increase/decrease nested tag count. Only do so if - # the tag's name match base tag's. - # - if (preg_match('{^</?'.$base_tag_name_re.'\b}', $tag)) { - if ($tag{1} == '/') $depth--; - else if ($tag{strlen($tag)-2} != '/') $depth++; - } - - # - # Check for `markdown="1"` attribute and handle it. - # - if ($md_attr && - preg_match($markdown_attr_re, $tag, $attr_m) && - preg_match('/^1|block|span$/', $attr_m[2] . $attr_m[3])) - { - # Remove `markdown` attribute from opening tag. - $tag = preg_replace($markdown_attr_re, '', $tag); - - # Check if text inside this tag must be parsed in span mode. - $this->mode = $attr_m[2] . $attr_m[3]; - $span_mode = $this->mode == 'span' || $this->mode != 'block' && - preg_match('{^<(?:'.$this->contain_span_tags_re.')\b}', $tag); - - # Calculate indent before tag. - if (preg_match('/(?:^|\n)( *?)(?! ).*?$/', $block_text, $matches)) { - $strlen = $this->utf8_strlen; - $indent = $strlen($matches[1], 'UTF-8'); - } else { - $indent = 0; - } - - # End preceding block with this tag. - $block_text .= $tag; - $parsed .= $this->$hash_method($block_text); - - # Get enclosing tag name for the ParseMarkdown function. - # (This pattern makes $tag_name_re safe without quoting.) - preg_match('/^<([\w:$]*)\b/', $tag, $matches); - $tag_name_re = $matches[1]; - - # Parse the content using the HTML-in-Markdown parser. - list ($block_text, $text) - = $this->_hashHTMLBlocks_inMarkdown($text, $indent, - $tag_name_re, $span_mode); - - # Outdent markdown text. - if ($indent > 0) { - $block_text = preg_replace("/^[ ]{1,$indent}/m", "", - $block_text); - } - - # Append tag content to parsed text. - if (!$span_mode) $parsed .= "\n\n$block_text\n\n"; - else $parsed .= "$block_text"; - - # Start over a new block. - $block_text = ""; - } - else $block_text .= $tag; - } - - } while ($depth > 0); - - # - # Hash last block text that wasn't processed inside the loop. - # - $parsed .= $this->$hash_method($block_text); - - return array($parsed, $text); - } - - - function hashClean($text) { - # - # Called whenever a tag must be hashed when a function insert a "clean" tag - # in $text, it pass through this function and is automaticaly escaped, - # blocking invalid nested overlap. - # - return $this->hashPart($text, 'C'); - } - - - function doHeaders($text) { - # - # Redefined to add id attribute support. - # - # Setext-style headers: - # Header 1 {#header1} - # ======== - # - # Header 2 {#header2} - # -------- - # - $text = preg_replace_callback( - '{ - (^.+?) # $1: Header text - (?:[ ]+\{\#([-_:a-zA-Z0-9]+)\})? # $2: Id attribute - [ ]*\n(=+|-+)[ ]*\n+ # $3: Header footer - }mx', - array(&$this, '_doHeaders_callback_setext'), $text); - - # atx-style headers: - # # Header 1 {#header1} - # ## Header 2 {#header2} - # ## Header 2 with closing hashes ## {#header3} - # ... - # ###### Header 6 {#header2} - # - $text = preg_replace_callback('{ - ^(\#{1,6}) # $1 = string of #\'s - [ ]* - (.+?) # $2 = Header text - [ ]* - \#* # optional closing #\'s (not counted) - (?:[ ]+\{\#([-_:a-zA-Z0-9]+)\})? # id attribute - [ ]* - \n+ - }xm', - array(&$this, '_doHeaders_callback_atx'), $text); - - return $text; - } - function _doHeaders_attr($attr) { - if (empty($attr)) return ""; - return " id=\"$attr\""; - } - function _doHeaders_callback_setext($matches) { - if ($matches[3] == '-' && preg_match('{^- }', $matches[1])) - return $matches[0]; - $level = $matches[3]{0} == '=' ? 1 : 2; - $attr = $this->_doHeaders_attr($id =& $matches[2]); - $block = "<h$level$attr>".$this->runSpanGamut($matches[1])."</h$level>"; - return "\n" . $this->hashBlock($block) . "\n\n"; - } - function _doHeaders_callback_atx($matches) { - $level = strlen($matches[1]); - $attr = $this->_doHeaders_attr($id =& $matches[3]); - $block = "<h$level$attr>".$this->runSpanGamut($matches[2])."</h$level>"; - return "\n" . $this->hashBlock($block) . "\n\n"; - } - - - function doTables($text) { - # - # Form HTML tables. - # - $less_than_tab = $this->tab_width - 1; - # - # Find tables with leading pipe. - # - # | Header 1 | Header 2 - # | -------- | -------- - # | Cell 1 | Cell 2 - # | Cell 3 | Cell 4 - # - $text = preg_replace_callback(' - { - ^ # Start of a line - [ ]{0,'.$less_than_tab.'} # Allowed whitespace. - [|] # Optional leading pipe (present) - (.+) \n # $1: Header row (at least one pipe) - - [ ]{0,'.$less_than_tab.'} # Allowed whitespace. - [|] ([ ]*[-:]+[-| :]*) \n # $2: Header underline - - ( # $3: Cells - (?> - [ ]* # Allowed whitespace. - [|] .* \n # Row content. - )* - ) - (?=\n|\Z) # Stop at final double newline. - }xm', - array(&$this, '_doTable_leadingPipe_callback'), $text); - - # - # Find tables without leading pipe. - # - # Header 1 | Header 2 - # -------- | -------- - # Cell 1 | Cell 2 - # Cell 3 | Cell 4 - # - $text = preg_replace_callback(' - { - ^ # Start of a line - [ ]{0,'.$less_than_tab.'} # Allowed whitespace. - (\S.*[|].*) \n # $1: Header row (at least one pipe) - - [ ]{0,'.$less_than_tab.'} # Allowed whitespace. - ([-:]+[ ]*[|][-| :]*) \n # $2: Header underline - - ( # $3: Cells - (?> - .* [|] .* \n # Row content - )* - ) - (?=\n|\Z) # Stop at final double newline. - }xm', - array(&$this, '_DoTable_callback'), $text); - - return $text; - } - function _doTable_leadingPipe_callback($matches) { - $head = $matches[1]; - $underline = $matches[2]; - $content = $matches[3]; - - # Remove leading pipe for each row. - $content = preg_replace('/^ *[|]/m', '', $content); - - return $this->_doTable_callback(array($matches[0], $head, $underline, $content)); - } - function _doTable_callback($matches) { - $head = $matches[1]; - $underline = $matches[2]; - $content = $matches[3]; - - # Remove any tailing pipes for each line. - $head = preg_replace('/[|] *$/m', '', $head); - $underline = preg_replace('/[|] *$/m', '', $underline); - $content = preg_replace('/[|] *$/m', '', $content); - - # Reading alignement from header underline. - $separators = preg_split('/ *[|] */', $underline); - foreach ($separators as $n => $s) { - if (preg_match('/^ *-+: *$/', $s)) $attr[$n] = ' align="right"'; - else if (preg_match('/^ *:-+: *$/', $s))$attr[$n] = ' align="center"'; - else if (preg_match('/^ *:-+ *$/', $s)) $attr[$n] = ' align="left"'; - else $attr[$n] = ''; - } - - # Parsing span elements, including code spans, character escapes, - # and inline HTML tags, so that pipes inside those gets ignored. - $head = $this->parseSpan($head); - $headers = preg_split('/ *[|] */', $head); - $col_count = count($headers); - - # Write column headers. - $text = "<table>\n"; - $text .= "<thead>\n"; - $text .= "<tr>\n"; - foreach ($headers as $n => $header) - $text .= " <th$attr[$n]>".$this->runSpanGamut(trim($header))."</th>\n"; - $text .= "</tr>\n"; - $text .= "</thead>\n"; - - # Split content by row. - $rows = explode("\n", trim($content, "\n")); - - $text .= "<tbody>\n"; - foreach ($rows as $row) { - # Parsing span elements, including code spans, character escapes, - # and inline HTML tags, so that pipes inside those gets ignored. - $row = $this->parseSpan($row); - - # Split row by cell. - $row_cells = preg_split('/ *[|] */', $row, $col_count); - $row_cells = array_pad($row_cells, $col_count, ''); - - $text .= "<tr>\n"; - foreach ($row_cells as $n => $cell) - $text .= " <td$attr[$n]>".$this->runSpanGamut(trim($cell))."</td>\n"; - $text .= "</tr>\n"; - } - $text .= "</tbody>\n"; - $text .= "</table>"; - - return $this->hashBlock($text) . "\n"; - } - - - function doDefLists($text) { - # - # Form HTML definition lists. - # - $less_than_tab = $this->tab_width - 1; - - # Re-usable pattern to match any entire dl list: - $whole_list_re = '(?> - ( # $1 = whole list - ( # $2 - [ ]{0,'.$less_than_tab.'} - ((?>.*\S.*\n)+) # $3 = defined term - \n? - [ ]{0,'.$less_than_tab.'}:[ ]+ # colon starting definition - ) - (?s:.+?) - ( # $4 - \z - | - \n{2,} - (?=\S) - (?! # Negative lookahead for another term - [ ]{0,'.$less_than_tab.'} - (?: \S.*\n )+? # defined term - \n? - [ ]{0,'.$less_than_tab.'}:[ ]+ # colon starting definition - ) - (?! # Negative lookahead for another definition - [ ]{0,'.$less_than_tab.'}:[ ]+ # colon starting definition - ) - ) - ) - )'; // mx - - $text = preg_replace_callback('{ - (?>\A\n?|(?<=\n\n)) - '.$whole_list_re.' - }mx', - array(&$this, '_doDefLists_callback'), $text); - - return $text; - } - function _doDefLists_callback($matches) { - # Re-usable patterns to match list item bullets and number markers: - $list = $matches[1]; - - # Turn double returns into triple returns, so that we can make a - # paragraph for the last item in a list, if necessary: - $result = trim($this->processDefListItems($list)); - $result = "<dl>\n" . $result . "\n</dl>"; - return $this->hashBlock($result) . "\n\n"; - } - - - function processDefListItems($list_str) { - # - # Process the contents of a single definition list, splitting it - # into individual term and definition list items. - # - $less_than_tab = $this->tab_width - 1; - - # trim trailing blank lines: - $list_str = preg_replace("/\n{2,}\\z/", "\n", $list_str); - - # Process definition terms. - $list_str = preg_replace_callback('{ - (?>\A\n?|\n\n+) # leading line - ( # definition terms = $1 - [ ]{0,'.$less_than_tab.'} # leading whitespace - (?![:][ ]|[ ]) # negative lookahead for a definition - # mark (colon) or more whitespace. - (?> \S.* \n)+? # actual term (not whitespace). - ) - (?=\n?[ ]{0,3}:[ ]) # lookahead for following line feed - # with a definition mark. - }xm', - array(&$this, '_processDefListItems_callback_dt'), $list_str); - - # Process actual definitions. - $list_str = preg_replace_callback('{ - \n(\n+)? # leading line = $1 - ( # marker space = $2 - [ ]{0,'.$less_than_tab.'} # whitespace before colon - [:][ ]+ # definition mark (colon) - ) - ((?s:.+?)) # definition text = $3 - (?= \n+ # stop at next definition mark, - (?: # next term or end of text - [ ]{0,'.$less_than_tab.'} [:][ ] | - <dt> | \z - ) - ) - }xm', - array(&$this, '_processDefListItems_callback_dd'), $list_str); - - return $list_str; - } - function _processDefListItems_callback_dt($matches) { - $terms = explode("\n", trim($matches[1])); - $text = ''; - foreach ($terms as $term) { - $term = $this->runSpanGamut(trim($term)); - $text .= "\n<dt>" . $term . "</dt>"; - } - return $text . "\n"; - } - function _processDefListItems_callback_dd($matches) { - $leading_line = $matches[1]; - $marker_space = $matches[2]; - $def = $matches[3]; - - if ($leading_line || preg_match('/\n{2,}/', $def)) { - # Replace marker with the appropriate whitespace indentation - $def = str_repeat(' ', strlen($marker_space)) . $def; - $def = $this->runBlockGamut($this->outdent($def . "\n\n")); - $def = "\n". $def ."\n"; - } - else { - $def = rtrim($def); - $def = $this->runSpanGamut($this->outdent($def)); - } - - return "\n<dd>" . $def . "</dd>\n"; - } - - - function doFencedCodeBlocks($text) { - # - # Adding the fenced code block syntax to regular Markdown: - # - # ~~~ - # Code block - # ~~~ - # - $less_than_tab = $this->tab_width; - - $text = preg_replace_callback('{ - (?:\n|\A) - # 1: Opening marker - ( - ~{3,} # Marker: three tilde or more. - ) - [ ]* \n # Whitespace and newline following marker. - - # 2: Content - ( - (?> - (?!\1 [ ]* \n) # Not a closing marker. - .*\n+ - )+ - ) - - # Closing marker. - \1 [ ]* \n - }xm', - array(&$this, '_doFencedCodeBlocks_callback'), $text); - - return $text; - } - function _doFencedCodeBlocks_callback($matches) { - $codeblock = $matches[2]; - $codeblock = htmlspecialchars($codeblock, ENT_NOQUOTES); - $codeblock = preg_replace_callback('/^\n+/', - array(&$this, '_doFencedCodeBlocks_newlines'), $codeblock); - $codeblock = "<pre><code>$codeblock</code></pre>"; - return "\n\n".$this->hashBlock($codeblock)."\n\n"; - } - function _doFencedCodeBlocks_newlines($matches) { - return str_repeat("<br$this->empty_element_suffix", - strlen($matches[0])); - } - - - # - # Redefining emphasis markers so that emphasis by underscore does not - # work in the middle of a word. - # - var $em_relist = array( - '' => '(?:(?<!\*)\*(?!\*)|(?<![a-zA-Z0-9_])_(?!_))(?=\S)(?![.,:;]\s)', - '*' => '(?<=\S)(?<!\*)\*(?!\*)', - '_' => '(?<=\S)(?<!_)_(?![a-zA-Z0-9_])', - ); - var $strong_relist = array( - '' => '(?:(?<!\*)\*\*(?!\*)|(?<![a-zA-Z0-9_])__(?!_))(?=\S)(?![.,:;]\s)', - '**' => '(?<=\S)(?<!\*)\*\*(?!\*)', - '__' => '(?<=\S)(?<!_)__(?![a-zA-Z0-9_])', - ); - var $em_strong_relist = array( - '' => '(?:(?<!\*)\*\*\*(?!\*)|(?<![a-zA-Z0-9_])___(?!_))(?=\S)(?![.,:;]\s)', - '***' => '(?<=\S)(?<!\*)\*\*\*(?!\*)', - '___' => '(?<=\S)(?<!_)___(?![a-zA-Z0-9_])', - ); - - - function formParagraphs($text) { - # - # Params: - # $text - string to process with html <p> tags - # - # Strip leading and trailing lines: - $text = preg_replace('/\A\n+|\n+\z/', '', $text); - - $grafs = preg_split('/\n{2,}/', $text, -1, PREG_SPLIT_NO_EMPTY); - - # - # Wrap <p> tags and unhashify HTML blocks - # - foreach ($grafs as $key => $value) { - $value = trim($this->runSpanGamut($value)); - - # Check if this should be enclosed in a paragraph. - # Clean tag hashes & block tag hashes are left alone. - $is_p = !preg_match('/^B\x1A[0-9]+B|^C\x1A[0-9]+C$/', $value); - - if ($is_p) { - $value = "<p>$value</p>"; - } - $grafs[$key] = $value; - } - - # Join grafs in one text, then unhash HTML tags. - $text = implode("\n\n", $grafs); - - # Finish by removing any tag hashes still present in $text. - $text = $this->unhash($text); - - return $text; - } - - - ### Footnotes - - function stripFootnotes($text) { - # - # Strips link definitions from text, stores the URLs and titles in - # hash references. - # - $less_than_tab = $this->tab_width - 1; - - # Link defs are in the form: [^id]: url "optional title" - $text = preg_replace_callback('{ - ^[ ]{0,'.$less_than_tab.'}\[\^(.+?)\][ ]?: # note_id = $1 - [ ]* - \n? # maybe *one* newline - ( # text = $2 (no blank lines allowed) - (?: - .+ # actual text - | - \n # newlines but - (?!\[\^.+?\]:\s)# negative lookahead for footnote marker. - (?!\n+[ ]{0,3}\S)# ensure line is not blank and followed - # by non-indented content - )* - ) - }xm', - array(&$this, '_stripFootnotes_callback'), - $text); - return $text; - } - function _stripFootnotes_callback($matches) { - $note_id = $this->fn_id_prefix . $matches[1]; - $this->footnotes[$note_id] = $this->outdent($matches[2]); - return ''; # String that will replace the block - } - - - function doFootnotes($text) { - # - # Replace footnote references in $text [^id] with a special text-token - # which will be replaced by the actual footnote marker in appendFootnotes. - # - if (!$this->in_anchor) { - $text = preg_replace('{\[\^(.+?)\]}', "F\x1Afn:\\1\x1A:", $text); - } - return $text; - } - - - function appendFootnotes($text) { - # - # Append footnote list to text. - # - $text = preg_replace_callback('{F\x1Afn:(.*?)\x1A:}', - array(&$this, '_appendFootnotes_callback'), $text); - - if (!empty($this->footnotes_ordered)) { - $text .= "\n\n"; - $text .= "<div class=\"footnotes\">\n"; - $text .= "<hr". MARKDOWN_EMPTY_ELEMENT_SUFFIX ."\n"; - $text .= "<ol>\n\n"; - - $attr = " rev=\"footnote\""; - if ($this->fn_backlink_class != "") { - $class = $this->fn_backlink_class; - $class = $this->encodeAttribute($class); - $attr .= " class=\"$class\""; - } - if ($this->fn_backlink_title != "") { - $title = $this->fn_backlink_title; - $title = $this->encodeAttribute($title); - $attr .= " title=\"$title\""; - } - $num = 0; - - while (!empty($this->footnotes_ordered)) { - $footnote = reset($this->footnotes_ordered); - $note_id = key($this->footnotes_ordered); - unset($this->footnotes_ordered[$note_id]); - - $footnote .= "\n"; # Need to append newline before parsing. - $footnote = $this->runBlockGamut("$footnote\n"); - $footnote = preg_replace_callback('{F\x1Afn:(.*?)\x1A:}', - array(&$this, '_appendFootnotes_callback'), $footnote); - - $attr = str_replace("%%", ++$num, $attr); - $note_id = $this->encodeAttribute($note_id); - - # Add backlink to last paragraph; create new paragraph if needed. - $backlink = "<a href=\"#fnref:$note_id\"$attr>↩</a>"; - if (preg_match('{</p>$}', $footnote)) { - $footnote = substr($footnote, 0, -4) . " $backlink</p>"; - } else { - $footnote .= "\n\n<p>$backlink</p>"; - } - - $text .= "<li id=\"fn:$note_id\">\n"; - $text .= $footnote . "\n"; - $text .= "</li>\n\n"; - } - - $text .= "</ol>\n"; - $text .= "</div>"; - } - return $text; - } - function _appendFootnotes_callback($matches) { - $node_id = $this->fn_id_prefix . $matches[1]; - - # Create footnote marker only if it has a corresponding footnote *and* - # the footnote hasn't been used by another marker. - if (isset($this->footnotes[$node_id])) { - # Transfert footnote content to the ordered list. - $this->footnotes_ordered[$node_id] = $this->footnotes[$node_id]; - unset($this->footnotes[$node_id]); - - $num = $this->footnote_counter++; - $attr = " rel=\"footnote\""; - if ($this->fn_link_class != "") { - $class = $this->fn_link_class; - $class = $this->encodeAttribute($class); - $attr .= " class=\"$class\""; - } - if ($this->fn_link_title != "") { - $title = $this->fn_link_title; - $title = $this->encodeAttribute($title); - $attr .= " title=\"$title\""; - } - - $attr = str_replace("%%", $num, $attr); - $node_id = $this->encodeAttribute($node_id); - - return - "<sup id=\"fnref:$node_id\">". - "<a href=\"#fn:$node_id\"$attr>$num</a>". - "</sup>"; - } - - return "[^".$matches[1]."]"; - } - - - ### Abbreviations ### - - function stripAbbreviations($text) { - # - # Strips abbreviations from text, stores titles in hash references. - # - $less_than_tab = $this->tab_width - 1; - - # Link defs are in the form: [id]*: url "optional title" - $text = preg_replace_callback('{ - ^[ ]{0,'.$less_than_tab.'}\*\[(.+?)\][ ]?: # abbr_id = $1 - (.*) # text = $2 (no blank lines allowed) - }xm', - array(&$this, '_stripAbbreviations_callback'), - $text); - return $text; - } - function _stripAbbreviations_callback($matches) { - $abbr_word = $matches[1]; - $abbr_desc = $matches[2]; - if ($this->abbr_word_re) - $this->abbr_word_re .= '|'; - $this->abbr_word_re .= preg_quote($abbr_word); - $this->abbr_desciptions[$abbr_word] = trim($abbr_desc); - return ''; # String that will replace the block - } - - - function doAbbreviations($text) { - # - # Find defined abbreviations in text and wrap them in <abbr> elements. - # - if ($this->abbr_word_re) { - // cannot use the /x modifier because abbr_word_re may - // contain significant spaces: - $text = preg_replace_callback('{'. - '(?<![\w\x1A])'. - '(?:'.$this->abbr_word_re.')'. - '(?![\w\x1A])'. - '}', - array(&$this, '_doAbbreviations_callback'), $text); - } - return $text; - } - function _doAbbreviations_callback($matches) { - $abbr = $matches[0]; - if (isset($this->abbr_desciptions[$abbr])) { - $desc = $this->abbr_desciptions[$abbr]; - if (empty($desc)) { - return $this->hashPart("<abbr>$abbr</abbr>"); - } else { - $desc = $this->encodeAttribute($desc); - return $this->hashPart("<abbr title=\"$desc\">$abbr</abbr>"); - } - } else { - return $matches[0]; - } - } - -} - - -/* - -PHP Markdown Extra -================== - -Description ------------ - -This is a PHP port of the original Markdown formatter written in Perl -by John Gruber. This special "Extra" version of PHP Markdown features -further enhancements to the syntax for making additional constructs -such as tables and definition list. - -Markdown is a text-to-HTML filter; it translates an easy-to-read / -easy-to-write structured text format into HTML. Markdown's text format -is most similar to that of plain text email, and supports features such -as headers, *emphasis*, code blocks, blockquotes, and links. - -Markdown's syntax is designed not as a generic markup language, but -specifically to serve as a front-end to (X)HTML. You can use span-level -HTML tags anywhere in a Markdown document, and you can use block level -HTML tags (like <div> and <table> as well). - -For more information about Markdown's syntax, see: - -<http://daringfireball.net/projects/markdown/> - - -Bugs ----- - -To file bug reports please send email to: - -<michel.fortin@michelf.com> - -Please include with your report: (1) the example input; (2) the output you -expected; (3) the output Markdown actually produced. - - -Version History ---------------- - -See the readme file for detailed release notes for this version. - - -Copyright and License ---------------------- - -PHP Markdown & Extra -Copyright (c) 2004-2008 Michel Fortin -<http://www.michelf.com/> -All rights reserved. - -Based on Markdown -Copyright (c) 2003-2006 John Gruber -<http://daringfireball.net/> -All rights reserved. - -Redistribution and use in source and binary forms, with or without -modification, are permitted provided that the following conditions are -met: - -* Redistributions of source code must retain the above copyright notice, - this list of conditions and the following disclaimer. - -* Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in the - documentation and/or other materials provided with the distribution. - -* Neither the name "Markdown" nor the names of its contributors may - be used to endorse or promote products derived from this software - without specific prior written permission. - -This software is provided by the copyright holders and contributors "as -is" and any express or implied warranties, including, but not limited -to, the implied warranties of merchantability and fitness for a -particular purpose are disclaimed. In no event shall the copyright owner -or contributors be liable for any direct, indirect, incidental, special, -exemplary, or consequential damages (including, but not limited to, -procurement of substitute goods or services; loss of use, data, or -profits; or business interruption) however caused and on any theory of -liability, whether in contract, strict liability, or tort (including -negligence or otherwise) arising in any way out of the use of this -software, even if advised of the possibility of such damage. - -*/ -?>
\ No newline at end of file @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -17,10 +18,10 @@ #include <string.h> #define VPX_CODEC_DISABLE_COMPAT 1 #include "vpx_config.h" -#include "vpx_decoder.h" +#include "vpx/vpx_decoder.h" #include "vpx_ports/vpx_timer.h" #if CONFIG_VP8_DECODER -#include "vp8dx.h" +#include "vpx/vp8dx.h" #endif #if CONFIG_MD5 #include "md5_utils.h" @@ -62,8 +63,10 @@ static const arg_def_t postprocarg = ARG_DEF(NULL, "postproc", 0, "Postprocess decoded frames"); static const arg_def_t summaryarg = ARG_DEF(NULL, "summary", 0, "Show timing summary"); -static const arg_def_t outputfile = ARG_DEF("o", "output-raw-file", 1, +static const arg_def_t outputfile = ARG_DEF("o", "output", 1, "Output raw yv12 file instead of images"); +static const arg_def_t usey4marg = ARG_DEF("y", "y4m", 0, + "Output file is YUV4MPEG2"); static const arg_def_t threadsarg = ARG_DEF("t", "threads", 1, "Max threads to use"); static const arg_def_t quietarg = ARG_DEF("q", "quiet", 0, @@ -77,7 +80,7 @@ static const arg_def_t *all_args[] = { &codecarg, &prefixarg, &use_yv12, &use_i420, &flipuvarg, &noblitarg, &progressarg, &limitarg, &postprocarg, &summaryarg, &outputfile, - &threadsarg, &quietarg, + &usey4marg, &threadsarg, &quietarg, #if CONFIG_MD5 &md5arg, #endif @@ -232,9 +235,9 @@ void *out_open(const char *out_fn, int do_md5) if (do_md5) { #if CONFIG_MD5 - md5_ctx_t *md5_ctx = out = malloc(sizeof(md5_ctx_t)); + MD5Context *md5_ctx = out = malloc(sizeof(MD5Context)); (void)out_fn; - md5_init(md5_ctx); + MD5Init(md5_ctx); #endif } else @@ -256,7 +259,7 @@ void out_put(void *out, const uint8_t *buf, unsigned int len, int do_md5) if (do_md5) { #if CONFIG_MD5 - md5_update(out, buf, len); + MD5Update(out, buf, len); #endif } else @@ -273,7 +276,7 @@ void out_close(void *out, const char *out_fn, int do_md5) uint8_t md5[16]; int i; - md5_finalize(out, md5); + MD5Final(md5, out); free(out); for (i = 0; i < 16; i++) @@ -288,7 +291,12 @@ void out_close(void *out, const char *out_fn, int do_md5) } } -unsigned int file_is_ivf(FILE *infile, unsigned int *fourcc) +unsigned int file_is_ivf(FILE *infile, + unsigned int *fourcc, + unsigned int *width, + unsigned int *height, + unsigned int *timebase_num, + unsigned int *timebase_den) { char raw_hdr[32]; int is_ivf = 0; @@ -305,6 +313,10 @@ unsigned int file_is_ivf(FILE *infile, unsigned int *fourcc) " decode properly."); *fourcc = mem_get_le32(raw_hdr + 8); + *width = mem_get_le16(raw_hdr + 12); + *height = mem_get_le16(raw_hdr + 14); + *timebase_den = mem_get_le32(raw_hdr + 16); + *timebase_num = mem_get_le32(raw_hdr + 20); } } @@ -330,6 +342,11 @@ int main(int argc, const char **argv_) struct arg arg; char **argv, **argi, **argj; const char *fn2 = 0; + int use_y4m = 0; + unsigned int width; + unsigned int height; + unsigned int timebase_num; + unsigned int timebase_den; void *out = NULL; vpx_codec_dec_cfg_t cfg = {0}; #if CONFIG_VP8_DECODER @@ -361,6 +378,8 @@ int main(int argc, const char **argv_) } else if (arg_match(&arg, &outputfile, argi)) fn2 = arg.val; + else if (arg_match(&arg, &usey4marg, argi)) + use_y4m = 1; else if (arg_match(&arg, &prefixarg, argi)) prefix = strdup(arg.val); else if (arg_match(&arg, &use_yv12, argi)) @@ -446,10 +465,31 @@ int main(int argc, const char **argv_) if (fn2) out = out_open(fn2, do_md5); - is_ivf = file_is_ivf(infile, &fourcc); + is_ivf = file_is_ivf(infile, &fourcc, &width, &height, + &timebase_num, &timebase_den); if (is_ivf) { + if (use_y4m) + { + char buffer[128]; + if (!fn2) + { + fprintf(stderr, "YUV4MPEG2 output only supported with -o.\n"); + return EXIT_FAILURE; + } + /*Correct for the factor of 2 applied to the timebase in the + encoder.*/ + if(timebase_den&1)timebase_num<<=1; + else timebase_den>>=1; + /*Note: We can't output an aspect ratio here because IVF doesn't + store one, and neither does VP8. + That will have to wait until these tools support WebM natively.*/ + sprintf(buffer, "YUV4MPEG2 C%s W%u H%u F%u:%u I%c\n", + "420jpeg", width, height, timebase_den, timebase_num, 'p'); + out_put(out, (unsigned char *)buffer, strlen(buffer), do_md5); + } + /* Try to determine the codec from the fourcc. */ for (i = 0; i < sizeof(ifaces) / sizeof(ifaces[0]); i++) if ((fourcc & ifaces[i].fourcc_mask) == ifaces[i].fourcc) @@ -465,6 +505,11 @@ int main(int argc, const char **argv_) break; } } + else if(use_y4m) + { + fprintf(stderr, "YUV4MPEG2 output only supported from IVF input.\n"); + return EXIT_FAILURE; + } if (vpx_codec_dec_init(&decoder, iface ? iface : ifaces[0].iface, &cfg, postproc ? VPX_CODEC_USE_POSTPROC : 0)) @@ -533,29 +578,31 @@ int main(int argc, const char **argv_) prefix, img->d_w, img->d_h, frame_in, sfx); out = out_open(out_fn, do_md5); } + else if(use_y4m) + out_put(out, (unsigned char *)"FRAME\n", 6, do_md5); - buf = img->planes[PLANE_Y]; + buf = img->planes[VPX_PLANE_Y]; for (y = 0; y < img->d_h; y++) { out_put(out, buf, img->d_w, do_md5); - buf += img->stride[PLANE_Y]; + buf += img->stride[VPX_PLANE_Y]; } - buf = img->planes[flipuv?PLANE_V:PLANE_U]; + buf = img->planes[flipuv?VPX_PLANE_V:VPX_PLANE_U]; for (y = 0; y < (1 + img->d_h) / 2; y++) { out_put(out, buf, (1 + img->d_w) / 2, do_md5); - buf += img->stride[PLANE_U]; + buf += img->stride[VPX_PLANE_U]; } - buf = img->planes[flipuv?PLANE_U:PLANE_V]; + buf = img->planes[flipuv?VPX_PLANE_U:VPX_PLANE_V]; for (y = 0; y < (1 + img->d_h) / 2; y++) { out_put(out, buf, (1 + img->d_w) / 2, do_md5); - buf += img->stride[PLANE_V]; + buf += img->stride[VPX_PLANE_V]; } if (!fn2) @@ -1,23 +1,28 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ /* This is a simple program that encodes YV12 files and generates ivf * files using the new interface. */ -#define USE_POSIX_MMAP HAVE_SYS_MMAN_H +#if defined(_WIN32) +#define USE_POSIX_MMAP 0 +#else +#define USE_POSIX_MMAP 1 +#endif #include <stdio.h> #include <stdlib.h> #include <stdarg.h> #include <string.h> -#include "vpx_encoder.h" +#include "vpx/vpx_encoder.h" #if USE_POSIX_MMAP #include <sys/types.h> #include <sys/stat.h> @@ -25,11 +30,10 @@ #include <fcntl.h> #include <unistd.h> #endif -#if CONFIG_VP8_ENCODER -#include "vp8cx.h" -#endif +#include "vpx/vp8cx.h" #include "vpx_ports/mem_ops.h" #include "vpx_ports/vpx_timer.h" +#include "y4minput.h" static const char *exec_name; @@ -51,8 +55,8 @@ void die(const char *fmt, ...) { va_list ap; va_start(ap, fmt); - vprintf(fmt, ap); - printf("\n"); + vfprintf(stderr, fmt, ap); + fprintf(stderr, "\n"); usage_exit(); } @@ -62,10 +66,10 @@ static void ctx_exit_on_error(vpx_codec_ctx_t *ctx, const char *s) { const char *detail = vpx_codec_error_detail(ctx); - printf("%s: %s\n", s, vpx_codec_error(ctx)); + fprintf(stderr, "%s: %s\n", s, vpx_codec_error(ctx)); if (detail) - printf(" %s\n", detail); + fprintf(stderr, " %s\n", detail); exit(EXIT_FAILURE); } @@ -215,49 +219,79 @@ vpx_fixed_buf_t stats_get(stats_io_t *stats) return stats->buf; } +enum video_file_type +{ + FILE_TYPE_RAW, + FILE_TYPE_IVF, + FILE_TYPE_Y4M +}; + +struct detect_buffer { + char buf[4]; + int valid; +}; + + #define IVF_FRAME_HDR_SZ (4+8) /* 4 byte size + 8 byte timestamp */ -static int read_frame(FILE *f, vpx_image_t *img, unsigned int is_ivf) +static int read_frame(FILE *f, vpx_image_t *img, unsigned int file_type, + y4m_input *y4m, struct detect_buffer *detect) { int plane = 0; - if (is_ivf) + if (file_type == FILE_TYPE_Y4M) { - char junk[IVF_FRAME_HDR_SZ]; - - /* Skip the frame header. We know how big the frame should be. See - * write_ivf_frame_header() for documentation on the frame header - * layout. - */ - fread(junk, 1, IVF_FRAME_HDR_SZ, f); + if (y4m_input_fetch_frame(y4m, f, img) < 0) + return 0; } - - for (plane = 0; plane < 3; plane++) + else { - unsigned char *ptr; - int w = (plane ? (1 + img->d_w) / 2 : img->d_w); - int h = (plane ? (1 + img->d_h) / 2 : img->d_h); - int r; - - /* Determine the correct plane based on the image format. The for-loop - * always counts in Y,U,V order, but this may not match the order of - * the data on disk. - */ - switch (plane) + if (file_type == FILE_TYPE_IVF) { - case 1: - ptr = img->planes[img->fmt==IMG_FMT_YV12? PLANE_V : PLANE_U]; - break; - case 2: - ptr = img->planes[img->fmt==IMG_FMT_YV12?PLANE_U : PLANE_V]; - break; - default: - ptr = img->planes[plane]; + char junk[IVF_FRAME_HDR_SZ]; + + /* Skip the frame header. We know how big the frame should be. See + * write_ivf_frame_header() for documentation on the frame header + * layout. + */ + fread(junk, 1, IVF_FRAME_HDR_SZ, f); } - for (r = 0; r < h; r++) + for (plane = 0; plane < 3; plane++) { - fread(ptr, 1, w, f); - ptr += img->stride[plane]; + unsigned char *ptr; + int w = (plane ? (1 + img->d_w) / 2 : img->d_w); + int h = (plane ? (1 + img->d_h) / 2 : img->d_h); + int r; + + /* Determine the correct plane based on the image format. The for-loop + * always counts in Y,U,V order, but this may not match the order of + * the data on disk. + */ + switch (plane) + { + case 1: + ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12? VPX_PLANE_V : VPX_PLANE_U]; + break; + case 2: + ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12?VPX_PLANE_U : VPX_PLANE_V]; + break; + default: + ptr = img->planes[plane]; + } + + for (r = 0; r < h; r++) + { + if (detect->valid) + { + memcpy(ptr, detect->buf, 4); + fread(ptr+4, 1, w-4, f); + detect->valid = 0; + } + else + fread(ptr, 1, w, f); + + ptr += img->stride[plane]; + } } } @@ -265,22 +299,36 @@ static int read_frame(FILE *f, vpx_image_t *img, unsigned int is_ivf) } +unsigned int file_is_y4m(FILE *infile, + y4m_input *y4m, + char detect[4]) +{ + if(memcmp(detect, "YUV4", 4) == 0) + { + return 1; + } + return 0; +} + #define IVF_FILE_HDR_SZ (32) unsigned int file_is_ivf(FILE *infile, unsigned int *fourcc, unsigned int *width, - unsigned int *height) + unsigned int *height, + char detect[4]) { char raw_hdr[IVF_FILE_HDR_SZ]; int is_ivf = 0; + if(memcmp(detect, "DKIF", 4) != 0) + return 0; + /* See write_ivf_file_header() for more documentation on the file header * layout. */ - if (fread(raw_hdr, 1, IVF_FILE_HDR_SZ, infile) == IVF_FILE_HDR_SZ) + if (fread(raw_hdr + 4, 1, IVF_FILE_HDR_SZ - 4, infile) + == IVF_FILE_HDR_SZ - 4) { - if (raw_hdr[0] == 'D' && raw_hdr[1] == 'K' - && raw_hdr[2] == 'I' && raw_hdr[3] == 'F') { is_ivf = 1; @@ -297,8 +345,6 @@ unsigned int file_is_ivf(FILE *infile, *width = mem_get_le16(raw_hdr + 12); *height = mem_get_le16(raw_hdr + 14); } - else - rewind(infile); return is_ivf; } @@ -459,9 +505,11 @@ static const arg_def_t kf_min_dist = ARG_DEF(NULL, "kf-min-dist", 1, "Minimum keyframe interval (frames)"); static const arg_def_t kf_max_dist = ARG_DEF(NULL, "kf-max-dist", 1, "Maximum keyframe interval (frames)"); +static const arg_def_t kf_disabled = ARG_DEF(NULL, "disable-kf", 0, + "Disable keyframe placement"); static const arg_def_t *kf_args[] = { - &kf_min_dist, &kf_max_dist, NULL + &kf_min_dist, &kf_max_dist, &kf_disabled, NULL }; @@ -512,28 +560,28 @@ static void usage_exit() { int i; - printf("Usage: %s <options> src_filename dst_filename\n", exec_name); + fprintf(stderr, "Usage: %s <options> src_filename dst_filename\n", exec_name); - printf("\n_options:\n"); + fprintf(stderr, "\n_options:\n"); arg_show_usage(stdout, main_args); - printf("\n_encoder Global Options:\n"); + fprintf(stderr, "\n_encoder Global Options:\n"); arg_show_usage(stdout, global_args); - printf("\n_rate Control Options:\n"); + fprintf(stderr, "\n_rate Control Options:\n"); arg_show_usage(stdout, rc_args); - printf("\n_twopass Rate Control Options:\n"); + fprintf(stderr, "\n_twopass Rate Control Options:\n"); arg_show_usage(stdout, rc_twopass_args); - printf("\n_keyframe Placement Options:\n"); + fprintf(stderr, "\n_keyframe Placement Options:\n"); arg_show_usage(stdout, kf_args); #if CONFIG_VP8_ENCODER - printf("\n_vp8 Specific Options:\n"); + fprintf(stderr, "\n_vp8 Specific Options:\n"); arg_show_usage(stdout, vp8_args); #endif - printf("\n" + fprintf(stderr, "\n" "Included encoders:\n" "\n"); for (i = 0; i < sizeof(codecs) / sizeof(codecs[0]); i++) - printf(" %-6s - %s\n", + fprintf(stderr, " %-6s - %s\n", codecs[i].name, vpx_codec_iface_name(codecs[i].iface)); @@ -566,8 +614,10 @@ int main(int argc, const char **argv_) static const int *ctrl_args_map = NULL; int verbose = 0, show_psnr = 0; int arg_use_i420 = 1; + int arg_have_timebase = 0; unsigned long cx_time = 0; - unsigned int is_ivf, fourcc; + unsigned int file_type, fourcc; + y4m_input y4m; exec_name = argv_[0]; @@ -651,7 +701,7 @@ int main(int argc, const char **argv_) /* DWIM: Assume the user meant passes=2 if pass=2 is specified */ if (one_pass_only > arg_passes) { - printf("Warning: Assuming --pass=%d implies --passes=%d\n", + fprintf(stderr, "Warning: Assuming --pass=%d implies --passes=%d\n", one_pass_only, one_pass_only); arg_passes = one_pass_only; } @@ -665,7 +715,8 @@ int main(int argc, const char **argv_) if (res) { - printf("Failed to get config: %s\n", vpx_codec_err_to_string(res)); + fprintf(stderr, "Failed to get config: %s\n", + vpx_codec_err_to_string(res)); return EXIT_FAILURE; } @@ -684,7 +735,10 @@ int main(int argc, const char **argv_) else if (arg_match(&arg, &height, argi)) cfg.g_h = arg_parse_uint(&arg); else if (arg_match(&arg, &timebase, argi)) + { cfg.g_timebase = arg_parse_rational(&arg); + arg_have_timebase = 1; + } else if (arg_match(&arg, &error_resilient, argi)) cfg.g_error_resilient = arg_parse_uint(&arg); else if (arg_match(&arg, &lag_in_frames, argi)) @@ -722,29 +776,34 @@ int main(int argc, const char **argv_) cfg.rc_2pass_vbr_bias_pct = arg_parse_uint(&arg); if (arg_passes < 2) - printf("Warning: option %s ignored in one-pass mode.\n", - arg.name); + fprintf(stderr, + "Warning: option %s ignored in one-pass mode.\n", + arg.name); } else if (arg_match(&arg, &minsection_pct, argi)) { cfg.rc_2pass_vbr_minsection_pct = arg_parse_uint(&arg); if (arg_passes < 2) - printf("Warning: option %s ignored in one-pass mode.\n", - arg.name); + fprintf(stderr, + "Warning: option %s ignored in one-pass mode.\n", + arg.name); } else if (arg_match(&arg, &maxsection_pct, argi)) { cfg.rc_2pass_vbr_maxsection_pct = arg_parse_uint(&arg); if (arg_passes < 2) - printf("Warning: option %s ignored in one-pass mode.\n", - arg.name); + fprintf(stderr, + "Warning: option %s ignored in one-pass mode.\n", + arg.name); } else if (arg_match(&arg, &kf_min_dist, argi)) cfg.kf_min_dist = arg_parse_uint(&arg); else if (arg_match(&arg, &kf_max_dist, argi)) cfg.kf_max_dist = arg_parse_uint(&arg); + else if (arg_match(&arg, &kf_disabled, argi)) + cfg.kf_mode = VPX_KF_DISABLED; else argj++; } @@ -787,7 +846,7 @@ int main(int argc, const char **argv_) /* Check for unrecognized options */ for (argi = argv; *argi; argi++) - if (argi[0][0] == '-') + if (argi[0][0] == '-' && argi[0][1]) die("Error: Unrecognized option %s\n", *argi); /* Handle non-option arguments */ @@ -797,106 +856,136 @@ int main(int argc, const char **argv_) if (!in_fn || !out_fn) usage_exit(); - /* Parse certain options from the input file, if possible */ - infile = fopen(in_fn, "rb"); - - if (!infile) - { - printf("Failed to open input file"); - return EXIT_FAILURE; - } - - is_ivf = file_is_ivf(infile, &fourcc, &cfg.g_w, &cfg.g_h); - - if (is_ivf) - { - switch (fourcc) - { - case 0x32315659: - arg_use_i420 = 0; - break; - case 0x30323449: - arg_use_i420 = 1; - break; - default: - printf("Unsupported fourcc (%08x) in IVF\n", fourcc); - return EXIT_FAILURE; - } - } - - fclose(infile); - - -#define SHOW(field) printf(" %-28s = %d\n", #field, cfg.field) - - if (verbose) - { - printf("Codec: %s\n", vpx_codec_iface_name(codec->iface)); - printf("Source file: %s Format: %s\n", in_fn, arg_use_i420 ? "I420" : "YV12"); - printf("Destination file: %s\n", out_fn); - printf("Encoder parameters:\n"); - - SHOW(g_usage); - SHOW(g_threads); - SHOW(g_profile); - SHOW(g_w); - SHOW(g_h); - SHOW(g_timebase.num); - SHOW(g_timebase.den); - SHOW(g_error_resilient); - SHOW(g_pass); - SHOW(g_lag_in_frames); - SHOW(rc_dropframe_thresh); - SHOW(rc_resize_allowed); - SHOW(rc_resize_up_thresh); - SHOW(rc_resize_down_thresh); - SHOW(rc_end_usage); - SHOW(rc_target_bitrate); - SHOW(rc_min_quantizer); - SHOW(rc_max_quantizer); - SHOW(rc_undershoot_pct); - SHOW(rc_overshoot_pct); - SHOW(rc_buf_sz); - SHOW(rc_buf_initial_sz); - SHOW(rc_buf_optimal_sz); - SHOW(rc_2pass_vbr_bias_pct); - SHOW(rc_2pass_vbr_minsection_pct); - SHOW(rc_2pass_vbr_maxsection_pct); - SHOW(kf_mode); - SHOW(kf_min_dist); - SHOW(kf_max_dist); - } - - vpx_img_alloc(&raw, arg_use_i420 ? IMG_FMT_I420 : IMG_FMT_YV12, - cfg.g_w, cfg.g_h, 1); - - // This was added so that ivfenc will create monotically increasing - // timestamps. Since we create new timestamps for alt-reference frames - // we need to make room in the series of timestamps. Since there can - // only be 1 alt-ref frame ( current bitstream) multiplying by 2 - // gives us enough room. - cfg.g_timebase.den *= 2; - memset(&stats, 0, sizeof(stats)); for (pass = one_pass_only ? one_pass_only - 1 : 0; pass < arg_passes; pass++) { int frames_in = 0, frames_out = 0; unsigned long nbytes = 0; + struct detect_buffer detect; - infile = fopen(in_fn, "rb"); + /* Parse certain options from the input file, if possible */ + infile = strcmp(in_fn, "-") ? fopen(in_fn, "rb") : stdin; if (!infile) { - printf("Failed to open input file"); + fprintf(stderr, "Failed to open input file\n"); return EXIT_FAILURE; } - outfile = fopen(out_fn, "wb"); + fread(detect.buf, 1, 4, infile); + detect.valid = 0; + + if (file_is_y4m(infile, &y4m, detect.buf)) + { + if (y4m_input_open(&y4m, infile, detect.buf, 4) >= 0) + { + file_type = FILE_TYPE_Y4M; + cfg.g_w = y4m.pic_w; + cfg.g_h = y4m.pic_h; + /* Use the frame rate from the file only if none was specified + * on the command-line. + */ + if (!arg_have_timebase) + { + cfg.g_timebase.num = y4m.fps_d; + cfg.g_timebase.den = y4m.fps_n; + /* And don't reset it in the second pass.*/ + arg_have_timebase = 1; + } + arg_use_i420 = 0; + } + else + { + fprintf(stderr, "Unsupported Y4M stream.\n"); + return EXIT_FAILURE; + } + } + else if (file_is_ivf(infile, &fourcc, &cfg.g_w, &cfg.g_h, detect.buf)) + { + file_type = FILE_TYPE_IVF; + switch (fourcc) + { + case 0x32315659: + arg_use_i420 = 0; + break; + case 0x30323449: + arg_use_i420 = 1; + break; + default: + fprintf(stderr, "Unsupported fourcc (%08x) in IVF\n", fourcc); + return EXIT_FAILURE; + } + } + else + { + file_type = FILE_TYPE_RAW; + detect.valid = 1; + } +#define SHOW(field) fprintf(stderr, " %-28s = %d\n", #field, cfg.field) + + if (verbose && pass == 0) + { + fprintf(stderr, "Codec: %s\n", vpx_codec_iface_name(codec->iface)); + fprintf(stderr, "Source file: %s Format: %s\n", in_fn, + arg_use_i420 ? "I420" : "YV12"); + fprintf(stderr, "Destination file: %s\n", out_fn); + fprintf(stderr, "Encoder parameters:\n"); + + SHOW(g_usage); + SHOW(g_threads); + SHOW(g_profile); + SHOW(g_w); + SHOW(g_h); + SHOW(g_timebase.num); + SHOW(g_timebase.den); + SHOW(g_error_resilient); + SHOW(g_pass); + SHOW(g_lag_in_frames); + SHOW(rc_dropframe_thresh); + SHOW(rc_resize_allowed); + SHOW(rc_resize_up_thresh); + SHOW(rc_resize_down_thresh); + SHOW(rc_end_usage); + SHOW(rc_target_bitrate); + SHOW(rc_min_quantizer); + SHOW(rc_max_quantizer); + SHOW(rc_undershoot_pct); + SHOW(rc_overshoot_pct); + SHOW(rc_buf_sz); + SHOW(rc_buf_initial_sz); + SHOW(rc_buf_optimal_sz); + SHOW(rc_2pass_vbr_bias_pct); + SHOW(rc_2pass_vbr_minsection_pct); + SHOW(rc_2pass_vbr_maxsection_pct); + SHOW(kf_mode); + SHOW(kf_min_dist); + SHOW(kf_max_dist); + } + + if(pass == (one_pass_only ? one_pass_only - 1 : 0)) { + if (file_type == FILE_TYPE_Y4M) + /*The Y4M reader does its own allocation. + Just initialize this here to avoid problems if we never read any + frames.*/ + memset(&raw, 0, sizeof(raw)); + else + vpx_img_alloc(&raw, arg_use_i420 ? VPX_IMG_FMT_I420 : VPX_IMG_FMT_YV12, + cfg.g_w, cfg.g_h, 1); + + // This was added so that ivfenc will create monotically increasing + // timestamps. Since we create new timestamps for alt-reference frames + // we need to make room in the series of timestamps. Since there can + // only be 1 alt-ref frame ( current bitstream) multiplying by 2 + // gives us enough room. + cfg.g_timebase.den *= 2; + } + + outfile = strcmp(out_fn, "-") ? fopen(out_fn, "wb") : stdout; if (!outfile) { - printf("Failed to open output file"); + fprintf(stderr, "Failed to open output file\n"); return EXIT_FAILURE; } @@ -904,7 +993,7 @@ int main(int argc, const char **argv_) { if (!stats_open_file(&stats, stats_fn, pass)) { - printf("Failed to open statistics store\n"); + fprintf(stderr, "Failed to open statistics store\n"); return EXIT_FAILURE; } } @@ -912,7 +1001,7 @@ int main(int argc, const char **argv_) { if (!stats_open_mem(&stats, pass)) { - printf("Failed to open statistics store\n"); + fprintf(stderr, "Failed to open statistics store\n"); return EXIT_FAILURE; } } @@ -933,9 +1022,6 @@ int main(int argc, const char **argv_) /* Construct Encoder Context */ - if (cfg.kf_min_dist == cfg.kf_max_dist) - cfg.kf_mode = VPX_KF_FIXED; - vpx_codec_enc_init(&encoder, codec->iface, &cfg, show_psnr ? VPX_CODEC_USE_PSNR : 0); ctx_exit_on_error(&encoder, "Failed to initialize encoder"); @@ -947,8 +1033,8 @@ int main(int argc, const char **argv_) for (i = 0; i < arg_ctrl_cnt; i++) { if (vpx_codec_control_(&encoder, arg_ctrls[i][0], arg_ctrls[i][1])) - printf("Error: Tried to set control %d = %d\n", - arg_ctrls[i][0], arg_ctrls[i][1]); + fprintf(stderr, "Error: Tried to set control %d = %d\n", + arg_ctrls[i][0], arg_ctrls[i][1]); ctx_exit_on_error(&encoder, "Failed to control codec"); } @@ -964,13 +1050,15 @@ int main(int argc, const char **argv_) if (!arg_limit || frames_in < arg_limit) { - frame_avail = read_frame(infile, &raw, is_ivf); + frame_avail = read_frame(infile, &raw, file_type, &y4m, + &detect); if (frame_avail) frames_in++; - printf("\rPass %d/%d frame %4d/%-4d %7ldB \033[K", pass + 1, - arg_passes, frames_in, frames_out, nbytes); + fprintf(stderr, + "\rPass %d/%d frame %4d/%-4d %7ldB \033[K", pass + 1, + arg_passes, frames_in, frames_out, nbytes); } else frame_avail = 0; @@ -989,24 +1077,25 @@ int main(int argc, const char **argv_) while ((pkt = vpx_codec_get_cx_data(&encoder, &iter))) { got_data = 1; - nbytes += pkt->data.raw.sz; switch (pkt->kind) { case VPX_CODEC_CX_FRAME_PKT: frames_out++; - printf(" %6luF", - (unsigned long)pkt->data.frame.sz); + fprintf(stderr, " %6luF", + (unsigned long)pkt->data.frame.sz); write_ivf_frame_header(outfile, pkt); fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz, outfile); + nbytes += pkt->data.raw.sz; break; case VPX_CODEC_STATS_PKT: frames_out++; - printf(" %6luS", + fprintf(stderr, " %6luS", (unsigned long)pkt->data.twopass_stats.sz); stats_write(&stats, pkt->data.twopass_stats.buf, pkt->data.twopass_stats.sz); + nbytes += pkt->data.raw.sz; break; case VPX_CODEC_PSNR_PKT: @@ -1015,7 +1104,7 @@ int main(int argc, const char **argv_) int i; for (i = 0; i < 4; i++) - printf("%.3lf ", pkt->data.psnr.psnr[i]); + fprintf(stderr, "%.3lf ", pkt->data.psnr.psnr[i]); } break; @@ -1030,10 +1119,11 @@ int main(int argc, const char **argv_) /* this bitrate calc is simplified and relies on the fact that this * application uses 1/timebase for framerate. */ - printf("\rPass %d/%d frame %4d/%-4d %7ldB %7ldb/f %7"PRId64"b/s" + fprintf(stderr, + "\rPass %d/%d frame %4d/%-4d %7ldB %7ldb/f %7"PRId64"b/s" " %7lu %s (%.2f fps)\033[K", pass + 1, arg_passes, frames_in, frames_out, nbytes, nbytes * 8 / frames_in, - nbytes * 8 *(int64_t)cfg.g_timebase.den / cfg.g_timebase.num / frames_in, + nbytes * 8 *(int64_t)cfg.g_timebase.den/2/ cfg.g_timebase.num / frames_in, cx_time > 9999999 ? cx_time / 1000 : cx_time, cx_time > 9999999 ? "ms" : "us", (float)frames_in * 1000000.0 / (float)cx_time); @@ -1047,7 +1137,7 @@ int main(int argc, const char **argv_) fclose(outfile); stats_close(&stats); - printf("\n"); + fprintf(stderr, "\n"); if (one_pass_only) break; diff --git a/libs.doxy_template b/libs.doxy_template index eb37dfc..02e2902 100644 --- a/libs.doxy_template +++ b/libs.doxy_template @@ -1,10 +1,11 @@ ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## @@ -1,17 +1,20 @@ ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## ASM:=$(if $(filter yes,$(CONFIG_GCC)),.asm.s,.asm) -include $(SRC_PATH_BARE)/vpx_codec/vpx_codec.mk -CODEC_SRCS-yes += $(addprefix vpx_codec/,$(call enabled,API_SRCS)) +CODEC_SRCS-yes += libs.mk + +include $(SRC_PATH_BARE)/vpx/vpx_codec.mk +CODEC_SRCS-yes += $(addprefix vpx/,$(call enabled,API_SRCS)) include $(SRC_PATH_BARE)/vpx_mem/vpx_mem.mk CODEC_SRCS-yes += $(addprefix vpx_mem/,$(call enabled,MEM_SRCS)) @@ -19,18 +22,16 @@ CODEC_SRCS-yes += $(addprefix vpx_mem/,$(call enabled,MEM_SRCS)) include $(SRC_PATH_BARE)/vpx_scale/vpx_scale.mk CODEC_SRCS-yes += $(addprefix vpx_scale/,$(call enabled,SCALE_SRCS)) -# Add vpx_codec/ to the include path to allow vp_n[cd]x.h to reference -# vpx_codec_impl_*.h without extra ifdeffery -CFLAGS += -I$(SRC_PATH_BARE)/vpx_codec ifeq ($(CONFIG_VP8_ENCODER),yes) VP8_PREFIX=vp8/ include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8cx.mk CODEC_SRCS-yes += $(addprefix $(VP8_PREFIX),$(call enabled,VP8_CX_SRCS)) CODEC_EXPORTS-yes += $(addprefix $(VP8_PREFIX),$(VP8_CX_EXPORTS)) - CODEC_SRCS-yes += $(VP8_PREFIX)vp8cx.mk - INSTALL_MAPS += include/% $(SRC_PATH_BARE)/$(VP8_PREFIX)/% - CODEC_DOC_SRCS += vp8/vp8.h vp8/vp8cx.h + CODEC_SRCS-yes += $(VP8_PREFIX)vp8cx.mk vpx/vp8.h vpx/vp8cx.h vpx/vp8e.h + INSTALL-LIBS-yes += include/vpx/vp8.h include/vpx/vp8e.h include/vpx/vp8cx.h + INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/$(VP8_PREFIX)/% + CODEC_DOC_SRCS += vpx/vp8.h vpx/vp8cx.h CODEC_DOC_SECTIONS += vp8 vp8_encoder endif @@ -39,9 +40,10 @@ ifeq ($(CONFIG_VP8_DECODER),yes) include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8dx.mk CODEC_SRCS-yes += $(addprefix $(VP8_PREFIX),$(call enabled,VP8_DX_SRCS)) CODEC_EXPORTS-yes += $(addprefix $(VP8_PREFIX),$(VP8_DX_EXPORTS)) - CODEC_SRCS-yes += $(VP8_PREFIX)vp8dx.mk - INSTALL_MAPS += include/% $(SRC_PATH_BARE)/$(VP8_PREFIX)/% - CODEC_DOC_SRCS += vp8/vp8.h vp8/vp8dx.h + CODEC_SRCS-yes += $(VP8_PREFIX)vp8dx.mk vpx/vp8.h vpx/vp8dx.h + INSTALL-LIBS-yes += include/vpx/vp8.h include/vpx/vp8dx.h + INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/$(VP8_PREFIX)/% + CODEC_DOC_SRCS += vpx/vp8.h vpx/vp8dx.h CODEC_DOC_SECTIONS += vp8 vp8_decoder endif @@ -59,18 +61,17 @@ CODEC_LIB=$(if $(CONFIG_STATIC_MSVCRT),vpxmt,vpxmd) # This variable uses deferred expansion intentionally, since the results of # $(wildcard) may change during the course of the Make. VS_PLATFORMS = $(foreach d,$(wildcard */Release/$(CODEC_LIB).lib),$(word 1,$(subst /, ,$(d)))) -CODEC_SRCS-yes += $(SRC_PATH_BARE)/libs.mk # to show up in the msvs workspace endif # The following pairs define a mapping of locations in the distribution # tree to locations in the source/build trees. -INSTALL_MAPS += include/% $(SRC_PATH_BARE)/vpx_codec/% -INSTALL_MAPS += include/% $(SRC_PATH_BARE)/vpx_ports/% -INSTALL_MAPS += lib/% % +INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/vpx/% +INSTALL_MAPS += include/vpx/% $(SRC_PATH_BARE)/vpx_ports/% +INSTALL_MAPS += $(LIBSUBDIR)/% % INSTALL_MAPS += src/% $(SRC_PATH_BARE)/% ifeq ($(CONFIG_MSVS),yes) -INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),lib/$(p)/% $(p)/Release/%) -INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),lib/$(p)/% $(p)/Debug/%) +INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/% $(p)/Release/%) +INSTALL_MAPS += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/% $(p)/Debug/%) endif # If this is a universal (fat) binary, then all the subarchitectures have @@ -81,7 +82,7 @@ endif $(eval $(if $(filter universal%,$(TOOLCHAIN)),LIPO_LIBVPX,BUILD_LIBVPX):=yes) CODEC_SRCS-$(BUILD_LIBVPX) += build/make/version.sh -CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/vpx_integer.h +CODEC_SRCS-$(BUILD_LIBVPX) += vpx/vpx_integer.h CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/vpx_timer.h CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/mem.h CODEC_SRCS-$(BUILD_LIBVPX) += $(BUILD_PFX)vpx_config.c @@ -92,26 +93,28 @@ CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/x86.h CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/x86_abi_support.asm endif CODEC_SRCS-$(ARCH_ARM) += $(BUILD_PFX)vpx_config.asm -CODEC_EXPORTS-$(BUILD_LIBVPX) += vpx_codec/exports - -INSTALL-LIBS-yes += include/vpx_codec.h -INSTALL-LIBS-yes += include/vpx_image.h -INSTALL-LIBS-yes += include/vpx_integer.h -INSTALL-LIBS-yes += include/vpx_codec_impl_top.h -INSTALL-LIBS-yes += include/vpx_codec_impl_bottom.h -INSTALL-LIBS-$(CONFIG_DECODERS) += include/vpx_decoder.h -INSTALL-LIBS-$(CONFIG_DECODERS) += include/vpx_decoder_compat.h -INSTALL-LIBS-$(CONFIG_ENCODERS) += include/vpx_encoder.h +CODEC_EXPORTS-$(BUILD_LIBVPX) += vpx/exports_com +CODEC_EXPORTS-$(CONFIG_ENCODERS) += vpx/exports_enc +CODEC_EXPORTS-$(CONFIG_DECODERS) += vpx/exports_dec + +INSTALL-LIBS-yes += include/vpx/vpx_codec.h +INSTALL-LIBS-yes += include/vpx/vpx_image.h +INSTALL-LIBS-yes += include/vpx/vpx_integer.h +INSTALL-LIBS-yes += include/vpx/vpx_codec_impl_top.h +INSTALL-LIBS-yes += include/vpx/vpx_codec_impl_bottom.h +INSTALL-LIBS-$(CONFIG_DECODERS) += include/vpx/vpx_decoder.h +INSTALL-LIBS-$(CONFIG_DECODERS) += include/vpx/vpx_decoder_compat.h +INSTALL-LIBS-$(CONFIG_ENCODERS) += include/vpx/vpx_encoder.h ifeq ($(CONFIG_EXTERNAL_BUILD),yes) ifeq ($(CONFIG_MSVS),yes) -INSTALL-LIBS-yes += $(foreach p,$(VS_PLATFORMS),lib/$(p)/$(CODEC_LIB).lib) -INSTALL-LIBS-$(CONFIG_DEBUG_LIBS) += $(foreach p,$(VS_PLATFORMS),lib/$(p)/$(CODEC_LIB)d.lib) -INSTALL-LIBS-$(CONFIG_SHARED) += $(foreach p,$(VS_PLATFORMS),lib/$(p)/vpx.dll) -INSTALL-LIBS-$(CONFIG_SHARED) += $(foreach p,$(VS_PLATFORMS),lib/$(p)/vpx.exp) +INSTALL-LIBS-yes += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/$(CODEC_LIB).lib) +INSTALL-LIBS-$(CONFIG_DEBUG_LIBS) += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/$(CODEC_LIB)d.lib) +INSTALL-LIBS-$(CONFIG_SHARED) += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/vpx.dll) +INSTALL-LIBS-$(CONFIG_SHARED) += $(foreach p,$(VS_PLATFORMS),$(LIBSUBDIR)/$(p)/vpx.exp) endif else -INSTALL-LIBS-yes += lib/libvpx.a -INSTALL-LIBS-$(CONFIG_DEBUG_LIBS) += lib/libvpx_g.a +INSTALL-LIBS-yes += $(LIBSUBDIR)/libvpx.a +INSTALL-LIBS-$(CONFIG_DEBUG_LIBS) += $(LIBSUBDIR)/libvpx_g.a endif CODEC_SRCS=$(call enabled,CODEC_SRCS) @@ -130,7 +133,6 @@ ARM_ARCH=v6 endif obj_int_extract.vcproj: $(SRC_PATH_BARE)/build/make/obj_int_extract.c @cp $(SRC_PATH_BARE)/build/arm-wince-vs8/obj_int_extract.bat . - @cp $(SRC_PATH_BARE)/build/arm-wince-vs8/armasm$(ARM_ARCH).rules . @echo " [CREATE] $@" $(SRC_PATH_BARE)/build/make/gen_msvs_proj.sh\ --exe\ @@ -143,7 +145,6 @@ obj_int_extract.vcproj: $(SRC_PATH_BARE)/build/make/obj_int_extract.c PROJECTS-$(BUILD_LIBVPX) += obj_int_extract.vcproj PROJECTS-$(BUILD_LIBVPX) += obj_int_extract.bat -PROJECTS-$(BUILD_LIBVPX) += armasm$(ARM_ARCH).rules endif vpx.def: $(call enabled,CODEC_EXPORTS) @@ -175,6 +176,31 @@ LIBVPX_OBJS=$(call objs,$(CODEC_SRCS)) OBJS-$(BUILD_LIBVPX) += $(LIBVPX_OBJS) LIBS-$(BUILD_LIBVPX) += $(BUILD_PFX)libvpx.a $(BUILD_PFX)libvpx_g.a $(BUILD_PFX)libvpx_g.a: $(LIBVPX_OBJS) + +BUILD_LIBVPX_SO := $(if $(BUILD_LIBVPX),$(CONFIG_SHARED)) +LIBVPX_SO := libvpx.so.$(VERSION_MAJOR).$(VERSION_MINOR).$(VERSION_PATCH) +LIBS-$(BUILD_LIBVPX_SO) += $(BUILD_PFX)$(LIBVPX_SO) +$(BUILD_PFX)$(LIBVPX_SO): $(LIBVPX_OBJS) libvpx.ver +$(BUILD_PFX)$(LIBVPX_SO): extralibs += -lm -pthread +$(BUILD_PFX)$(LIBVPX_SO): SONAME = libvpx.so.$(VERSION_MAJOR) +$(BUILD_PFX)$(LIBVPX_SO): SO_VERSION_SCRIPT = libvpx.ver +LIBVPX_SO_SYMLINKS := $(addprefix $(LIBSUBDIR)/, \ + libvpx.so libvpx.so.$(VERSION_MAJOR) \ + libvpx.so.$(VERSION_MAJOR).$(VERSION_MINOR)) + +libvpx.ver: $(call enabled,CODEC_EXPORTS) + @echo " [CREATE] $@" + $(qexec)echo "{ global:" > $@ + $(qexec)for f in $?; do awk '{print $$2";"}' < $$f >>$@; done + $(qexec)echo "local: *; };" >> $@ +CLEAN-OBJS += libvpx.ver + +$(addprefix $(DIST_DIR)/,$(LIBVPX_SO_SYMLINKS)): + @echo " [LN] $@" + $(qexec)ln -sf $(LIBVPX_SO) $@ + +INSTALL-LIBS-$(CONFIG_SHARED) += $(LIBVPX_SO_SYMLINKS) +INSTALL-LIBS-$(CONFIG_SHARED) += $(LIBSUBDIR)/$(LIBVPX_SO) endif LIBS-$(LIPO_LIBVPX) += libvpx.a @@ -209,10 +235,10 @@ $(filter %.asm.o,$(OBJS-yes)): $(BUILD_PFX)vpx_config.asm $(shell $(SRC_PATH_BARE)/build/make/version.sh "$(SRC_PATH_BARE)" $(BUILD_PFX)vpx_version.h) CLEAN-OBJS += $(BUILD_PFX)vpx_version.h -CODEC_DOC_SRCS += vpx_codec/vpx_codec.h \ - vpx_codec/vpx_decoder.h \ - vpx_codec/vpx_encoder.h \ - vpx_codec/vpx_image.h +CODEC_DOC_SRCS += vpx/vpx_codec.h \ + vpx/vpx_decoder.h \ + vpx/vpx_encoder.h \ + vpx/vpx_image.h CLEAN-OBJS += libs.doxy DOCS-yes += libs.doxy diff --git a/mainpage.dox b/mainpage.dox index 3596ce0..49dff7b 100644 --- a/mainpage.dox +++ b/mainpage.dox @@ -11,7 +11,7 @@ source codec deployed on millions of computers and devices worldwide. This distribution of the WebM VP8 Codec SDK includes the following support: - + \if vp8_encoder - \ref vp8_encoder \endif \if vp8_decoder - \ref vp8_decoder \endif diff --git a/md5_utils.c b/md5_utils.c index 16c6f7e..455d9cd 100644 --- a/md5_utils.c +++ b/md5_utils.c @@ -1,298 +1,253 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * This code implements the MD5 message-digest algorithm. + * The algorithm is due to Ron Rivest. This code was + * written by Colin Plumb in 1993, no copyright is claimed. + * This code is in the public domain; do with it what you wish. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Equivalent code is available from RSA Data Security, Inc. + * This code has been tested against that, and is equivalent, + * except that you don't need to include two pages of legalese + * with every copy. + * + * To compute the message digest of a chunk of bytes, declare an + * MD5Context structure, pass it to MD5Init, call MD5Update as + * needed on buffers full of bytes, and then call MD5Final, which + * will fill a supplied 16-byte array with the digest. + * + * Changed so as no longer to depend on Colin Plumb's `usual.h' header + * definitions + * - Ian Jackson <ian@chiark.greenend.org.uk>. + * Still in the public domain. */ +#include <sys/types.h> /* for stupid systems */ -/* -Copyright (C) 1991-2, RSA Data Security, Inc. Created 1991. All -rights reserved. - -License to copy and use this software is granted provided that it -is identified as the "RSA Data Security, Inc. MD5 Message-Digest -Algorithm" in all material mentioning or referencing this software -or this function. - -License is also granted to make and use derivative works provided -that such works are identified as "derived from the RSA Data -Security, Inc. MD5 Message-Digest Algorithm" in all material -mentioning or referencing the derived work. - -RSA Data Security, Inc. makes no representations concerning either -the merchantability of this software or the suitability of this -software for any particular purpose. It is provided "as is" -without express or implied warranty of any kind. - -These notices must be retained in any copies of any part of this -documentation and/or software. -*/ +#include <string.h> /* for memcpy() */ #include "md5_utils.h" -#include <string.h> -/* Constants for md5_transform routine. - */ -#define S11 7 -#define S12 12 -#define S13 17 -#define S14 22 -#define S21 5 -#define S22 9 -#define S23 14 -#define S24 20 -#define S31 4 -#define S32 11 -#define S33 16 -#define S34 23 -#define S41 6 -#define S42 10 -#define S43 15 -#define S44 21 - -static void md5_transform(uint32_t state[4], const uint8_t block[64]); -static void Encode(uint8_t *output, const uint32_t *input, unsigned int len); -static void Decode(uint32_t *output, const uint8_t *input, unsigned int len); -#define md5_memset memset -#define md5_memcpy memcpy - -static unsigned char PADDING[64] = +void +byteSwap(UWORD32 *buf, unsigned words) { - 0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 -}; + md5byte *p; -/* F, G, H and I are basic MD5 functions. - */ -#define F(x, y, z) (((x) & (y)) | ((~x) & (z))) -#define G(x, y, z) (((x) & (z)) | ((y) & (~z))) -#define H(x, y, z) ((x) ^ (y) ^ (z)) -#define I(x, y, z) ((y) ^ ((x) | (~z))) + /* Only swap bytes for big endian machines */ + int i = 1; -/* ROTATE_LEFT rotates x left n bits. - */ -#define ROTATE_LEFT(x, n) (((x) << (n)) | ((x) >> (32-(n)))) + if (*(char *)&i == 1) + return; -/* FF, GG, HH, and II transformations for rounds 1, 2, 3, and 4. -Rotation is separate from addition to prevent recomputation. - */ -#define FF(a, b, c, d, x, s, ac) { \ - (a) += F ((b), (c), (d)) + (x) + (uint32_t)(ac); \ - (a) = ROTATE_LEFT ((a), (s)); \ - (a) += (b); \ - } -#define GG(a, b, c, d, x, s, ac) { \ - (a) += G ((b), (c), (d)) + (x) + (uint32_t)(ac); \ - (a) = ROTATE_LEFT ((a), (s)); \ - (a) += (b); \ - } -#define HH(a, b, c, d, x, s, ac) { \ - (a) += H ((b), (c), (d)) + (x) + (uint32_t)(ac); \ - (a) = ROTATE_LEFT ((a), (s)); \ - (a) += (b); \ - } -#define II(a, b, c, d, x, s, ac) { \ - (a) += I ((b), (c), (d)) + (x) + (uint32_t)(ac); \ - (a) = ROTATE_LEFT ((a), (s)); \ - (a) += (b); \ + p = (md5byte *)buf; + + do + { + *buf++ = (UWORD32)((unsigned)p[3] << 8 | p[2]) << 16 | + ((unsigned)p[1] << 8 | p[0]); + p += 4; } + while (--words); +} -/* MD5 initialization. Begins an MD5 operation, writing a new context. +/* + * Start MD5 accumulation. Set bit count to 0 and buffer to mysterious + * initialization constants. */ -void md5_init(md5_ctx_t *context) +void +MD5Init(struct MD5Context *ctx) { - context->count[0] = context->count[1] = 0; - /* Load magic initialization constants. - */ - context->state[0] = 0x67452301; - context->state[1] = 0xefcdab89; - context->state[2] = 0x98badcfe; - context->state[3] = 0x10325476; + ctx->buf[0] = 0x67452301; + ctx->buf[1] = 0xefcdab89; + ctx->buf[2] = 0x98badcfe; + ctx->buf[3] = 0x10325476; + + ctx->bytes[0] = 0; + ctx->bytes[1] = 0; } -/* MD5 block update operation. Continues an MD5 message-digest - operation, processing another message block, and updating the - context. +/* + * Update context to reflect the concatenation of another buffer full + * of bytes. */ -void md5_update(md5_ctx_t *context, const uint8_t *input, unsigned int input_len) +void +MD5Update(struct MD5Context *ctx, md5byte const *buf, unsigned len) { - unsigned int i, index, part_len; + UWORD32 t; - /* Compute number of bytes mod 64 */ - index = (unsigned int)((context->count[0] >> 3) & 0x3F); + /* Update byte count */ - /* Update number of bits */ - if ((context->count[0] += ((uint32_t)input_len << 3)) - < ((uint32_t)input_len << 3)) - context->count[1]++; + t = ctx->bytes[0]; - context->count[1] += ((uint32_t)input_len >> 29); + if ((ctx->bytes[0] = t + len) < t) + ctx->bytes[1]++; /* Carry from low to high */ - part_len = 64 - index; + t = 64 - (t & 0x3f); /* Space available in ctx->in (at least 1) */ - /* Transform as many times as possible. */ - if (input_len >= part_len) + if (t > len) { - memcpy(&context->buffer[index], input, part_len); - md5_transform(context->state, context->buffer); + memcpy((md5byte *)ctx->in + 64 - t, buf, len); + return; + } - for (i = part_len; i + 63 < input_len; i += 64) - md5_transform(context->state, &input[i]); + /* First chunk is an odd size */ + memcpy((md5byte *)ctx->in + 64 - t, buf, t); + byteSwap(ctx->in, 16); + MD5Transform(ctx->buf, ctx->in); + buf += t; + len -= t; - index = 0; + /* Process data in 64-byte chunks */ + while (len >= 64) + { + memcpy(ctx->in, buf, 64); + byteSwap(ctx->in, 16); + MD5Transform(ctx->buf, ctx->in); + buf += 64; + len -= 64; } - else - i = 0; - /* Buffer remaining input */ - memcpy(&context->buffer[index], &input[i], input_len - i); + /* Handle any remaining bytes of data. */ + memcpy(ctx->in, buf, len); } -/* MD5 finalization. Ends an MD5 message-digest operation, writing the - the message digest and zeroizing the context. +/* + * Final wrapup - pad to 64-byte boundary with the bit pattern + * 1 0* (64-bit count of bits processed, MSB-first) */ -void md5_finalize(md5_ctx_t *context, uint8_t digest[16]) +void +MD5Final(md5byte digest[16], struct MD5Context *ctx) { - unsigned char bits[8]; - unsigned int index, pad_len; - - /* Save number of bits */ - Encode(bits, context->count, 8); - - /* Pad out to 56 mod 64. - */ - index = (unsigned int)((context->count[0] >> 3) & 0x3f); - pad_len = (index < 56) ? (56 - index) : (120 - index); - md5_update(context, PADDING, pad_len); - - /* Append length (before padding) */ - md5_update(context, bits, 8); - /* Store state in digest */ - Encode(digest, context->state, 16); - - /* Zeroize sensitive information. - */ - memset(context, 0, sizeof(*context)); -} + int count = ctx->bytes[0] & 0x3f; /* Number of bytes in ctx->in */ + md5byte *p = (md5byte *)ctx->in + count; -/* MD5 basic transformation. Transforms state based on block. - */ -static void md5_transform(uint32_t state[4], const uint8_t block[64]) -{ - uint32_t a = state[0], b = state[1], c = state[2], d = state[3], x[16]; - - Decode(x, block, 64); - - /* Round 1 */ - FF(a, b, c, d, x[ 0], S11, 0xd76aa478); /* 1 */ - FF(d, a, b, c, x[ 1], S12, 0xe8c7b756); /* 2 */ - FF(c, d, a, b, x[ 2], S13, 0x242070db); /* 3 */ - FF(b, c, d, a, x[ 3], S14, 0xc1bdceee); /* 4 */ - FF(a, b, c, d, x[ 4], S11, 0xf57c0faf); /* 5 */ - FF(d, a, b, c, x[ 5], S12, 0x4787c62a); /* 6 */ - FF(c, d, a, b, x[ 6], S13, 0xa8304613); /* 7 */ - FF(b, c, d, a, x[ 7], S14, 0xfd469501); /* 8 */ - FF(a, b, c, d, x[ 8], S11, 0x698098d8); /* 9 */ - FF(d, a, b, c, x[ 9], S12, 0x8b44f7af); /* 10 */ - FF(c, d, a, b, x[10], S13, 0xffff5bb1); /* 11 */ - FF(b, c, d, a, x[11], S14, 0x895cd7be); /* 12 */ - FF(a, b, c, d, x[12], S11, 0x6b901122); /* 13 */ - FF(d, a, b, c, x[13], S12, 0xfd987193); /* 14 */ - FF(c, d, a, b, x[14], S13, 0xa679438e); /* 15 */ - FF(b, c, d, a, x[15], S14, 0x49b40821); /* 16 */ - - /* Round 2 */ - GG(a, b, c, d, x[ 1], S21, 0xf61e2562); /* 17 */ - GG(d, a, b, c, x[ 6], S22, 0xc040b340); /* 18 */ - GG(c, d, a, b, x[11], S23, 0x265e5a51); /* 19 */ - GG(b, c, d, a, x[ 0], S24, 0xe9b6c7aa); /* 20 */ - GG(a, b, c, d, x[ 5], S21, 0xd62f105d); /* 21 */ - GG(d, a, b, c, x[10], S22, 0x2441453); /* 22 */ - GG(c, d, a, b, x[15], S23, 0xd8a1e681); /* 23 */ - GG(b, c, d, a, x[ 4], S24, 0xe7d3fbc8); /* 24 */ - GG(a, b, c, d, x[ 9], S21, 0x21e1cde6); /* 25 */ - GG(d, a, b, c, x[14], S22, 0xc33707d6); /* 26 */ - GG(c, d, a, b, x[ 3], S23, 0xf4d50d87); /* 27 */ - GG(b, c, d, a, x[ 8], S24, 0x455a14ed); /* 28 */ - GG(a, b, c, d, x[13], S21, 0xa9e3e905); /* 29 */ - GG(d, a, b, c, x[ 2], S22, 0xfcefa3f8); /* 30 */ - GG(c, d, a, b, x[ 7], S23, 0x676f02d9); /* 31 */ - GG(b, c, d, a, x[12], S24, 0x8d2a4c8a); /* 32 */ - - /* Round 3 */ - HH(a, b, c, d, x[ 5], S31, 0xfffa3942); /* 33 */ - HH(d, a, b, c, x[ 8], S32, 0x8771f681); /* 34 */ - HH(c, d, a, b, x[11], S33, 0x6d9d6122); /* 35 */ - HH(b, c, d, a, x[14], S34, 0xfde5380c); /* 36 */ - HH(a, b, c, d, x[ 1], S31, 0xa4beea44); /* 37 */ - HH(d, a, b, c, x[ 4], S32, 0x4bdecfa9); /* 38 */ - HH(c, d, a, b, x[ 7], S33, 0xf6bb4b60); /* 39 */ - HH(b, c, d, a, x[10], S34, 0xbebfbc70); /* 40 */ - HH(a, b, c, d, x[13], S31, 0x289b7ec6); /* 41 */ - HH(d, a, b, c, x[ 0], S32, 0xeaa127fa); /* 42 */ - HH(c, d, a, b, x[ 3], S33, 0xd4ef3085); /* 43 */ - HH(b, c, d, a, x[ 6], S34, 0x4881d05); /* 44 */ - HH(a, b, c, d, x[ 9], S31, 0xd9d4d039); /* 45 */ - HH(d, a, b, c, x[12], S32, 0xe6db99e5); /* 46 */ - HH(c, d, a, b, x[15], S33, 0x1fa27cf8); /* 47 */ - HH(b, c, d, a, x[ 2], S34, 0xc4ac5665); /* 48 */ - - /* Round 4 */ - II(a, b, c, d, x[ 0], S41, 0xf4292244); /* 49 */ - II(d, a, b, c, x[ 7], S42, 0x432aff97); /* 50 */ - II(c, d, a, b, x[14], S43, 0xab9423a7); /* 51 */ - II(b, c, d, a, x[ 5], S44, 0xfc93a039); /* 52 */ - II(a, b, c, d, x[12], S41, 0x655b59c3); /* 53 */ - II(d, a, b, c, x[ 3], S42, 0x8f0ccc92); /* 54 */ - II(c, d, a, b, x[10], S43, 0xffeff47d); /* 55 */ - II(b, c, d, a, x[ 1], S44, 0x85845dd1); /* 56 */ - II(a, b, c, d, x[ 8], S41, 0x6fa87e4f); /* 57 */ - II(d, a, b, c, x[15], S42, 0xfe2ce6e0); /* 58 */ - II(c, d, a, b, x[ 6], S43, 0xa3014314); /* 59 */ - II(b, c, d, a, x[13], S44, 0x4e0811a1); /* 60 */ - II(a, b, c, d, x[ 4], S41, 0xf7537e82); /* 61 */ - II(d, a, b, c, x[11], S42, 0xbd3af235); /* 62 */ - II(c, d, a, b, x[ 2], S43, 0x2ad7d2bb); /* 63 */ - II(b, c, d, a, x[ 9], S44, 0xeb86d391); /* 64 */ - - state[0] += a; - state[1] += b; - state[2] += c; - state[3] += d; - - /* Zeroize sensitive information. - */ - memset(x, 0, sizeof(x)); -} + /* Set the first char of padding to 0x80. There is always room. */ + *p++ = 0x80; -/* Encodes input (uint32_t) into output (unsigned char). Assumes len is - a multiple of 4. - */ -static void Encode(uint8_t *output, const uint32_t *input, unsigned int len) -{ - unsigned int i, j; + /* Bytes of padding needed to make 56 bytes (-8..55) */ + count = 56 - 1 - count; - for (i = 0, j = 0; j < len; i++, j += 4) + if (count < 0) /* Padding forces an extra block */ { - output[j] = (unsigned char)(input[i] & 0xff); - output[j+1] = (unsigned char)((input[i] >> 8) & 0xff); - output[j+2] = (unsigned char)((input[i] >> 16) & 0xff); - output[j+3] = (unsigned char)((input[i] >> 24) & 0xff); + memset(p, 0, count + 8); + byteSwap(ctx->in, 16); + MD5Transform(ctx->buf, ctx->in); + p = (md5byte *)ctx->in; + count = 56; } + + memset(p, 0, count); + byteSwap(ctx->in, 14); + + /* Append length in bits and transform */ + ctx->in[14] = ctx->bytes[0] << 3; + ctx->in[15] = ctx->bytes[1] << 3 | ctx->bytes[0] >> 29; + MD5Transform(ctx->buf, ctx->in); + + byteSwap(ctx->buf, 4); + memcpy(digest, ctx->buf, 16); + memset(ctx, 0, sizeof(*ctx)); /* In case it's sensitive */ } -/* Decodes input (unsigned char) into output (uint32_t). Assumes len is - a multiple of 4. +#ifndef ASM_MD5 + +/* The four core functions - F1 is optimized somewhat */ + +/* #define F1(x, y, z) (x & y | ~x & z) */ +#define F1(x, y, z) (z ^ (x & (y ^ z))) +#define F2(x, y, z) F1(z, x, y) +#define F3(x, y, z) (x ^ y ^ z) +#define F4(x, y, z) (y ^ (x | ~z)) + +/* This is the central step in the MD5 algorithm. */ +#define MD5STEP(f,w,x,y,z,in,s) \ + (w += f(x,y,z) + in, w = (w<<s | w>>(32-s)) + x) + +/* + * The core of the MD5 algorithm, this alters an existing MD5 hash to + * reflect the addition of 16 longwords of new data. MD5Update blocks + * the data and converts bytes into longwords for this routine. */ -static void Decode(uint32_t *output, const uint8_t *input, unsigned int len) +void +MD5Transform(UWORD32 buf[4], UWORD32 const in[16]) { - unsigned int i, j; - - for (i = 0, j = 0; j < len; i++, j += 4) - output[i] = ((uint32_t)input[j]) | (((uint32_t)input[j+1]) << 8) | - (((uint32_t)input[j+2]) << 16) | (((uint32_t)input[j+3]) << 24); + register UWORD32 a, b, c, d; + + a = buf[0]; + b = buf[1]; + c = buf[2]; + d = buf[3]; + + MD5STEP(F1, a, b, c, d, in[0] + 0xd76aa478, 7); + MD5STEP(F1, d, a, b, c, in[1] + 0xe8c7b756, 12); + MD5STEP(F1, c, d, a, b, in[2] + 0x242070db, 17); + MD5STEP(F1, b, c, d, a, in[3] + 0xc1bdceee, 22); + MD5STEP(F1, a, b, c, d, in[4] + 0xf57c0faf, 7); + MD5STEP(F1, d, a, b, c, in[5] + 0x4787c62a, 12); + MD5STEP(F1, c, d, a, b, in[6] + 0xa8304613, 17); + MD5STEP(F1, b, c, d, a, in[7] + 0xfd469501, 22); + MD5STEP(F1, a, b, c, d, in[8] + 0x698098d8, 7); + MD5STEP(F1, d, a, b, c, in[9] + 0x8b44f7af, 12); + MD5STEP(F1, c, d, a, b, in[10] + 0xffff5bb1, 17); + MD5STEP(F1, b, c, d, a, in[11] + 0x895cd7be, 22); + MD5STEP(F1, a, b, c, d, in[12] + 0x6b901122, 7); + MD5STEP(F1, d, a, b, c, in[13] + 0xfd987193, 12); + MD5STEP(F1, c, d, a, b, in[14] + 0xa679438e, 17); + MD5STEP(F1, b, c, d, a, in[15] + 0x49b40821, 22); + + MD5STEP(F2, a, b, c, d, in[1] + 0xf61e2562, 5); + MD5STEP(F2, d, a, b, c, in[6] + 0xc040b340, 9); + MD5STEP(F2, c, d, a, b, in[11] + 0x265e5a51, 14); + MD5STEP(F2, b, c, d, a, in[0] + 0xe9b6c7aa, 20); + MD5STEP(F2, a, b, c, d, in[5] + 0xd62f105d, 5); + MD5STEP(F2, d, a, b, c, in[10] + 0x02441453, 9); + MD5STEP(F2, c, d, a, b, in[15] + 0xd8a1e681, 14); + MD5STEP(F2, b, c, d, a, in[4] + 0xe7d3fbc8, 20); + MD5STEP(F2, a, b, c, d, in[9] + 0x21e1cde6, 5); + MD5STEP(F2, d, a, b, c, in[14] + 0xc33707d6, 9); + MD5STEP(F2, c, d, a, b, in[3] + 0xf4d50d87, 14); + MD5STEP(F2, b, c, d, a, in[8] + 0x455a14ed, 20); + MD5STEP(F2, a, b, c, d, in[13] + 0xa9e3e905, 5); + MD5STEP(F2, d, a, b, c, in[2] + 0xfcefa3f8, 9); + MD5STEP(F2, c, d, a, b, in[7] + 0x676f02d9, 14); + MD5STEP(F2, b, c, d, a, in[12] + 0x8d2a4c8a, 20); + + MD5STEP(F3, a, b, c, d, in[5] + 0xfffa3942, 4); + MD5STEP(F3, d, a, b, c, in[8] + 0x8771f681, 11); + MD5STEP(F3, c, d, a, b, in[11] + 0x6d9d6122, 16); + MD5STEP(F3, b, c, d, a, in[14] + 0xfde5380c, 23); + MD5STEP(F3, a, b, c, d, in[1] + 0xa4beea44, 4); + MD5STEP(F3, d, a, b, c, in[4] + 0x4bdecfa9, 11); + MD5STEP(F3, c, d, a, b, in[7] + 0xf6bb4b60, 16); + MD5STEP(F3, b, c, d, a, in[10] + 0xbebfbc70, 23); + MD5STEP(F3, a, b, c, d, in[13] + 0x289b7ec6, 4); + MD5STEP(F3, d, a, b, c, in[0] + 0xeaa127fa, 11); + MD5STEP(F3, c, d, a, b, in[3] + 0xd4ef3085, 16); + MD5STEP(F3, b, c, d, a, in[6] + 0x04881d05, 23); + MD5STEP(F3, a, b, c, d, in[9] + 0xd9d4d039, 4); + MD5STEP(F3, d, a, b, c, in[12] + 0xe6db99e5, 11); + MD5STEP(F3, c, d, a, b, in[15] + 0x1fa27cf8, 16); + MD5STEP(F3, b, c, d, a, in[2] + 0xc4ac5665, 23); + + MD5STEP(F4, a, b, c, d, in[0] + 0xf4292244, 6); + MD5STEP(F4, d, a, b, c, in[7] + 0x432aff97, 10); + MD5STEP(F4, c, d, a, b, in[14] + 0xab9423a7, 15); + MD5STEP(F4, b, c, d, a, in[5] + 0xfc93a039, 21); + MD5STEP(F4, a, b, c, d, in[12] + 0x655b59c3, 6); + MD5STEP(F4, d, a, b, c, in[3] + 0x8f0ccc92, 10); + MD5STEP(F4, c, d, a, b, in[10] + 0xffeff47d, 15); + MD5STEP(F4, b, c, d, a, in[1] + 0x85845dd1, 21); + MD5STEP(F4, a, b, c, d, in[8] + 0x6fa87e4f, 6); + MD5STEP(F4, d, a, b, c, in[15] + 0xfe2ce6e0, 10); + MD5STEP(F4, c, d, a, b, in[6] + 0xa3014314, 15); + MD5STEP(F4, b, c, d, a, in[13] + 0x4e0811a1, 21); + MD5STEP(F4, a, b, c, d, in[4] + 0xf7537e82, 6); + MD5STEP(F4, d, a, b, c, in[11] + 0xbd3af235, 10); + MD5STEP(F4, c, d, a, b, in[2] + 0x2ad7d2bb, 15); + MD5STEP(F4, b, c, d, a, in[9] + 0xeb86d391, 21); + + buf[0] += a; + buf[1] += b; + buf[2] += c; + buf[3] += d; } + +#endif diff --git a/md5_utils.h b/md5_utils.h index 8dda8d1..5ca1b5f 100644 --- a/md5_utils.h +++ b/md5_utils.h @@ -1,51 +1,42 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * This is the header file for the MD5 message-digest algorithm. + * The algorithm is due to Ron Rivest. This code was + * written by Colin Plumb in 1993, no copyright is claimed. + * This code is in the public domain; do with it what you wish. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Equivalent code is available from RSA Data Security, Inc. + * This code has been tested against that, and is equivalent, + * except that you don't need to include two pages of legalese + * with every copy. + * + * To compute the message digest of a chunk of bytes, declare an + * MD5Context structure, pass it to MD5Init, call MD5Update as + * needed on buffers full of bytes, and then call MD5Final, which + * will fill a supplied 16-byte array with the digest. + * + * Changed so as no longer to depend on Colin Plumb's `usual.h' + * header definitions + * - Ian Jackson <ian@chiark.greenend.org.uk>. + * Still in the public domain. */ -/* -Copyright (C) 1991-2, RSA Data Security, Inc. Created 1991. All -rights reserved. - -License to copy and use this software is granted provided that it -is identified as the "RSA Data Security, Inc. MD5 Message-Digest -Algorithm" in all material mentioning or referencing this software -or this function. +#ifndef MD5_H +#define MD5_H -License is also granted to make and use derivative works provided -that such works are identified as "derived from the RSA Data -Security, Inc. MD5 Message-Digest Algorithm" in all material -mentioning or referencing the derived work. +#define md5byte unsigned char +#define UWORD32 unsigned int -RSA Data Security, Inc. makes no representations concerning either -the merchantability of this software or the suitability of this -software for any particular purpose. It is provided "as is" -without express or implied warranty of any kind. - -These notices must be retained in any copies of any part of this -documentation and/or software. -*/ -#ifndef HAVE_VPX_PORTS -#define HAVE_VPX_PORTS 0 -#endif -#if HAVE_VPX_PORTS -#include "vpx_ports/vpx_integer.h" -#else -#include "vpx_integer.h" -#endif - -/* MD5 context. */ -typedef struct +typedef struct MD5Context MD5Context; +struct MD5Context { - uint32_t state[4]; /* state (ABCD) */ - uint32_t count[2]; /* number of bits, modulo 2^64 (lsb first) */ - uint8_t buffer[64]; /* input buffer */ -} md5_ctx_t; + UWORD32 buf[4]; + UWORD32 bytes[2]; + UWORD32 in[16]; +}; + +void MD5Init(struct MD5Context *context); +void MD5Update(struct MD5Context *context, md5byte const *buf, unsigned len); +void MD5Final(unsigned char digest[16], struct MD5Context *context); +void MD5Transform(UWORD32 buf[4], UWORD32 const in[16]); -void md5_init(md5_ctx_t *ctx); -void md5_update(md5_ctx_t *ctx, const uint8_t *buf, unsigned int len); -void md5_finalize(md5_ctx_t *ctx, uint8_t md5[16]); +#endif /* !MD5_H */ @@ -1,11 +1,12 @@ -#!/bin/bash +#!/bin/sh ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## @@ -20,7 +21,7 @@ for opt; do esac done -TAB=$'\t' +TAB="$(printf '\t')" cat > release.mk << EOF %\$(BUILD_SFX).tar.bz2: %/.done ${TAB}@echo "\$(subst .tar.bz2,,\$@): tarball" @@ -185,7 +186,7 @@ for cfg in $CONFIGS; do esac opts="$opts --enable-postproc" - [ "x${clean}" == "xyes" ] \ + [ "x${clean}" = "xyes" ] \ && rm -rf ${full_cfg}${BUILD_SFX}${TAR_SFX} \ && rm -rf logs/${full_cfg}${BUILD_SFX}.log.bz2 diff --git a/solution.mk b/solution.mk index 783c6f8..8e852ec 100644 --- a/solution.mk +++ b/solution.mk @@ -1,10 +1,11 @@ ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## diff --git a/tools/gen_authors.sh b/tools/gen_authors.sh new file mode 100755 index 0000000..e1246f0 --- /dev/null +++ b/tools/gen_authors.sh @@ -0,0 +1,13 @@ +#!/bin/bash + +# Add organization names manually. + +cat <<EOF +# This file is automatically generated from the git commit history +# by tools/gen_authors.sh. + +$(git log --pretty=format:"%aN <%aE>" | sort | uniq) +Google Inc. +The Mozilla Foundation +The Xiph.Org Foundation +EOF diff --git a/vp8/common/alloccommon.c b/vp8/common/alloccommon.c index ac110f7..408c253 100644 --- a/vp8/common/alloccommon.c +++ b/vp8/common/alloccommon.c @@ -1,13 +1,15 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ +#include "vpx_ports/config.h" #include "blockd.h" #include "vpx_mem/vpx_mem.h" #include "onyxc_int.h" @@ -16,52 +18,42 @@ #include "systemdependent.h" #include "vpxerrors.h" -#ifdef HAVE_CONFIG_H -#include "vpx_config.h" -#endif extern void vp8_init_scan_order_mask(); void vp8_update_mode_info_border(MODE_INFO *mi, int rows, int cols) { int i; - vpx_memset(mi - cols - 1, 0, sizeof(MODE_INFO) * cols + 1); + vpx_memset(mi - cols - 2, 0, sizeof(MODE_INFO) * (cols + 1)); for (i = 0; i < rows; i++) { vpx_memset(&mi[i*cols-1], 0, sizeof(MODE_INFO)); } } + void vp8_de_alloc_frame_buffers(VP8_COMMON *oci) { + int i; + + for (i = 0; i < NUM_YV12_BUFFERS; i++) + vp8_yv12_de_alloc_frame_buffer(&oci->yv12_fb[i]); + vp8_yv12_de_alloc_frame_buffer(&oci->temp_scale_frame); - vp8_yv12_de_alloc_frame_buffer(&oci->new_frame); - vp8_yv12_de_alloc_frame_buffer(&oci->last_frame); - vp8_yv12_de_alloc_frame_buffer(&oci->golden_frame); - vp8_yv12_de_alloc_frame_buffer(&oci->alt_ref_frame); vp8_yv12_de_alloc_frame_buffer(&oci->post_proc_buffer); - vpx_free(oci->above_context[Y1CONTEXT]); - vpx_free(oci->above_context[UCONTEXT]); - vpx_free(oci->above_context[VCONTEXT]); - vpx_free(oci->above_context[Y2CONTEXT]); + vpx_free(oci->above_context); vpx_free(oci->mip); - oci->above_context[Y1CONTEXT] = 0; - oci->above_context[UCONTEXT] = 0; - oci->above_context[VCONTEXT] = 0; - oci->above_context[Y2CONTEXT] = 0; + oci->above_context = 0; oci->mip = 0; - // Structure used to minitor GF useage - if (oci->gf_active_flags != 0) - vpx_free(oci->gf_active_flags); - - oci->gf_active_flags = 0; } int vp8_alloc_frame_buffers(VP8_COMMON *oci, int width, int height) { + int i; + vp8_de_alloc_frame_buffers(oci); // our internal buffers are always multiples of 16 @@ -72,32 +64,28 @@ int vp8_alloc_frame_buffers(VP8_COMMON *oci, int width, int height) height += 16 - (height & 0xf); - if (vp8_yv12_alloc_frame_buffer(&oci->temp_scale_frame, width, 16, VP8BORDERINPIXELS) < 0) + for (i = 0; i < NUM_YV12_BUFFERS; i++) { - vp8_de_alloc_frame_buffers(oci); - return ALLOC_FAILURE; - } - + oci->fb_idx_ref_cnt[0] = 0; - if (vp8_yv12_alloc_frame_buffer(&oci->new_frame, width, height, VP8BORDERINPIXELS) < 0) - { - vp8_de_alloc_frame_buffers(oci); - return ALLOC_FAILURE; + if (vp8_yv12_alloc_frame_buffer(&oci->yv12_fb[i], width, height, VP8BORDERINPIXELS) < 0) + { + vp8_de_alloc_frame_buffers(oci); + return ALLOC_FAILURE; + } } - if (vp8_yv12_alloc_frame_buffer(&oci->last_frame, width, height, VP8BORDERINPIXELS) < 0) - { - vp8_de_alloc_frame_buffers(oci); - return ALLOC_FAILURE; - } + oci->new_fb_idx = 0; + oci->lst_fb_idx = 1; + oci->gld_fb_idx = 2; + oci->alt_fb_idx = 3; - if (vp8_yv12_alloc_frame_buffer(&oci->golden_frame, width, height, VP8BORDERINPIXELS) < 0) - { - vp8_de_alloc_frame_buffers(oci); - return ALLOC_FAILURE; - } + oci->fb_idx_ref_cnt[0] = 1; + oci->fb_idx_ref_cnt[1] = 1; + oci->fb_idx_ref_cnt[2] = 1; + oci->fb_idx_ref_cnt[3] = 1; - if (vp8_yv12_alloc_frame_buffer(&oci->alt_ref_frame, width, height, VP8BORDERINPIXELS) < 0) + if (vp8_yv12_alloc_frame_buffer(&oci->temp_scale_frame, width, 16, VP8BORDERINPIXELS) < 0) { vp8_de_alloc_frame_buffers(oci); return ALLOC_FAILURE; @@ -124,33 +112,9 @@ int vp8_alloc_frame_buffers(VP8_COMMON *oci, int width, int height) oci->mi = oci->mip + oci->mode_info_stride + 1; - oci->above_context[Y1CONTEXT] = vpx_calloc(sizeof(ENTROPY_CONTEXT) * oci->mb_cols * 4 , 1); - - if (!oci->above_context[Y1CONTEXT]) - { - vp8_de_alloc_frame_buffers(oci); - return ALLOC_FAILURE; - } + oci->above_context = vpx_calloc(sizeof(ENTROPY_CONTEXT_PLANES) * oci->mb_cols, 1); - oci->above_context[UCONTEXT] = vpx_calloc(sizeof(ENTROPY_CONTEXT) * oci->mb_cols * 2 , 1); - - if (!oci->above_context[UCONTEXT]) - { - vp8_de_alloc_frame_buffers(oci); - return ALLOC_FAILURE; - } - - oci->above_context[VCONTEXT] = vpx_calloc(sizeof(ENTROPY_CONTEXT) * oci->mb_cols * 2 , 1); - - if (!oci->above_context[VCONTEXT]) - { - vp8_de_alloc_frame_buffers(oci); - return ALLOC_FAILURE; - } - - oci->above_context[Y2CONTEXT] = vpx_calloc(sizeof(ENTROPY_CONTEXT) * oci->mb_cols , 1); - - if (!oci->above_context[Y2CONTEXT]) + if (!oci->above_context) { vp8_de_alloc_frame_buffers(oci); return ALLOC_FAILURE; @@ -158,20 +122,6 @@ int vp8_alloc_frame_buffers(VP8_COMMON *oci, int width, int height) vp8_update_mode_info_border(oci->mi, oci->mb_rows, oci->mb_cols); - // Structures used to minitor GF usage - if (oci->gf_active_flags != 0) - vpx_free(oci->gf_active_flags); - - oci->gf_active_flags = (unsigned char *)vpx_calloc(oci->mb_rows * oci->mb_cols, 1); - - if (!oci->gf_active_flags) - { - vp8_de_alloc_frame_buffers(oci); - return ALLOC_FAILURE; - } - - oci->gf_active_count = oci->mb_rows * oci->mb_cols; - return 0; } void vp8_setup_version(VP8_COMMON *cm) diff --git a/vp8/common/alloccommon.h b/vp8/common/alloccommon.h index 73c7383..ea93c25 100644 --- a/vp8/common/alloccommon.h +++ b/vp8/common/alloccommon.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/arm/armv6/bilinearfilter_v6.asm b/vp8/common/arm/armv6/bilinearfilter_v6.asm index 4428cf8..09d7338 100644 --- a/vp8/common/arm/armv6/bilinearfilter_v6.asm +++ b/vp8/common/arm/armv6/bilinearfilter_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/armv6/copymem16x16_v6.asm b/vp8/common/arm/armv6/copymem16x16_v6.asm index 00e9739..fca91a0 100644 --- a/vp8/common/arm/armv6/copymem16x16_v6.asm +++ b/vp8/common/arm/armv6/copymem16x16_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/armv6/copymem8x4_v6.asm b/vp8/common/arm/armv6/copymem8x4_v6.asm index 94473ca..d8362ef 100644 --- a/vp8/common/arm/armv6/copymem8x4_v6.asm +++ b/vp8/common/arm/armv6/copymem8x4_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/armv6/copymem8x8_v6.asm b/vp8/common/arm/armv6/copymem8x8_v6.asm index 7cfa533..c6a60c6 100644 --- a/vp8/common/arm/armv6/copymem8x8_v6.asm +++ b/vp8/common/arm/armv6/copymem8x8_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/armv6/dc_only_idct_add_v6.asm b/vp8/common/arm/armv6/dc_only_idct_add_v6.asm new file mode 100644 index 0000000..e0660e9 --- /dev/null +++ b/vp8/common/arm/armv6/dc_only_idct_add_v6.asm @@ -0,0 +1,67 @@ +; +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. +; +; Use of this source code is governed by a BSD-style license and patent +; grant that can be found in the LICENSE file in the root of the source +; tree. All contributing project authors may be found in the AUTHORS +; file in the root of the source tree. +; + + EXPORT |vp8_dc_only_idct_add_v6| + + AREA |.text|, CODE, READONLY + +;void vp8_dc_only_idct_add_v6(short input_dc, unsigned char *pred_ptr, +; unsigned char *dst_ptr, int pitch, int stride) +; r0 input_dc +; r1 pred_ptr +; r2 dest_ptr +; r3 pitch +; sp stride + +|vp8_dc_only_idct_add_v6| PROC + stmdb sp!, {r4 - r7, lr} + + add r0, r0, #4 ; input_dc += 4 + ldr r12, c0x0000FFFF + ldr r4, [r1], r3 + ldr r6, [r1], r3 + and r0, r12, r0, asr #3 ; input_dc >> 3 + mask + ldr lr, [sp, #20] + orr r0, r0, r0, lsl #16 ; a1 | a1 + + uxtab16 r5, r0, r4 ; a1+2 | a1+0 + uxtab16 r4, r0, r4, ror #8 ; a1+3 | a1+1 + uxtab16 r7, r0, r6 + uxtab16 r6, r0, r6, ror #8 + usat16 r5, #8, r5 + usat16 r4, #8, r4 + usat16 r7, #8, r7 + usat16 r6, #8, r6 + orr r5, r5, r4, lsl #8 + orr r7, r7, r6, lsl #8 + ldr r4, [r1], r3 + ldr r6, [r1] + str r5, [r2], lr + str r7, [r2], lr + + uxtab16 r5, r0, r4 + uxtab16 r4, r0, r4, ror #8 + uxtab16 r7, r0, r6 + uxtab16 r6, r0, r6, ror #8 + usat16 r5, #8, r5 + usat16 r4, #8, r4 + usat16 r7, #8, r7 + usat16 r6, #8, r6 + orr r5, r5, r4, lsl #8 + orr r7, r7, r6, lsl #8 + str r5, [r2], lr + str r7, [r2] + + ldmia sp!, {r4 - r7, pc} + + ENDP ; |vp8_dc_only_idct_add_v6| + +; Constant Pool +c0x0000FFFF DCD 0x0000FFFF + END diff --git a/vp8/common/arm/armv6/filter_v6.asm b/vp8/common/arm/armv6/filter_v6.asm index a7863fc..8bc6d77 100644 --- a/vp8/common/arm/armv6/filter_v6.asm +++ b/vp8/common/arm/armv6/filter_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/armv6/idct_v6.asm b/vp8/common/arm/armv6/idct_v6.asm index 25c5165..27215af 100644 --- a/vp8/common/arm/armv6/idct_v6.asm +++ b/vp8/common/arm/armv6/idct_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -14,8 +15,6 @@ EXPORT |vp8_short_idct4x4llm_v6_scott| EXPORT |vp8_short_idct4x4llm_v6_dual| - EXPORT |vp8_dc_only_idct_armv6| - AREA |.text|, CODE, READONLY ;******************************************************************************** @@ -343,34 +342,4 @@ loop2_dual ldmia sp!, {r4 - r11, pc} ; replace vars, return restore ENDP - -; sjl added 10/17/08 -;void dc_only_idct_armv6(short input_dc, short *output, int pitch) -|vp8_dc_only_idct_armv6| PROC - stmdb sp!, {r4 - r6, lr} - - add r0, r0, #0x4 - add r4, r1, r2 ; output + shortpitch - mov r0, r0, ASR #0x3 ;aka a1 - add r5, r1, r2, LSL #1 ; output + shortpitch * 2 - pkhbt r0, r0, r0, lsl #16 ; a1 | a1 - add r6, r5, r2 ; output + shortpitch * 3 - - str r0, [r1, #0] - str r0, [r1, #4] - - str r0, [r4, #0] - str r0, [r4, #4] - - str r0, [r5, #0] - str r0, [r5, #4] - - str r0, [r6, #0] - str r0, [r6, #4] - - - ldmia sp!, {r4 - r6, pc} - - ENDP ; |vp8_dc_only_idct_armv6| - END diff --git a/vp8/common/arm/armv6/iwalsh_v6.asm b/vp8/common/arm/armv6/iwalsh_v6.asm index 8747568..463bff0 100644 --- a/vp8/common/arm/armv6/iwalsh_v6.asm +++ b/vp8/common/arm/armv6/iwalsh_v6.asm @@ -1,14 +1,15 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; - EXPORT |vp8_short_inv_walsh4x4_armv6| - EXPORT |vp8_short_inv_walsh4x4_1_armv6| + EXPORT |vp8_short_inv_walsh4x4_v6| + EXPORT |vp8_short_inv_walsh4x4_1_v6| ARM REQUIRE8 @@ -16,8 +17,8 @@ AREA |.text|, CODE, READONLY ; name this block of code -;short vp8_short_inv_walsh4x4_armv6(short *input, short *output) -|vp8_short_inv_walsh4x4_armv6| PROC +;short vp8_short_inv_walsh4x4_v6(short *input, short *output) +|vp8_short_inv_walsh4x4_v6| PROC stmdb sp!, {r4 - r11, lr} @@ -122,11 +123,11 @@ str r5, [r1] ldmia sp!, {r4 - r11, pc} - ENDP ; |vp8_short_inv_walsh4x4_armv6| + ENDP ; |vp8_short_inv_walsh4x4_v6| -;short vp8_short_inv_walsh4x4_1_armv6(short *input, short *output) -|vp8_short_inv_walsh4x4_1_armv6| PROC +;short vp8_short_inv_walsh4x4_1_v6(short *input, short *output) +|vp8_short_inv_walsh4x4_1_v6| PROC ldrsh r2, [r0] ; [0] add r2, r2, #3 ; [0] + 3 @@ -144,7 +145,7 @@ str r2, [r1] bx lr - ENDP ; |vp8_short_inv_walsh4x4_1_armv6| + ENDP ; |vp8_short_inv_walsh4x4_1_v6| ; Constant Pool c0x00030003 DCD 0x00030003 diff --git a/vp8/common/arm/armv6/loopfilter_v6.asm b/vp8/common/arm/armv6/loopfilter_v6.asm index c2b02dc..b6417de 100644 --- a/vp8/common/arm/armv6/loopfilter_v6.asm +++ b/vp8/common/arm/armv6/loopfilter_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/armv6/recon_v6.asm b/vp8/common/arm/armv6/recon_v6.asm index 085ff80..99c7bcf 100644 --- a/vp8/common/arm/armv6/recon_v6.asm +++ b/vp8/common/arm/armv6/recon_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/armv6/simpleloopfilter_v6.asm b/vp8/common/arm/armv6/simpleloopfilter_v6.asm index 15c6c7d..0137120 100644 --- a/vp8/common/arm/armv6/simpleloopfilter_v6.asm +++ b/vp8/common/arm/armv6/simpleloopfilter_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -54,113 +55,87 @@ pstep RN r1 ;stack const char *thresh, ;stack int count -;Note: All 16 elements in flimit are equal. So, in the code, only one load is needed -;for flimit. Same way applies to limit and thresh. +; All 16 elements in flimit are equal. So, in the code, only one load is needed +; for flimit. Same applies to limit. thresh is not used in simple looopfilter ;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- |vp8_loop_filter_simple_horizontal_edge_armv6| PROC ;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- stmdb sp!, {r4 - r11, lr} - sub src, src, pstep, lsl #1 ; move src pointer down by 2 lines - - ldr r12, [r3], #4 ; limit - ldr r3, [src], pstep ; p1 - - ldr r9, [sp, #36] ; count for 8-in-parallel - ldr r4, [src], pstep ; p0 - - ldr r7, [r2], #4 ; flimit - ldr r5, [src], pstep ; q0 + ldr r12, [r3] ; limit + ldr r3, [src, -pstep, lsl #1] ; p1 + ldr r4, [src, -pstep] ; p0 + ldr r5, [src] ; q0 + ldr r6, [src, pstep] ; q1 + ldr r7, [r2] ; flimit ldr r2, c0x80808080 - - ldr r6, [src] ; q1 - + ldr r9, [sp, #40] ; count for 8-in-parallel uadd8 r7, r7, r7 ; flimit * 2 - mov r9, r9, lsl #1 ; 4-in-parallel + mov r9, r9, lsl #1 ; double the count. we're doing 4 at a time uadd8 r12, r7, r12 ; flimit * 2 + limit + mov lr, #0 ; need 0 in a couple places |simple_hnext8| - ; vp8_simple_filter_mask() function + ; vp8_simple_filter_mask() uqsub8 r7, r3, r6 ; p1 - q1 uqsub8 r8, r6, r3 ; q1 - p1 uqsub8 r10, r4, r5 ; p0 - q0 uqsub8 r11, r5, r4 ; q0 - p0 orr r8, r8, r7 ; abs(p1 - q1) - ldr lr, c0x7F7F7F7F ; 01111111 mask orr r10, r10, r11 ; abs(p0 - q0) - and r8, lr, r8, lsr #1 ; abs(p1 - q1) / 2 uqadd8 r10, r10, r10 ; abs(p0 - q0) * 2 - mvn lr, #0 ; r10 == -1 + uhadd8 r8, r8, lr ; abs(p1 - q2) >> 1 uqadd8 r10, r10, r8 ; abs(p0 - q0)*2 + abs(p1 - q1)/2 - ; STALL waiting on r10 :( - uqsub8 r10, r10, r12 ; compare to flimit - mov r8, #0 - - usub8 r10, r8, r10 ; use usub8 instead of ssub8 - ; STALL (maybe?) when are flags set? :/ - sel r10, lr, r8 ; filter mask: lr - + mvn r8, #0 + usub8 r10, r12, r10 ; compare to flimit. usub8 sets GE flags + sel r10, r8, lr ; filter mask: F or 0 cmp r10, #0 - beq simple_hskip_filter ; skip filtering + beq simple_hskip_filter ; skip filtering if all masks are 0x00 - ;vp8_simple_filter() function + ;vp8_simple_filter() eor r3, r3, r2 ; p1 offset to convert to a signed value eor r6, r6, r2 ; q1 offset to convert to a signed value eor r4, r4, r2 ; p0 offset to convert to a signed value eor r5, r5, r2 ; q0 offset to convert to a signed value - qsub8 r3, r3, r6 ; vp8_filter (r3) = vp8_signed_char_clamp(p1-q1) - qsub8 r6, r5, r4 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( q0 - p0)) - - qadd8 r3, r3, r6 - ldr r8, c0x03030303 ; r8 = 3 - - qadd8 r3, r3, r6 + qsub8 r3, r3, r6 ; vp8_filter = p1 - q1 + qsub8 r6, r5, r4 ; q0 - p0 + qadd8 r3, r3, r6 ; += q0 - p0 ldr r7, c0x04040404 - - qadd8 r3, r3, r6 - and r3, r3, lr ; vp8_filter &= mask; - - ;save bottom 3 bits so that we round one side +4 and the other +3 - qadd8 r8 , r3 , r8 ; Filter2 (r8) = vp8_signed_char_clamp(vp8_filter+3) - qadd8 r3 , r3 , r7 ; Filter1 (r3) = vp8_signed_char_clamp(vp8_filter+4) - - mov r7, #0 - shadd8 r8 , r8 , r7 ; Filter2 >>= 3 - shadd8 r3 , r3 , r7 ; Filter1 >>= 3 - shadd8 r8 , r8 , r7 - shadd8 r3 , r3 , r7 - shadd8 r8 , r8 , r7 ; r8: Filter2 - shadd8 r3 , r3 , r7 ; r7: filter1 - - ;calculate output - sub src, src, pstep, lsl #1 - - qadd8 r4, r4, r8 ; u = vp8_signed_char_clamp(p0 + Filter2) - qsub8 r5 ,r5, r3 ; u = vp8_signed_char_clamp(q0 - Filter1) - eor r4, r4, r2 ; *op0 = u^0x80 - str r4, [src], pstep ; store op0 result + qadd8 r3, r3, r6 ; += q0 - p0 + ldr r8, c0x03030303 + qadd8 r3, r3, r6 ; vp8_filter = p1-q1 + 3*(q0-p0)) + ;STALL + and r3, r3, r10 ; vp8_filter &= mask + + qadd8 r7 , r3 , r7 ; Filter1 = vp8_filter + 4 + qadd8 r8 , r3 , r8 ; Filter2 = vp8_filter + 3 + + shadd8 r7 , r7 , lr + shadd8 r8 , r8 , lr + shadd8 r7 , r7 , lr + shadd8 r8 , r8 , lr + shadd8 r7 , r7 , lr ; Filter1 >>= 3 + shadd8 r8 , r8 , lr ; Filter2 >>= 3 + + qsub8 r5 ,r5, r7 ; u = q0 - Filter1 + qadd8 r4, r4, r8 ; u = p0 + Filter2 eor r5, r5, r2 ; *oq0 = u^0x80 - str r5, [src], pstep ; store oq0 result + str r5, [src] ; store oq0 result + eor r4, r4, r2 ; *op0 = u^0x80 + str r4, [src, -pstep] ; store op0 result |simple_hskip_filter| - add src, src, #4 - sub src, src, pstep - sub src, src, pstep, lsl #1 - subs r9, r9, #1 + addne src, src, #4 ; next row - ;pld [src] - ;pld [src, pstep] - ;pld [src, pstep, lsl #1] - - ldrne r3, [src], pstep ; p1 - ldrne r4, [src], pstep ; p0 - ldrne r5, [src], pstep ; q0 - ldrne r6, [src] ; q1 + ldrne r3, [src, -pstep, lsl #1] ; p1 + ldrne r4, [src, -pstep] ; p0 + ldrne r5, [src] ; q0 + ldrne r6, [src, pstep] ; q1 bne simple_hnext8 @@ -173,9 +148,9 @@ pstep RN r1 ;-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- stmdb sp!, {r4 - r11, lr} - ldr r12, [r2], #4 ; r12: flimit + ldr r12, [r2] ; r12: flimit ldr r2, c0x80808080 - ldr r7, [r3], #4 ; limit + ldr r7, [r3] ; limit ; load soure data to r7, r8, r9, r10 ldrh r3, [src, #-2] @@ -212,16 +187,14 @@ pstep RN r1 uqsub8 r10, r5, r4 ; q0 - p0 orr r7, r7, r8 ; abs(p1 - q1) orr r9, r9, r10 ; abs(p0 - q0) - ldr lr, c0x7F7F7F7F ; 0111 1111 mask - uqadd8 r9, r9, r9 ; abs(p0 - q0) * 2 - and r7, lr, r7, lsr #1 ; abs(p1 - q1) / 2 mov r8, #0 + uqadd8 r9, r9, r9 ; abs(p0 - q0) * 2 + uhadd8 r7, r7, r8 ; abs(p1 - q1) / 2 uqadd8 r7, r7, r9 ; abs(p0 - q0)*2 + abs(p1 - q1)/2 mvn r10, #0 ; r10 == -1 - uqsub8 r7, r7, r12 ; compare to flimit - usub8 r7, r8, r7 - sel r7, r10, r8 ; filter mask: lr + usub8 r7, r12, r7 ; compare to flimit + sel lr, r10, r8 ; filter mask cmp lr, #0 beq simple_vskip_filter ; skip filtering @@ -232,35 +205,34 @@ pstep RN r1 eor r4, r4, r2 ; p0 offset to convert to a signed value eor r5, r5, r2 ; q0 offset to convert to a signed value - qsub8 r3, r3, r6 ; vp8_filter (r3) = vp8_signed_char_clamp(p1-q1) - qsub8 r6, r5, r4 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( q0 - p0)) + qsub8 r3, r3, r6 ; vp8_filter = p1 - q1 + qsub8 r6, r5, r4 ; q0 - p0 - qadd8 r3, r3, r6 - ldr r8, c0x03030303 ; r8 = 3 + qadd8 r3, r3, r6 ; vp8_filter += q0 - p0 + ldr r9, c0x03030303 ; r9 = 3 - qadd8 r3, r3, r6 + qadd8 r3, r3, r6 ; vp8_filter += q0 - p0 ldr r7, c0x04040404 - qadd8 r3, r3, r6 + qadd8 r3, r3, r6 ; vp8_filter = p1-q1 + 3*(q0-p0)) + ;STALL and r3, r3, lr ; vp8_filter &= mask - ;save bottom 3 bits so that we round one side +4 and the other +3 - qadd8 r8 , r3 , r8 ; Filter2 (r8) = vp8_signed_char_clamp(vp8_filter+3) - qadd8 r3 , r3 , r7 ; Filter1 (r3) = vp8_signed_char_clamp(vp8_filter+4) + qadd8 r9 , r3 , r9 ; Filter2 = vp8_filter + 3 + qadd8 r3 , r3 , r7 ; Filter1 = vp8_filter + 4 - mov r7, #0 - shadd8 r8 , r8 , r7 ; Filter2 >>= 3 - shadd8 r3 , r3 , r7 ; Filter1 >>= 3 - shadd8 r8 , r8 , r7 - shadd8 r3 , r3 , r7 - shadd8 r8 , r8 , r7 ; r8: filter2 - shadd8 r3 , r3 , r7 ; r7: filter1 + shadd8 r9 , r9 , r8 + shadd8 r3 , r3 , r8 + shadd8 r9 , r9 , r8 + shadd8 r3 , r3 , r8 + shadd8 r9 , r9 , r8 ; Filter2 >>= 3 + shadd8 r3 , r3 , r8 ; Filter1 >>= 3 ;calculate output sub src, src, pstep, lsl #2 - qadd8 r4, r4, r8 ; u = vp8_signed_char_clamp(p0 + Filter2) - qsub8 r5, r5, r3 ; u = vp8_signed_char_clamp(q0 - Filter1) + qadd8 r4, r4, r9 ; u = p0 + Filter2 + qsub8 r5, r5, r3 ; u = q0 - Filter1 eor r4, r4, r2 ; *op0 = u^0x80 eor r5, r5, r2 ; *oq0 = u^0x80 @@ -285,10 +257,6 @@ pstep RN r1 |simple_vskip_filter| subs r11, r11, #1 - ;pld [src] - ;pld [src, pstep] - ;pld [src, pstep, lsl #1] - ; load soure data to r7, r8, r9, r10 ldrneh r3, [src, #-2] ldrneh r4, [src], pstep @@ -308,14 +276,12 @@ pstep RN r1 bne simple_vnext8 - ldmia sp!, {r4 - r12, pc} + ldmia sp!, {r4 - r11, pc} ENDP ; |vp8_loop_filter_simple_vertical_edge_armv6| ; Constant Pool c0x80808080 DCD 0x80808080 c0x03030303 DCD 0x03030303 c0x04040404 DCD 0x04040404 -c0x01010101 DCD 0x01010101 -c0x7F7F7F7F DCD 0x7F7F7F7F END diff --git a/vp8/common/arm/armv6/sixtappredict8x4_v6.asm b/vp8/common/arm/armv6/sixtappredict8x4_v6.asm index 551d863..8fb80ef 100644 --- a/vp8/common/arm/armv6/sixtappredict8x4_v6.asm +++ b/vp8/common/arm/armv6/sixtappredict8x4_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/bilinearfilter_arm.c b/vp8/common/arm/bilinearfilter_arm.c index bf972a3..247f95b 100644 --- a/vp8/common/arm/bilinearfilter_arm.c +++ b/vp8/common/arm/bilinearfilter_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/arm/filter_arm.c b/vp8/common/arm/filter_arm.c index 2a4640c..5ed4f80 100644 --- a/vp8/common/arm/filter_arm.c +++ b/vp8/common/arm/filter_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/arm/idct_arm.h b/vp8/common/arm/idct_arm.h index f9ed21e..f28d7f6 100644 --- a/vp8/common/arm/idct_arm.h +++ b/vp8/common/arm/idct_arm.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -14,9 +15,9 @@ #if HAVE_ARMV6 extern prototype_idct(vp8_short_idct4x4llm_1_v6); extern prototype_idct(vp8_short_idct4x4llm_v6_dual); -extern prototype_idct_scalar(vp8_dc_only_idct_armv6); -extern prototype_second_order(vp8_short_inv_walsh4x4_1_armv6); -extern prototype_second_order(vp8_short_inv_walsh4x4_armv6); +extern prototype_idct_scalar_add(vp8_dc_only_idct_add_v6); +extern prototype_second_order(vp8_short_inv_walsh4x4_1_v6); +extern prototype_second_order(vp8_short_inv_walsh4x4_v6); #undef vp8_idct_idct1 #define vp8_idct_idct1 vp8_short_idct4x4llm_1_v6 @@ -24,20 +25,20 @@ extern prototype_second_order(vp8_short_inv_walsh4x4_armv6); #undef vp8_idct_idct16 #define vp8_idct_idct16 vp8_short_idct4x4llm_v6_dual -#undef vp8_idct_idct1_scalar -#define vp8_idct_idct1_scalar vp8_dc_only_idct_armv6 +#undef vp8_idct_idct1_scalar_add +#define vp8_idct_idct1_scalar_add vp8_dc_only_idct_add_v6 #undef vp8_idct_iwalsh1 -#define vp8_idct_iwalsh1 vp8_short_inv_walsh4x4_1_armv6 +#define vp8_idct_iwalsh1 vp8_short_inv_walsh4x4_1_v6 #undef vp8_idct_iwalsh16 -#define vp8_idct_iwalsh16 vp8_short_inv_walsh4x4_armv6 +#define vp8_idct_iwalsh16 vp8_short_inv_walsh4x4_v6 #endif #if HAVE_ARMV7 extern prototype_idct(vp8_short_idct4x4llm_1_neon); extern prototype_idct(vp8_short_idct4x4llm_neon); -extern prototype_idct_scalar(vp8_dc_only_idct_neon); +extern prototype_idct_scalar_add(vp8_dc_only_idct_add_neon); extern prototype_second_order(vp8_short_inv_walsh4x4_1_neon); extern prototype_second_order(vp8_short_inv_walsh4x4_neon); @@ -47,8 +48,8 @@ extern prototype_second_order(vp8_short_inv_walsh4x4_neon); #undef vp8_idct_idct16 #define vp8_idct_idct16 vp8_short_idct4x4llm_neon -#undef vp8_idct_idct1_scalar -#define vp8_idct_idct1_scalar vp8_dc_only_idct_neon +#undef vp8_idct_idct1_scalar_add +#define vp8_idct_idct1_scalar_add vp8_dc_only_idct_add_neon #undef vp8_idct_iwalsh1 #define vp8_idct_iwalsh1 vp8_short_inv_walsh4x4_1_neon diff --git a/vp8/common/arm/loopfilter_arm.c b/vp8/common/arm/loopfilter_arm.c index fa7c626..f86bca1 100644 --- a/vp8/common/arm/loopfilter_arm.c +++ b/vp8/common/arm/loopfilter_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -13,16 +14,6 @@ #include "loopfilter.h" #include "onyxc_int.h" -typedef void loop_filter_uvfunction -( - unsigned char *u, // source pointer - int p, // pitch - const signed char *flimit, - const signed char *limit, - const signed char *thresh, - unsigned char *v -); - extern prototype_loopfilter(vp8_loop_filter_horizontal_edge_armv6); extern prototype_loopfilter(vp8_loop_filter_vertical_edge_armv6); extern prototype_loopfilter(vp8_mbloop_filter_horizontal_edge_armv6); diff --git a/vp8/common/arm/loopfilter_arm.h b/vp8/common/arm/loopfilter_arm.h index 4bb4945..6c3628a 100644 --- a/vp8/common/arm/loopfilter_arm.h +++ b/vp8/common/arm/loopfilter_arm.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/arm/neon/bilinearpredict16x16_neon.asm b/vp8/common/arm/neon/bilinearpredict16x16_neon.asm index a2fea2b..bb72bad 100644 --- a/vp8/common/arm/neon/bilinearpredict16x16_neon.asm +++ b/vp8/common/arm/neon/bilinearpredict16x16_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/bilinearpredict4x4_neon.asm b/vp8/common/arm/neon/bilinearpredict4x4_neon.asm index 74d2db5..6d4820b 100644 --- a/vp8/common/arm/neon/bilinearpredict4x4_neon.asm +++ b/vp8/common/arm/neon/bilinearpredict4x4_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/bilinearpredict8x4_neon.asm b/vp8/common/arm/neon/bilinearpredict8x4_neon.asm index 46ebb0e..b9f3ce0 100644 --- a/vp8/common/arm/neon/bilinearpredict8x4_neon.asm +++ b/vp8/common/arm/neon/bilinearpredict8x4_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/bilinearpredict8x8_neon.asm b/vp8/common/arm/neon/bilinearpredict8x8_neon.asm index 80728d4..f7a7d14 100644 --- a/vp8/common/arm/neon/bilinearpredict8x8_neon.asm +++ b/vp8/common/arm/neon/bilinearpredict8x8_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/buildintrapredictorsmby_neon.asm b/vp8/common/arm/neon/buildintrapredictorsmby_neon.asm index f42ac63..e3ea91f 100644 --- a/vp8/common/arm/neon/buildintrapredictorsmby_neon.asm +++ b/vp8/common/arm/neon/buildintrapredictorsmby_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/copymem16x16_neon.asm b/vp8/common/arm/neon/copymem16x16_neon.asm index 89d5e10..bda4b96 100644 --- a/vp8/common/arm/neon/copymem16x16_neon.asm +++ b/vp8/common/arm/neon/copymem16x16_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/copymem8x4_neon.asm b/vp8/common/arm/neon/copymem8x4_neon.asm index 302f734..35c0f67 100644 --- a/vp8/common/arm/neon/copymem8x4_neon.asm +++ b/vp8/common/arm/neon/copymem8x4_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/copymem8x8_neon.asm b/vp8/common/arm/neon/copymem8x8_neon.asm index 50d39ef..1f5b941 100644 --- a/vp8/common/arm/neon/copymem8x8_neon.asm +++ b/vp8/common/arm/neon/copymem8x8_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/dc_only_idct_add_neon.asm b/vp8/common/arm/neon/dc_only_idct_add_neon.asm new file mode 100644 index 0000000..49ba05f --- /dev/null +++ b/vp8/common/arm/neon/dc_only_idct_add_neon.asm @@ -0,0 +1,49 @@ +; +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. +; +; Use of this source code is governed by a BSD-style license and patent +; grant that can be found in the LICENSE file in the root of the source +; tree. All contributing project authors may be found in the AUTHORS +; file in the root of the source tree. +; + + + EXPORT |vp8_dc_only_idct_add_neon| + ARM + REQUIRE8 + PRESERVE8 + + AREA ||.text||, CODE, READONLY, ALIGN=2 +;void vp8_dc_only_idct_add_neon(short input_dc, unsigned char *pred_ptr, +; unsigned char *dst_ptr, int pitch, int stride) +; r0 input_dc +; r1 pred_ptr +; r2 dst_ptr +; r3 pitch +; sp stride +|vp8_dc_only_idct_add_neon| PROC + add r0, r0, #4 + asr r0, r0, #3 + ldr r12, [sp] + vdup.16 q0, r0 + + vld1.32 {d2[0]}, [r1], r3 + vld1.32 {d2[1]}, [r1], r3 + vld1.32 {d4[0]}, [r1], r3 + vld1.32 {d4[1]}, [r1] + + vaddw.u8 q1, q0, d2 + vaddw.u8 q2, q0, d4 + + vqmovun.s16 d2, q1 + vqmovun.s16 d4, q2 + + vst1.32 {d2[0]}, [r2], r12 + vst1.32 {d2[1]}, [r2], r12 + vst1.32 {d4[0]}, [r2], r12 + vst1.32 {d4[1]}, [r2] + + bx lr + + ENDP + END diff --git a/vp8/common/arm/neon/iwalsh_neon.asm b/vp8/common/arm/neon/iwalsh_neon.asm index 4fc744c..663bf39 100644 --- a/vp8/common/arm/neon/iwalsh_neon.asm +++ b/vp8/common/arm/neon/iwalsh_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; EXPORT |vp8_short_inv_walsh4x4_neon| EXPORT |vp8_short_inv_walsh4x4_1_neon| diff --git a/vp8/common/arm/neon/loopfilterhorizontaledge_uv_neon.asm b/vp8/common/arm/neon/loopfilterhorizontaledge_uv_neon.asm index e3e8e8a..c0c3e33 100644 --- a/vp8/common/arm/neon/loopfilterhorizontaledge_uv_neon.asm +++ b/vp8/common/arm/neon/loopfilterhorizontaledge_uv_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/loopfilterhorizontaledge_y_neon.asm b/vp8/common/arm/neon/loopfilterhorizontaledge_y_neon.asm index f11055d..a8314cd 100644 --- a/vp8/common/arm/neon/loopfilterhorizontaledge_y_neon.asm +++ b/vp8/common/arm/neon/loopfilterhorizontaledge_y_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.asm b/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.asm index 6d74fab..0b84dc7 100644 --- a/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.asm +++ b/vp8/common/arm/neon/loopfiltersimplehorizontaledge_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.asm b/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.asm index 2bb6222..a793d09 100644 --- a/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.asm +++ b/vp8/common/arm/neon/loopfiltersimpleverticaledge_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/loopfilterverticaledge_uv_neon.asm b/vp8/common/arm/neon/loopfilterverticaledge_uv_neon.asm index d79cc68..57913d2 100644 --- a/vp8/common/arm/neon/loopfilterverticaledge_uv_neon.asm +++ b/vp8/common/arm/neon/loopfilterverticaledge_uv_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/loopfilterverticaledge_y_neon.asm b/vp8/common/arm/neon/loopfilterverticaledge_y_neon.asm index 3a230a9..2eb695f 100644 --- a/vp8/common/arm/neon/loopfilterverticaledge_y_neon.asm +++ b/vp8/common/arm/neon/loopfilterverticaledge_y_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/mbloopfilterhorizontaledge_uv_neon.asm b/vp8/common/arm/neon/mbloopfilterhorizontaledge_uv_neon.asm index 86eddaa..4576a6a 100644 --- a/vp8/common/arm/neon/mbloopfilterhorizontaledge_uv_neon.asm +++ b/vp8/common/arm/neon/mbloopfilterhorizontaledge_uv_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/mbloopfilterhorizontaledge_y_neon.asm b/vp8/common/arm/neon/mbloopfilterhorizontaledge_y_neon.asm index 2ab0fc2..8e85caa 100644 --- a/vp8/common/arm/neon/mbloopfilterhorizontaledge_y_neon.asm +++ b/vp8/common/arm/neon/mbloopfilterhorizontaledge_y_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/mbloopfilterverticaledge_uv_neon.asm b/vp8/common/arm/neon/mbloopfilterverticaledge_uv_neon.asm index ad5afba..d9dbdcf 100644 --- a/vp8/common/arm/neon/mbloopfilterverticaledge_uv_neon.asm +++ b/vp8/common/arm/neon/mbloopfilterverticaledge_uv_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/mbloopfilterverticaledge_y_neon.asm b/vp8/common/arm/neon/mbloopfilterverticaledge_y_neon.asm index 60e5175..bdffc62 100644 --- a/vp8/common/arm/neon/mbloopfilterverticaledge_y_neon.asm +++ b/vp8/common/arm/neon/mbloopfilterverticaledge_y_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/recon16x16mb_neon.asm b/vp8/common/arm/neon/recon16x16mb_neon.asm index b9ba1cb..3f1a30f 100644 --- a/vp8/common/arm/neon/recon16x16mb_neon.asm +++ b/vp8/common/arm/neon/recon16x16mb_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/recon2b_neon.asm b/vp8/common/arm/neon/recon2b_neon.asm index 25aaf8c..99b251c 100644 --- a/vp8/common/arm/neon/recon2b_neon.asm +++ b/vp8/common/arm/neon/recon2b_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/recon4b_neon.asm b/vp8/common/arm/neon/recon4b_neon.asm index a4f5b80..9917277 100644 --- a/vp8/common/arm/neon/recon4b_neon.asm +++ b/vp8/common/arm/neon/recon4b_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/reconb_neon.asm b/vp8/common/arm/neon/reconb_neon.asm index 16d85a0..288c0ef 100644 --- a/vp8/common/arm/neon/reconb_neon.asm +++ b/vp8/common/arm/neon/reconb_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/save_neon_reg.asm b/vp8/common/arm/neon/save_neon_reg.asm index 4873e44..fd7002e 100644 --- a/vp8/common/arm/neon/save_neon_reg.asm +++ b/vp8/common/arm/neon/save_neon_reg.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/shortidct4x4llm_1_neon.asm b/vp8/common/arm/neon/shortidct4x4llm_1_neon.asm index 7d06ff9..d7bdbae 100644 --- a/vp8/common/arm/neon/shortidct4x4llm_1_neon.asm +++ b/vp8/common/arm/neon/shortidct4x4llm_1_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/shortidct4x4llm_neon.asm b/vp8/common/arm/neon/shortidct4x4llm_neon.asm index ffecfbf..d77a287 100644 --- a/vp8/common/arm/neon/shortidct4x4llm_neon.asm +++ b/vp8/common/arm/neon/shortidct4x4llm_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/sixtappredict16x16_neon.asm b/vp8/common/arm/neon/sixtappredict16x16_neon.asm index 9f5f0d2..e434a70 100644 --- a/vp8/common/arm/neon/sixtappredict16x16_neon.asm +++ b/vp8/common/arm/neon/sixtappredict16x16_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/sixtappredict4x4_neon.asm b/vp8/common/arm/neon/sixtappredict4x4_neon.asm index c23a9db..3d22d77 100644 --- a/vp8/common/arm/neon/sixtappredict4x4_neon.asm +++ b/vp8/common/arm/neon/sixtappredict4x4_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/sixtappredict8x4_neon.asm b/vp8/common/arm/neon/sixtappredict8x4_neon.asm index 18e19f9..1dd6b1b 100644 --- a/vp8/common/arm/neon/sixtappredict8x4_neon.asm +++ b/vp8/common/arm/neon/sixtappredict8x4_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/neon/sixtappredict8x8_neon.asm b/vp8/common/arm/neon/sixtappredict8x8_neon.asm index d27485e..37255c7 100644 --- a/vp8/common/arm/neon/sixtappredict8x8_neon.asm +++ b/vp8/common/arm/neon/sixtappredict8x8_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/arm/recon_arm.c b/vp8/common/arm/recon_arm.c index 130059e..218898b 100644 --- a/vp8/common/arm/recon_arm.c +++ b/vp8/common/arm/recon_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/arm/recon_arm.h b/vp8/common/arm/recon_arm.h index fd9f85e..18855a3 100644 --- a/vp8/common/arm/recon_arm.h +++ b/vp8/common/arm/recon_arm.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/arm/reconintra4x4_arm.c b/vp8/common/arm/reconintra4x4_arm.c index 334d352..8d968d7 100644 --- a/vp8/common/arm/reconintra4x4_arm.c +++ b/vp8/common/arm/reconintra4x4_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/arm/reconintra_arm.c b/vp8/common/arm/reconintra_arm.c index d7ee1dd..4cc93d1 100644 --- a/vp8/common/arm/reconintra_arm.c +++ b/vp8/common/arm/reconintra_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -28,7 +29,7 @@ void vp8_build_intra_predictors_mby_neon(MACROBLOCKD *x) unsigned char *y_buffer = x->dst.y_buffer; unsigned char *ypred_ptr = x->predictor; int y_stride = x->dst.y_stride; - int mode = x->mbmi.mode; + int mode = x->mode_info_context->mbmi.mode; int Up = x->up_available; int Left = x->left_available; @@ -51,7 +52,7 @@ void vp8_build_intra_predictors_mby_s_neon(MACROBLOCKD *x) unsigned char *y_buffer = x->dst.y_buffer; unsigned char *ypred_ptr = x->predictor; int y_stride = x->dst.y_stride; - int mode = x->mbmi.mode; + int mode = x->mode_info_context->mbmi.mode; int Up = x->up_available; int Left = x->left_available; diff --git a/vp8/common/arm/subpixel_arm.h b/vp8/common/arm/subpixel_arm.h index 56aec55..53600e5 100644 --- a/vp8/common/arm/subpixel_arm.h +++ b/vp8/common/arm/subpixel_arm.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/arm/systemdependent.c b/vp8/common/arm/systemdependent.c index ecc6929..1eed97e 100644 --- a/vp8/common/arm/systemdependent.c +++ b/vp8/common/arm/systemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -42,7 +43,6 @@ void vp8_machine_specific_config(VP8_COMMON *ctx) rtcd->idct.idct1 = vp8_short_idct4x4llm_1_neon; rtcd->idct.idct16 = vp8_short_idct4x4llm_neon; - rtcd->idct.idct1_scalar = vp8_dc_only_idct_neon; rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_neon; rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_neon; @@ -74,7 +74,6 @@ void vp8_machine_specific_config(VP8_COMMON *ctx) rtcd->idct.idct1 = vp8_short_idct4x4llm_1_v6; rtcd->idct.idct16 = vp8_short_idct4x4llm_v6_dual; - rtcd->idct.idct1_scalar = vp8_dc_only_idct_armv6; rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_armv6; rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_armv6; @@ -127,11 +126,13 @@ void vp8_machine_specific_config(VP8_COMMON *ctx) rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_c; #endif +#if CONFIG_POSTPROC || (CONFIG_VP8_ENCODER && CONFIG_PSNR) rtcd->postproc.down = vp8_mbpost_proc_down_c; rtcd->postproc.across = vp8_mbpost_proc_across_ip_c; rtcd->postproc.downacross = vp8_post_proc_down_and_across_c; rtcd->postproc.addnoise = vp8_plane_add_noise_c; #endif +#endif #if HAVE_ARMV7 vp8_build_intra_predictors_mby_ptr = vp8_build_intra_predictors_mby_neon; diff --git a/vp8/common/arm/vpx_asm_offsets.c b/vp8/common/arm/vpx_asm_offsets.c index 68634bf..5baf8cc 100644 --- a/vp8/common/arm/vpx_asm_offsets.c +++ b/vp8/common/arm/vpx_asm_offsets.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -31,55 +32,50 @@ */ #if CONFIG_VP8_DECODER || CONFIG_VP8_ENCODER -DEFINE(yv12_buffer_config_y_width, offsetof(YV12_BUFFER_CONFIG, y_width)); -DEFINE(yv12_buffer_config_y_height, offsetof(YV12_BUFFER_CONFIG, y_height)); -DEFINE(yv12_buffer_config_y_stride, offsetof(YV12_BUFFER_CONFIG, y_stride)); -DEFINE(yv12_buffer_config_uv_width, offsetof(YV12_BUFFER_CONFIG, uv_width)); -DEFINE(yv12_buffer_config_uv_height, offsetof(YV12_BUFFER_CONFIG, uv_height)); -DEFINE(yv12_buffer_config_uv_stride, offsetof(YV12_BUFFER_CONFIG, uv_stride)); -DEFINE(yv12_buffer_config_y_buffer, offsetof(YV12_BUFFER_CONFIG, y_buffer)); -DEFINE(yv12_buffer_config_u_buffer, offsetof(YV12_BUFFER_CONFIG, u_buffer)); -DEFINE(yv12_buffer_config_v_buffer, offsetof(YV12_BUFFER_CONFIG, v_buffer)); +DEFINE(yv12_buffer_config_y_width, offsetof(YV12_BUFFER_CONFIG, y_width)); +DEFINE(yv12_buffer_config_y_height, offsetof(YV12_BUFFER_CONFIG, y_height)); +DEFINE(yv12_buffer_config_y_stride, offsetof(YV12_BUFFER_CONFIG, y_stride)); +DEFINE(yv12_buffer_config_uv_width, offsetof(YV12_BUFFER_CONFIG, uv_width)); +DEFINE(yv12_buffer_config_uv_height, offsetof(YV12_BUFFER_CONFIG, uv_height)); +DEFINE(yv12_buffer_config_uv_stride, offsetof(YV12_BUFFER_CONFIG, uv_stride)); +DEFINE(yv12_buffer_config_y_buffer, offsetof(YV12_BUFFER_CONFIG, y_buffer)); +DEFINE(yv12_buffer_config_u_buffer, offsetof(YV12_BUFFER_CONFIG, u_buffer)); +DEFINE(yv12_buffer_config_v_buffer, offsetof(YV12_BUFFER_CONFIG, v_buffer)); DEFINE(yv12_buffer_config_border, offsetof(YV12_BUFFER_CONFIG, border)); #endif #if CONFIG_VP8_DECODER DEFINE(mb_diff, offsetof(MACROBLOCKD, diff)); DEFINE(mb_predictor, offsetof(MACROBLOCKD, predictor)); -DEFINE(mb_dst_y_stride, offsetof(MACROBLOCKD, dst.y_stride)); -DEFINE(mb_dst_y_buffer, offsetof(MACROBLOCKD, dst.y_buffer)); -DEFINE(mb_dst_u_buffer, offsetof(MACROBLOCKD, dst.u_buffer)); -DEFINE(mb_dst_v_buffer, offsetof(MACROBLOCKD, dst.v_buffer)); -DEFINE(mb_mbmi_mode, offsetof(MACROBLOCKD, mbmi.mode)); -DEFINE(mb_up_available, offsetof(MACROBLOCKD, up_available)); -DEFINE(mb_left_available, offsetof(MACROBLOCKD, left_available)); +DEFINE(mb_dst_y_stride, offsetof(MACROBLOCKD, dst.y_stride)); +DEFINE(mb_dst_y_buffer, offsetof(MACROBLOCKD, dst.y_buffer)); +DEFINE(mb_dst_u_buffer, offsetof(MACROBLOCKD, dst.u_buffer)); +DEFINE(mb_dst_v_buffer, offsetof(MACROBLOCKD, dst.v_buffer)); +DEFINE(mb_up_available, offsetof(MACROBLOCKD, up_available)); +DEFINE(mb_left_available, offsetof(MACROBLOCKD, left_available)); DEFINE(detok_scan, offsetof(DETOK, scan)); -DEFINE(detok_ptr_onyxblock2context_leftabove, offsetof(DETOK, ptr_onyxblock2context_leftabove)); -DEFINE(detok_onyx_coef_tree_ptr, offsetof(DETOK, vp8_coef_tree_ptr)); -DEFINE(detok_teb_base_ptr, offsetof(DETOK, teb_base_ptr)); -DEFINE(detok_norm_ptr, offsetof(DETOK, norm_ptr)); -DEFINE(detok_ptr_onyx_coef_bands_x, offsetof(DETOK, ptr_onyx_coef_bands_x)); +DEFINE(detok_ptr_block2leftabove, offsetof(DETOK, ptr_block2leftabove)); +DEFINE(detok_coef_tree_ptr, offsetof(DETOK, vp8_coef_tree_ptr)); +DEFINE(detok_teb_base_ptr, offsetof(DETOK, teb_base_ptr)); +DEFINE(detok_norm_ptr, offsetof(DETOK, norm_ptr)); +DEFINE(detok_ptr_coef_bands_x, offsetof(DETOK, ptr_coef_bands_x)); -DEFINE(DETOK_A, offsetof(DETOK, A)); -DEFINE(DETOK_L, offsetof(DETOK, L)); +DEFINE(detok_A, offsetof(DETOK, A)); +DEFINE(detok_L, offsetof(DETOK, L)); -DEFINE(detok_qcoeff_start_ptr, offsetof(DETOK, qcoeff_start_ptr)); -DEFINE(detok_current_bc, offsetof(DETOK, current_bc)); -DEFINE(detok_coef_probs, offsetof(DETOK, coef_probs)); +DEFINE(detok_qcoeff_start_ptr, offsetof(DETOK, qcoeff_start_ptr)); +DEFINE(detok_current_bc, offsetof(DETOK, current_bc)); +DEFINE(detok_coef_probs, offsetof(DETOK, coef_probs)); DEFINE(detok_eob, offsetof(DETOK, eob)); -DEFINE(bool_decoder_lowvalue, offsetof(BOOL_DECODER, lowvalue)); -DEFINE(bool_decoder_range, offsetof(BOOL_DECODER, range)); +DEFINE(bool_decoder_user_buffer_end, offsetof(BOOL_DECODER, user_buffer_end)); +DEFINE(bool_decoder_user_buffer, offsetof(BOOL_DECODER, user_buffer)); DEFINE(bool_decoder_value, offsetof(BOOL_DECODER, value)); DEFINE(bool_decoder_count, offsetof(BOOL_DECODER, count)); -DEFINE(bool_decoder_user_buffer, offsetof(BOOL_DECODER, user_buffer)); -DEFINE(bool_decoder_user_buffer_sz, offsetof(BOOL_DECODER, user_buffer_sz)); -DEFINE(bool_decoder_decode_buffer, offsetof(BOOL_DECODER, decode_buffer)); -DEFINE(bool_decoder_read_ptr, offsetof(BOOL_DECODER, read_ptr)); -DEFINE(bool_decoder_write_ptr, offsetof(BOOL_DECODER, write_ptr)); +DEFINE(bool_decoder_range, offsetof(BOOL_DECODER, range)); -DEFINE(tokenextrabits_min_val, offsetof(TOKENEXTRABITS, min_val)); +DEFINE(tokenextrabits_min_val, offsetof(TOKENEXTRABITS, min_val)); DEFINE(tokenextrabits_length, offsetof(TOKENEXTRABITS, Length)); #endif diff --git a/vp8/common/bigend.h b/vp8/common/bigend.h index 6a91ba1..6ac3f8b 100644 --- a/vp8/common/bigend.h +++ b/vp8/common/bigend.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/blockd.c b/vp8/common/blockd.c index 53f5e72..7f75a72 100644 --- a/vp8/common/blockd.c +++ b/vp8/common/blockd.c @@ -1,23 +1,24 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ #include "blockd.h" #include "vpx_mem/vpx_mem.h" -void vp8_setup_temp_context(TEMP_CONTEXT *t, ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l, int count) -{ - vpx_memcpy(t->l, l, sizeof(ENTROPY_CONTEXT) * count); - vpx_memcpy(t->a, a, sizeof(ENTROPY_CONTEXT) * count); -} - -const int vp8_block2left[25] = { 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 0, 0, 1, 1, 0, 0, 1, 1, 0}; -const int vp8_block2above[25] = { 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 0, 1, 0, 1, 0, 1, 0}; const int vp8_block2type[25] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 2, 2, 2, 2, 2, 2, 2, 1}; -const int vp8_block2context[25] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3}; + +const unsigned char vp8_block2left[25] = +{ + 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8 +}; +const unsigned char vp8_block2above[25] = +{ + 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 4, 5, 6, 7, 6, 7, 8 +}; diff --git a/vp8/common/blockd.h b/vp8/common/blockd.h index 84ed53a..4b7f1a3 100644 --- a/vp8/common/blockd.h +++ b/vp8/common/blockd.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -48,19 +49,19 @@ typedef struct } POS; -typedef int ENTROPY_CONTEXT; - +typedef char ENTROPY_CONTEXT; typedef struct { - ENTROPY_CONTEXT l[4]; - ENTROPY_CONTEXT a[4]; -} TEMP_CONTEXT; + ENTROPY_CONTEXT y1[4]; + ENTROPY_CONTEXT u[2]; + ENTROPY_CONTEXT v[2]; + ENTROPY_CONTEXT y2; +} ENTROPY_CONTEXT_PLANES; -extern void vp8_setup_temp_context(TEMP_CONTEXT *t, ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l, int count); -extern const int vp8_block2left[25]; -extern const int vp8_block2above[25]; extern const int vp8_block2type[25]; -extern const int vp8_block2context[25]; + +extern const unsigned char vp8_block2left[25]; +extern const unsigned char vp8_block2above[25]; #define VP8_COMBINEENTROPYCONTEXTS( Dest, A, B) \ Dest = ((A)!=0) + ((B)!=0); @@ -167,15 +168,15 @@ typedef struct int as_int; MV as_mv; } mv; - int partitioning; - int partition_count; - int mb_skip_coeff; //does this mb has coefficients at all, 1=no coefficients, 0=need decode tokens - int dc_diff; - unsigned char segment_id; // Which set of segmentation parameters should be used for this MB - int force_no_skip; - B_MODE_INFO partition_bmi[16]; + char partitioning; + unsigned char mb_skip_coeff; //does this mb has coefficients at all, 1=no coefficients, 0=need decode tokens + unsigned char dc_diff; + unsigned char need_to_clamp_mvs; + unsigned char segment_id; // Which set of segmentation parameters should be used for this MB + + unsigned char force_no_skip; //encoder only } MB_MODE_INFO; @@ -215,9 +216,10 @@ typedef struct { DECLARE_ALIGNED(16, short, diff[400]); // from idct diff DECLARE_ALIGNED(16, unsigned char, predictor[384]); - DECLARE_ALIGNED(16, short, reference[384]); +//not used DECLARE_ALIGNED(16, short, reference[384]); DECLARE_ALIGNED(16, short, qcoeff[400]); DECLARE_ALIGNED(16, short, dqcoeff[400]); + DECLARE_ALIGNED(16, char, eobs[25]); // 16 Y blocks, 4 U, 4 V, 1 DC 2nd order block, each with 16 entries. BLOCKD block[25]; @@ -226,20 +228,16 @@ typedef struct YV12_BUFFER_CONFIG dst; MODE_INFO *mode_info_context; - MODE_INFO *mode_info; - int mode_info_stride; FRAME_TYPE frame_type; - MB_MODE_INFO mbmi; - int up_available; int left_available; // Y,U,V,Y2 - ENTROPY_CONTEXT *above_context[4]; // row of context for each plane - ENTROPY_CONTEXT(*left_context)[4]; // (up to) 4 contexts "" + ENTROPY_CONTEXT_PLANES *above_context; + ENTROPY_CONTEXT_PLANES *left_context; // 0 indicates segmentation at MB level is not enabled. Otherwise the individual bits indicate which features are active. unsigned char segmentation_enabled; @@ -275,9 +273,6 @@ typedef struct int mb_to_top_edge; int mb_to_bottom_edge; - //char * gf_active_ptr; - signed char *gf_active_ptr; - unsigned int frames_since_golden; unsigned int frames_till_alt_ref_frame; vp8_subpix_fn_t subpixel_predict; diff --git a/vp8/common/boolcoder.h b/vp8/common/boolcoder.h index 0659d48..5658868 100644 --- a/vp8/common/boolcoder.h +++ b/vp8/common/boolcoder.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/codec_common_interface.h b/vp8/common/codec_common_interface.h index 7881b0a..7a7db38 100644 --- a/vp8/common/codec_common_interface.h +++ b/vp8/common/codec_common_interface.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ #ifndef CODEC_COMMON_INTERFACE_H diff --git a/vp8/common/coefupdateprobs.h b/vp8/common/coefupdateprobs.h index 99affd6..785e3ff 100644 --- a/vp8/common/coefupdateprobs.h +++ b/vp8/common/coefupdateprobs.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/common.h b/vp8/common/common.h index 29f6d37..9a93da9 100644 --- a/vp8/common/common.h +++ b/vp8/common/common.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/common_types.h b/vp8/common/common_types.h index deb5ed8..4e62486 100644 --- a/vp8/common/common_types.h +++ b/vp8/common/common_types.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/context.c b/vp8/common/context.c index 17ee8c3..99e95d3 100644 --- a/vp8/common/context.c +++ b/vp8/common/context.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/debugmodes.c b/vp8/common/debugmodes.c index e2d2d2c..c3ac88f 100644 --- a/vp8/common/debugmodes.c +++ b/vp8/common/debugmodes.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/defaultcoefcounts.h b/vp8/common/defaultcoefcounts.h index ccdf326..b85f59b 100644 --- a/vp8/common/defaultcoefcounts.h +++ b/vp8/common/defaultcoefcounts.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/dma_desc.h b/vp8/common/dma_desc.h index 5e6fa0c..b923da6 100644 --- a/vp8/common/dma_desc.h +++ b/vp8/common/dma_desc.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/duck_io.h b/vp8/common/duck_io.h index f63a5cd..43daa65 100644 --- a/vp8/common/duck_io.h +++ b/vp8/common/duck_io.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/entropy.c b/vp8/common/entropy.c index e524c2a..1438e7e 100644 --- a/vp8/common/entropy.c +++ b/vp8/common/entropy.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/entropy.h b/vp8/common/entropy.h index 1415832..80b4817 100644 --- a/vp8/common/entropy.h +++ b/vp8/common/entropy.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/entropymode.c b/vp8/common/entropymode.c index 7dc1acd..e9dc668 100644 --- a/vp8/common/entropymode.c +++ b/vp8/common/entropymode.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -263,8 +264,10 @@ void vp8_entropy_mode_init() vp8_tokens_from_tree(vp8_uv_mode_encodings, vp8_uv_mode_tree); vp8_tokens_from_tree(vp8_mbsplit_encodings, vp8_mbsplit_tree); - vp8_tokens_from_tree(VP8_MVREFENCODINGS, vp8_mv_ref_tree); - vp8_tokens_from_tree(VP8_SUBMVREFENCODINGS, vp8_sub_mv_ref_tree); + vp8_tokens_from_tree_offset(vp8_mv_ref_encoding_array, + vp8_mv_ref_tree, NEARESTMV); + vp8_tokens_from_tree_offset(vp8_sub_mv_ref_encoding_array, + vp8_sub_mv_ref_tree, LEFT4X4); vp8_tokens_from_tree(vp8_small_mvencodings, vp8_small_mvtree); } diff --git a/vp8/common/entropymode.h b/vp8/common/entropymode.h index ff630a4..da6ae8e 100644 --- a/vp8/common/entropymode.h +++ b/vp8/common/entropymode.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -53,10 +54,6 @@ extern struct vp8_token_struct vp8_mbsplit_encodings [VP8_NUMMBSPLITS]; extern struct vp8_token_struct vp8_mv_ref_encoding_array [VP8_MVREFS]; extern struct vp8_token_struct vp8_sub_mv_ref_encoding_array [VP8_SUBMVREFS]; -#define VP8_MVREFENCODINGS (vp8_mv_ref_encoding_array - NEARESTMV) -#define VP8_SUBMVREFENCODINGS (vp8_sub_mv_ref_encoding_array - LEFT4X4) - - extern const vp8_tree_index vp8_small_mvtree[]; extern struct vp8_token_struct vp8_small_mvencodings [8]; diff --git a/vp8/common/entropymv.c b/vp8/common/entropymv.c index 2b00c17..8e72881 100644 --- a/vp8/common/entropymv.c +++ b/vp8/common/entropymv.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/entropymv.h b/vp8/common/entropymv.h index d940c59..911507d 100644 --- a/vp8/common/entropymv.h +++ b/vp8/common/entropymv.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/extend.c b/vp8/common/extend.c index 7407952..7e06ac3 100644 --- a/vp8/common/extend.c +++ b/vp8/common/extend.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -38,7 +39,10 @@ static void extend_plane_borders for (i = 0; i < h - 0 + 1; i++) { - vpx_memset(dest_ptr1, src_ptr1[0], el); + // Some linkers will complain if we call vpx_memset with el set to a + // constant 0. + if (el) + vpx_memset(dest_ptr1, src_ptr1[0], el); vpx_memset(dest_ptr2, src_ptr2[0], er); src_ptr1 += sp; src_ptr2 += sp; diff --git a/vp8/common/extend.h b/vp8/common/extend.h index 6809ae7..fd0a608 100644 --- a/vp8/common/extend.h +++ b/vp8/common/extend.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/filter_c.c b/vp8/common/filter_c.c index 38991cb..3d18d81 100644 --- a/vp8/common/filter_c.c +++ b/vp8/common/filter_c.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/findnearmv.c b/vp8/common/findnearmv.c index fcb1f20..41037f7 100644 --- a/vp8/common/findnearmv.c +++ b/vp8/common/findnearmv.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/findnearmv.h b/vp8/common/findnearmv.h index 2c02033..1a6c72b 100644 --- a/vp8/common/findnearmv.h +++ b/vp8/common/findnearmv.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/fourcc.hpp b/vp8/common/fourcc.hpp index 5f1faed..c582628 100644 --- a/vp8/common/fourcc.hpp +++ b/vp8/common/fourcc.hpp @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/g_common.h b/vp8/common/g_common.h index e68c53e..5f52398 100644 --- a/vp8/common/g_common.h +++ b/vp8/common/g_common.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/generic/systemdependent.c b/vp8/common/generic/systemdependent.c index 0011ae0..c04e31f 100644 --- a/vp8/common/generic/systemdependent.c +++ b/vp8/common/generic/systemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -31,7 +32,7 @@ void vp8_machine_specific_config(VP8_COMMON *ctx) rtcd->idct.idct1 = vp8_short_idct4x4llm_1_c; rtcd->idct.idct16 = vp8_short_idct4x4llm_c; - rtcd->idct.idct1_scalar = vp8_dc_only_idct_c; + rtcd->idct.idct1_scalar_add = vp8_dc_only_idct_add_c; rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_c; rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_c; @@ -60,7 +61,7 @@ void vp8_machine_specific_config(VP8_COMMON *ctx) rtcd->loopfilter.simple_mb_h = vp8_loop_filter_mbhs_c; rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_c; -#if CONFIG_POSTPROC || CONFIG_VP8_ENCODER +#if CONFIG_POSTPROC || (CONFIG_VP8_ENCODER && CONFIG_PSNR) rtcd->postproc.down = vp8_mbpost_proc_down_c; rtcd->postproc.across = vp8_mbpost_proc_across_ip_c; rtcd->postproc.downacross = vp8_post_proc_down_and_across_c; diff --git a/vp8/common/header.h b/vp8/common/header.h index 8b2b009..3e98eeb 100644 --- a/vp8/common/header.h +++ b/vp8/common/header.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/idct.h b/vp8/common/idct.h index 47b5f05..f5fd94d 100644 --- a/vp8/common/idct.h +++ b/vp8/common/idct.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -17,8 +18,10 @@ #define prototype_idct(sym) \ void sym(short *input, short *output, int pitch) -#define prototype_idct_scalar(sym) \ - void sym(short input, short *output, int pitch) +#define prototype_idct_scalar_add(sym) \ + void sym(short input, \ + unsigned char *pred, unsigned char *output, \ + int pitch, int stride) #if ARCH_X86 || ARCH_X86_64 #include "x86/idct_x86.h" @@ -38,10 +41,10 @@ extern prototype_idct(vp8_idct_idct1); #endif extern prototype_idct(vp8_idct_idct16); -#ifndef vp8_idct_idct1_scalar -#define vp8_idct_idct1_scalar vp8_dc_only_idct_c +#ifndef vp8_idct_idct1_scalar_add +#define vp8_idct_idct1_scalar_add vp8_dc_only_idct_add_c #endif -extern prototype_idct_scalar(vp8_idct_idct1_scalar); +extern prototype_idct_scalar_add(vp8_idct_idct1_scalar_add); #ifndef vp8_idct_iwalsh1 @@ -55,14 +58,14 @@ extern prototype_second_order(vp8_idct_iwalsh1); extern prototype_second_order(vp8_idct_iwalsh16); typedef prototype_idct((*vp8_idct_fn_t)); -typedef prototype_idct_scalar((*vp8_idct_scalar_fn_t)); +typedef prototype_idct_scalar_add((*vp8_idct_scalar_add_fn_t)); typedef prototype_second_order((*vp8_second_order_fn_t)); typedef struct { - vp8_idct_fn_t idct1; - vp8_idct_fn_t idct16; - vp8_idct_scalar_fn_t idct1_scalar; + vp8_idct_fn_t idct1; + vp8_idct_fn_t idct16; + vp8_idct_scalar_add_fn_t idct1_scalar_add; vp8_second_order_fn_t iwalsh1; vp8_second_order_fn_t iwalsh16; diff --git a/vp8/common/idctllm.c b/vp8/common/idctllm.c index 57cf858..196062d 100644 --- a/vp8/common/idctllm.c +++ b/vp8/common/idctllm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -103,23 +104,30 @@ void vp8_short_idct4x4llm_1_c(short *input, short *output, int pitch) } } - -void vp8_dc_only_idct_c(short input_dc, short *output, int pitch) +void vp8_dc_only_idct_add_c(short input_dc, unsigned char *pred_ptr, unsigned char *dst_ptr, int pitch, int stride) { - int i; - int a1; - short *op = output; - int shortpitch = pitch >> 1; - a1 = ((input_dc + 4) >> 3); + int a1 = ((input_dc + 4) >> 3); + int r, c; - for (i = 0; i < 4; i++) + for (r = 0; r < 4; r++) { - op[0] = a1; - op[1] = a1; - op[2] = a1; - op[3] = a1; - op += shortpitch; + for (c = 0; c < 4; c++) + { + int a = a1 + pred_ptr[c] ; + + if (a < 0) + a = 0; + + if (a > 255) + a = 255; + + dst_ptr[c] = (unsigned char) a ; + } + + dst_ptr += stride; + pred_ptr += pitch; } + } void vp8_short_inv_walsh4x4_c(short *input, short *output) diff --git a/vp8/common/invtrans.c b/vp8/common/invtrans.c index 1ff596e..4cb433a 100644 --- a/vp8/common/invtrans.c +++ b/vp8/common/invtrans.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -64,7 +65,8 @@ void vp8_inverse_transform_mb(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x { int i; - if (x->mbmi.mode != B_PRED && x->mbmi.mode != SPLITMV) + if (x->mode_info_context->mbmi.mode != B_PRED && + x->mode_info_context->mbmi.mode != SPLITMV) { // do 2nd order transform on the dc block diff --git a/vp8/common/invtrans.h b/vp8/common/invtrans.h index 93a40f9..b3ffb70 100644 --- a/vp8/common/invtrans.h +++ b/vp8/common/invtrans.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/littlend.h b/vp8/common/littlend.h index 08c525c..99df116 100644 --- a/vp8/common/littlend.h +++ b/vp8/common/littlend.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/loopfilter.c b/vp8/common/loopfilter.c index 79e6177..da9ca28 100644 --- a/vp8/common/loopfilter.c +++ b/vp8/common/loopfilter.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/loopfilter.h b/vp8/common/loopfilter.h index c6ce508..a2049bf 100644 --- a/vp8/common/loopfilter.h +++ b/vp8/common/loopfilter.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -116,5 +117,14 @@ typedef struct #define LF_INVOKE(ctx,fn) vp8_lf_##fn #endif +typedef void loop_filter_uvfunction +( + unsigned char *u, // source pointer + int p, // pitch + const signed char *flimit, + const signed char *limit, + const signed char *thresh, + unsigned char *v +); #endif diff --git a/vp8/common/loopfilter_filters.c b/vp8/common/loopfilter_filters.c index 7d16e48..ea82e2a 100644 --- a/vp8/common/loopfilter_filters.c +++ b/vp8/common/loopfilter_filters.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -17,7 +18,7 @@ typedef unsigned char uc; -__inline signed char vp8_signed_char_clamp(int t) +static __inline signed char vp8_signed_char_clamp(int t) { t = (t < -128 ? -128 : t); t = (t > 127 ? 127 : t); @@ -26,7 +27,7 @@ __inline signed char vp8_signed_char_clamp(int t) // should we apply any filter at all ( 11111111 yes, 00000000 no) -__inline signed char vp8_filter_mask(signed char limit, signed char flimit, +static __inline signed char vp8_filter_mask(signed char limit, signed char flimit, uc p3, uc p2, uc p1, uc p0, uc q0, uc q1, uc q2, uc q3) { signed char mask = 0; @@ -46,7 +47,7 @@ __inline signed char vp8_filter_mask(signed char limit, signed char flimit, } // is there high variance internal edge ( 11111111 yes, 00000000 no) -__inline signed char vp8_hevmask(signed char thresh, uc p1, uc p0, uc q0, uc q1) +static __inline signed char vp8_hevmask(signed char thresh, uc p1, uc p0, uc q0, uc q1) { signed char hev = 0; hev |= (abs(p1 - p0) > thresh) * -1; @@ -54,7 +55,7 @@ __inline signed char vp8_hevmask(signed char thresh, uc p1, uc p0, uc q0, uc q1) return hev; } -__inline void vp8_filter(signed char mask, signed char hev, uc *op1, uc *op0, uc *oq0, uc *oq1) +static __inline void vp8_filter(signed char mask, signed char hev, uc *op1, uc *op0, uc *oq0, uc *oq1) { signed char ps0, qs0; @@ -160,7 +161,7 @@ void vp8_loop_filter_vertical_edge_c while (++i < count * 8); } -__inline void vp8_mbfilter(signed char mask, signed char hev, +static __inline void vp8_mbfilter(signed char mask, signed char hev, uc *op2, uc *op1, uc *op0, uc *oq0, uc *oq1, uc *oq2) { signed char s, u; @@ -280,7 +281,7 @@ void vp8_mbloop_filter_vertical_edge_c } // should we apply any filter at all ( 11111111 yes, 00000000 no) -__inline signed char vp8_simple_filter_mask(signed char limit, signed char flimit, uc p1, uc p0, uc q0, uc q1) +static __inline signed char vp8_simple_filter_mask(signed char limit, signed char flimit, uc p1, uc p0, uc q0, uc q1) { // Why does this cause problems for win32? // error C2143: syntax error : missing ';' before 'type' @@ -293,7 +294,7 @@ __inline signed char vp8_simple_filter_mask(signed char limit, signed char flimi return mask; } -__inline void vp8_simple_filter(signed char mask, uc *op1, uc *op0, uc *oq0, uc *oq1) +static __inline void vp8_simple_filter(signed char mask, uc *op1, uc *op0, uc *oq0, uc *oq1) { signed char vp8_filter, Filter1, Filter2; signed char p1 = (signed char) * op1 ^ 0x80; diff --git a/vp8/common/mac_specs.h b/vp8/common/mac_specs.h index 97bffc7..4b8ee58 100644 --- a/vp8/common/mac_specs.h +++ b/vp8/common/mac_specs.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/mbpitch.c b/vp8/common/mbpitch.c index a7e0ce9..ce40d16 100644 --- a/vp8/common/mbpitch.c +++ b/vp8/common/mbpitch.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/modecont.c b/vp8/common/modecont.c index 9301a25..0fa2995 100644 --- a/vp8/common/modecont.c +++ b/vp8/common/modecont.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/modecont.h b/vp8/common/modecont.h index 0c57651..24db882 100644 --- a/vp8/common/modecont.h +++ b/vp8/common/modecont.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/modecontext.c b/vp8/common/modecontext.c index ceee74c..8e483b8 100644 --- a/vp8/common/modecontext.c +++ b/vp8/common/modecontext.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/mv.h b/vp8/common/mv.h index 3d84181..73c91b9 100644 --- a/vp8/common/mv.h +++ b/vp8/common/mv.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/onyx.h b/vp8/common/onyx.h index b66c400..a006306 100644 --- a/vp8/common/onyx.h +++ b/vp8/common/onyx.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -16,7 +17,7 @@ extern "C" { #endif -#include "vpx_codec/internal/vpx_codec_internal.h" +#include "vpx/internal/vpx_codec_internal.h" #include "vpx_scale/yv12config.h" #include "type_aliases.h" #include "ppflags.h" diff --git a/vp8/common/onyxc_int.h b/vp8/common/onyxc_int.h index a40ffb9..132765d 100644 --- a/vp8/common/onyxc_int.h +++ b/vp8/common/onyxc_int.h @@ -1,18 +1,19 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ #ifndef __INC_VP8C_INT_H #define __INC_VP8C_INT_H -#include "vpx_ports/config.h" -#include "vpx_codec/internal/vpx_codec_internal.h" +#include "vpx_config.h" +#include "vpx/internal/vpx_codec_internal.h" #include "loopfilter.h" #include "entropymv.h" #include "entropy.h" @@ -32,6 +33,7 @@ void vp8_initialize_common(void); #define MAXQ 127 #define QINDEX_RANGE (MAXQ + 1) +#define NUM_YV12_BUFFERS 4 typedef struct frame_contexts { @@ -93,15 +95,16 @@ typedef struct VP8Common YUV_TYPE clr_type; CLAMP_TYPE clamp_type; - YV12_BUFFER_CONFIG last_frame; - YV12_BUFFER_CONFIG golden_frame; - YV12_BUFFER_CONFIG alt_ref_frame; - YV12_BUFFER_CONFIG new_frame; YV12_BUFFER_CONFIG *frame_to_show; + + YV12_BUFFER_CONFIG yv12_fb[NUM_YV12_BUFFERS]; + int fb_idx_ref_cnt[NUM_YV12_BUFFERS]; + int new_fb_idx, lst_fb_idx, gld_fb_idx, alt_fb_idx; + YV12_BUFFER_CONFIG post_proc_buffer; YV12_BUFFER_CONFIG temp_scale_frame; - FRAME_TYPE last_frame_type; //Add to check if vp8_frame_init_loop_filter() can be skiped. + FRAME_TYPE last_frame_type; //Add to check if vp8_frame_init_loop_filter() can be skipped. FRAME_TYPE frame_type; int show_frame; @@ -130,8 +133,6 @@ typedef struct VP8Common unsigned int frames_since_golden; unsigned int frames_till_alt_ref_frame; - unsigned char *gf_active_flags; // Record of which MBs still refer to last golden frame either directly or through 0,0 - int gf_active_count; /* We allocate a MODE_INFO struct for each macroblock, together with an extra row on top and column on the left to simplify prediction. */ @@ -164,8 +165,8 @@ typedef struct VP8Common int ref_frame_sign_bias[MAX_REF_FRAMES]; // Two state 0, 1 // Y,U,V,Y2 - ENTROPY_CONTEXT *above_context[4]; // row of context for each plane - ENTROPY_CONTEXT left_context[4][4]; // (up to) 4 contexts "" + ENTROPY_CONTEXT_PLANES *above_context; // row of context for each plane + ENTROPY_CONTEXT_PLANES left_context; // (up to) 4 contexts "" // keyframe block modes are predicted by their above, left neighbors @@ -200,6 +201,7 @@ typedef struct VP8Common void vp8_adjust_mb_lf_value(MACROBLOCKD *mbd, int *filter_level); void vp8_init_loop_filter(VP8_COMMON *cm); +void vp8_frame_init_loop_filter(loop_filter_info *lfi, int frame_type); extern void vp8_loop_filter_frame(VP8_COMMON *cm, MACROBLOCKD *mbd, int filt_val); #endif diff --git a/vp8/common/onyxd.h b/vp8/common/onyxd.h index 644c0ec..00a97d9 100644 --- a/vp8/common/onyxd.h +++ b/vp8/common/onyxd.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/partialgfupdate.h b/vp8/common/partialgfupdate.h index 32a55ee..115134a 100644 --- a/vp8/common/partialgfupdate.h +++ b/vp8/common/partialgfupdate.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/postproc.c b/vp8/common/postproc.c index f019925..0c8cf13 100644 --- a/vp8/common/postproc.c +++ b/vp8/common/postproc.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -274,7 +275,7 @@ static void vp8_deblock_and_de_macro_block(YV12_BUFFER_CONFIG *source, } -extern void vp8_deblock(YV12_BUFFER_CONFIG *source, +void vp8_deblock(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *post, int q, int low_var_thresh, @@ -329,13 +330,6 @@ void vp8_de_noise(YV12_BUFFER_CONFIG *source, } - -//Notes: It is better to change CHAR to unsigned or signed to -//avoid error on ARM platform. -char vp8_an[8][64][3072]; -int vp8_cd[8][64]; - - double vp8_gaussian(double sigma, double mu, double x) { return 1 / (sigma * sqrt(2.0 * 3.14159265)) * diff --git a/vp8/common/postproc.h b/vp8/common/postproc.h index c45fe92..80337fc 100644 --- a/vp8/common/postproc.h +++ b/vp8/common/postproc.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -87,4 +88,11 @@ void vp8_de_noise(YV12_BUFFER_CONFIG *source, int low_var_thresh, int flag, vp8_postproc_rtcd_vtable_t *rtcd); + +void vp8_deblock(YV12_BUFFER_CONFIG *source, + YV12_BUFFER_CONFIG *post, + int q, + int low_var_thresh, + int flag, + vp8_postproc_rtcd_vtable_t *rtcd); #endif diff --git a/vp8/common/ppc/copy_altivec.asm b/vp8/common/ppc/copy_altivec.asm index e87eb21..a4ce915 100644 --- a/vp8/common/ppc/copy_altivec.asm +++ b/vp8/common/ppc/copy_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/ppc/filter_altivec.asm b/vp8/common/ppc/filter_altivec.asm index 2a35507..4da2e94 100644 --- a/vp8/common/ppc/filter_altivec.asm +++ b/vp8/common/ppc/filter_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/ppc/filter_bilinear_altivec.asm b/vp8/common/ppc/filter_bilinear_altivec.asm index 27e02a8..fd8aa66 100644 --- a/vp8/common/ppc/filter_bilinear_altivec.asm +++ b/vp8/common/ppc/filter_bilinear_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/ppc/idctllm_altivec.asm b/vp8/common/ppc/idctllm_altivec.asm index e88af8d..117d9cf 100644 --- a/vp8/common/ppc/idctllm_altivec.asm +++ b/vp8/common/ppc/idctllm_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/ppc/loopfilter_altivec.c b/vp8/common/ppc/loopfilter_altivec.c index 586eed4..bad3cf3 100644 --- a/vp8/common/ppc/loopfilter_altivec.c +++ b/vp8/common/ppc/loopfilter_altivec.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/ppc/loopfilter_filters_altivec.asm b/vp8/common/ppc/loopfilter_filters_altivec.asm index 78a5cf9..61df4e9 100644 --- a/vp8/common/ppc/loopfilter_filters_altivec.asm +++ b/vp8/common/ppc/loopfilter_filters_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/ppc/platform_altivec.asm b/vp8/common/ppc/platform_altivec.asm index 227ef2a..f81d86f 100644 --- a/vp8/common/ppc/platform_altivec.asm +++ b/vp8/common/ppc/platform_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/ppc/recon_altivec.asm b/vp8/common/ppc/recon_altivec.asm index f478b95..dd39e05 100644 --- a/vp8/common/ppc/recon_altivec.asm +++ b/vp8/common/ppc/recon_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/ppc/systemdependent.c b/vp8/common/ppc/systemdependent.c index 2847310..1f5d790 100644 --- a/vp8/common/ppc/systemdependent.c +++ b/vp8/common/ppc/systemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/ppflags.h b/vp8/common/ppflags.h index c663976..b1f925c 100644 --- a/vp8/common/ppflags.h +++ b/vp8/common/ppflags.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/pragmas.h b/vp8/common/pragmas.h index 25a4b77..99fee5a 100644 --- a/vp8/common/pragmas.h +++ b/vp8/common/pragmas.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/predictdc.c b/vp8/common/predictdc.c index df4c96e..f315f50 100644 --- a/vp8/common/predictdc.c +++ b/vp8/common/predictdc.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/predictdc.h b/vp8/common/predictdc.h index b8871e4..fa85968 100644 --- a/vp8/common/predictdc.h +++ b/vp8/common/predictdc.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/preproc.h b/vp8/common/preproc.h index 00ec9a8..0b142bd 100644 --- a/vp8/common/preproc.h +++ b/vp8/common/preproc.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/preprocif.h b/vp8/common/preprocif.h index 986c45b..7d554b5 100644 --- a/vp8/common/preprocif.h +++ b/vp8/common/preprocif.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/proposed.h b/vp8/common/proposed.h index 1171ede..c965990 100644 --- a/vp8/common/proposed.h +++ b/vp8/common/proposed.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/quant_common.c b/vp8/common/quant_common.c index 09fe31f..e9833fe 100644 --- a/vp8/common/quant_common.c +++ b/vp8/common/quant_common.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/quant_common.h b/vp8/common/quant_common.h index 0c92ce8..cb64d8e 100644 --- a/vp8/common/quant_common.h +++ b/vp8/common/quant_common.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/recon.c b/vp8/common/recon.c index d1268ea..0b439e0 100644 --- a/vp8/common/recon.c +++ b/vp8/common/recon.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/recon.h b/vp8/common/recon.h index f65a90f..e34a63c 100644 --- a/vp8/common/recon.h +++ b/vp8/common/recon.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/reconinter.c b/vp8/common/reconinter.c index c48886d..ffdc660 100644 --- a/vp8/common/reconinter.c +++ b/vp8/common/reconinter.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -209,7 +210,8 @@ void vp8_build_inter_predictors_mbuv(MACROBLOCKD *x) { int i; - if (x->mbmi.ref_frame != INTRA_FRAME && x->mbmi.mode != SPLITMV) + if (x->mode_info_context->mbmi.ref_frame != INTRA_FRAME && + x->mode_info_context->mbmi.mode != SPLITMV) { unsigned char *uptr, *vptr; unsigned char *upred_ptr = &x->predictor[256]; @@ -253,16 +255,18 @@ void vp8_build_inter_predictors_mbuv(MACROBLOCKD *x) } } - +//encoder only void vp8_build_inter_predictors_mby(MACROBLOCKD *x) { - if (x->mbmi.ref_frame != INTRA_FRAME && x->mbmi.mode != SPLITMV) + + if (x->mode_info_context->mbmi.ref_frame != INTRA_FRAME && + x->mode_info_context->mbmi.mode != SPLITMV) { unsigned char *ptr_base; unsigned char *ptr; unsigned char *pred_ptr = x->predictor; - int mv_row = x->mbmi.mv.as_mv.row; - int mv_col = x->mbmi.mv.as_mv.col; + int mv_row = x->mode_info_context->mbmi.mv.as_mv.row; + int mv_col = x->mode_info_context->mbmi.mv.as_mv.col; int pre_stride = x->block[0].pre_stride; ptr_base = x->pre.y_buffer; @@ -281,7 +285,7 @@ void vp8_build_inter_predictors_mby(MACROBLOCKD *x) { int i; - if (x->mbmi.partitioning < 3) + if (x->mode_info_context->mbmi.partitioning < 3) { for (i = 0; i < 4; i++) { @@ -312,7 +316,9 @@ void vp8_build_inter_predictors_mby(MACROBLOCKD *x) void vp8_build_inter_predictors_mb(MACROBLOCKD *x) { - if (x->mbmi.ref_frame != INTRA_FRAME && x->mbmi.mode != SPLITMV) + + if (x->mode_info_context->mbmi.ref_frame != INTRA_FRAME && + x->mode_info_context->mbmi.mode != SPLITMV) { int offset; unsigned char *ptr_base; @@ -322,8 +328,8 @@ void vp8_build_inter_predictors_mb(MACROBLOCKD *x) unsigned char *upred_ptr = &x->predictor[256]; unsigned char *vpred_ptr = &x->predictor[320]; - int mv_row = x->mbmi.mv.as_mv.row; - int mv_col = x->mbmi.mv.as_mv.col; + int mv_row = x->mode_info_context->mbmi.mv.as_mv.row; + int mv_col = x->mode_info_context->mbmi.mv.as_mv.col; int pre_stride = x->block[0].pre_stride; ptr_base = x->pre.y_buffer; @@ -360,7 +366,7 @@ void vp8_build_inter_predictors_mb(MACROBLOCKD *x) { int i; - if (x->mbmi.partitioning < 3) + if (x->mode_info_context->mbmi.partitioning < 3) { for (i = 0; i < 4; i++) { @@ -409,7 +415,7 @@ void vp8_build_uvmvs(MACROBLOCKD *x, int fullpixel) { int i, j; - if (x->mbmi.mode == SPLITMV) + if (x->mode_info_context->mbmi.mode == SPLITMV) { for (i = 0; i < 2; i++) { @@ -454,8 +460,8 @@ void vp8_build_uvmvs(MACROBLOCKD *x, int fullpixel) } else { - int mvrow = x->mbmi.mv.as_mv.row; - int mvcol = x->mbmi.mv.as_mv.col; + int mvrow = x->mode_info_context->mbmi.mv.as_mv.row; + int mvcol = x->mode_info_context->mbmi.mv.as_mv.col; if (mvrow < 0) mvrow -= 1; @@ -534,7 +540,7 @@ void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x) unsigned char *pred_ptr = x->predictor; unsigned char *dst_ptr = x->dst.y_buffer; - if (x->mbmi.mode != SPLITMV) + if (x->mode_info_context->mbmi.mode != SPLITMV) { int offset; unsigned char *ptr_base; @@ -546,8 +552,8 @@ void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x) unsigned char *udst_ptr = x->dst.u_buffer; unsigned char *vdst_ptr = x->dst.v_buffer; - int mv_row = x->mbmi.mv.as_mv.row; - int mv_col = x->mbmi.mv.as_mv.col; + int mv_row = x->mode_info_context->mbmi.mv.as_mv.row; + int mv_col = x->mode_info_context->mbmi.mv.as_mv.col; int pre_stride = x->dst.y_stride; //x->block[0].pre_stride; ptr_base = x->pre.y_buffer; @@ -586,7 +592,7 @@ void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x) //if sth is wrong, go back to what it is in build_inter_predictors_mb. int i; - if (x->mbmi.partitioning < 3) + if (x->mode_info_context->mbmi.partitioning < 3) { for (i = 0; i < 4; i++) { diff --git a/vp8/common/reconinter.h b/vp8/common/reconinter.h index b2d1ae9..7c1dee4 100644 --- a/vp8/common/reconinter.h +++ b/vp8/common/reconinter.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/reconintra.c b/vp8/common/reconintra.c index e33bce3..ce0b1b8 100644 --- a/vp8/common/reconintra.c +++ b/vp8/common/reconintra.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -42,7 +43,7 @@ void vp8_build_intra_predictors_mby(MACROBLOCKD *x) } // for Y - switch (x->mbmi.mode) + switch (x->mode_info_context->mbmi.mode) { case DC_PRED: { @@ -163,7 +164,7 @@ void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x) } // for Y - switch (x->mbmi.mode) + switch (x->mode_info_context->mbmi.mode) { case DC_PRED: { @@ -289,7 +290,7 @@ void vp8_build_intra_predictors_mbuv(MACROBLOCKD *x) vleft_col[i] = x->dst.v_buffer [i* x->dst.uv_stride -1]; } - switch (x->mbmi.uv_mode) + switch (x->mode_info_context->mbmi.uv_mode) { case DC_PRED: { @@ -429,7 +430,7 @@ void vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x) vleft_col[i] = x->dst.v_buffer [i* x->dst.uv_stride -1]; } - switch (x->mbmi.uv_mode) + switch (x->mode_info_context->mbmi.uv_mode) { case DC_PRED: { diff --git a/vp8/common/reconintra.h b/vp8/common/reconintra.h index d63aa15..988b43a 100644 --- a/vp8/common/reconintra.h +++ b/vp8/common/reconintra.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/reconintra4x4.c b/vp8/common/reconintra4x4.c index d92d5c9..c6e5fe7 100644 --- a/vp8/common/reconintra4x4.c +++ b/vp8/common/reconintra4x4.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/reconintra4x4.h b/vp8/common/reconintra4x4.h index 788c8c4..6ac2b71 100644 --- a/vp8/common/reconintra4x4.h +++ b/vp8/common/reconintra4x4.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/segmentation_common.h b/vp8/common/segmentation_common.h deleted file mode 100644 index bb93533..0000000 --- a/vp8/common/segmentation_common.h +++ /dev/null @@ -1,15 +0,0 @@ -/* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. - * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. - */ - - -#include "string.h" -#include "blockd.h" -#include "onyxc_int.h" - -extern void vp8_update_gf_useage_maps(VP8_COMMON *cm, MACROBLOCKD *xd); diff --git a/vp8/common/setupintrarecon.c b/vp8/common/setupintrarecon.c index dcaafe6..8647ae2 100644 --- a/vp8/common/setupintrarecon.c +++ b/vp8/common/setupintrarecon.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -16,21 +17,15 @@ void vp8_setup_intra_recon(YV12_BUFFER_CONFIG *ybf) int i; // set up frame new frame for intra coded blocks - vpx_memset(ybf->y_buffer - 1 - 2 * ybf->y_stride, 127, ybf->y_width + 5); vpx_memset(ybf->y_buffer - 1 - ybf->y_stride, 127, ybf->y_width + 5); - for (i = 0; i < ybf->y_height; i++) ybf->y_buffer[ybf->y_stride *i - 1] = (unsigned char) 129; - vpx_memset(ybf->u_buffer - 1 - 2 * ybf->uv_stride, 127, ybf->uv_width + 5); vpx_memset(ybf->u_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5); - for (i = 0; i < ybf->uv_height; i++) ybf->u_buffer[ybf->uv_stride *i - 1] = (unsigned char) 129; - vpx_memset(ybf->v_buffer - 1 - 2 * ybf->uv_stride, 127, ybf->uv_width + 5); vpx_memset(ybf->v_buffer - 1 - ybf->uv_stride, 127, ybf->uv_width + 5); - for (i = 0; i < ybf->uv_height; i++) ybf->v_buffer[ybf->uv_stride *i - 1] = (unsigned char) 129; diff --git a/vp8/common/setupintrarecon.h b/vp8/common/setupintrarecon.h index 6ec79b2..5264fd0 100644 --- a/vp8/common/setupintrarecon.h +++ b/vp8/common/setupintrarecon.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/subpixel.h b/vp8/common/subpixel.h index fbd5f4d..acdeec3 100644 --- a/vp8/common/subpixel.h +++ b/vp8/common/subpixel.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/swapyv12buffer.c b/vp8/common/swapyv12buffer.c index afe6a88..73656b3 100644 --- a/vp8/common/swapyv12buffer.c +++ b/vp8/common/swapyv12buffer.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/swapyv12buffer.h b/vp8/common/swapyv12buffer.h index caf9499..a6473ed 100644 --- a/vp8/common/swapyv12buffer.h +++ b/vp8/common/swapyv12buffer.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/systemdependent.h b/vp8/common/systemdependent.h index 1829b64..db99698 100644 --- a/vp8/common/systemdependent.h +++ b/vp8/common/systemdependent.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/textblit.c b/vp8/common/textblit.c index a45937b..da40f93 100644 --- a/vp8/common/textblit.c +++ b/vp8/common/textblit.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/threading.h b/vp8/common/threading.h index a02cb24..f9a2574 100644 --- a/vp8/common/threading.h +++ b/vp8/common/threading.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -74,7 +75,8 @@ #define thread_sleep(nms) // { struct timespec ts;ts.tv_sec=0; ts.tv_nsec = 1000*nms;nanosleep(&ts, NULL);} #else #include <unistd.h> -#define thread_sleep(nms) usleep(nms*1000);// {struct timespec ts;ts.tv_sec=0; ts.tv_nsec = 1000*nms;nanosleep(&ts, NULL);} +#include <sched.h> +#define thread_sleep(nms) sched_yield();// {struct timespec ts;ts.tv_sec=0; ts.tv_nsec = 1000*nms;nanosleep(&ts, NULL);} #endif /* Not Windows. Assume pthreads */ diff --git a/vp8/common/treecoder.c b/vp8/common/treecoder.c index 4ad018d..d80c64b 100644 --- a/vp8/common/treecoder.c +++ b/vp8/common/treecoder.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -46,6 +47,12 @@ void vp8_tokens_from_tree(struct vp8_token_struct *p, vp8_tree t) tree2tok(p, t, 0, 0, 0); } +void vp8_tokens_from_tree_offset(struct vp8_token_struct *p, vp8_tree t, + int offset) +{ + tree2tok(p - offset, t, 0, 0, 0); +} + static void branch_counts( int n, /* n = size of alphabet */ vp8_token tok [ /* n */ ], diff --git a/vp8/common/treecoder.h b/vp8/common/treecoder.h index 0356d2b..35e5be1 100644 --- a/vp8/common/treecoder.h +++ b/vp8/common/treecoder.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -53,6 +54,8 @@ typedef const struct vp8_token_struct /* Construct encoding array from tree. */ void vp8_tokens_from_tree(struct vp8_token_struct *, vp8_tree); +void vp8_tokens_from_tree_offset(struct vp8_token_struct *, vp8_tree, + int offset); /* Convert array of token occurrence counts into a table of probabilities diff --git a/vp8/common/type_aliases.h b/vp8/common/type_aliases.h index addd264..f2a3702 100644 --- a/vp8/common/type_aliases.h +++ b/vp8/common/type_aliases.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/vfwsetting.hpp b/vp8/common/vfwsetting.hpp index e352e7a..44869ec 100644 --- a/vp8/common/vfwsetting.hpp +++ b/vp8/common/vfwsetting.hpp @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/vpx_ref_build_prefix.h b/vp8/common/vpx_ref_build_prefix.h index 40608c6..a2fce65 100644 --- a/vp8/common/vpx_ref_build_prefix.h +++ b/vp8/common/vpx_ref_build_prefix.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/vpxblit.h b/vp8/common/vpxblit.h index d03e0bd..a95d905 100644 --- a/vp8/common/vpxblit.h +++ b/vp8/common/vpxblit.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/vpxblit_c64.h b/vp8/common/vpxblit_c64.h index a8e28f5..4ee617f 100644 --- a/vp8/common/vpxblit_c64.h +++ b/vp8/common/vpxblit_c64.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/vpxerrors.h b/vp8/common/vpxerrors.h index e4c9f3e..b70f296 100644 --- a/vp8/common/vpxerrors.h +++ b/vp8/common/vpxerrors.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/x86/boolcoder.cxx b/vp8/common/x86/boolcoder.cxx index 06faca6..faddf1f 100644 --- a/vp8/common/x86/boolcoder.cxx +++ b/vp8/common/x86/boolcoder.cxx @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/x86/idct_x86.h b/vp8/common/x86/idct_x86.h index 5dfb212..f6e568c 100644 --- a/vp8/common/x86/idct_x86.h +++ b/vp8/common/x86/idct_x86.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -21,7 +22,7 @@ #if HAVE_MMX extern prototype_idct(vp8_short_idct4x4llm_1_mmx); extern prototype_idct(vp8_short_idct4x4llm_mmx); -extern prototype_idct_scalar(vp8_dc_only_idct_mmx); +extern prototype_idct_scalar_add(vp8_dc_only_idct_add_mmx); extern prototype_second_order(vp8_short_inv_walsh4x4_mmx); extern prototype_second_order(vp8_short_inv_walsh4x4_1_mmx); @@ -33,8 +34,8 @@ extern prototype_second_order(vp8_short_inv_walsh4x4_1_mmx); #undef vp8_idct_idct16 #define vp8_idct_idct16 vp8_short_idct4x4llm_mmx -#undef vp8_idct_idct1_scalar -#define vp8_idct_idct1_scalar vp8_dc_only_idct_mmx +#undef vp8_idct_idct1_scalar_add +#define vp8_idct_idct1_scalar_add vp8_dc_only_idct_add_mmx #undef vp8_idct_iwalsh16 #define vp8_idct_iwalsh16 vp8_short_inv_walsh4x4_mmx diff --git a/vp8/common/x86/idctllm_mmx.asm b/vp8/common/x86/idctllm_mmx.asm index 2751c69..99e09a5 100644 --- a/vp8/common/x86/idctllm_mmx.asm +++ b/vp8/common/x86/idctllm_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -219,35 +220,61 @@ sym(vp8_short_idct4x4llm_1_mmx): pop rbp ret -;void dc_only_idct_mmx(short input_dc, short *output, int pitch) -global sym(vp8_dc_only_idct_mmx) -sym(vp8_dc_only_idct_mmx): +;void vp8_dc_only_idct_add_mmx(short input_dc, unsigned char *pred_ptr, unsigned char *dst_ptr, int pitch, int stride) +global sym(vp8_dc_only_idct_add_mmx) +sym(vp8_dc_only_idct_add_mmx): push rbp mov rbp, rsp - SHADOW_ARGS_TO_STACK 3 + SHADOW_ARGS_TO_STACK 5 GET_GOT rbx + push rsi + push rdi ; end prolog - movd mm0, arg(0) ;input_dc + mov rsi, arg(1) ;s -- prediction + mov rdi, arg(2) ;d -- destination + movsxd rax, dword ptr arg(4) ;stride + movsxd rdx, dword ptr arg(3) ;pitch + pxor mm0, mm0 - paddw mm0, [fours GLOBAL] - mov rdx, arg(1) ;output + movd mm5, arg(0) ;input_dc - psraw mm0, 3 - movsxd rax, dword ptr arg(2) ;pitch + paddw mm5, [fours GLOBAL] - punpcklwd mm0, mm0 - punpckldq mm0, mm0 + psraw mm5, 3 - movq [rdx], mm0 - movq [rdx+rax], mm0 + punpcklwd mm5, mm5 + punpckldq mm5, mm5 - movq [rdx+rax*2], mm0 - add rdx, rax + movd mm1, [rsi] + punpcklbw mm1, mm0 + paddsw mm1, mm5 + packuswb mm1, mm0 ; pack and unpack to saturate + movd [rdi], mm1 - movq [rdx+rax*2], mm0 + movd mm2, [rsi+rdx] + punpcklbw mm2, mm0 + paddsw mm2, mm5 + packuswb mm2, mm0 ; pack and unpack to saturate + movd [rdi+rax], mm2 + + movd mm3, [rsi+2*rdx] + punpcklbw mm3, mm0 + paddsw mm3, mm5 + packuswb mm3, mm0 ; pack and unpack to saturate + movd [rdi+2*rax], mm3 + + add rdi, rax + add rsi, rdx + movd mm4, [rsi+2*rdx] + punpcklbw mm4, mm0 + paddsw mm4, mm5 + packuswb mm4, mm0 ; pack and unpack to saturate + movd [rdi+2*rax], mm4 ; begin epilog + pop rdi + pop rsi RESTORE_GOT UNSHADOW_ARGS pop rbp diff --git a/vp8/common/x86/idctllm_sse2.asm b/vp8/common/x86/idctllm_sse2.asm new file mode 100644 index 0000000..ac94185 --- /dev/null +++ b/vp8/common/x86/idctllm_sse2.asm @@ -0,0 +1,708 @@ +; +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. +; +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. +; + + +%include "vpx_ports/x86_abi_support.asm" + +;void idct_dequant_0_2x_sse2 +; ( +; short *qcoeff - 0 +; short *dequant - 1 +; unsigned char *pre - 2 +; unsigned char *dst - 3 +; int dst_stride - 4 +; int blk_stride - 5 +; ) + +global sym(idct_dequant_0_2x_sse2) +sym(idct_dequant_0_2x_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + GET_GOT rbx + ; end prolog + + mov rdx, arg(1) ; dequant + mov rax, arg(0) ; qcoeff + + ; Zero out xmm7, for use unpacking + pxor xmm7, xmm7 + + movd xmm4, [rax] + movd xmm5, [rdx] + + pinsrw xmm4, [rax+32], 4 + pinsrw xmm5, [rdx], 4 + + pmullw xmm4, xmm5 + + ; clear coeffs + movd [rax], xmm7 + movd [rax+32], xmm7 +;pshufb + pshuflw xmm4, xmm4, 00000000b + pshufhw xmm4, xmm4, 00000000b + + mov rax, arg(2) ; pre + paddw xmm4, [fours GLOBAL] + + movsxd rcx, dword ptr arg(5) ; blk_stride + psraw xmm4, 3 + + movq xmm0, [rax] + movq xmm1, [rax+rcx] + movq xmm2, [rax+2*rcx] + lea rcx, [3*rcx] + movq xmm3, [rax+rcx] + + punpcklbw xmm0, xmm7 + punpcklbw xmm1, xmm7 + punpcklbw xmm2, xmm7 + punpcklbw xmm3, xmm7 + + mov rax, arg(3) ; dst + movsxd rdx, dword ptr arg(4) ; dst_stride + + ; Add to predict buffer + paddw xmm0, xmm4 + paddw xmm1, xmm4 + paddw xmm2, xmm4 + paddw xmm3, xmm4 + + ; pack up before storing + packuswb xmm0, xmm7 + packuswb xmm1, xmm7 + packuswb xmm2, xmm7 + packuswb xmm3, xmm7 + + ; store blocks back out + movq [rax], xmm0 + movq [rax + rdx], xmm1 + + lea rax, [rax + 2*rdx] + + movq [rax], xmm2 + movq [rax + rdx], xmm3 + + ; begin epilog + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +global sym(idct_dequant_full_2x_sse2) +sym(idct_dequant_full_2x_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 7 + GET_GOT rbx + push rsi + push rdi + ; end prolog + + ; special case when 2 blocks have 0 or 1 coeffs + ; dc is set as first coeff, so no need to load qcoeff + mov rax, arg(0) ; qcoeff + mov rsi, arg(2) ; pre + mov rdi, arg(3) ; dst + movsxd rcx, dword ptr arg(5) ; blk_stride + + ; Zero out xmm7, for use unpacking + pxor xmm7, xmm7 + + mov rdx, arg(1) ; dequant + + ; note the transpose of xmm1 and xmm2, necessary for shuffle + ; to spit out sensicle data + movdqa xmm0, [rax] + movdqa xmm2, [rax+16] + movdqa xmm1, [rax+32] + movdqa xmm3, [rax+48] + + ; Clear out coeffs + movdqa [rax], xmm7 + movdqa [rax+16], xmm7 + movdqa [rax+32], xmm7 + movdqa [rax+48], xmm7 + + ; dequantize qcoeff buffer + pmullw xmm0, [rdx] + pmullw xmm2, [rdx+16] + pmullw xmm1, [rdx] + pmullw xmm3, [rdx+16] + + ; repack so block 0 row x and block 1 row x are together + movdqa xmm4, xmm0 + punpckldq xmm0, xmm1 + punpckhdq xmm4, xmm1 + + pshufd xmm0, xmm0, 11011000b + pshufd xmm1, xmm4, 11011000b + + movdqa xmm4, xmm2 + punpckldq xmm2, xmm3 + punpckhdq xmm4, xmm3 + + pshufd xmm2, xmm2, 11011000b + pshufd xmm3, xmm4, 11011000b + + ; first pass + psubw xmm0, xmm2 ; b1 = 0-2 + paddw xmm2, xmm2 ; + + movdqa xmm5, xmm1 + paddw xmm2, xmm0 ; a1 = 0+2 + + pmulhw xmm5, [x_s1sqr2 GLOBAL] + paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2) + + movdqa xmm7, xmm3 + pmulhw xmm7, [x_c1sqr2less1 GLOBAL] + + paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2) + psubw xmm7, xmm5 ; c1 + + movdqa xmm5, xmm1 + movdqa xmm4, xmm3 + + pmulhw xmm5, [x_c1sqr2less1 GLOBAL] + paddw xmm5, xmm1 + + pmulhw xmm3, [x_s1sqr2 GLOBAL] + paddw xmm3, xmm4 + + paddw xmm3, xmm5 ; d1 + movdqa xmm6, xmm2 ; a1 + + movdqa xmm4, xmm0 ; b1 + paddw xmm2, xmm3 ;0 + + paddw xmm4, xmm7 ;1 + psubw xmm0, xmm7 ;2 + + psubw xmm6, xmm3 ;3 + + ; transpose for the second pass + movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000 + punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000 + punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100 + + movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008 + punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008 + punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108 + + + movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000 + punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000 + punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002 + + movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100 + punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100 + punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102 + + + movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000 + punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000 + punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001 + + movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002 + punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002 + punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003 + + pshufd xmm0, xmm2, 11011000b + pshufd xmm2, xmm1, 11011000b + + pshufd xmm1, xmm5, 11011000b + pshufd xmm3, xmm7, 11011000b + + ; second pass + psubw xmm0, xmm2 ; b1 = 0-2 + paddw xmm2, xmm2 + + movdqa xmm5, xmm1 + paddw xmm2, xmm0 ; a1 = 0+2 + + pmulhw xmm5, [x_s1sqr2 GLOBAL] + paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2) + + movdqa xmm7, xmm3 + pmulhw xmm7, [x_c1sqr2less1 GLOBAL] + + paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2) + psubw xmm7, xmm5 ; c1 + + movdqa xmm5, xmm1 + movdqa xmm4, xmm3 + + pmulhw xmm5, [x_c1sqr2less1 GLOBAL] + paddw xmm5, xmm1 + + pmulhw xmm3, [x_s1sqr2 GLOBAL] + paddw xmm3, xmm4 + + paddw xmm3, xmm5 ; d1 + paddw xmm0, [fours GLOBAL] + + paddw xmm2, [fours GLOBAL] + movdqa xmm6, xmm2 ; a1 + + movdqa xmm4, xmm0 ; b1 + paddw xmm2, xmm3 ;0 + + paddw xmm4, xmm7 ;1 + psubw xmm0, xmm7 ;2 + + psubw xmm6, xmm3 ;3 + psraw xmm2, 3 + + psraw xmm0, 3 + psraw xmm4, 3 + + psraw xmm6, 3 + + ; transpose to save + movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000 + punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000 + punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100 + + movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008 + punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008 + punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108 + + + movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000 + punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000 + punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002 + + movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100 + punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100 + punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102 + + + movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000 + punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000 + punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001 + + movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002 + punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002 + punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003 + + pshufd xmm0, xmm2, 11011000b + pshufd xmm2, xmm1, 11011000b + + pshufd xmm1, xmm5, 11011000b + pshufd xmm3, xmm7, 11011000b + + pxor xmm7, xmm7 + + ; Load up predict blocks + movq xmm4, [rsi] + movq xmm5, [rsi+rcx] + + punpcklbw xmm4, xmm7 + punpcklbw xmm5, xmm7 + + paddw xmm0, xmm4 + paddw xmm1, xmm5 + + movq xmm4, [rsi+2*rcx] + lea rcx, [3*rcx] + movq xmm5, [rsi+rcx] + + punpcklbw xmm4, xmm7 + punpcklbw xmm5, xmm7 + + paddw xmm2, xmm4 + paddw xmm3, xmm5 + +.finish: + + ; pack up before storing + packuswb xmm0, xmm7 + packuswb xmm1, xmm7 + packuswb xmm2, xmm7 + packuswb xmm3, xmm7 + + ; Load destination stride before writing out, + ; doesn't need to persist + movsxd rdx, dword ptr arg(4) ; dst_stride + + ; store blocks back out + movq [rdi], xmm0 + movq [rdi + rdx], xmm1 + + lea rdi, [rdi + 2*rdx] + + movq [rdi], xmm2 + movq [rdi + rdx], xmm3 + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +;void idct_dequant_dc_0_2x_sse2 +; ( +; short *qcoeff - 0 +; short *dequant - 1 +; unsigned char *pre - 2 +; unsigned char *dst - 3 +; int dst_stride - 4 +; short *dc - 5 +; ) +global sym(idct_dequant_dc_0_2x_sse2) +sym(idct_dequant_dc_0_2x_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 7 + GET_GOT rbx + push rsi + push rdi + ; end prolog + + ; special case when 2 blocks have 0 or 1 coeffs + ; dc is set as first coeff, so no need to load qcoeff + mov rax, arg(0) ; qcoeff + mov rsi, arg(2) ; pre + mov rdi, arg(3) ; dst + mov rdx, arg(5) ; dc + + ; Zero out xmm7, for use unpacking + pxor xmm7, xmm7 + + ; load up 2 dc words here == 2*16 = doubleword + movd xmm4, [rdx] + + ; Load up predict blocks + movq xmm0, [rsi] + movq xmm1, [rsi+16] + movq xmm2, [rsi+32] + movq xmm3, [rsi+48] + + ; Duplicate and expand dc across + punpcklwd xmm4, xmm4 + punpckldq xmm4, xmm4 + + ; Rounding to dequant and downshift + paddw xmm4, [fours GLOBAL] + psraw xmm4, 3 + + ; Predict buffer needs to be expanded from bytes to words + punpcklbw xmm0, xmm7 + punpcklbw xmm1, xmm7 + punpcklbw xmm2, xmm7 + punpcklbw xmm3, xmm7 + + ; Add to predict buffer + paddw xmm0, xmm4 + paddw xmm1, xmm4 + paddw xmm2, xmm4 + paddw xmm3, xmm4 + + ; pack up before storing + packuswb xmm0, xmm7 + packuswb xmm1, xmm7 + packuswb xmm2, xmm7 + packuswb xmm3, xmm7 + + ; Load destination stride before writing out, + ; doesn't need to persist + movsxd rdx, dword ptr arg(4) ; dst_stride + + ; store blocks back out + movq [rdi], xmm0 + movq [rdi + rdx], xmm1 + + lea rdi, [rdi + 2*rdx] + + movq [rdi], xmm2 + movq [rdi + rdx], xmm3 + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +global sym(idct_dequant_dc_full_2x_sse2) +sym(idct_dequant_dc_full_2x_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 7 + GET_GOT rbx + push rsi + push rdi + ; end prolog + + ; special case when 2 blocks have 0 or 1 coeffs + ; dc is set as first coeff, so no need to load qcoeff + mov rax, arg(0) ; qcoeff + mov rsi, arg(2) ; pre + mov rdi, arg(3) ; dst + + ; Zero out xmm7, for use unpacking + pxor xmm7, xmm7 + + mov rdx, arg(1) ; dequant + + ; note the transpose of xmm1 and xmm2, necessary for shuffle + ; to spit out sensicle data + movdqa xmm0, [rax] + movdqa xmm2, [rax+16] + movdqa xmm1, [rax+32] + movdqa xmm3, [rax+48] + + ; Clear out coeffs + movdqa [rax], xmm7 + movdqa [rax+16], xmm7 + movdqa [rax+32], xmm7 + movdqa [rax+48], xmm7 + + ; dequantize qcoeff buffer + pmullw xmm0, [rdx] + pmullw xmm2, [rdx+16] + pmullw xmm1, [rdx] + pmullw xmm3, [rdx+16] + + ; DC component + mov rdx, arg(5) + + ; repack so block 0 row x and block 1 row x are together + movdqa xmm4, xmm0 + punpckldq xmm0, xmm1 + punpckhdq xmm4, xmm1 + + pshufd xmm0, xmm0, 11011000b + pshufd xmm1, xmm4, 11011000b + + movdqa xmm4, xmm2 + punpckldq xmm2, xmm3 + punpckhdq xmm4, xmm3 + + pshufd xmm2, xmm2, 11011000b + pshufd xmm3, xmm4, 11011000b + + ; insert DC component + pinsrw xmm0, [rdx], 0 + pinsrw xmm0, [rdx+2], 4 + + ; first pass + psubw xmm0, xmm2 ; b1 = 0-2 + paddw xmm2, xmm2 ; + + movdqa xmm5, xmm1 + paddw xmm2, xmm0 ; a1 = 0+2 + + pmulhw xmm5, [x_s1sqr2 GLOBAL] + paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2) + + movdqa xmm7, xmm3 + pmulhw xmm7, [x_c1sqr2less1 GLOBAL] + + paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2) + psubw xmm7, xmm5 ; c1 + + movdqa xmm5, xmm1 + movdqa xmm4, xmm3 + + pmulhw xmm5, [x_c1sqr2less1 GLOBAL] + paddw xmm5, xmm1 + + pmulhw xmm3, [x_s1sqr2 GLOBAL] + paddw xmm3, xmm4 + + paddw xmm3, xmm5 ; d1 + movdqa xmm6, xmm2 ; a1 + + movdqa xmm4, xmm0 ; b1 + paddw xmm2, xmm3 ;0 + + paddw xmm4, xmm7 ;1 + psubw xmm0, xmm7 ;2 + + psubw xmm6, xmm3 ;3 + + ; transpose for the second pass + movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000 + punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000 + punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100 + + movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008 + punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008 + punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108 + + + movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000 + punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000 + punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002 + + movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100 + punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100 + punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102 + + + movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000 + punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000 + punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001 + + movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002 + punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002 + punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003 + + pshufd xmm0, xmm2, 11011000b + pshufd xmm2, xmm1, 11011000b + + pshufd xmm1, xmm5, 11011000b + pshufd xmm3, xmm7, 11011000b + + ; second pass + psubw xmm0, xmm2 ; b1 = 0-2 + paddw xmm2, xmm2 + + movdqa xmm5, xmm1 + paddw xmm2, xmm0 ; a1 = 0+2 + + pmulhw xmm5, [x_s1sqr2 GLOBAL] + paddw xmm5, xmm1 ; ip1 * sin(pi/8) * sqrt(2) + + movdqa xmm7, xmm3 + pmulhw xmm7, [x_c1sqr2less1 GLOBAL] + + paddw xmm7, xmm3 ; ip3 * cos(pi/8) * sqrt(2) + psubw xmm7, xmm5 ; c1 + + movdqa xmm5, xmm1 + movdqa xmm4, xmm3 + + pmulhw xmm5, [x_c1sqr2less1 GLOBAL] + paddw xmm5, xmm1 + + pmulhw xmm3, [x_s1sqr2 GLOBAL] + paddw xmm3, xmm4 + + paddw xmm3, xmm5 ; d1 + paddw xmm0, [fours GLOBAL] + + paddw xmm2, [fours GLOBAL] + movdqa xmm6, xmm2 ; a1 + + movdqa xmm4, xmm0 ; b1 + paddw xmm2, xmm3 ;0 + + paddw xmm4, xmm7 ;1 + psubw xmm0, xmm7 ;2 + + psubw xmm6, xmm3 ;3 + psraw xmm2, 3 + + psraw xmm0, 3 + psraw xmm4, 3 + + psraw xmm6, 3 + + ; transpose to save + movdqa xmm7, xmm2 ; 103 102 101 100 003 002 001 000 + punpcklwd xmm2, xmm0 ; 007 003 006 002 005 001 004 000 + punpckhwd xmm7, xmm0 ; 107 103 106 102 105 101 104 100 + + movdqa xmm5, xmm4 ; 111 110 109 108 011 010 009 008 + punpcklwd xmm4, xmm6 ; 015 011 014 010 013 009 012 008 + punpckhwd xmm5, xmm6 ; 115 111 114 110 113 109 112 108 + + + movdqa xmm1, xmm2 ; 007 003 006 002 005 001 004 000 + punpckldq xmm2, xmm4 ; 013 009 005 001 012 008 004 000 + punpckhdq xmm1, xmm4 ; 015 011 007 003 014 010 006 002 + + movdqa xmm6, xmm7 ; 107 103 106 102 105 101 104 100 + punpckldq xmm7, xmm5 ; 113 109 105 101 112 108 104 100 + punpckhdq xmm6, xmm5 ; 115 111 107 103 114 110 106 102 + + + movdqa xmm5, xmm2 ; 013 009 005 001 012 008 004 000 + punpckldq xmm2, xmm7 ; 112 108 012 008 104 100 004 000 + punpckhdq xmm5, xmm7 ; 113 109 013 009 105 101 005 001 + + movdqa xmm7, xmm1 ; 015 011 007 003 014 010 006 002 + punpckldq xmm1, xmm6 ; 114 110 014 010 106 102 006 002 + punpckhdq xmm7, xmm6 ; 115 111 015 011 107 103 007 003 + + pshufd xmm0, xmm2, 11011000b + pshufd xmm2, xmm1, 11011000b + + pshufd xmm1, xmm5, 11011000b + pshufd xmm3, xmm7, 11011000b + + pxor xmm7, xmm7 + + ; Load up predict blocks + movq xmm4, [rsi] + movq xmm5, [rsi+16] + + punpcklbw xmm4, xmm7 + punpcklbw xmm5, xmm7 + + paddw xmm0, xmm4 + paddw xmm1, xmm5 + + movq xmm4, [rsi+32] + movq xmm5, [rsi+48] + + punpcklbw xmm4, xmm7 + punpcklbw xmm5, xmm7 + + paddw xmm2, xmm4 + paddw xmm3, xmm5 + +.finish: + + ; pack up before storing + packuswb xmm0, xmm7 + packuswb xmm1, xmm7 + packuswb xmm2, xmm7 + packuswb xmm3, xmm7 + + ; Load destination stride before writing out, + ; doesn't need to persist + movsxd rdx, dword ptr arg(4) ; dst_stride + + ; store blocks back out + movq [rdi], xmm0 + movq [rdi + rdx], xmm1 + + lea rdi, [rdi + 2*rdx] + + movq [rdi], xmm2 + movq [rdi + rdx], xmm3 + + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +SECTION_RODATA +align 16 +fours: + times 8 dw 0x0004 +align 16 +x_s1sqr2: + times 8 dw 0x8A8C +align 16 +x_c1sqr2less1: + times 8 dw 0x4E7B diff --git a/vp8/common/x86/iwalsh_mmx.asm b/vp8/common/x86/iwalsh_mmx.asm index 562e590..3f0671c 100644 --- a/vp8/common/x86/iwalsh_mmx.asm +++ b/vp8/common/x86/iwalsh_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/x86/iwalsh_sse2.asm b/vp8/common/x86/iwalsh_sse2.asm index 96943df..83c97df 100644 --- a/vp8/common/x86/iwalsh_sse2.asm +++ b/vp8/common/x86/iwalsh_sse2.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -16,6 +17,7 @@ sym(vp8_short_inv_walsh4x4_sse2): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 2 + SAVE_XMM push rsi push rdi ; end prolog @@ -100,6 +102,7 @@ sym(vp8_short_inv_walsh4x4_sse2): ; begin epilog pop rdi pop rsi + RESTORE_XMM UNSHADOW_ARGS pop rbp ret diff --git a/vp8/common/x86/loopfilter_mmx.asm b/vp8/common/x86/loopfilter_mmx.asm index 6e4d2b6..0b39e62 100644 --- a/vp8/common/x86/loopfilter_mmx.asm +++ b/vp8/common/x86/loopfilter_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/x86/loopfilter_sse2.asm b/vp8/common/x86/loopfilter_sse2.asm index 5275dfa..57276b6 100644 --- a/vp8/common/x86/loopfilter_sse2.asm +++ b/vp8/common/x86/loopfilter_sse2.asm @@ -1,777 +1,147 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; %include "vpx_ports/x86_abi_support.asm" -;void vp8_loop_filter_horizontal_edge_sse2 -;( -; unsigned char *src_ptr, -; int src_pixel_step, -; const char *flimit, -; const char *limit, -; const char *thresh, -; int count -;) -global sym(vp8_loop_filter_horizontal_edge_sse2) -sym(vp8_loop_filter_horizontal_edge_sse2): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 6 - GET_GOT rbx - push rsi - push rdi - ; end prolog - - ALIGN_STACK 16, rax - sub rsp, 32 ; reserve 32 bytes - %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16]; - %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16]; - - mov rsi, arg(0) ;src_ptr - movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch? - - mov rdx, arg(3) ;limit - movdqa xmm7, XMMWORD PTR [rdx] - mov rdi, rsi ; rdi points to row +1 for indirect addressing - add rdi, rax - - ; calculate breakout conditions - movdqu xmm2, [rdi+2*rax] ; q3 - movdqu xmm1, [rsi+2*rax] ; q2 - movdqa xmm6, xmm1 ; q2 - psubusb xmm1, xmm2 ; q2-=q3 - psubusb xmm2, xmm6 ; q3-=q2 - por xmm1, xmm2 ; abs(q3-q2) - psubusb xmm1, xmm7 ; - - - movdqu xmm4, [rsi+rax] ; q1 - movdqa xmm3, xmm4 ; q1 - psubusb xmm4, xmm6 ; q1-=q2 - psubusb xmm6, xmm3 ; q2-=q1 - por xmm4, xmm6 ; abs(q2-q1) - - psubusb xmm4, xmm7 - por xmm1, xmm4 - - movdqu xmm4, [rsi] ; q0 - movdqa xmm0, xmm4 ; q0 - psubusb xmm4, xmm3 ; q0-=q1 - psubusb xmm3, xmm0 ; q1-=q0 - por xmm4, xmm3 ; abs(q0-q1) - movdqa t0, xmm4 ; save to t0 - psubusb xmm4, xmm7 - por xmm1, xmm4 - - neg rax ; negate pitch to deal with above border - movdqu xmm2, [rsi+4*rax] ; p3 - movdqu xmm4, [rdi+4*rax] ; p2 - movdqa xmm5, xmm4 ; p2 - psubusb xmm4, xmm2 ; p2-=p3 - psubusb xmm2, xmm5 ; p3-=p2 - por xmm4, xmm2 ; abs(p3 - p2) - psubusb xmm4, xmm7 - por xmm1, xmm4 - - - movdqu xmm4, [rsi+2*rax] ; p1 - movdqa xmm3, xmm4 ; p1 - psubusb xmm4, xmm5 ; p1-=p2 - psubusb xmm5, xmm3 ; p2-=p1 - por xmm4, xmm5 ; abs(p2 - p1) - psubusb xmm4, xmm7 - por xmm1, xmm4 - - movdqa xmm2, xmm3 ; p1 - - movdqu xmm4, [rsi+rax] ; p0 - movdqa xmm5, xmm4 ; p0 - psubusb xmm4, xmm3 ; p0-=p1 - psubusb xmm3, xmm5 ; p1-=p0 - por xmm4, xmm3 ; abs(p1 - p0) - movdqa t1, xmm4 ; save to t1 - psubusb xmm4, xmm7 - por xmm1, xmm4 - - movdqu xmm3, [rdi] ; q1 - movdqa xmm4, xmm3 ; q1 - psubusb xmm3, xmm2 ; q1-=p1 - psubusb xmm2, xmm4 ; p1-=q1 - por xmm2, xmm3 ; abs(p1-q1) - pand xmm2, [tfe GLOBAL] ; set lsb of each byte to zero - psrlw xmm2, 1 ; abs(p1-q1)/2 - - movdqa xmm6, xmm5 ; p0 - movdqu xmm3, [rsi] ; q0 - psubusb xmm5, xmm3 ; p0-=q0 - psubusb xmm3, xmm6 ; q0-=p0 - por xmm5, xmm3 ; abs(p0 - q0) - paddusb xmm5, xmm5 ; abs(p0-q0)*2 - paddusb xmm5, xmm2 ; abs (p0 - q0) *2 + abs(p1-q1)/2 - - mov rdx, arg(2) ;flimit ; get flimit - movdqa xmm2, [rdx] ; - - paddb xmm2, xmm2 ; flimit*2 (less than 255) - paddb xmm7, xmm2 ; flimit * 2 + limit (less than 255) - - psubusb xmm5, xmm7 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit - por xmm1, xmm5 - pxor xmm5, xmm5 - pcmpeqb xmm1, xmm5 ; mask mm1 - - - ; calculate high edge variance - mov rdx, arg(4) ;thresh ; get thresh - movdqa xmm7, [rdx] ; - movdqa xmm4, t0 ; get abs (q1 - q0) - psubusb xmm4, xmm7 - movdqa xmm3, t1 ; get abs (p1 - p0) - psubusb xmm3, xmm7 - paddb xmm4, xmm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh - pcmpeqb xmm4, xmm5 - pcmpeqb xmm5, xmm5 - pxor xmm4, xmm5 - - - ; start work on filters - movdqu xmm2, [rsi+2*rax] ; p1 - movdqu xmm7, [rdi] ; q1 - pxor xmm2, [t80 GLOBAL] ; p1 offset to convert to signed values - pxor xmm7, [t80 GLOBAL] ; q1 offset to convert to signed values - psubsb xmm2, xmm7 ; p1 - q1 - pand xmm2, xmm4 ; high var mask (hvm)(p1 - q1) - pxor xmm6, [t80 GLOBAL] ; offset to convert to signed values - pxor xmm0, [t80 GLOBAL] ; offset to convert to signed values - movdqa xmm3, xmm0 ; q0 - psubsb xmm0, xmm6 ; q0 - p0 - paddsb xmm2, xmm0 ; 1 * (q0 - p0) + hvm(p1 - q1) - paddsb xmm2, xmm0 ; 2 * (q0 - p0) + hvm(p1 - q1) - paddsb xmm2, xmm0 ; 3 * (q0 - p0) + hvm(p1 - q1) - pand xmm1, xmm2 ; mask filter values we don't care about - movdqa xmm2, xmm1 - paddsb xmm1, [t4 GLOBAL] ; 3* (q0 - p0) + hvm(p1 - q1) + 4 - paddsb xmm2, [t3 GLOBAL] ; 3* (q0 - p0) + hvm(p1 - q1) + 3 - - pxor xmm0, xmm0 ; - pxor xmm5, xmm5 - punpcklbw xmm0, xmm2 ; - punpckhbw xmm5, xmm2 ; - psraw xmm0, 11 ; - psraw xmm5, 11 - packsswb xmm0, xmm5 - movdqa xmm2, xmm0 ; (3* (q0 - p0) + hvm(p1 - q1) + 3) >> 3; - - pxor xmm0, xmm0 ; 0 - movdqa xmm5, xmm1 ; abcdefgh - punpcklbw xmm0, xmm1 ; e0f0g0h0 - psraw xmm0, 11 ; sign extended shift right by 3 - pxor xmm1, xmm1 ; 0 - punpckhbw xmm1, xmm5 ; a0b0c0d0 - psraw xmm1, 11 ; sign extended shift right by 3 - movdqa xmm5, xmm0 ; save results - - packsswb xmm0, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>3 - paddsw xmm5, [ones GLOBAL] - paddsw xmm1, [ones GLOBAL] - psraw xmm5, 1 ; partial shifted one more time for 2nd tap - psraw xmm1, 1 ; partial shifted one more time for 2nd tap - packsswb xmm5, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>4 - pandn xmm4, xmm5 ; high edge variance additive - - paddsb xmm6, xmm2 ; p0+= p0 add - pxor xmm6, [t80 GLOBAL] ; unoffset - movdqu [rsi+rax], xmm6 ; write back - - movdqu xmm6, [rsi+2*rax] ; p1 - pxor xmm6, [t80 GLOBAL] ; reoffset - paddsb xmm6, xmm4 ; p1+= p1 add - pxor xmm6, [t80 GLOBAL] ; unoffset - movdqu [rsi+2*rax], xmm6 ; write back - - psubsb xmm3, xmm0 ; q0-= q0 add - pxor xmm3, [t80 GLOBAL] ; unoffset - movdqu [rsi], xmm3 ; write back - - psubsb xmm7, xmm4 ; q1-= q1 add - pxor xmm7, [t80 GLOBAL] ; unoffset - movdqu [rdi], xmm7 ; write back - - add rsp, 32 - pop rsp - ; begin epilog - pop rdi - pop rsi - RESTORE_GOT - UNSHADOW_ARGS - pop rbp - ret - - -;void vp8_loop_filter_vertical_edge_sse2 -;( -; unsigned char *src_ptr, -; int src_pixel_step, -; const char *flimit, -; const char *limit, -; const char *thresh, -; int count -;) -global sym(vp8_loop_filter_vertical_edge_sse2) -sym(vp8_loop_filter_vertical_edge_sse2): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 6 - GET_GOT rbx - push rsi - push rdi - ; end prolog - - ALIGN_STACK 16, rax - sub rsp, 96 ; reserve 96 bytes - %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16]; - %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16]; - %define srct [rsp + 32] ;__declspec(align(16)) char srct[64]; - - mov rsi, arg(0) ;src_ptr - movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch? - - lea rsi, [rsi + rax*4 - 4] - mov rdi, rsi ; rdi points to row +1 for indirect addressing - - add rdi, rax - lea rcx, [rdi + rax *8] - - ;transpose - movq xmm7, QWORD PTR [rsi+2*rax] ; 67 66 65 64 63 62 61 60 - movq xmm6, QWORD PTR [rdi+2*rax] ; 77 76 75 74 73 72 71 70 - - punpcklbw xmm7, xmm6 ; 77 67 76 66 75 65 74 64 73 63 72 62 71 61 70 60 - movq xmm5, QWORD PTR [rsi] ; 47 46 45 44 43 42 41 40 - - movq xmm4, QWORD PTR [rsi+rax] ; 57 56 55 54 53 52 51 50 - punpcklbw xmm5, xmm4 ; 57 47 56 46 55 45 54 44 53 43 52 42 51 41 50 40 - - movdqa xmm3, xmm5 ; 57 47 56 46 55 45 54 44 53 43 52 42 51 41 50 40 - punpckhwd xmm5, xmm7 ; 77 67 57 47 76 66 56 46 75 65 55 45 74 64 54 44 - - lea rsi, [rsi+ rax*8] - - punpcklwd xmm3, xmm7 ; 73 63 53 43 72 62 52 42 71 61 51 41 70 60 50 40 - movq xmm6, QWORD PTR [rsi + 2*rax] ; e7 e6 e5 e4 e3 e2 e1 e0 - - movq xmm7, QWORD PTR [rcx + 2*rax] ; f7 f6 f5 f4 f3 f2 f1 f0 - punpcklbw xmm6, xmm7 ; f7 e7 f6 e6 f5 e5 f4 e4 f3 e3 f2 e2 f1 e1 f0 e0 - - movq xmm4, QWORD PTR [rsi] ; c7 c6 c5 c4 c3 c2 c1 c0 - movq xmm7, QWORD PTR [rsi + rax] ; d7 d6 d5 d4 d3 d2 d1 d0 - - punpcklbw xmm4, xmm7 ; d7 c7 d6 c6 d5 c5 d4 c4 d3 c3 d2 c2 d1 c1 d0 c0 - movdqa xmm7, xmm4 ; d7 c7 d6 c6 d5 c5 d4 c4 d3 c3 d2 c2 d1 c1 d0 c0 - - punpckhwd xmm7, xmm6 ; f7 e7 d7 c7 f6 e6 d6 c6 f5 e5 d5 c5 f4 e4 d4 c4 - punpcklwd xmm4, xmm6 ; f3 e3 d3 c3 f2 e2 d2 c2 f1 e1 d1 c1 f0 e0 d0 c0 - - ; xmm3 xmm4, xmm5 xmm7 in use - neg rax - - lea rsi, [rsi+rax*8] - movq xmm6, QWORD PTR [rsi+rax*2] ; 27 26 25 24 23 22 21 20 - - movq xmm1, QWORD PTR [rsi+rax ] ; 37 36 35 34 33 32 31 30 - punpcklbw xmm6, xmm1 ; 37 27 36 26 35 25 34 24 33 23 32 22 31 21 30 20 - - movq xmm2, QWORD PTR [rsi+rax*4] ; 07 06 05 04 03 02 01 00 - movq xmm1, QWORD PTR [rdi+rax*4] ; 17 16 15 14 13 12 11 10 - - punpcklbw xmm2, xmm1 ; 17 07 16 06 15 05 14 04 13 03 12 02 11 01 10 00 - movdqa xmm0, xmm2 - - punpckhwd xmm2, xmm6 ; 37 27 17 07 36 26 16 06 35 25 15 05 34 24 14 04 - punpcklwd xmm0, xmm6 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00 - - movdqa xmm6, xmm2 - punpckldq xmm2, xmm5 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04 - - punpckhdq xmm6, xmm5 ; 77 67 57 47 37 27 17 07 76 66 56 46 36 26 16 06 - ;xmm0 xmm2 xmm3 xmm4, xmm6, xmm7 - - movdqa xmm5, xmm0 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00 - punpckhdq xmm5, xmm3 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02 - - punpckldq xmm0, xmm3 ; 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00 - lea rsi, [rcx+rax] - ; xmm1, xmm3 free - movq xmm1, QWORD PTR [rsi+rax*2] ; a7 a6 a5 a4 a3 a2 a1 a0 - movq xmm3, QWORD PTR [rsi+rax] ; b7 b6 b5 b4 b3 b2 b1 b0 - - punpcklbw xmm1, xmm3 ; - lea rdx, srct ; - - movdqa [rdx+16], xmm1 ; b7 a7 b6 a6 b5 a5 b4 a4 b3 a3 b2 a2 b1 a1 b0 a0 - movq xmm3, QWORD PTR [rsi+rax*4] ; 87 86 85 84 83 82 81 80 - - movq xmm1, QWORD PTR [rcx+rax*4] - punpcklbw xmm3, xmm1 ; 97 87 96 86 95 85 94 84 93 83 92 82 91 81 90 80 - - movdqa [rdx], xmm3 ; 97 87 96 86 95 85 94 84 93 83 92 82 91 81 90 80 - - punpckhwd xmm3, [rdx+16] ; b7 a7 97 87 b6 a6 96 86 b5 a5 95 85 b4 a4 94 84 - movdqa xmm1, xmm3 ; b7 a7 97 87 b6 a6 96 86 b5 a5 95 85 b4 a4 94 84 - - punpckhdq xmm1, xmm7 ; f7 e7 d7 c7 b7 a7 97 87 f6 e6 d6 c6 b6 a6 96 86 - punpckldq xmm3, xmm7 ; f5 e5 d5 c5 b5 a5 95 85 f4 e4 d4 c4 b4 a4 94 84 - - movdqa xmm7, xmm2 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04 - punpcklqdq xmm7, xmm3 ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 - - punpckhqdq xmm2, xmm3 ; f5 e5 d5 c5 b5 a5 95 85 75 65 55 45 35 25 15 05 - movdqa [rdx+32], xmm7 ; save 4s - - movdqa [rdx+48], xmm2 ; save 5s - movdqa xmm7, xmm6 ; 77 67 57 47 37 27 17 07 76 66 56 46 36 26 16 06 - - punpckhqdq xmm7, xmm1 ; f7 e7 d7 c7 b7 a7 97 87 77 67 57 47 37 27 17 07 = q3 - punpcklqdq xmm6, xmm1 ; f6 e6 d6 c6 b6 a6 96 86 76 66 56 46 36 26 16 06 = q2 - - ; free 1, 3 xmm7-7s xmm6-6s, xmm2-5s - movq xmm1, QWORD PTR [rdx] ; 93 83 92 82 91 81 90 80 - movq xmm3, QWORD PTR [rdx+16] ; b3 a3 b2 a2 b1 a1 b0 a0 - - punpcklwd xmm1, xmm3 ; b3 a3 93 83 b2 a2 92 82 b1 a1 91 81 b0 a0 90 80 - movdqa xmm3, xmm1 ; b3 a3 93 83 b2 a2 92 82 b1 a1 91 81 b0 a0 90 80 - - punpckhdq xmm3, xmm4 ; f3 e3 d3 c3 b3 a3 93 83 f2 e2 d2 c2 b2 a2 92 82 - punpckldq xmm1, xmm4 ; f1 e1 d1 c1 b1 a1 91 81 f0 e0 d0 c0 b0 a0 90 80 - - movdqa xmm4, xmm5 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02 - punpcklqdq xmm5, xmm3 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 - - punpckhqdq xmm4, xmm3 ; f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03 - movdqa [rdx], xmm5 ; save 2s - - movdqa [rdx+16], xmm4 ; save 3s - - movdqa xmm3, xmm6 ; - psubusb xmm3, xmm7 ; q3 - q2 - - psubusb xmm7, xmm6 ; q2 - q3 - por xmm7, xmm3 ; abs(q3-q2) - - movdqa xmm3, xmm2 ; q1 - psubusb xmm3, xmm6 ; q1 - q2 - - psubusb xmm6, xmm2 ; q2 - q1 - por xmm6, xmm3 ; abs(q2-q1) - - - movdqa xmm3, xmm0 ; 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00 - punpcklqdq xmm0, xmm1 ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00 - - punpckhqdq xmm3, xmm1 ; f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01 - movdqa xmm1, xmm3 - - psubusb xmm3, xmm0 ; p2-p3 - psubusb xmm0, xmm1 ; p3-p2 - - por xmm0, xmm3 ; abs(p3-p2) - movdqa xmm3, xmm5 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 - - psubusb xmm3, xmm1 ; p1-p2 - psubusb xmm1, xmm5 ; p2-p1 - - por xmm1, xmm3 ; abs(p1-p2) - mov rdx, arg(3) ;limit - - movdqa xmm3, [rdx] ; limit - - psubusb xmm7, xmm3 - psubusb xmm0, xmm3 - - psubusb xmm1, xmm3 - psubusb xmm6, xmm3 - - por xmm7, xmm6 - por xmm0, xmm1 - - por xmm0, xmm7 ; abs(q3-q2) > limit || abs(p3-p2) > limit ||abs(p2-p1) > limit || abs(q2-q1) > limit - - movdqa xmm1, xmm5 ; p1 - - movdqa xmm7, xmm4 ; xmm4 xmm7 = p0 - - psubusb xmm7, xmm5 ; p0 - p1 - psubusb xmm5, xmm4 ; p1 - p0 - - por xmm5, xmm7 ; abs(p1-p0) - movdqa t0, xmm5 ; save abs(p1-p0) - - lea rdx, srct - psubusb xmm5, xmm3 - - por xmm0, xmm5 ; xmm0=mask - movdqa xmm5, [rdx+32] ; xmm5=q0 - - movdqa xmm7, [rdx+48] ; xmm7=q1 - movdqa xmm6, xmm5 ; mm6=q0 - - movdqa xmm2, xmm7 ; q1 - - psubusb xmm5, xmm7 ; q0-q1 - psubusb xmm7, xmm6 ; q1-q0 - - por xmm7, xmm5 ; abs(q1-q0) - movdqa t1, xmm7 ; save abs(q1-q0) - - psubusb xmm7, xmm3 - por xmm0, xmm7 ; mask - - movdqa xmm5, xmm2 ; q1 - psubusb xmm5, xmm1 ; q1-=p1 - psubusb xmm1, xmm2 ; p1-=q1 - por xmm5, xmm1 ; abs(p1-q1) - pand xmm5, [tfe GLOBAL] ; set lsb of each byte to zero - psrlw xmm5, 1 ; abs(p1-q1)/2 - - mov rdx, arg(2) ;flimit ; - movdqa xmm2, [rdx] ;flimit xmm2 - - movdqa xmm1, xmm4 ; xmm1=xmm4=p0 - - movdqa xmm7, xmm6 ; xmm7=xmm6=q0 - psubusb xmm1, xmm7 ; p0-q0 - - psubusb xmm7, xmm4 ; q0-p0 - por xmm1, xmm7 ; abs(q0-p0) - paddusb xmm1, xmm1 ; abs(q0-p0)*2 - paddusb xmm1, xmm5 ; abs (p0 - q0) *2 + abs(p1-q1)/2 - - paddb xmm2, xmm2 ; flimit*2 (less than 255) - paddb xmm3, xmm2 ; flimit * 2 + limit (less than 255) - - psubusb xmm1, xmm3 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit - - por xmm1, xmm0; ; mask - - pxor xmm0, xmm0 - pcmpeqb xmm1, xmm0 - ; calculate high edge variance - mov rdx, arg(4) ;thresh ; get thresh - movdqa xmm7, [rdx] - - ; - movdqa xmm4, t0 ; get abs (q1 - q0) - psubusb xmm4, xmm7 - - movdqa xmm3, t1 ; get abs (p1 - p0) - psubusb xmm3, xmm7 - - por xmm4, xmm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh - pcmpeqb xmm4, xmm0 - - pcmpeqb xmm0, xmm0 - pxor xmm4, xmm0 - - ; start work on filters - lea rdx, srct - - movdqa xmm2, [rdx] ; p1 - movdqa xmm7, [rdx+48] ; q1 - - movdqa xmm6, [rdx+16] ; p0 - movdqa xmm0, [rdx+32] ; q0 - - pxor xmm2, [t80 GLOBAL] ; p1 offset to convert to signed values - pxor xmm7, [t80 GLOBAL] ; q1 offset to convert to signed values - - psubsb xmm2, xmm7 ; p1 - q1 - pand xmm2, xmm4 ; high var mask (hvm)(p1 - q1) - - pxor xmm6, [t80 GLOBAL] ; offset to convert to signed values - pxor xmm0, [t80 GLOBAL] ; offset to convert to signed values - - movdqa xmm3, xmm0 ; q0 - psubsb xmm0, xmm6 ; q0 - p0 - - paddsb xmm2, xmm0 ; 1 * (q0 - p0) + hvm(p1 - q1) - paddsb xmm2, xmm0 ; 2 * (q0 - p0) + hvm(p1 - q1) - - paddsb xmm2, xmm0 ; 3 * (q0 - p0) + hvm(p1 - q1) - pand xmm1, xmm2 ; mask filter values we don't care about - - movdqa xmm2, xmm1 - paddsb xmm1, [t4 GLOBAL] ; 3* (q0 - p0) + hvm(p1 - q1) + 4 - - paddsb xmm2, [t3 GLOBAL] ; 3* (q0 - p0) + hvm(p1 - q1) + 3 - pxor xmm0, xmm0 ; - - pxor xmm5, xmm5 - punpcklbw xmm0, xmm2 ; - - punpckhbw xmm5, xmm2 ; - psraw xmm0, 11 ; - - psraw xmm5, 11 - packsswb xmm0, xmm5 - - movdqa xmm2, xmm0 ; (3* (q0 - p0) + hvm(p1 - q1) + 3) >> 3; - - pxor xmm0, xmm0 ; 0 - movdqa xmm5, xmm1 ; abcdefgh - - punpcklbw xmm0, xmm1 ; e0f0g0h0 - psraw xmm0, 11 ; sign extended shift right by 3 - - pxor xmm1, xmm1 ; 0 - punpckhbw xmm1, xmm5 ; a0b0c0d0 - - psraw xmm1, 11 ; sign extended shift right by 3 - movdqa xmm5, xmm0 ; save results - - packsswb xmm0, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>3 - paddsw xmm5, [ones GLOBAL] - - paddsw xmm1, [ones GLOBAL] - psraw xmm5, 1 ; partial shifted one more time for 2nd tap - - psraw xmm1, 1 ; partial shifted one more time for 2nd tap - packsswb xmm5, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>4 - - pandn xmm4, xmm5 ; high edge variance additive - - paddsb xmm6, xmm2 ; p0+= p0 add - pxor xmm6, [t80 GLOBAL] ; unoffset - - ; mm6=p0 ; - movdqa xmm1, [rdx] ; p1 - pxor xmm1, [t80 GLOBAL] ; reoffset - - paddsb xmm1, xmm4 ; p1+= p1 add - pxor xmm1, [t80 GLOBAL] ; unoffset - ; mm6 = p0 mm1 = p1 - - psubsb xmm3, xmm0 ; q0-= q0 add - pxor xmm3, [t80 GLOBAL] ; unoffset - - ; mm3 = q0 - psubsb xmm7, xmm4 ; q1-= q1 add - pxor xmm7, [t80 GLOBAL] ; unoffset - ; mm7 = q1 - - ; tranpose and write back - ; xmm1 = f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 - ; xmm6 = f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03 - ; xmm3 = f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 - ; xmm7 = f5 e5 d5 c5 b5 a5 95 85 75 65 55 45 35 25 15 05 - movdqa xmm2, xmm1 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 - punpcklbw xmm2, xmm6 ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02 - - movdqa xmm4, xmm3 ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 - punpckhbw xmm1, xmm6 ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82 - - punpcklbw xmm4, xmm7 ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04 - punpckhbw xmm3, xmm7 ; f5 f4 e5 e4 d5 d4 c5 c4 b5 b4 a5 a4 95 94 85 84 - - movdqa xmm6, xmm2 ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02 - punpcklwd xmm2, xmm4 ; 35 34 33 32 25 24 23 22 15 14 13 12 05 04 03 02 - - punpckhwd xmm6, xmm4 ; 75 74 73 72 65 64 63 62 55 54 53 52 45 44 43 42 - movdqa xmm5, xmm1 ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82 - - punpcklwd xmm1, xmm3 ; f5 f4 f3 f2 e5 e4 e3 e2 d5 d4 d3 d2 c5 c4 c3 c2 - punpckhwd xmm5, xmm3 ; b5 b4 b3 b2 a5 a4 a3 a2 95 94 93 92 85 84 83 82 - - ; xmm2 = 35 34 33 32 25 24 23 22 15 14 13 12 05 04 03 02 - ; xmm6 = 75 74 73 72 65 64 63 62 55 54 53 52 45 44 43 42 - ; xmm5 = f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82 - ; xmm1 = b5 b4 b3 b2 a5 a4 a3 a2 95 94 93 92 85 84 83 82 - lea rsi, [rsi+rax*8] - - movd [rsi+rax*4+2], xmm2 - psrldq xmm2, 4 - - movd [rdi+rax*4+2], xmm2 - psrldq xmm2, 4 - - movd [rsi+rax*2+2], xmm2 - psrldq xmm2, 4 - - movd [rdi+rax*2+2], xmm2 - movd [rsi+2], xmm6 - - psrldq xmm6, 4 - movd [rdi+2], xmm6 - - psrldq xmm6, 4 - neg rax - - movd [rdi+rax+2], xmm6 - psrldq xmm6, 4 - - movd [rdi+rax*2+2], xmm6 - lea rsi, [rsi+rax*8] - - neg rax - ;;;;;;;;;;;;;;;;;;;;/ - movd [rsi+rax*4+2], xmm1 - psrldq xmm1, 4 - - movd [rcx+rax*4+2], xmm1 - psrldq xmm1, 4 - - movd [rsi+rax*2+2], xmm1 - psrldq xmm1, 4 - - movd [rcx+rax*2+2], xmm1 - psrldq xmm1, 4 - - movd [rsi+2], xmm5 - psrldq xmm5, 4 - - movd [rcx+2], xmm5 - psrldq xmm5, 4 - - neg rax - movd [rcx+rax+2], xmm5 - - psrldq xmm5, 4 - movd [rcx+rax*2+2], xmm5 - - add rsp, 96 - pop rsp - ; begin epilog - pop rdi - pop rsi - RESTORE_GOT - UNSHADOW_ARGS - pop rbp - ret - - -;void vp8_mbloop_filter_horizontal_edge_sse2 -;( -; unsigned char *src_ptr, -; int src_pixel_step, -; const char *flimit, -; const char *limit, -; const char *thresh, -; int count -;) -global sym(vp8_mbloop_filter_horizontal_edge_sse2) -sym(vp8_mbloop_filter_horizontal_edge_sse2): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 6 - GET_GOT rbx - push rsi - push rdi - ; end prolog - - ALIGN_STACK 16, rax - sub rsp, 32 ; reserve 32 bytes - %define t0 [rsp + 0] ;__declspec(align(16)) char t0[8]; - %define t1 [rsp + 16] ;__declspec(align(16)) char t1[8]; - - mov rsi, arg(0) ;src_ptr - movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch? - - mov rdx, arg(3) ;limit - movdqa xmm7, XMMWORD PTR [rdx] - - mov rdi, rsi ; rdi points to row +1 for indirect addressing - add rdi, rax - - ; calculate breakout conditions - movdqa xmm2, XMMWORD PTR [rdi+2*rax] ; q3 - movdqa xmm1, XMMWORD PTR [rsi+2*rax] ; q2 +%macro LFH_FILTER_MASK 1 +%if %1 + movdqa xmm2, [rdi+2*rax] ; q3 + movdqa xmm1, [rsi+2*rax] ; q2 +%else + movq xmm0, [rsi + rcx*2] ; q3 + movq xmm2, [rdi + rcx*2] + pslldq xmm2, 8 + por xmm2, xmm0 + movq xmm1, [rsi + rcx] ; q2 + movq xmm3, [rdi + rcx] + pslldq xmm3, 8 + por xmm1, xmm3 + movdqa XMMWORD PTR [rsp], xmm1 ; store q2 +%endif movdqa xmm6, xmm1 ; q2 psubusb xmm1, xmm2 ; q2-=q3 - - psubusb xmm2, xmm6 ; q3-=q2 por xmm1, xmm2 ; abs(q3-q2) psubusb xmm1, xmm7 - ; mm1 = abs(q3-q2), mm6 =q2, mm7 = limit - movdqa xmm4, XMMWORD PTR [rsi+rax] ; q1 - movdqa xmm3, xmm4 ; q1 +%if %1 + movdqa xmm4, [rsi+rax] ; q1 +%else + movq xmm0, [rsi] ; q1 + movq xmm4, [rdi] + pslldq xmm4, 8 + por xmm4, xmm0 + movdqa XMMWORD PTR [rsp + 16], xmm4 ; store q1 +%endif + movdqa xmm3, xmm4 ; q1 psubusb xmm4, xmm6 ; q1-=q2 psubusb xmm6, xmm3 ; q2-=q1 - por xmm4, xmm6 ; abs(q2-q1) psubusb xmm4, xmm7 por xmm1, xmm4 - ; mm1 = mask, mm3=q1, mm7 = limit - movdqa xmm4, XMMWORD PTR [rsi] ; q0 - movdqa xmm0, xmm4 ; q0 +%if %1 + movdqa xmm4, [rsi] ; q0 +%else + movq xmm4, [rsi + rax] ; q0 + movq xmm0, [rdi + rax] + pslldq xmm0, 8 + por xmm4, xmm0 +%endif + movdqa xmm0, xmm4 ; q0 psubusb xmm4, xmm3 ; q0-=q1 psubusb xmm3, xmm0 ; q1-=q0 - por xmm4, xmm3 ; abs(q0-q1) - movdqa t0, xmm4 ; save to t0 + movdqa t0, xmm4 ; save to t0 psubusb xmm4, xmm7 por xmm1, xmm4 - ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1) - neg rax ; negate pitch to deal with above border - - movdqa xmm2, XMMWORD PTR [rsi+4*rax] ; p3 - movdqa xmm4, XMMWORD PTR [rdi+4*rax] ; p2 +%if %1 + neg rax ; negate pitch to deal with above border + + movdqa xmm2, [rsi+4*rax] ; p3 + movdqa xmm4, [rdi+4*rax] ; p2 +%else + lea rsi, [rsi + rax*4] + lea rdi, [rdi + rax*4] + + movq xmm2, [rsi + rax] ; p3 + movq xmm3, [rdi + rax] + pslldq xmm3, 8 + por xmm2, xmm3 + movq xmm4, [rsi] ; p2 + movq xmm5, [rdi] + pslldq xmm5, 8 + por xmm4, xmm5 + movdqa XMMWORD PTR [rsp + 32], xmm4 ; store p2 +%endif movdqa xmm5, xmm4 ; p2 psubusb xmm4, xmm2 ; p2-=p3 - psubusb xmm2, xmm5 ; p3-=p2 por xmm4, xmm2 ; abs(p3 - p2) psubusb xmm4, xmm7 por xmm1, xmm4 - ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1) - movdqa xmm4, XMMWORD PTR [rsi+2*rax] ; p1 - movdqa xmm3, xmm4 ; p1 +%if %1 + movdqa xmm4, [rsi+2*rax] ; p1 +%else + movq xmm4, [rsi + rcx] ; p1 + movq xmm3, [rdi + rcx] + pslldq xmm3, 8 + por xmm4, xmm3 + movdqa XMMWORD PTR [rsp + 48], xmm4 ; store p1 +%endif + movdqa xmm3, xmm4 ; p1 psubusb xmm4, xmm5 ; p1-=p2 psubusb xmm5, xmm3 ; p2-=p1 - por xmm4, xmm5 ; abs(p2 - p1) psubusb xmm4, xmm7 por xmm1, xmm4 - movdqa xmm2, xmm3 ; p1 - ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1) - movdqa xmm4, XMMWORD PTR [rsi+rax] ; p0 - movdqa xmm5, xmm4 ; p0 +%if %1 + movdqa xmm4, [rsi+rax] ; p0 +%else + movq xmm4, [rsi + rcx*2] ; p0 + movq xmm5, [rdi + rcx*2] + pslldq xmm5, 8 + por xmm4, xmm5 +%endif + movdqa xmm5, xmm4 ; p0 psubusb xmm4, xmm3 ; p0-=p1 psubusb xmm3, xmm5 ; p1-=p0 - por xmm4, xmm3 ; abs(p1 - p0) - movdqa t1, xmm4 ; save to t1 + movdqa t1, xmm4 ; save to t1 psubusb xmm4, xmm7 por xmm1, xmm4 - ; mm1 = mask, mm0=q0, mm7 = limit, t0 = abs(q0-q1) t1 = abs(p1-p0) - ; mm5 = p0 - movdqa xmm3, XMMWORD PTR [rdi] ; q1 +%if %1 + movdqa xmm3, [rdi] ; q1 +%else + movdqa xmm3, q1 ; q1 +%endif + movdqa xmm4, xmm3 ; q1 psubusb xmm3, xmm2 ; q1-=p1 psubusb xmm2, xmm4 ; p1-=q1 @@ -781,113 +151,317 @@ sym(vp8_mbloop_filter_horizontal_edge_sse2): movdqa xmm6, xmm5 ; p0 movdqa xmm3, xmm0 ; q0 - psubusb xmm5, xmm3 ; p0-=q0 psubusb xmm3, xmm6 ; q0-=p0 - por xmm5, xmm3 ; abs(p0 - q0) paddusb xmm5, xmm5 ; abs(p0-q0)*2 paddusb xmm5, xmm2 ; abs (p0 - q0) *2 + abs(p1-q1)/2 - mov rdx, arg(2) ;flimit ; get flimit - movdqa xmm2, XMMWORD PTR [rdx] ; + mov rdx, arg(2) ; get flimit + movdqa xmm2, XMMWORD PTR [rdx] paddb xmm2, xmm2 ; flimit*2 (less than 255) paddb xmm7, xmm2 ; flimit * 2 + limit (less than 255) psubusb xmm5, xmm7 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit por xmm1, xmm5 pxor xmm5, xmm5 - pcmpeqb xmm1, xmm5 ; mask mm1 - ; mm1 = mask, mm0=q0, mm7 = flimit, t0 = abs(q0-q1) t1 = abs(p1-p0) - ; mm6 = p0, + pcmpeqb xmm1, xmm5 ; mask mm1 +%endmacro - ; calculate high edge variance - mov rdx, arg(4) ;thresh ; get thresh - movdqa xmm7, XMMWORD PTR [rdx] ; +%macro LFH_HEV_MASK 0 + mov rdx, arg(4) ; get thresh + movdqa xmm7, XMMWORD PTR [rdx] movdqa xmm4, t0 ; get abs (q1 - q0) psubusb xmm4, xmm7 - movdqa xmm3, t1 ; get abs (p1 - p0) psubusb xmm3, xmm7 - paddb xmm4, xmm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh pcmpeqb xmm4, xmm5 pcmpeqb xmm5, xmm5 pxor xmm4, xmm5 - ; mm1 = mask, mm0=q0, mm7 = thresh, t0 = abs(q0-q1) t1 = abs(p1-p0) - ; mm6 = p0, mm4=hev - ; start work on filters - movdqa xmm2, XMMWORD PTR [rsi+2*rax] ; p1 - movdqa xmm7, XMMWORD PTR [rdi] ; q1 +%endmacro - pxor xmm2, [t80 GLOBAL] ; p1 offset to convert to signed values - pxor xmm7, [t80 GLOBAL] ; q1 offset to convert to signed values +%macro BH_FILTER 1 +%if %1 + movdqa xmm2, [rsi+2*rax] ; p1 + movdqa xmm7, [rdi] ; q1 +%else + movdqa xmm2, p1 ; p1 + movdqa xmm7, q1 ; q1 +%endif + + pxor xmm2, [t80 GLOBAL] ; p1 offset to convert to signed values + pxor xmm7, [t80 GLOBAL] ; q1 offset to convert to signed values psubsb xmm2, xmm7 ; p1 - q1 - pxor xmm6, [t80 GLOBAL] ; offset to convert to signed values + pxor xmm6, [t80 GLOBAL] ; offset to convert to signed values + + pand xmm2, xmm4 ; high var mask (hvm)(p1 - q1) + pxor xmm0, [t80 GLOBAL] ; offset to convert to signed values - pxor xmm0, [t80 GLOBAL] ; offset to convert to signed values movdqa xmm3, xmm0 ; q0 + psubsb xmm0, xmm6 ; q0 - p0 + paddsb xmm2, xmm0 ; 1 * (q0 - p0) + hvm(p1 - q1) + paddsb xmm2, xmm0 ; 2 * (q0 - p0) + hvm(p1 - q1) + paddsb xmm2, xmm0 ; 3 * (q0 - p0) + hvm(p1 - q1) + pand xmm1, xmm2 ; mask filter values we don't care about + movdqa xmm2, xmm1 + paddsb xmm1, [t4 GLOBAL] ; 3* (q0 - p0) + hvm(p1 - q1) + 4 + paddsb xmm2, [t3 GLOBAL] ; 3* (q0 - p0) + hvm(p1 - q1) + 3 + + punpckhbw xmm5, xmm2 ; axbxcxdx + punpcklbw xmm2, xmm2 ; exfxgxhx + + psraw xmm5, 11 ; sign extended shift right by 3 + psraw xmm2, 11 ; sign extended shift right by 3 + packsswb xmm2, xmm5 ; (3* (q0 - p0) + hvm(p1 - q1) + 3) >> 3; + + punpcklbw xmm0, xmm1 ; exfxgxhx + punpckhbw xmm1, xmm1 ; axbxcxdx + + psraw xmm0, 11 ; sign extended shift right by 3 + psraw xmm1, 11 ; sign extended shift right by 3 + + movdqa xmm5, xmm0 ; save results + packsswb xmm0, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>3 + + paddsw xmm5, [ones GLOBAL] + paddsw xmm1, [ones GLOBAL] + + psraw xmm5, 1 ; partial shifted one more time for 2nd tap + psraw xmm1, 1 ; partial shifted one more time for 2nd tap + + packsswb xmm5, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>4 + pandn xmm4, xmm5 ; high edge variance additive +%endmacro + +%macro BH_WRITEBACK 1 + paddsb xmm6, xmm2 ; p0+= p0 add + pxor xmm6, [t80 GLOBAL] ; unoffset +%if %1 + movdqa [rsi+rax], xmm6 ; write back +%else + lea rsi, [rsi + rcx*2] + lea rdi, [rdi + rcx*2] + movq MMWORD PTR [rsi], xmm6 ; p0 + psrldq xmm6, 8 + movq MMWORD PTR [rdi], xmm6 +%endif + +%if %1 + movdqa xmm6, [rsi+2*rax] ; p1 +%else + movdqa xmm6, p1 ; p1 +%endif + pxor xmm6, [t80 GLOBAL] ; reoffset + paddsb xmm6, xmm4 ; p1+= p1 add + pxor xmm6, [t80 GLOBAL] ; unoffset +%if %1 + movdqa [rsi+2*rax], xmm6 ; write back +%else + movq MMWORD PTR [rsi + rax], xmm6 ; p1 + psrldq xmm6, 8 + movq MMWORD PTR [rdi + rax], xmm6 +%endif + + psubsb xmm3, xmm0 ; q0-= q0 add + pxor xmm3, [t80 GLOBAL] ; unoffset +%if %1 + movdqa [rsi], xmm3 ; write back +%else + movq MMWORD PTR [rsi + rcx], xmm3 ; q0 + psrldq xmm3, 8 + movq MMWORD PTR [rdi + rcx], xmm3 +%endif + + psubsb xmm7, xmm4 ; q1-= q1 add + pxor xmm7, [t80 GLOBAL] ; unoffset +%if %1 + movdqa [rdi], xmm7 ; write back +%else + movq MMWORD PTR [rsi + rcx*2],xmm7 ; q1 + psrldq xmm7, 8 + movq MMWORD PTR [rdi + rcx*2],xmm7 +%endif +%endmacro + + +;void vp8_loop_filter_horizontal_edge_sse2 +;( +; unsigned char *src_ptr, +; int src_pixel_step, +; const char *flimit, +; const char *limit, +; const char *thresh, +; int count +;) +global sym(vp8_loop_filter_horizontal_edge_sse2) +sym(vp8_loop_filter_horizontal_edge_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM + GET_GOT rbx + push rsi + push rdi + ; end prolog + + ALIGN_STACK 16, rax + sub rsp, 32 ; reserve 32 bytes + %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16]; + %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16]; + + mov rsi, arg(0) ;src_ptr + movsxd rax, dword ptr arg(1) ;src_pixel_step + + mov rdx, arg(3) ;limit + movdqa xmm7, XMMWORD PTR [rdx] + + lea rdi, [rsi+rax] ; rdi points to row +1 for indirect addressing + + ; calculate breakout conditions + LFH_FILTER_MASK 1 + + ; calculate high edge variance + LFH_HEV_MASK + + ; start work on filters + BH_FILTER 1 + ; write back the result + BH_WRITEBACK 1 + + add rsp, 32 + pop rsp + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + RESTORE_XMM + UNSHADOW_ARGS + pop rbp + ret + + +;void vp8_loop_filter_horizontal_edge_uv_sse2 +;( +; unsigned char *src_ptr, +; int src_pixel_step, +; const char *flimit, +; const char *limit, +; const char *thresh, +; int count +;) +global sym(vp8_loop_filter_horizontal_edge_uv_sse2) +sym(vp8_loop_filter_horizontal_edge_uv_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM + GET_GOT rbx + push rsi + push rdi + ; end prolog + + ALIGN_STACK 16, rax + sub rsp, 96 ; reserve 96 bytes + %define q2 [rsp + 0] ;__declspec(align(16)) char q2[16]; + %define q1 [rsp + 16] ;__declspec(align(16)) char q1[16]; + %define p2 [rsp + 32] ;__declspec(align(16)) char p2[16]; + %define p1 [rsp + 48] ;__declspec(align(16)) char p1[16]; + %define t0 [rsp + 64] ;__declspec(align(16)) char t0[16]; + %define t1 [rsp + 80] ;__declspec(align(16)) char t1[16]; + + mov rsi, arg(0) ; u + mov rdi, arg(5) ; v + movsxd rax, dword ptr arg(1) ; src_pixel_step + mov rcx, rax + neg rax ; negate pitch to deal with above border + + mov rdx, arg(3) ;limit + movdqa xmm7, XMMWORD PTR [rdx] + + lea rsi, [rsi + rcx] + lea rdi, [rdi + rcx] + ; calculate breakout conditions + LFH_FILTER_MASK 0 + ; calculate high edge variance + LFH_HEV_MASK + + ; start work on filters + BH_FILTER 0 + ; write back the result + BH_WRITEBACK 0 + + add rsp, 96 + pop rsp + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + RESTORE_XMM + UNSHADOW_ARGS + pop rbp + ret + + +%macro MBH_FILTER 1 +%if %1 + movdqa xmm2, [rsi+2*rax] ; p1 + movdqa xmm7, [rdi] ; q1 +%else + movdqa xmm2, p1 ; p1 + movdqa xmm7, q1 ; q1 +%endif + pxor xmm2, [t80 GLOBAL] ; p1 offset to convert to signed values + pxor xmm7, [t80 GLOBAL] ; q1 offset to convert to signed values + + psubsb xmm2, xmm7 ; p1 - q1 + pxor xmm6, [t80 GLOBAL] ; offset to convert to signed values + pxor xmm0, [t80 GLOBAL] ; offset to convert to signed values + movdqa xmm3, xmm0 ; q0 psubsb xmm0, xmm6 ; q0 - p0 paddsb xmm2, xmm0 ; 1 * (q0 - p0) + (p1 - q1) - paddsb xmm2, xmm0 ; 2 * (q0 - p0) paddsb xmm2, xmm0 ; 3 * (q0 - p0) + (p1 - q1) pand xmm1, xmm2 ; mask filter values we don't care about - ; mm1 = vp8_filter, mm4=hev, mm6=ps0, mm3=qs0 movdqa xmm2, xmm1 ; vp8_filter pand xmm2, xmm4; ; Filter2 = vp8_filter & hev + movdqa xmm5, xmm2 + paddsb xmm5, [t3 GLOBAL] ; vp8_signed_char_clamp(Filter2 + 3) - movdqa xmm5, xmm2 ; - paddsb xmm5, [t3 GLOBAL]; - - pxor xmm0, xmm0 ; 0 - pxor xmm7, xmm7 ; 0 - - punpcklbw xmm0, xmm5 ; e0f0g0h0 - psraw xmm0, 11 ; sign extended shift right by 3 + punpckhbw xmm7, xmm5 ; axbxcxdx + punpcklbw xmm5, xmm5 ; exfxgxhx - punpckhbw xmm7, xmm5 ; a0b0c0d0 psraw xmm7, 11 ; sign extended shift right by 3 + psraw xmm5, 11 ; sign extended shift right by 3 - packsswb xmm0, xmm7 ; Filter2 >>=3; - movdqa xmm5, xmm0 ; Filter2 - - paddsb xmm2, [t4 GLOBAL] ; vp8_signed_char_clamp(Filter2 + 4) - pxor xmm0, xmm0 ; 0 + packsswb xmm5, xmm7 ; Filter2 >>=3; + paddsb xmm2, [t4 GLOBAL] ; vp8_signed_char_clamp(Filter2 + 4) - pxor xmm7, xmm7 ; 0 - punpcklbw xmm0, xmm2 ; e0f0g0h0 + punpckhbw xmm7, xmm2 ; axbxcxdx + punpcklbw xmm0, xmm2 ; exfxgxhx + psraw xmm7, 11 ; sign extended shift right by 3 psraw xmm0, 11 ; sign extended shift right by 3 - punpckhbw xmm7, xmm2 ; a0b0c0d0 - psraw xmm7, 11 ; sign extended shift right by 3 packsswb xmm0, xmm7 ; Filter2 >>=3; - - ; mm0= filter2 mm1 = vp8_filter, mm3 =qs0 mm5=s mm4 =hev mm6=ps0 - psubsb xmm3, xmm0 ; qs0 =qs0 - filter1 paddsb xmm6, xmm5 ; ps0 =ps0 + Fitler2 - ; mm1=vp8_filter, mm3=qs0, mm4 =hev mm6=ps0 - ; vp8_filter &= ~hev; - ; Filter2 = vp8_filter; + psubsb xmm3, xmm0 ; qs0 =qs0 - filter1 pandn xmm4, xmm1 ; vp8_filter&=~hev +%endmacro - - ; mm3=qs0, mm4=filter2, mm6=ps0 - +%macro MBH_WRITEBACK 1 ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7); ; s = vp8_signed_char_clamp(qs0 - u); ; *oq0 = s^0x80; ; s = vp8_signed_char_clamp(ps0 + u); ; *op0 = s^0x80; - pxor xmm0, xmm0 pxor xmm1, xmm1 pxor xmm2, xmm2 @@ -911,8 +485,20 @@ sym(vp8_mbloop_filter_horizontal_edge_sse2): pxor xmm3, [t80 GLOBAL] pxor xmm6, [t80 GLOBAL] +%if %1 movdqa XMMWORD PTR [rsi+rax], xmm6 movdqa XMMWORD PTR [rsi], xmm3 +%else + lea rsi, [rsi + rcx*2] + lea rdi, [rdi + rcx*2] + + movq MMWORD PTR [rsi], xmm6 ; p0 + psrldq xmm6, 8 + movq MMWORD PTR [rdi], xmm6 + movq MMWORD PTR [rsi + rcx], xmm3 ; q0 + psrldq xmm3, 8 + movq MMWORD PTR [rdi + rcx], xmm3 +%endif ; roughly 2/7th difference across boundary ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7); @@ -937,8 +523,13 @@ sym(vp8_mbloop_filter_horizontal_edge_sse2): packsswb xmm1, xmm2 +%if %1 movdqa xmm3, XMMWORD PTR [rdi] - movdqa xmm6, XMMWORD PTR [rsi+rax*2] ; p1 + movdqa xmm6, XMMWORD PTR [rsi+rax*2] ; p1 +%else + movdqa xmm3, q1 ; q1 + movdqa xmm6, p1 ; p1 +%endif pxor xmm3, [t80 GLOBAL] pxor xmm6, [t80 GLOBAL] @@ -949,9 +540,18 @@ sym(vp8_mbloop_filter_horizontal_edge_sse2): pxor xmm6, [t80 GLOBAL] pxor xmm3, [t80 GLOBAL] +%if %1 movdqa XMMWORD PTR [rdi], xmm3 movdqa XMMWORD PTR [rsi+rax*2],xmm6 - +%else + movq MMWORD PTR [rsi + rcx*2],xmm3 ; q1 + psrldq xmm3, 8 + movq MMWORD PTR [rdi + rcx*2],xmm3 + + movq MMWORD PTR [rsi + rax], xmm6 ; p1 + psrldq xmm6, 8 + movq MMWORD PTR [rdi + rax], xmm6 +%endif ; roughly 1/7th difference across boundary ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7); ; s = vp8_signed_char_clamp(qs2 - u); @@ -975,11 +575,15 @@ sym(vp8_mbloop_filter_horizontal_edge_sse2): packsswb xmm1, xmm2 - +%if %1 movdqa xmm6, XMMWORD PTR [rdi+rax*4] neg rax - movdqa xmm3, XMMWORD PTR [rdi+rax ] + movdqa xmm3, XMMWORD PTR [rdi+rax] +%else + movdqa xmm6, p2 ; p2 + movdqa xmm3, q2 ; q2 +%endif pxor xmm6, [t80 GLOBAL] pxor xmm3, [t80 GLOBAL] @@ -989,11 +593,68 @@ sym(vp8_mbloop_filter_horizontal_edge_sse2): pxor xmm6, [t80 GLOBAL] pxor xmm3, [t80 GLOBAL] - - movdqa XMMWORD PTR [rdi+rax ], xmm3 +%if %1 + movdqa XMMWORD PTR [rdi+rax ],xmm3 neg rax - movdqa XMMWORD PTR [rdi+rax*4], xmm6 + movdqa XMMWORD PTR [rdi+rax*4],xmm6 +%else + movq MMWORD PTR [rsi+rax*2], xmm6 ; p2 + psrldq xmm6, 8 + movq MMWORD PTR [rdi+rax*2], xmm6 + + lea rsi, [rsi + rcx] + lea rdi, [rdi + rcx] + movq MMWORD PTR [rsi+rcx*2 ],xmm3 ; q2 + psrldq xmm3, 8 + movq MMWORD PTR [rdi+rcx*2 ],xmm3 +%endif +%endmacro + + +;void vp8_mbloop_filter_horizontal_edge_sse2 +;( +; unsigned char *src_ptr, +; int src_pixel_step, +; const char *flimit, +; const char *limit, +; const char *thresh, +; int count +;) +global sym(vp8_mbloop_filter_horizontal_edge_sse2) +sym(vp8_mbloop_filter_horizontal_edge_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM + GET_GOT rbx + push rsi + push rdi + ; end prolog + + ALIGN_STACK 16, rax + sub rsp, 32 ; reserve 32 bytes + %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16]; + %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16]; + + mov rsi, arg(0) ;src_ptr + movsxd rax, dword ptr arg(1) ;src_pixel_step + + mov rdx, arg(3) ;limit + movdqa xmm7, XMMWORD PTR [rdx] + + lea rdi, [rsi+rax] ; rdi points to row +1 for indirect addressing + + ; calculate breakout conditions + LFH_FILTER_MASK 1 + + ; calculate high edge variance + LFH_HEV_MASK + + ; start work on filters + MBH_FILTER 1 + ; write back the result + MBH_WRITEBACK 1 add rsp, 32 pop rsp @@ -1001,383 +662,694 @@ sym(vp8_mbloop_filter_horizontal_edge_sse2): pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret -;void vp8_mbloop_filter_vertical_edge_sse2 +;void vp8_mbloop_filter_horizontal_edge_uv_sse2 ;( -; unsigned char *src_ptr, +; unsigned char *u, ; int src_pixel_step, ; const char *flimit, ; const char *limit, ; const char *thresh, -; int count +; unsigned char *v ;) -global sym(vp8_mbloop_filter_vertical_edge_sse2) -sym(vp8_mbloop_filter_vertical_edge_sse2): +global sym(vp8_mbloop_filter_horizontal_edge_uv_sse2) +sym(vp8_mbloop_filter_horizontal_edge_uv_sse2): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 6 + SAVE_XMM GET_GOT rbx push rsi push rdi ; end prolog ALIGN_STACK 16, rax - sub rsp, 160 ; reserve 160 bytes - %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16]; - %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16]; - %define srct [rsp + 32] ;__declspec(align(16)) char srct[128]; + sub rsp, 96 ; reserve 96 bytes + %define q2 [rsp + 0] ;__declspec(align(16)) char q2[16]; + %define q1 [rsp + 16] ;__declspec(align(16)) char q1[16]; + %define p2 [rsp + 32] ;__declspec(align(16)) char p2[16]; + %define p1 [rsp + 48] ;__declspec(align(16)) char p1[16]; + %define t0 [rsp + 64] ;__declspec(align(16)) char t0[16]; + %define t1 [rsp + 80] ;__declspec(align(16)) char t1[16]; + + mov rsi, arg(0) ; u + mov rdi, arg(5) ; v + movsxd rax, dword ptr arg(1) ; src_pixel_step + mov rcx, rax + neg rax ; negate pitch to deal with above border + + mov rdx, arg(3) ;limit + movdqa xmm7, XMMWORD PTR [rdx] + lea rsi, [rsi + rcx] + lea rdi, [rdi + rcx] - mov rsi, arg(0) ;src_ptr - movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch? + ; calculate breakout conditions + LFH_FILTER_MASK 0 - lea rsi, [rsi + rax*4 - 4] - lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing + ; calculate high edge variance + LFH_HEV_MASK - mov rcx, rax - neg rcx + ; start work on filters + MBH_FILTER 0 + ; write back the result + MBH_WRITEBACK 0 - ; Transpose - movq xmm0, QWORD PTR [rdi+rax*2] ; xx xx xx xx xx xx xx xx 77 76 75 74 73 72 71 70 - movq xmm7, QWORD PTR [rsi+rax*2] ; xx xx xx xx xx xx xx xx 67 66 65 64 63 62 61 60 + add rsp, 96 + pop rsp + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + RESTORE_XMM + UNSHADOW_ARGS + pop rbp + ret - punpcklbw xmm7, xmm0 ; 77 67 76 66 75 65 74 64 73 63 72 62 71 61 70 60 - movq xmm0, QWORD PTR [rsi+rax] ; - movq xmm5, QWORD PTR [rsi] ; - punpcklbw xmm5, xmm0 ; 57 47 56 46 55 45 54 44 53 43 52 42 51 41 50 40 +%macro TRANSPOSE_16X8_1 0 + movq xmm4, QWORD PTR [rsi] ; xx xx xx xx xx xx xx xx 07 06 05 04 03 02 01 00 + movq xmm7, QWORD PTR [rdi] ; xx xx xx xx xx xx xx xx 17 16 15 14 13 12 11 10 - movdqa xmm6, xmm5 ; 57 47 56 46 55 45 54 44 53 43 52 42 51 41 50 40 - punpcklwd xmm5, xmm7 ; 73 63 53 43 72 62 52 42 71 61 51 41 70 60 50 40 + punpcklbw xmm4, xmm7 ; 17 07 16 06 15 05 14 04 13 03 12 02 11 01 10 00 + movq xmm0, QWORD PTR [rsi+2*rax] ; xx xx xx xx xx xx xx xx 27 26 25 24 23 22 21 20 - punpckhwd xmm6, xmm7 ; 77 67 57 47 76 66 56 46 75 65 55 45 74 64 54 44 - movq xmm7, QWORD PTR [rsi + rcx] ; xx xx xx xx xx xx xx xx 37 36 35 34 33 32 31 30 + movdqa xmm3, xmm4 ; 17 07 16 06 15 05 14 04 13 03 12 02 11 01 10 00 - movq xmm0, QWORD PTR [rsi + rcx*2] ; xx xx xx xx xx xx xx xx 27 26 25 24 23 22 21 20 - punpcklbw xmm0, xmm7 ; 37 27 36 36 35 25 34 24 33 23 32 22 31 21 30 20 + movq xmm7, QWORD PTR [rdi+2*rax] ; xx xx xx xx xx xx xx xx 37 36 35 34 33 32 31 30 + punpcklbw xmm0, xmm7 ; 37 27 36 36 35 25 34 24 33 23 32 22 31 21 30 20 - movq xmm4, QWORD PTR [rsi + rcx*4] ; xx xx xx xx xx xx xx xx 07 06 05 04 03 02 01 00 - movq xmm7, QWORD PTR [rdi + rcx*4] ; xx xx xx xx xx xx xx xx 17 16 15 14 13 12 11 10 + movq xmm5, QWORD PTR [rsi+4*rax] ; xx xx xx xx xx xx xx xx 47 46 45 44 43 42 41 40 + movq xmm2, QWORD PTR [rdi+4*rax] ; xx xx xx xx xx xx xx xx 57 56 55 54 53 52 51 50 - punpcklbw xmm4, xmm7 ; 17 07 16 06 15 05 14 04 13 03 12 02 11 01 10 00 - movdqa xmm3, xmm4 ; 17 07 16 06 15 05 14 04 13 03 12 02 11 01 10 00 + punpcklbw xmm5, xmm2 ; 57 47 56 46 55 45 54 44 53 43 52 42 51 41 50 40 + movq xmm7, QWORD PTR [rsi+2*rcx] ; xx xx xx xx xx xx xx xx 67 66 65 64 63 62 61 60 - punpcklwd xmm3, xmm0 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00 - punpckhwd xmm4, xmm0 ; 37 27 17 07 36 26 16 06 35 25 15 05 34 24 14 04 + movq xmm1, QWORD PTR [rdi+2*rcx] ; xx xx xx xx xx xx xx xx 77 76 75 74 73 72 71 70 + movdqa xmm6, xmm5 ; 57 47 56 46 55 45 54 44 53 43 52 42 51 41 50 40 - movdqa xmm7, xmm4 ; 37 27 17 07 36 26 16 06 35 25 15 05 34 24 14 04 - movdqa xmm2, xmm3 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00 + punpcklbw xmm7, xmm1 ; 77 67 76 66 75 65 74 64 73 63 72 62 71 61 70 60 + punpcklwd xmm5, xmm7 ; 73 63 53 43 72 62 52 42 71 61 51 41 70 60 50 40 - punpckhdq xmm7, xmm6 ; 77 67 57 47 37 27 17 07 76 66 56 46 36 26 16 06 - punpckldq xmm4, xmm6 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04 + punpckhwd xmm6, xmm7 ; 77 67 57 47 76 66 56 46 75 65 55 45 74 64 54 44 - punpckhdq xmm3, xmm5 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02 - punpckldq xmm2, xmm5 ; 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00 + punpcklwd xmm3, xmm0 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00 + punpckhwd xmm4, xmm0 ; 37 27 17 07 36 26 16 06 35 25 15 05 34 24 14 04 - movdqa t0, xmm2 ; save to free XMM2 - ;movdqa t1, xmm3 + movdqa xmm7, xmm4 ; 37 27 17 07 36 26 16 06 35 25 15 05 34 24 14 04 + movdqa xmm2, xmm3 ; 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00 - ; XMM3 XMM4 XMM7 in use - lea rsi, [rsi+rax*8] - lea rdi, [rdi+rax*8] + punpckhdq xmm7, xmm6 ; 77 67 57 47 37 27 17 07 76 66 56 46 36 26 16 06 + punpckldq xmm4, xmm6 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04 + + punpckhdq xmm3, xmm5 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02 + punpckldq xmm2, xmm5 ; 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00 + + movdqa t0, xmm2 ; save to free XMM2 +%endmacro - movq xmm6, QWORD PTR [rdi+rax*2] ; xx xx xx xx xx xx xx xx f7 f6 f5 f4 f3 f2 f1 f0 - movq xmm5, QWORD PTR [rsi+rax*2] ; xx xx xx xx xx xx xx xx e7 e6 e5 e4 e3 e2 e1 e0 +%macro TRANSPOSE_16X8_2 1 + movq xmm2, QWORD PTR [rsi] ; xx xx xx xx xx xx xx xx 87 86 85 84 83 82 81 80 + movq xmm5, QWORD PTR [rdi] ; xx xx xx xx xx xx xx xx 97 96 95 94 93 92 91 90 - punpcklbw xmm5, xmm6 ; f7 e7 f6 e6 f5 e5 f4 e4 f3 e3 f2 e2 f1 e1 f0 e0 - movq xmm6, QWORD PTR [rsi+rax] ; xx xx xx xx xx xx xx xx d7 d6 d5 d4 d3 d2 d1 d0 + punpcklbw xmm2, xmm5 ; 97 87 96 86 95 85 94 84 93 83 92 82 91 81 90 80 + movq xmm0, QWORD PTR [rsi+2*rax] ; xx xx xx xx xx xx xx xx a7 a6 a5 a4 a3 a2 a1 a0 - movq xmm1, QWORD PTR [rsi] ; xx xx xx xx xx xx xx xx c7 c6 c5 c4 c3 c2 c1 c0 - punpcklbw xmm1, xmm6 ; d7 c7 d6 c6 d5 c5 d4 c4 d3 c3 d2 c2 d1 e1 d0 c0 + movq xmm5, QWORD PTR [rdi+2*rax] ; xx xx xx xx xx xx xx xx b7 b6 b5 b4 b3 b2 b1 b0 + punpcklbw xmm0, xmm5 ; b7 a7 b6 a6 b5 a5 b4 a4 b3 a3 b2 a2 b1 a1 b0 a0 - movdqa xmm6, xmm1 ; - punpckhwd xmm6, xmm5 ; f7 e7 d7 c7 f6 e6 d6 c6 f5 e5 d5 c5 f4 e4 d4 c4 + movq xmm1, QWORD PTR [rsi+4*rax] ; xx xx xx xx xx xx xx xx c7 c6 c5 c4 c3 c2 c1 c0 + movq xmm6, QWORD PTR [rdi+4*rax] ; xx xx xx xx xx xx xx xx d7 d6 d5 d4 d3 d2 d1 d0 - punpcklwd xmm1, xmm5 ; f3 e3 d3 c3 f2 e2 d2 c2 f1 e1 d1 c1 f0 e0 d0 c0 - movq xmm5, QWORD PTR [rsi+rcx] ; xx xx xx xx xx xx xx xx b7 b6 b5 b4 b3 b2 b1 b0 + punpcklbw xmm1, xmm6 ; d7 c7 d6 c6 d5 c5 d4 c4 d3 c3 d2 c2 d1 e1 d0 c0 + movq xmm5, QWORD PTR [rsi+2*rcx] ; xx xx xx xx xx xx xx xx e7 e6 e5 e4 e3 e2 e1 e0 - movq xmm0, QWORD PTR [rsi+rcx*2] ; xx xx xx xx xx xx xx xx a7 a6 a5 a4 a3 a2 a1 a0 - punpcklbw xmm0, xmm5 ; b7 a7 b6 a6 b5 a5 b4 a4 b3 a3 b2 a2 b1 a1 b0 a0 + movq xmm6, QWORD PTR [rdi+2*rcx] ; xx xx xx xx xx xx xx xx f7 f6 f5 f4 f3 f2 f1 f0 + punpcklbw xmm5, xmm6 ; f7 e7 f6 e6 f5 e5 f4 e4 f3 e3 f2 e2 f1 e1 f0 e0 - movq xmm2, QWORD PTR [rsi+rcx*4] ; xx xx xx xx xx xx xx xx 87 86 85 84 83 82 81 80 - movq xmm5, QWORD PTR [rdi+rcx*4] ; xx xx xx xx xx xx xx xx 97 96 95 94 93 92 91 90 + movdqa xmm6, xmm1 ; + punpckhwd xmm6, xmm5 ; f7 e7 d7 c7 f6 e6 d6 c6 f5 e5 d5 c5 f4 e4 d4 c4 - punpcklbw xmm2, xmm5 ; 97 87 96 86 95 85 94 84 93 83 92 82 91 81 90 80 - movdqa xmm5, xmm2 ; 97 87 96 86 95 85 94 84 93 83 92 82 91 81 90 80 + punpcklwd xmm1, xmm5 ; f3 e3 d3 c3 f2 e2 d2 c2 f1 e1 d1 c1 f0 e0 d0 c0 + movdqa xmm5, xmm2 ; 97 87 96 86 95 85 94 84 93 83 92 82 91 81 90 80 - punpcklwd xmm5, xmm0 ; b3 a3 93 83 b2 a2 92 82 b1 a1 91 81 b0 a0 90 80 - punpckhwd xmm2, xmm0 ; b7 a7 97 87 b6 a6 96 86 b5 a5 95 85 b4 a4 94 84 + punpcklwd xmm5, xmm0 ; b3 a3 93 83 b2 a2 92 82 b1 a1 91 81 b0 a0 90 80 + punpckhwd xmm2, xmm0 ; b7 a7 97 87 b6 a6 96 86 b5 a5 95 85 b4 a4 94 84 movdqa xmm0, xmm5 - punpckldq xmm0, xmm1 ; f1 e1 d1 c1 b1 a1 91 81 f0 e0 d0 c0 b0 a0 90 80 + punpckldq xmm0, xmm1 ; f1 e1 d1 c1 b1 a1 91 81 f0 e0 d0 c0 b0 a0 90 80 - punpckhdq xmm5, xmm1 ; f3 e3 d3 c3 b3 a3 93 83 f2 e2 d2 c2 b2 a2 92 82 - movdqa xmm1, xmm2 ; b7 a7 97 87 b6 a6 96 86 b5 a5 95 85 b4 a4 94 84 + punpckhdq xmm5, xmm1 ; f3 e3 d3 c3 b3 a3 93 83 f2 e2 d2 c2 b2 a2 92 82 + movdqa xmm1, xmm2 ; b7 a7 97 87 b6 a6 96 86 b5 a5 95 85 b4 a4 94 84 - punpckldq xmm1, xmm6 ; f5 e5 d5 c5 b5 a5 95 85 f4 e4 d4 c4 b4 a4 94 84 - punpckhdq xmm2, xmm6 ; f7 e7 d7 c7 b7 a7 97 87 f6 e6 d6 c6 b6 a6 96 86 + punpckldq xmm1, xmm6 ; f5 e5 d5 c5 b5 a5 95 85 f4 e4 d4 c4 b4 a4 94 84 + punpckhdq xmm2, xmm6 ; f7 e7 d7 c7 b7 a7 97 87 f6 e6 d6 c6 b6 a6 96 86 - movdqa xmm6, xmm7 ; 77 67 57 47 37 27 17 07 76 66 56 46 36 26 16 06 - punpcklqdq xmm6, xmm2 ; f6 e6 d6 c6 b6 a6 96 86 76 66 56 46 36 26 16 06 + movdqa xmm6, xmm7 ; 77 67 57 47 37 27 17 07 76 66 56 46 36 26 16 06 + punpcklqdq xmm6, xmm2 ; f6 e6 d6 c6 b6 a6 96 86 76 66 56 46 36 26 16 06 + punpckhqdq xmm7, xmm2 ; f7 e7 d7 c7 b7 a7 97 87 77 67 57 47 37 27 17 07 +%if %1 + movdqa xmm2, xmm3 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02 - lea rdx, srct - punpckhqdq xmm7, xmm2 ; f7 e7 d7 c7 b7 a7 97 87 77 67 57 47 37 27 17 07 + punpcklqdq xmm2, xmm5 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 + + punpckhqdq xmm3, xmm5 ; f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03 + movdqa [rdx], xmm2 ; save 2 + + movdqa xmm5, xmm4 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04 + punpcklqdq xmm4, xmm1 ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 + + movdqa [rdx+16], xmm3 ; save 3 + punpckhqdq xmm5, xmm1 ; f5 e5 d5 c5 b5 a5 95 85 75 65 55 45 35 25 15 05 + + movdqa [rdx+32], xmm4 ; save 4 + movdqa [rdx+48], xmm5 ; save 5 + + movdqa xmm1, t0 ; get + movdqa xmm2, xmm1 ; - movdqa [rdx+112], xmm7 ; save 7 - movdqa xmm2, xmm3 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02 + punpckhqdq xmm1, xmm0 ; f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01 + punpcklqdq xmm2, xmm0 ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00 +%else + movdqa [rdx+112], xmm7 ; save 7 + movdqa xmm2, xmm3 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02 - movdqa [rdx+96], xmm6 ; save 6 - punpcklqdq xmm2, xmm5 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 + movdqa [rdx+96], xmm6 ; save 6 + punpcklqdq xmm2, xmm5 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 - punpckhqdq xmm3, xmm5 ; f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03 - movdqa [rdx+32], xmm2 ; save 2 + punpckhqdq xmm3, xmm5 ; f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03 + movdqa [rdx+32], xmm2 ; save 2 - movdqa xmm5, xmm4 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04 - punpcklqdq xmm4, xmm1 ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 + movdqa xmm5, xmm4 ; 75 65 55 45 35 25 15 05 74 64 54 44 34 24 14 04 + punpcklqdq xmm4, xmm1 ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 - movdqa [rdx+48], xmm3 ; save 3 - punpckhqdq xmm5, xmm1 ; f5 e5 d5 c5 b5 a5 95 85 75 65 55 45 35 25 15 05 + movdqa [rdx+48], xmm3 ; save 3 + punpckhqdq xmm5, xmm1 ; f5 e5 d5 c5 b5 a5 95 85 75 65 55 45 35 25 15 05 - movdqa [rdx+64], xmm4 ; save 4 - movdqa [rdx+80], xmm5 ; save 5 + movdqa [rdx+64], xmm4 ; save 4 + movdqa [rdx+80], xmm5 ; save 5 - movdqa xmm1, t0 ; get - movdqa xmm2, xmm1 ; + movdqa xmm1, t0 ; get + movdqa xmm2, xmm1 - punpckhqdq xmm1, xmm0 ; f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01 - punpcklqdq xmm2, xmm0 ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00 + punpckhqdq xmm1, xmm0 ; f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01 + punpcklqdq xmm2, xmm0 ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00 movdqa [rdx+16], xmm1 movdqa [rdx], xmm2 +%endif +%endmacro - movdqa xmm0, xmm6 ; q2 - psubusb xmm0, xmm7 ; q2-q3 +%macro LFV_FILTER_MASK 1 + movdqa xmm0, xmm6 ; q2 + psubusb xmm0, xmm7 ; q2-q3 - psubusb xmm7, xmm6 ; q3-q2 - por xmm7, xmm0 ; abs (q3-q2) + psubusb xmm7, xmm6 ; q3-q2 + por xmm7, xmm0 ; abs (q3-q2) - movdqa xmm1, xmm5 ; q1 - psubusb xmm1, xmm6 ; q1-q2 + movdqa xmm4, xmm5 ; q1 + psubusb xmm4, xmm6 ; q1-q2 - psubusb xmm6, xmm5 ; q2-q1 - por xmm6, xmm1 ; abs (q2-q1) + psubusb xmm6, xmm5 ; q2-q1 + por xmm6, xmm4 ; abs (q2-q1) - ;/* - ;movdqa xmm0, xmm4 ; q0 - ;psubusb xmm0 xmm5 ; q0-q1 - ; - ;pusbusb xmm5, xmm4 ; q1-q0 - ;por xmm5, xmm0 ; abs (q1-q0) - ;*/ - - movdqa xmm1, [rdx+16] ; p2 movdqa xmm0, xmm1 - psubusb xmm0, xmm2 ; p2 - p3; - psubusb xmm2, xmm1 ; p3 - p2; + psubusb xmm0, xmm2 ; p2 - p3; + psubusb xmm2, xmm1 ; p3 - p2; + + por xmm0, xmm2 ; abs(p2-p3) +%if %1 + movdqa xmm2, [rdx] ; p1 +%else + movdqa xmm2, [rdx+32] ; p1 +%endif + movdqa xmm5, xmm2 ; p1 + + psubusb xmm5, xmm1 ; p1-p2 + psubusb xmm1, xmm2 ; p2-p1 + + por xmm1, xmm5 ; abs(p2-p1) + + mov rdx, arg(3) ; limit + movdqa xmm4, [rdx] ; limit - por xmm0, xmm2 ; abs(p2-p3) + psubusb xmm7, xmm4 - movdqa xmm2, [rdx+32] ; p1 - movdqa xmm5, xmm2 ; p1 + psubusb xmm0, xmm4 ; abs(p3-p2) > limit + psubusb xmm1, xmm4 ; abs(p2-p1) > limit - psubusb xmm5, xmm1 ; p1-p2 - psubusb xmm1, xmm2 ; p2-p1 + psubusb xmm6, xmm4 ; abs(q2-q1) > limit + por xmm7, xmm6 ; or - por xmm1, xmm5 ; abs(p2-p1) - mov rdx, arg(3) ;limit + por xmm0, xmm1 + por xmm0, xmm7 ; abs(q3-q2) > limit || abs(p3-p2) > limit ||abs(p2-p1) > limit || abs(q2-q1) > limit - movdqa xmm4, [rdx] ; limit - psubusb xmm7, xmm4 ; + movdqa xmm1, xmm2 ; p1 + movdqa xmm7, xmm3 ; p0 + psubusb xmm7, xmm2 ; p0-p1 - psubusb xmm0, xmm4 ; abs(p3-p2) > limit - psubusb xmm1, xmm4 ; abs(p2-p1) > limit + psubusb xmm2, xmm3 ; p1-p0 + por xmm2, xmm7 ; abs(p1-p0) - psubusb xmm6, xmm4 ; abs(q2-q1) > limit - por xmm7, xmm6 ; or + movdqa t0, xmm2 ; save abs(p1-p0) + lea rdx, srct + + psubusb xmm2, xmm4 ; abs(p1-p0)>limit + por xmm0, xmm2 ; mask +%if %1 + movdqa xmm5, [rdx+32] ; q0 + movdqa xmm7, [rdx+48] ; q1 +%else + movdqa xmm5, [rdx+64] ; q0 + movdqa xmm7, [rdx+80] ; q1 +%endif + movdqa xmm6, xmm5 ; q0 + movdqa xmm2, xmm7 ; q1 + psubusb xmm5, xmm7 ; q0-q1 + + psubusb xmm7, xmm6 ; q1-q0 + por xmm7, xmm5 ; abs(q1-q0) + + movdqa t1, xmm7 ; save abs(q1-q0) + psubusb xmm7, xmm4 ; abs(q1-q0)> limit + + por xmm0, xmm7 ; mask + + movdqa xmm5, xmm2 ; q1 + psubusb xmm5, xmm1 ; q1-=p1 + psubusb xmm1, xmm2 ; p1-=q1 + por xmm5, xmm1 ; abs(p1-q1) + pand xmm5, [tfe GLOBAL] ; set lsb of each byte to zero + psrlw xmm5, 1 ; abs(p1-q1)/2 + + mov rdx, arg(2) ; flimit + movdqa xmm2, [rdx] ; flimit + + movdqa xmm1, xmm3 ; p0 + movdqa xmm7, xmm6 ; q0 + psubusb xmm1, xmm7 ; p0-q0 + psubusb xmm7, xmm3 ; q0-p0 + por xmm1, xmm7 ; abs(q0-p0) + paddusb xmm1, xmm1 ; abs(q0-p0)*2 + paddusb xmm1, xmm5 ; abs (p0 - q0) *2 + abs(p1-q1)/2 + + paddb xmm2, xmm2 ; flimit*2 (less than 255) + paddb xmm4, xmm2 ; flimit * 2 + limit (less than 255) + + psubusb xmm1, xmm4 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit + por xmm1, xmm0; ; mask + pxor xmm0, xmm0 + pcmpeqb xmm1, xmm0 +%endmacro - por xmm0, xmm1 ; - por xmm0, xmm7 ; abs(q3-q2) > limit || abs(p3-p2) > limit ||abs(p2-p1) > limit || abs(q2-q1) > limit +%macro LFV_HEV_MASK 0 + mov rdx, arg(4) ; get thresh + movdqa xmm7, XMMWORD PTR [rdx] - movdqa xmm1, xmm2 ; p1 + movdqa xmm4, t0 ; get abs (q1 - q0) + psubusb xmm4, xmm7 ; abs(q1 - q0) > thresh - movdqa xmm7, xmm3 ; p0 - psubusb xmm7, xmm2 ; p0-p1 + movdqa xmm3, t1 ; get abs (p1 - p0) + psubusb xmm3, xmm7 ; abs(p1 - p0)> thresh - psubusb xmm2, xmm3 ; p1-p0 - por xmm2, xmm7 ; abs(p1-p0) + por xmm4, xmm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh + pcmpeqb xmm4, xmm0 - movdqa t0, xmm2 ; save abs(p1-p0) + pcmpeqb xmm0, xmm0 + pxor xmm4, xmm0 +%endmacro + +%macro BV_FILTER 0 lea rdx, srct - psubusb xmm2, xmm4 ; abs(p1-p0)>limit - por xmm0, xmm2 ; mask + movdqa xmm2, [rdx] ; p1 lea rsi, [rsi+rcx*8] + movdqa xmm7, [rdx+48] ; q1 + movdqa xmm6, [rdx+16] ; p0 + movdqa xmm0, [rdx+32] ; q0 - movdqa xmm5, [rdx+64] ; q0 - movdqa xmm7, [rdx+80] ; q1 + pxor xmm2, [t80 GLOBAL] ; p1 offset to convert to signed values + pxor xmm7, [t80 GLOBAL] ; q1 offset to convert to signed values - movdqa xmm6, xmm5 ; q0 - movdqa xmm2, xmm7 ; q1 - psubusb xmm5, xmm7 ; q0-q1 + psubsb xmm2, xmm7 ; p1 - q1 + pand xmm2, xmm4 ; high var mask (hvm)(p1 - q1) - psubusb xmm7, xmm6 ; q1-q0 - por xmm7, xmm5 ; abs(q1-q0) + pxor xmm6, [t80 GLOBAL] ; offset to convert to signed values + pxor xmm0, [t80 GLOBAL] ; offset to convert to signed values - movdqa t1, xmm7 ; save abs(q1-q0) - psubusb xmm7, xmm4 ; abs(q1-q0)> limit + movdqa xmm3, xmm0 ; q0 + psubsb xmm0, xmm6 ; q0 - p0 - por xmm0, xmm7 ; mask + paddsb xmm2, xmm0 ; 1 * (q0 - p0) + hvm(p1 - q1) + paddsb xmm2, xmm0 ; 2 * (q0 - p0) + hvm(p1 - q1) - movdqa xmm5, xmm2 ; q1 - psubusb xmm5, xmm1 ; q1-=p1 - psubusb xmm1, xmm2 ; p1-=q1 - por xmm5, xmm1 ; abs(p1-q1) - pand xmm5, [tfe GLOBAL] ; set lsb of each byte to zero - psrlw xmm5, 1 ; abs(p1-q1)/2 + paddsb xmm2, xmm0 ; 3 * (q0 - p0) + hvm(p1 - q1) + pand xmm1, xmm2 ; mask filter values we don't care about - mov rdx, arg(2) ;flimit ; - movdqa xmm2, [rdx] ; flimit + movdqa xmm2, xmm1 + paddsb xmm1, [t4 GLOBAL] ; 3* (q0 - p0) + hvm(p1 - q1) + 4 - movdqa xmm1, xmm3 ; p0 - movdqa xmm7, xmm6 ; q0 - psubusb xmm1, xmm7 ; p0-q0 - psubusb xmm7, xmm3 ; q0-p0 - por xmm1, xmm7 ; abs(q0-p0) - paddusb xmm1, xmm1 ; abs(q0-p0)*2 - paddusb xmm1, xmm5 ; abs (p0 - q0) *2 + abs(p1-q1)/2 + paddsb xmm2, [t3 GLOBAL] ; 3* (q0 - p0) + hvm(p1 - q1) + 3 - paddb xmm2, xmm2 ; flimit*2 (less than 255) - paddb xmm4, xmm2 ; flimit * 2 + limit (less than 255) + punpckhbw xmm5, xmm2 + punpcklbw xmm2, xmm2 - psubusb xmm1, xmm4 ; abs (p0 - q0) *2 + abs(p1-q1)/2 > flimit * 2 + limit - por xmm1, xmm0; ; mask - pxor xmm0, xmm0 - pcmpeqb xmm1, xmm0 + psraw xmm5, 11 + psraw xmm2, 11 - ; calculate high edge variance - mov rdx, arg(4) ;thresh ; get thresh - movdqa xmm7, [rdx] + packsswb xmm2, xmm5 ; (3* (q0 - p0) + hvm(p1 - q1) + 3) >> 3; + punpcklbw xmm0, xmm1 ; exfxgxhx - movdqa xmm4, t0 ; get abs (q1 - q0) - psubusb xmm4, xmm7 ; abs(q1 - q0) > thresh + punpckhbw xmm1, xmm1 ; axbxcxdx + psraw xmm0, 11 ; sign extended shift right by 3 - movdqa xmm3, t1 ; get abs (p1 - p0) - psubusb xmm3, xmm7 ; abs(p1 - p0)> thresh + psraw xmm1, 11 ; sign extended shift right by 3 + movdqa xmm5, xmm0 ; save results - por xmm4, xmm3 ; abs(q1 - q0) > thresh || abs(p1 - p0) > thresh - pcmpeqb xmm4, xmm0 + packsswb xmm0, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>3 + paddsw xmm5, [ones GLOBAL] - pcmpeqb xmm0, xmm0 - pxor xmm4, xmm0 + paddsw xmm1, [ones GLOBAL] + psraw xmm5, 1 ; partial shifted one more time for 2nd tap + + psraw xmm1, 1 ; partial shifted one more time for 2nd tap + packsswb xmm5, xmm1 ; (3* (q0 - p0) + hvm(p1 - q1) + 4) >>4 + + pandn xmm4, xmm5 ; high edge variance additive + + paddsb xmm6, xmm2 ; p0+= p0 add + pxor xmm6, [t80 GLOBAL] ; unoffset + + movdqa xmm1, [rdx] ; p1 + pxor xmm1, [t80 GLOBAL] ; reoffset + + paddsb xmm1, xmm4 ; p1+= p1 add + pxor xmm1, [t80 GLOBAL] ; unoffset + + psubsb xmm3, xmm0 ; q0-= q0 add + pxor xmm3, [t80 GLOBAL] ; unoffset + + psubsb xmm7, xmm4 ; q1-= q1 add + pxor xmm7, [t80 GLOBAL] ; unoffset +%endmacro + +%macro BV_TRANSPOSE 0 + ; xmm1 = f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 + ; xmm6 = f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03 + ; xmm3 = f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 + ; xmm7 = f5 e5 d5 c5 b5 a5 95 85 75 65 55 45 35 25 15 05 + movdqa xmm2, xmm1 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 + punpcklbw xmm2, xmm6 ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02 + + movdqa xmm4, xmm3 ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 + punpckhbw xmm1, xmm6 ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82 + + punpcklbw xmm4, xmm7 ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04 + punpckhbw xmm3, xmm7 ; f5 f4 e5 e4 d5 d4 c5 c4 b5 b4 a5 a4 95 94 85 84 + + movdqa xmm6, xmm2 ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02 + punpcklwd xmm2, xmm4 ; 35 34 33 32 25 24 23 22 15 14 13 12 05 04 03 02 + + punpckhwd xmm6, xmm4 ; 75 74 73 72 65 64 63 62 55 54 53 52 45 44 43 42 + movdqa xmm5, xmm1 ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82 + + punpcklwd xmm1, xmm3 ; b5 b4 b3 b2 a5 a4 a3 a2 95 94 93 92 85 84 83 82 + punpckhwd xmm5, xmm3 ; f5 f4 f3 f2 e5 e4 e3 e2 d5 d4 d3 d2 c5 c4 c3 c2 + ; xmm2 = 35 34 33 32 25 24 23 22 15 14 13 12 05 04 03 02 + ; xmm6 = 75 74 73 72 65 64 63 62 55 54 53 52 45 44 43 42 + ; xmm1 = b5 b4 b3 b2 a5 a4 a3 a2 95 94 93 92 85 84 83 82 + ; xmm5 = f5 f4 f3 f2 e5 e4 e3 e2 d5 d4 d3 d2 c5 c4 c3 c2 +%endmacro + +%macro BV_WRITEBACK 2 + movd [rsi+2], %1 + psrldq %1, 4 + + movd [rdi+2], %1 + psrldq %1, 4 + + movd [rsi+2*rax+2], %1 + psrldq %1, 4 + + movd [rdi+2*rax+2], %1 + + movd [rsi+4*rax+2], %2 + psrldq %2, 4 + + movd [rdi+4*rax+2], %2 + psrldq %2, 4 + + movd [rsi+2*rcx+2], %2 + psrldq %2, 4 + + movd [rdi+2*rcx+2], %2 +%endmacro + + +;void vp8_loop_filter_vertical_edge_sse2 +;( +; unsigned char *src_ptr, +; int src_pixel_step, +; const char *flimit, +; const char *limit, +; const char *thresh, +; int count +;) +global sym(vp8_loop_filter_vertical_edge_sse2) +sym(vp8_loop_filter_vertical_edge_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM + GET_GOT rbx + push rsi + push rdi + ; end prolog + + ALIGN_STACK 16, rax + sub rsp, 96 ; reserve 96 bytes + %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16]; + %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16]; + %define srct [rsp + 32] ;__declspec(align(16)) char srct[64]; + + mov rsi, arg(0) ; src_ptr + movsxd rax, dword ptr arg(1) ; src_pixel_step + + lea rsi, [rsi - 4] + lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing + lea rcx, [rax*2+rax] + ;transpose 16x8 to 8x16, and store the 8-line result on stack. + TRANSPOSE_16X8_1 + + lea rsi, [rsi+rax*8] + lea rdi, [rdi+rax*8] + lea rdx, srct + TRANSPOSE_16X8_2 1 + + ; calculate filter mask + LFV_FILTER_MASK 1 + ; calculate high edge variance + LFV_HEV_MASK ; start work on filters - lea rdx, srct + BV_FILTER + + ; tranpose and write back - only work on q1, q0, p0, p1 + BV_TRANSPOSE + ; store 16-line result + + lea rdx, [rax] + neg rdx + + BV_WRITEBACK xmm1, xmm5 + + lea rsi, [rsi+rdx*8] + lea rdi, [rdi+rdx*8] + BV_WRITEBACK xmm2, xmm6 + + add rsp, 96 + pop rsp + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + RESTORE_XMM + UNSHADOW_ARGS + pop rbp + ret + + +;void vp8_loop_filter_vertical_edge_uv_sse2 +;( +; unsigned char *u, +; int src_pixel_step, +; const char *flimit, +; const char *limit, +; const char *thresh, +; unsigned char *v +;) +global sym(vp8_loop_filter_vertical_edge_uv_sse2) +sym(vp8_loop_filter_vertical_edge_uv_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM + GET_GOT rbx + push rsi + push rdi + ; end prolog + + ALIGN_STACK 16, rax + sub rsp, 96 ; reserve 96 bytes + %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16]; + %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16]; + %define srct [rsp + 32] ;__declspec(align(16)) char srct[64]; + + mov rsi, arg(0) ; u_ptr + movsxd rax, dword ptr arg(1) ; src_pixel_step + + lea rsi, [rsi - 4] + lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing + lea rcx, [rax+2*rax] + + ;transpose 16x8 to 8x16, and store the 8-line result on stack. + TRANSPOSE_16X8_1 + + mov rsi, arg(5) ; v_ptr + lea rsi, [rsi - 4] + lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing + + lea rdx, srct + TRANSPOSE_16X8_2 1 + + ; calculate filter mask + LFV_FILTER_MASK 1 + ; calculate high edge variance + LFV_HEV_MASK ; start work on filters - movdqa xmm2, [rdx+32] ; p1 - movdqa xmm7, [rdx+80] ; q1 + BV_FILTER + + ; tranpose and write back - only work on q1, q0, p0, p1 + BV_TRANSPOSE + + lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing + + ; store 16-line result + BV_WRITEBACK xmm1, xmm5 + + mov rsi, arg(0) ; u_ptr + lea rsi, [rsi - 4] + lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing + BV_WRITEBACK xmm2, xmm6 + + add rsp, 96 + pop rsp + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + RESTORE_XMM + UNSHADOW_ARGS + pop rbp + ret - pxor xmm2, [t80 GLOBAL] ; p1 offset to convert to signed values - pxor xmm7, [t80 GLOBAL] ; q1 offset to convert to signed values - psubsb xmm2, xmm7 ; p1 - q1 - movdqa xmm6, [rdx+48] ; p0 +%macro MBV_FILTER 0 + lea rdx, srct - movdqa xmm0, [rdx+64] ; q0 - pxor xmm6, [t80 GLOBAL] ; offset to convert to signed values + movdqa xmm2, [rdx+32] ; p1 + movdqa xmm7, [rdx+80] ; q1 + movdqa xmm6, [rdx+48] ; p0 + movdqa xmm0, [rdx+64] ; q0 - pxor xmm0, [t80 GLOBAL] ; offset to convert to signed values - movdqa xmm3, xmm0 ; q0 + pxor xmm2, [t80 GLOBAL] ; p1 offset to convert to signed values + pxor xmm7, [t80 GLOBAL] ; q1 offset to convert to signed values + pxor xmm6, [t80 GLOBAL] ; offset to convert to signed values + pxor xmm0, [t80 GLOBAL] ; offset to convert to signed values - psubsb xmm0, xmm6 ; q0 - p0 - paddsb xmm2, xmm0 ; 1 * (q0 - p0) + (p1 - q1) + psubsb xmm2, xmm7 ; p1 - q1 - paddsb xmm2, xmm0 ; 2 * (q0 - p0) - paddsb xmm2, xmm0 ; 3 * (q0 - p0)+ (p1 - q1) + movdqa xmm3, xmm0 ; q0 - pand xmm1, xmm2 ; mask filter values we don't care about + psubsb xmm0, xmm6 ; q0 - p0 + paddsb xmm2, xmm0 ; 1 * (q0 - p0) + (p1 - q1) - ; xmm1 = vp8_filter, xmm4=hev, xmm6=ps0, xmm3=qs0 - movdqa xmm2, xmm1 ; vp8_filter - pand xmm2, xmm4; ; Filter2 = vp8_filter & hev + paddsb xmm2, xmm0 ; 2 * (q0 - p0) + paddsb xmm2, xmm0 ; 3 * (q0 - p0)+ (p1 - q1) - movdqa xmm5, xmm2 - paddsb xmm5, [t3 GLOBAL] + pand xmm1, xmm2 ; mask filter values we don't care about - pxor xmm0, xmm0 ; 0 - pxor xmm7, xmm7 ; 0 + movdqa xmm2, xmm1 ; vp8_filter + pand xmm2, xmm4; ; Filter2 = vp8_filter & hev - punpcklbw xmm0, xmm5 ; e0f0g0h0 - psraw xmm0, 11 ; sign extended shift right by 3 + movdqa xmm5, xmm2 + paddsb xmm5, [t3 GLOBAL] - punpckhbw xmm7, xmm5 ; a0b0c0d0 - psraw xmm7, 11 ; sign extended shift right by 3 + punpckhbw xmm7, xmm5 ; axbxcxdx + punpcklbw xmm5, xmm5 ; exfxgxhx - packsswb xmm0, xmm7 ; Filter2 >>=3; - movdqa xmm5, xmm0 ; Filter2 + psraw xmm7, 11 ; sign extended shift right by 3 + psraw xmm5, 11 ; sign extended shift right by 3 - paddsb xmm2, [t4 GLOBAL] ; vp8_signed_char_clamp(Filter2 + 4) - pxor xmm0, xmm0 ; 0 + packsswb xmm5, xmm7 ; Filter2 >>=3; - pxor xmm7, xmm7 ; 0 - punpcklbw xmm0, xmm2 ; e0f0g0h0 + paddsb xmm2, [t4 GLOBAL] ; vp8_signed_char_clamp(Filter2 + 4) - psraw xmm0, 11 ; sign extended shift right by 3 - punpckhbw xmm7, xmm2 ; a0b0c0d0 + punpcklbw xmm0, xmm2 ; exfxgxhx + punpckhbw xmm7, xmm2 ; axbxcxdx - psraw xmm7, 11 ; sign extended shift right by 3 - packsswb xmm0, xmm7 ; Filter2 >>=3; + psraw xmm0, 11 ; sign extended shift right by 3 + psraw xmm7, 11 ; sign extended shift right by 3 - ; xmm0= filter2 xmm1 = vp8_filter, xmm3 =qs0 xmm5=s xmm4 =hev xmm6=ps0 - psubsb xmm3, xmm0 ; qs0 =qs0 - filter1 - paddsb xmm6, xmm5 ; ps0 =ps0 + Fitler2 + packsswb xmm0, xmm7 ; Filter2 >>=3; + psubsb xmm3, xmm0 ; qs0 =qs0 - filter1 + paddsb xmm6, xmm5 ; ps0 =ps0 + Fitler2 - ; xmm1=vp8_filter, xmm3=qs0, xmm4 =hev xmm6=ps0 ; vp8_filter &= ~hev; ; Filter2 = vp8_filter; - pandn xmm4, xmm1 ; vp8_filter&=~hev + pandn xmm4, xmm1 ; vp8_filter&=~hev - ; xmm3=qs0, xmm4=filter2, xmm6=ps0 ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7); ; s = vp8_signed_char_clamp(qs0 - u); ; *oq0 = s^0x80; ; s = vp8_signed_char_clamp(ps0 + u); ; *op0 = s^0x80; - pxor xmm0, xmm0 - pxor xmm1, xmm1 + pxor xmm1, xmm1 - pxor xmm2, xmm2 - punpcklbw xmm1, xmm4 + pxor xmm2, xmm2 + punpcklbw xmm1, xmm4 - punpckhbw xmm2, xmm4 - pmulhw xmm1, [s27 GLOBAL] + punpckhbw xmm2, xmm4 + pmulhw xmm1, [s27 GLOBAL] - pmulhw xmm2, [s27 GLOBAL] - paddw xmm1, [s63 GLOBAL] + pmulhw xmm2, [s27 GLOBAL] + paddw xmm1, [s63 GLOBAL] - paddw xmm2, [s63 GLOBAL] - psraw xmm1, 7 + paddw xmm2, [s63 GLOBAL] + psraw xmm1, 7 - psraw xmm2, 7 - packsswb xmm1, xmm2 + psraw xmm2, 7 + packsswb xmm1, xmm2 - psubsb xmm3, xmm1 - paddsb xmm6, xmm1 + psubsb xmm3, xmm1 + paddsb xmm6, xmm1 - pxor xmm3, [t80 GLOBAL] - pxor xmm6, [t80 GLOBAL] + pxor xmm3, [t80 GLOBAL] + pxor xmm6, [t80 GLOBAL] - movdqa [rdx+48], xmm6 - movdqa [rdx+64], xmm3 + movdqa [rdx+48], xmm6 + movdqa [rdx+64], xmm3 ; roughly 2/7th difference across boundary ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7); @@ -1385,38 +1357,37 @@ sym(vp8_mbloop_filter_vertical_edge_sse2): ; *oq1 = s^0x80; ; s = vp8_signed_char_clamp(ps1 + u); ; *op1 = s^0x80; - pxor xmm1, xmm1 - pxor xmm2, xmm2 - - punpcklbw xmm1, xmm4 - punpckhbw xmm2, xmm4 + pxor xmm1, xmm1 + pxor xmm2, xmm2 - pmulhw xmm1, [s18 GLOBAL] - pmulhw xmm2, [s18 GLOBAL] + punpcklbw xmm1, xmm4 + punpckhbw xmm2, xmm4 - paddw xmm1, [s63 GLOBAL] - paddw xmm2, [s63 GLOBAL] + pmulhw xmm1, [s18 GLOBAL] + pmulhw xmm2, [s18 GLOBAL] - psraw xmm1, 7 - psraw xmm2, 7 + paddw xmm1, [s63 GLOBAL] + paddw xmm2, [s63 GLOBAL] - packsswb xmm1, xmm2 + psraw xmm1, 7 + psraw xmm2, 7 - movdqa xmm3, [rdx + 80] ;/q1 - movdqa xmm6, [rdx + 32] ; p1 + packsswb xmm1, xmm2 - pxor xmm3, [t80 GLOBAL] - pxor xmm6, [t80 GLOBAL] + movdqa xmm3, [rdx + 80] ; q1 + movdqa xmm6, [rdx + 32] ; p1 - paddsb xmm6, xmm1 - psubsb xmm3, xmm1 + pxor xmm3, [t80 GLOBAL] + pxor xmm6, [t80 GLOBAL] - pxor xmm6, [t80 GLOBAL] - pxor xmm3, [t80 GLOBAL] + paddsb xmm6, xmm1 + psubsb xmm3, xmm1 - movdqa [rdx + 80], xmm3 - movdqa [rdx + 32], xmm6 + pxor xmm6, [t80 GLOBAL] + pxor xmm3, [t80 GLOBAL] + movdqa [rdx + 80], xmm3 + movdqa [rdx + 32], xmm6 ; roughly 1/7th difference across boundary ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7); @@ -1424,138 +1395,281 @@ sym(vp8_mbloop_filter_vertical_edge_sse2): ; *oq2 = s^0x80; ; s = vp8_signed_char_clamp(ps2 + u); ; *op2 = s^0x80; - pxor xmm1, xmm1 - pxor xmm2, xmm2 + pxor xmm1, xmm1 + pxor xmm2, xmm2 - punpcklbw xmm1, xmm4 - punpckhbw xmm2, xmm4 + punpcklbw xmm1, xmm4 + punpckhbw xmm2, xmm4 - pmulhw xmm1, [s9 GLOBAL] - pmulhw xmm2, [s9 GLOBAL] + pmulhw xmm1, [s9 GLOBAL] + pmulhw xmm2, [s9 GLOBAL] - paddw xmm1, [s63 GLOBAL] - paddw xmm2, [s63 GLOBAL] + paddw xmm1, [s63 GLOBAL] + paddw xmm2, [s63 GLOBAL] - psraw xmm1, 7 - psraw xmm2, 7 + psraw xmm1, 7 + psraw xmm2, 7 - packsswb xmm1, xmm2 + packsswb xmm1, xmm2 - movdqa xmm6, [rdx+16] - movdqa xmm3, [rdx+96] + movdqa xmm6, [rdx+16] + movdqa xmm3, [rdx+96] - pxor xmm6, [t80 GLOBAL] - pxor xmm3, [t80 GLOBAL] + pxor xmm6, [t80 GLOBAL] + pxor xmm3, [t80 GLOBAL] - paddsb xmm6, xmm1 - psubsb xmm3, xmm1 + paddsb xmm6, xmm1 + psubsb xmm3, xmm1 - pxor xmm6, [t80 GLOBAL] ; xmm6 = f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01 - pxor xmm3, [t80 GLOBAL] ; xmm3 = f6 e6 d6 c6 b6 a6 96 86 76 66 56 46 36 26 15 06 + pxor xmm6, [t80 GLOBAL] ; xmm6 = f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01 + pxor xmm3, [t80 GLOBAL] ; xmm3 = f6 e6 d6 c6 b6 a6 96 86 76 66 56 46 36 26 15 06 +%endmacro +%macro MBV_TRANSPOSE 0 + movdqa xmm0, [rdx] ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00 + movdqa xmm1, xmm0 ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00 - ; transpose and write back - movdqa xmm0, [rdx] ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00 - movdqa xmm1, xmm0 ; f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00 + punpcklbw xmm0, xmm6 ; 71 70 61 60 51 50 41 40 31 30 21 20 11 10 01 00 + punpckhbw xmm1, xmm6 ; f1 f0 e1 e0 d1 d0 c1 c0 b1 b0 a1 a0 91 90 81 80 + + movdqa xmm2, [rdx+32] ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 + movdqa xmm6, xmm2 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 + + punpcklbw xmm2, [rdx+48] ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02 + punpckhbw xmm6, [rdx+48] ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82 + + movdqa xmm5, xmm0 ; 71 70 61 60 51 50 41 40 31 30 21 20 11 10 01 00 + punpcklwd xmm0, xmm2 ; 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00 + + punpckhwd xmm5, xmm2 ; 73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40 + movdqa xmm4, xmm1 ; f1 f0 e1 e0 d1 d0 c1 c0 b1 b0 a1 a0 91 90 81 80 + + punpcklwd xmm1, xmm6 ; b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80 + punpckhwd xmm4, xmm6 ; f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0 + + movdqa xmm2, [rdx+64] ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 + punpcklbw xmm2, [rdx+80] ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04 + + movdqa xmm6, xmm3 ; f6 e6 d6 c6 b6 a6 96 86 76 66 56 46 36 26 16 06 + punpcklbw xmm6, [rdx+112] ; 77 76 67 66 57 56 47 46 37 36 27 26 17 16 07 06 - punpcklbw xmm0, xmm6 ; 71 70 61 60 51 50 41 40 31 30 21 20 11 10 01 00 - punpckhbw xmm1, xmm6 ; f1 f0 e1 e0 d1 d0 c1 c0 b1 b0 a1 a0 91 90 81 80 + movdqa xmm7, xmm2 ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04 + punpcklwd xmm2, xmm6 ; 37 36 35 34 27 26 25 24 17 16 15 14 07 06 05 04 - movdqa xmm2, [rdx+32] ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 - movdqa xmm6, xmm2 ; f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02 + punpckhwd xmm7, xmm6 ; 77 76 75 74 67 66 65 64 57 56 55 54 47 46 45 44 + movdqa xmm6, xmm0 ; 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00 - punpcklbw xmm2, [rdx+48] ; 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02 - punpckhbw xmm6, [rdx+48] ; f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82 + punpckldq xmm0, xmm2 ; 17 16 15 14 13 12 11 10 07 06 05 04 03 02 01 00 + punpckhdq xmm6, xmm2 ; 37 36 35 34 33 32 31 30 27 26 25 24 23 22 21 20 +%endmacro - movdqa xmm5, xmm0 ; 71 70 61 60 51 50 41 40 31 30 21 20 11 10 01 00 - punpcklwd xmm0, xmm2 ; 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00 +%macro MBV_WRITEBACK_1 0 + movq QWORD PTR [rsi], xmm0 + psrldq xmm0, 8 - punpckhwd xmm5, xmm2 ; 73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40 - movdqa xmm4, xmm1 ; f1 f0 e1 e0 d1 d0 c1 c0 b1 b0 a1 a0 91 90 81 80 + movq QWORD PTR [rdi], xmm0 - punpcklwd xmm1, xmm6 ; b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80 - punpckhwd xmm4, xmm6 ; f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0 + movq QWORD PTR [rsi+2*rax], xmm6 + psrldq xmm6, 8 - movdqa xmm2, [rdx+64] ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 - punpcklbw xmm2, [rdx+80] ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04 + movq QWORD PTR [rdi+2*rax], xmm6 - movdqa xmm6, xmm3 ; f6 e6 d6 c6 b6 a6 96 86 76 66 56 46 36 26 16 06 - punpcklbw xmm6, [rdx+112] ; 77 76 67 66 57 56 47 46 37 36 27 26 17 16 07 06 + movdqa xmm0, xmm5 ; 73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40 + punpckldq xmm0, xmm7 ; 57 56 55 54 53 52 51 50 47 46 45 44 43 42 41 40 - movdqa xmm7, xmm2 ; 75 74 65 64 55 54 45 44 35 34 25 24 15 14 05 04 - punpcklwd xmm2, xmm6 ; 37 36 35 34 27 26 25 24 17 16 15 14 07 06 05 04 + punpckhdq xmm5, xmm7 ; 77 76 75 74 73 72 71 70 67 66 65 64 63 62 61 60 - punpckhwd xmm7, xmm6 ; 77 76 75 74 67 66 65 64 57 56 55 54 47 46 45 44 - movdqa xmm6, xmm0 ; 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00 + movq QWORD PTR [rsi+4*rax], xmm0 + psrldq xmm0, 8 - punpckldq xmm0, xmm2 ; 17 16 15 14 13 12 11 10 07 06 05 04 03 02 01 00 - punpckhdq xmm6, xmm2 ; 37 36 35 34 33 32 31 30 27 26 25 24 23 22 21 20 + movq QWORD PTR [rdi+4*rax], xmm0 - lea rsi, [rsi+rcx*8] - lea rdi, [rdi+rcx*8] + movq QWORD PTR [rsi+2*rcx], xmm5 + psrldq xmm5, 8 - movq QWORD PTR [rsi+rcx*4], xmm0 - psrldq xmm0, 8 + movq QWORD PTR [rdi+2*rcx], xmm5 - movq QWORD PTR [rsi+rcx*2], xmm6 - psrldq xmm6, 8 + movdqa xmm2, [rdx+64] ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 + punpckhbw xmm2, [rdx+80] ; f5 f4 e5 e4 d5 d4 c5 c4 b5 b4 a5 a4 95 94 85 84 - movq QWORD PTR [rdi+rcx*4], xmm0 - movq QWORD PTR [rsi+rcx], xmm6 + punpckhbw xmm3, [rdx+112] ; f7 f6 e7 e6 d7 d6 c7 c6 b7 b6 a7 a6 97 96 87 86 + movdqa xmm0, xmm2 - movdqa xmm0, xmm5 ; 73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40 - punpckldq xmm0, xmm7 ; 57 56 55 54 53 52 51 50 47 46 45 44 43 42 41 40 + punpcklwd xmm0, xmm3 ; b7 b6 b4 b4 a7 a6 a5 a4 97 96 95 94 87 86 85 84 + punpckhwd xmm2, xmm3 ; f7 f6 f5 f4 e7 e6 e5 e4 d7 d6 d5 d4 c7 c6 c5 c4 - punpckhdq xmm5, xmm7 ; 77 76 75 74 73 72 71 70 67 66 65 64 63 62 61 60 + movdqa xmm3, xmm1 ; b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80 + punpckldq xmm1, xmm0 ; 97 96 95 94 93 92 91 90 87 86 85 83 84 82 81 80 - movq QWORD PTR [rsi], xmm0 - psrldq xmm0, 8 + punpckhdq xmm3, xmm0 ; b7 b6 b5 b4 b3 b2 b1 b0 a7 a6 a5 a4 a3 a2 a1 a0 +%endmacro - movq QWORD PTR [rsi+rax*2], xmm5 - psrldq xmm5, 8 +%macro MBV_WRITEBACK_2 0 + movq QWORD PTR [rsi], xmm1 + psrldq xmm1, 8 - movq QWORD PTR [rsi+rax], xmm0 - movq QWORD PTR [rdi+rax*2], xmm5 + movq QWORD PTR [rdi], xmm1 - movdqa xmm2, [rdx+64] ; f4 e4 d4 c4 b4 a4 94 84 74 64 54 44 34 24 14 04 - punpckhbw xmm2, [rdx+80] ; f5 f4 e5 e4 d5 d4 c5 c4 b5 b4 a5 a4 95 94 85 84 + movq QWORD PTR [rsi+2*rax], xmm3 + psrldq xmm3, 8 - punpckhbw xmm3, [rdx+112] ; f7 f6 e7 e6 d7 d6 c7 c6 b7 b6 a7 a6 97 96 87 86 - movdqa xmm0, xmm2 + movq QWORD PTR [rdi+2*rax], xmm3 - punpcklwd xmm0, xmm3 ; b7 b6 b4 b4 a7 a6 a5 a4 97 96 95 94 87 86 85 84 - punpckhwd xmm2, xmm3 ; f7 f6 f5 f4 e7 e6 e5 e4 d7 d6 d5 d4 c7 c6 c5 c4 + movdqa xmm1, xmm4 ; f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0 + punpckldq xmm1, xmm2 ; d7 d6 d5 d4 d3 d2 d1 d0 c7 c6 c5 c4 c3 c2 c1 c0 - movdqa xmm3, xmm1 ; b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80 - punpckldq xmm1, xmm0 ; 97 96 95 94 93 92 91 90 87 86 85 83 84 82 81 80 + punpckhdq xmm4, xmm2 ; f7 f6 f4 f4 f3 f2 f1 f0 e7 e6 e5 e4 e3 e2 e1 e0 + movq QWORD PTR [rsi+4*rax], xmm1 - punpckhdq xmm3, xmm0 ; b7 b6 b5 b4 b3 b2 b1 b0 a7 a6 a5 a4 a3 a2 a1 a0 + psrldq xmm1, 8 - lea rsi, [rsi+rax*8] - lea rdi, [rdi+rax*8] + movq QWORD PTR [rdi+4*rax], xmm1 - movq QWORD PTR [rsi+rcx*4], xmm1 - psrldq xmm1, 8 + movq QWORD PTR [rsi+2*rcx], xmm4 + psrldq xmm4, 8 - movq QWORD PTR [rsi+rcx*2], xmm3 - psrldq xmm3, 8 + movq QWORD PTR [rdi+2*rcx], xmm4 +%endmacro - movq QWORD PTR [rdi+rcx*4], xmm1 - movq QWORD PTR [rsi+rcx], xmm3 - movdqa xmm1, xmm4 ; f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0 - punpckldq xmm1, xmm2 ; d7 d6 d5 d4 d3 d2 d1 d0 c7 c6 c5 c4 c3 c2 c1 c0 +;void vp8_mbloop_filter_vertical_edge_sse2 +;( +; unsigned char *src_ptr, +; int src_pixel_step, +; const char *flimit, +; const char *limit, +; const char *thresh, +; int count +;) +global sym(vp8_mbloop_filter_vertical_edge_sse2) +sym(vp8_mbloop_filter_vertical_edge_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM + GET_GOT rbx + push rsi + push rdi + ; end prolog + + ALIGN_STACK 16, rax + sub rsp, 160 ; reserve 160 bytes + %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16]; + %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16]; + %define srct [rsp + 32] ;__declspec(align(16)) char srct[128]; + + mov rsi, arg(0) ; src_ptr + movsxd rax, dword ptr arg(1) ; src_pixel_step + + lea rsi, [rsi - 4] + lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing + lea rcx, [rax*2+rax] + + ; Transpose + TRANSPOSE_16X8_1 + + lea rsi, [rsi+rax*8] + lea rdi, [rdi+rax*8] + lea rdx, srct + TRANSPOSE_16X8_2 0 + + ; calculate filter mask + LFV_FILTER_MASK 0 + ; calculate high edge variance + LFV_HEV_MASK + + neg rax + ; start work on filters + MBV_FILTER + + lea rsi, [rsi+rax*8] + lea rdi, [rdi+rax*8] + + ; transpose and write back + MBV_TRANSPOSE + + neg rax + + MBV_WRITEBACK_1 - punpckhdq xmm4, xmm2 ; f7 f6 f4 f4 f3 f2 f1 f0 e7 e6 e5 e4 e3 e2 e1 e0 - movq QWORD PTR [rsi], xmm1 + lea rsi, [rsi+rax*8] + lea rdi, [rdi+rax*8] + MBV_WRITEBACK_2 + + add rsp, 160 + pop rsp + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + RESTORE_XMM + UNSHADOW_ARGS + pop rbp + ret + + +;void vp8_mbloop_filter_vertical_edge_uv_sse2 +;( +; unsigned char *u, +; int src_pixel_step, +; const char *flimit, +; const char *limit, +; const char *thresh, +; unsigned char *v +;) +global sym(vp8_mbloop_filter_vertical_edge_uv_sse2) +sym(vp8_mbloop_filter_vertical_edge_uv_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM + GET_GOT rbx + push rsi + push rdi + ; end prolog + + ALIGN_STACK 16, rax + sub rsp, 160 ; reserve 160 bytes + %define t0 [rsp + 0] ;__declspec(align(16)) char t0[16]; + %define t1 [rsp + 16] ;__declspec(align(16)) char t1[16]; + %define srct [rsp + 32] ;__declspec(align(16)) char srct[128]; - psrldq xmm1, 8 + mov rsi, arg(0) ; u_ptr + movsxd rax, dword ptr arg(1) ; src_pixel_step - movq QWORD PTR [rsi+rax*2], xmm4 - psrldq xmm4, 8 + lea rsi, [rsi - 4] + lea rdi, [rsi + rax] ; rdi points to row +1 for indirect addressing + lea rcx, [rax+2*rax] + + ; Transpose + TRANSPOSE_16X8_1 + + ; XMM3 XMM4 XMM7 in use + mov rsi, arg(5) ; v_ptr + lea rsi, [rsi - 4] + lea rdi, [rsi + rax] + lea rdx, srct + TRANSPOSE_16X8_2 0 + + ; calculate filter mask + LFV_FILTER_MASK 0 + ; calculate high edge variance + LFV_HEV_MASK + + ; start work on filters + MBV_FILTER + + ; transpose and write back + MBV_TRANSPOSE - movq QWORD PTR [rsi+rax], xmm1 - movq QWORD PTR [rdi+rax*2], xmm4 + mov rsi, arg(0) ;u_ptr + lea rsi, [rsi - 4] + lea rdi, [rsi + rax] + MBV_WRITEBACK_1 + mov rsi, arg(5) ;v_ptr + lea rsi, [rsi - 4] + lea rdi, [rsi + rax] + MBV_WRITEBACK_2 add rsp, 160 pop rsp @@ -1563,6 +1677,7 @@ sym(vp8_mbloop_filter_vertical_edge_sse2): pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret @@ -1582,6 +1697,7 @@ sym(vp8_loop_filter_simple_horizontal_edge_sse2): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 6 + SAVE_XMM GET_GOT rbx push rsi push rdi @@ -1678,6 +1794,7 @@ sym(vp8_loop_filter_simple_horizontal_edge_sse2): pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret @@ -1697,6 +1814,7 @@ sym(vp8_loop_filter_simple_vertical_edge_sse2): push rbp ; save old base pointer value. mov rbp, rsp ; set new base pointer value. SHADOW_ARGS_TO_STACK 6 + SAVE_XMM GET_GOT rbx ; save callee-saved reg push rsi push rdi @@ -1941,6 +2059,7 @@ sym(vp8_loop_filter_simple_vertical_edge_sse2): pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret diff --git a/vp8/common/x86/loopfilter_x86.c b/vp8/common/x86/loopfilter_x86.c index 143ee74..3ff8c4e 100644 --- a/vp8/common/x86/loopfilter_x86.c +++ b/vp8/common/x86/loopfilter_x86.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -33,6 +34,11 @@ prototype_loopfilter(vp8_loop_filter_simple_vertical_edge_sse2); prototype_loopfilter(vp8_loop_filter_simple_horizontal_edge_sse2); prototype_loopfilter(vp8_fast_loop_filter_vertical_edges_sse2); +extern loop_filter_uvfunction vp8_loop_filter_horizontal_edge_uv_sse2; +extern loop_filter_uvfunction vp8_loop_filter_vertical_edge_uv_sse2; +extern loop_filter_uvfunction vp8_mbloop_filter_horizontal_edge_uv_sse2; +extern loop_filter_uvfunction vp8_mbloop_filter_vertical_edge_uv_sse2; + #if HAVE_MMX // Horizontal MB filtering void vp8_loop_filter_mbh_mmx(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, @@ -156,10 +162,7 @@ void vp8_loop_filter_mbh_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsign vp8_mbloop_filter_horizontal_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2); if (u_ptr) - vp8_mbloop_filter_horizontal_edge_mmx(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1); - - if (v_ptr) - vp8_mbloop_filter_horizontal_edge_mmx(v_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1); + vp8_mbloop_filter_horizontal_edge_uv_sse2(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, v_ptr); } @@ -182,10 +185,7 @@ void vp8_loop_filter_mbv_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsign vp8_mbloop_filter_vertical_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2); if (u_ptr) - vp8_mbloop_filter_vertical_edge_mmx(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1); - - if (v_ptr) - vp8_mbloop_filter_vertical_edge_mmx(v_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1); + vp8_mbloop_filter_vertical_edge_uv_sse2(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, v_ptr); } @@ -210,10 +210,7 @@ void vp8_loop_filter_bh_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigne vp8_loop_filter_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2); if (u_ptr) - vp8_loop_filter_horizontal_edge_mmx(u_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1); - - if (v_ptr) - vp8_loop_filter_horizontal_edge_mmx(v_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1); + vp8_loop_filter_horizontal_edge_uv_sse2(u_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, v_ptr + 4 * uv_stride); } @@ -240,10 +237,7 @@ void vp8_loop_filter_bv_sse2(unsigned char *y_ptr, unsigned char *u_ptr, unsigne vp8_loop_filter_vertical_edge_sse2(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2); if (u_ptr) - vp8_loop_filter_vertical_edge_mmx(u_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1); - - if (v_ptr) - vp8_loop_filter_vertical_edge_mmx(v_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1); + vp8_loop_filter_vertical_edge_uv_sse2(u_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, v_ptr + 4); } diff --git a/vp8/common/x86/loopfilter_x86.h b/vp8/common/x86/loopfilter_x86.h index c87f38a..80dbebc 100644 --- a/vp8/common/x86/loopfilter_x86.h +++ b/vp8/common/x86/loopfilter_x86.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/x86/postproc_mmx.asm b/vp8/common/x86/postproc_mmx.asm index 721c8d6..349ac0d 100644 --- a/vp8/common/x86/postproc_mmx.asm +++ b/vp8/common/x86/postproc_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/x86/postproc_mmx.c b/vp8/common/x86/postproc_mmx.c index 095797b..6b6321a 100644 --- a/vp8/common/x86/postproc_mmx.c +++ b/vp8/common/x86/postproc_mmx.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/x86/postproc_sse2.asm b/vp8/common/x86/postproc_sse2.asm index bfa36fa..276f208 100644 --- a/vp8/common/x86/postproc_sse2.asm +++ b/vp8/common/x86/postproc_sse2.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -25,6 +26,7 @@ sym(vp8_post_proc_down_and_across_xmm): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 7 + SAVE_XMM GET_GOT rbx push rsi push rdi @@ -239,6 +241,7 @@ acrossnextcol: pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret @@ -253,6 +256,7 @@ sym(vp8_mbpost_proc_down_xmm): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 5 + SAVE_XMM GET_GOT rbx push rsi push rdi @@ -438,6 +442,7 @@ loop_row: pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret @@ -451,6 +456,7 @@ sym(vp8_mbpost_proc_across_ip_xmm): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 5 + SAVE_XMM GET_GOT rbx push rsi push rdi @@ -611,6 +617,7 @@ nextcol4: pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret diff --git a/vp8/common/x86/postproc_x86.h b/vp8/common/x86/postproc_x86.h index 49a1907..899dd2f 100644 --- a/vp8/common/x86/postproc_x86.h +++ b/vp8/common/x86/postproc_x86.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/x86/recon_mmx.asm b/vp8/common/x86/recon_mmx.asm index ba60c5d..e7211fc 100644 --- a/vp8/common/x86/recon_mmx.asm +++ b/vp8/common/x86/recon_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/common/x86/recon_sse2.asm b/vp8/common/x86/recon_sse2.asm index f2685a7..4ad3973 100644 --- a/vp8/common/x86/recon_sse2.asm +++ b/vp8/common/x86/recon_sse2.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -66,6 +67,7 @@ sym(vp8_recon4b_sse2): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 4 + SAVE_XMM push rsi push rdi ; end prolog @@ -118,6 +120,7 @@ sym(vp8_recon4b_sse2): ; begin epilog pop rdi pop rsi + RESTORE_XMM UNSHADOW_ARGS pop rbp ret diff --git a/vp8/common/x86/recon_x86.h b/vp8/common/x86/recon_x86.h index c469778..40ee65a 100644 --- a/vp8/common/x86/recon_x86.h +++ b/vp8/common/x86/recon_x86.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/x86/subpixel_mmx.asm b/vp8/common/x86/subpixel_mmx.asm index c502118..06db0c6 100644 --- a/vp8/common/x86/subpixel_mmx.asm +++ b/vp8/common/x86/subpixel_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -730,7 +731,7 @@ rd: times 4 dw 0x40 align 16 -global sym(vp8_six_tap_mmx) +global HIDDEN_DATA(sym(vp8_six_tap_mmx)) sym(vp8_six_tap_mmx): times 8 dw 0 times 8 dw 0 @@ -790,7 +791,7 @@ sym(vp8_six_tap_mmx): align 16 -global sym(vp8_bilinear_filters_mmx) +global HIDDEN_DATA(sym(vp8_bilinear_filters_mmx)) sym(vp8_bilinear_filters_mmx): times 8 dw 128 times 8 dw 0 diff --git a/vp8/common/x86/subpixel_sse2.asm b/vp8/common/x86/subpixel_sse2.asm index dee04f2..2385abf 100644 --- a/vp8/common/x86/subpixel_sse2.asm +++ b/vp8/common/x86/subpixel_sse2.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -36,6 +37,7 @@ sym(vp8_filter_block1d8_h6_sse2): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 7 + SAVE_XMM GET_GOT rbx push rsi push rdi @@ -128,6 +130,7 @@ filter_block1d8_h6_rowloop: pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret @@ -154,6 +157,7 @@ sym(vp8_filter_block1d16_h6_sse2): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 7 + SAVE_XMM GET_GOT rbx push rsi push rdi @@ -303,6 +307,7 @@ filter_block1d16_h6_sse2_rowloop: pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret @@ -328,6 +333,7 @@ sym(vp8_filter_block1d8_v6_sse2): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 8 + SAVE_XMM GET_GOT rbx push rsi push rdi @@ -396,221 +402,553 @@ vp8_filter_block1d8_v6_sse2_loop: pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret -;void vp8_unpack_block1d16_h6_sse2 +;void vp8_filter_block1d16_v6_sse2 +;( +; unsigned short *src_ptr, +; unsigned char *output_ptr, +; int dst_ptich, +; unsigned int pixels_per_line, +; unsigned int pixel_step, +; unsigned int output_height, +; unsigned int output_width, +; const short *vp8_filter +;) +;/************************************************************************************ +; Notes: filter_block1d16_v6 applies a 6 tap filter vertically to the input pixels. The +; input pixel array has output_height rows. +;*************************************************************************************/ +global sym(vp8_filter_block1d16_v6_sse2) +sym(vp8_filter_block1d16_v6_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 8 + SAVE_XMM + GET_GOT rbx + push rsi + push rdi + ; end prolog + + mov rax, arg(7) ;vp8_filter + movsxd rdx, dword ptr arg(3) ;pixels_per_line + + mov rdi, arg(1) ;output_ptr + mov rsi, arg(0) ;src_ptr + + sub rsi, rdx + sub rsi, rdx + + movsxd rcx, DWORD PTR arg(5) ;[output_height] +%if ABI_IS_32BIT=0 + movsxd r8, dword ptr arg(2) ; dst_ptich +%endif + +vp8_filter_block1d16_v6_sse2_loop: +; The order for adding 6-tap is 2 5 3 1 4 6. Read in data in that order. + movdqa xmm1, XMMWORD PTR [rsi + rdx] ; line 2 + movdqa xmm2, XMMWORD PTR [rsi + rdx + 16] + pmullw xmm1, [rax + 16] + pmullw xmm2, [rax + 16] + + movdqa xmm3, XMMWORD PTR [rsi + rdx * 4] ; line 5 + movdqa xmm4, XMMWORD PTR [rsi + rdx * 4 + 16] + pmullw xmm3, [rax + 64] + pmullw xmm4, [rax + 64] + + movdqa xmm5, XMMWORD PTR [rsi + rdx * 2] ; line 3 + movdqa xmm6, XMMWORD PTR [rsi + rdx * 2 + 16] + pmullw xmm5, [rax + 32] + pmullw xmm6, [rax + 32] + + movdqa xmm7, XMMWORD PTR [rsi] ; line 1 + movdqa xmm0, XMMWORD PTR [rsi + 16] + pmullw xmm7, [rax] + pmullw xmm0, [rax] + + paddsw xmm1, xmm3 + paddsw xmm2, xmm4 + paddsw xmm1, xmm5 + paddsw xmm2, xmm6 + paddsw xmm1, xmm7 + paddsw xmm2, xmm0 + + add rsi, rdx + + movdqa xmm3, XMMWORD PTR [rsi + rdx * 2] ; line 4 + movdqa xmm4, XMMWORD PTR [rsi + rdx * 2 + 16] + pmullw xmm3, [rax + 48] + pmullw xmm4, [rax + 48] + + movdqa xmm5, XMMWORD PTR [rsi + rdx * 4] ; line 6 + movdqa xmm6, XMMWORD PTR [rsi + rdx * 4 + 16] + pmullw xmm5, [rax + 80] + pmullw xmm6, [rax + 80] + + movdqa xmm7, XMMWORD PTR [rd GLOBAL] + pxor xmm0, xmm0 ; clear xmm0 + + paddsw xmm1, xmm3 + paddsw xmm2, xmm4 + paddsw xmm1, xmm5 + paddsw xmm2, xmm6 + + paddsw xmm1, xmm7 + paddsw xmm2, xmm7 + + psraw xmm1, 7 + psraw xmm2, 7 + + packuswb xmm1, xmm2 ; pack and saturate + movdqa XMMWORD PTR [rdi], xmm1 ; store the results in the destination +%if ABI_IS_32BIT + add rdi, DWORD PTR arg(2) ;[dst_ptich] +%else + add rdi, r8 +%endif + dec rcx ; decrement count + jnz vp8_filter_block1d16_v6_sse2_loop ; next row + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + RESTORE_XMM + UNSHADOW_ARGS + pop rbp + ret + + +;void vp8_filter_block1d8_h6_only_sse2 ;( ; unsigned char *src_ptr, -; unsigned short *output_ptr, ; unsigned int src_pixels_per_line, +; unsigned char *output_ptr, +; int dst_ptich, ; unsigned int output_height, -; unsigned int output_width +; const short *vp8_filter ;) -global sym(vp8_unpack_block1d16_h6_sse2) -sym(vp8_unpack_block1d16_h6_sse2): +; First-pass filter only when yoffset==0 +global sym(vp8_filter_block1d8_h6_only_sse2) +sym(vp8_filter_block1d8_h6_only_sse2): push rbp mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM GET_GOT rbx push rsi push rdi ; end prolog + mov rdx, arg(5) ;vp8_filter mov rsi, arg(0) ;src_ptr - mov rdi, arg(1) ;output_ptr - movsxd rcx, dword ptr arg(3) ;output_height - movsxd rax, dword ptr arg(2) ;src_pixels_per_line ; Pitch for Source + mov rdi, arg(2) ;output_ptr - pxor xmm0, xmm0 ; clear xmm0 for unpack + movsxd rcx, dword ptr arg(4) ;output_height + movsxd rax, dword ptr arg(1) ;src_pixels_per_line ; Pitch for Source %if ABI_IS_32BIT=0 - movsxd r8, dword ptr arg(4) ;output_width ; Pitch for Source + movsxd r8, dword ptr arg(3) ;dst_ptich %endif + pxor xmm0, xmm0 ; clear xmm0 for unpack -unpack_block1d16_h6_sse2_rowloop: - movq xmm1, MMWORD PTR [rsi] ; 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1 -2 - movq xmm3, MMWORD PTR [rsi+8] ; make copy of xmm1 +filter_block1d8_h6_only_rowloop: + movq xmm3, MMWORD PTR [rsi - 2] + movq xmm1, MMWORD PTR [rsi + 6] + + prefetcht2 [rsi+rax-2] + + pslldq xmm1, 8 + por xmm1, xmm3 + + movdqa xmm4, xmm1 + movdqa xmm5, xmm1 + + movdqa xmm6, xmm1 + movdqa xmm7, xmm1 punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2 - punpcklbw xmm1, xmm0 + psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1 + + pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1 + punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1 + + psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 + pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2 + + + punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00 + psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 + + pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3 + + punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01 + psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 + + pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4 + + punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02 + psrldq xmm1, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 + + + pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5 + + punpcklbw xmm1, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03 + pmullw xmm1, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6 - movdqa XMMWORD Ptr [rdi], xmm1 - movdqa XMMWORD Ptr [rdi + 16], xmm3 + paddsw xmm4, xmm7 + paddsw xmm4, xmm5 + + paddsw xmm4, xmm3 + paddsw xmm4, xmm6 + + paddsw xmm4, xmm1 + paddsw xmm4, [rd GLOBAL] + + psraw xmm4, 7 + + packuswb xmm4, xmm0 + + movq QWORD PTR [rdi], xmm4 ; store the results in the destination lea rsi, [rsi + rax] + %if ABI_IS_32BIT - add rdi, DWORD Ptr arg(4) ;[output_width] + add rdi, DWORD Ptr arg(3) ;dst_ptich %else add rdi, r8 %endif dec rcx - jnz unpack_block1d16_h6_sse2_rowloop ; next row + + jnz filter_block1d8_h6_only_rowloop ; next row ; begin epilog pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret -;void vp8_unpack_block1d8_h6_sse2 +;void vp8_filter_block1d16_h6_only_sse2 ;( ; unsigned char *src_ptr, -; unsigned short *output_ptr, ; unsigned int src_pixels_per_line, +; unsigned char *output_ptr, +; int dst_ptich, ; unsigned int output_height, -; unsigned int output_width +; const short *vp8_filter ;) -global sym(vp8_unpack_block1d8_h6_sse2) -sym(vp8_unpack_block1d8_h6_sse2): +; First-pass filter only when yoffset==0 +global sym(vp8_filter_block1d16_h6_only_sse2) +sym(vp8_filter_block1d16_h6_only_sse2): push rbp mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM GET_GOT rbx push rsi push rdi ; end prolog + mov rdx, arg(5) ;vp8_filter mov rsi, arg(0) ;src_ptr - mov rdi, arg(1) ;output_ptr - movsxd rcx, dword ptr arg(3) ;output_height - movsxd rax, dword ptr arg(2) ;src_pixels_per_line ; Pitch for Source + mov rdi, arg(2) ;output_ptr - pxor xmm0, xmm0 ; clear xmm0 for unpack + movsxd rcx, dword ptr arg(4) ;output_height + movsxd rax, dword ptr arg(1) ;src_pixels_per_line ; Pitch for Source %if ABI_IS_32BIT=0 - movsxd r8, dword ptr arg(4) ;output_width ; Pitch for Source + movsxd r8, dword ptr arg(3) ;dst_ptich %endif -unpack_block1d8_h6_sse2_rowloop: - movq xmm1, MMWORD PTR [rsi] ; 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1 -2 - lea rsi, [rsi + rax] + pxor xmm0, xmm0 ; clear xmm0 for unpack - punpcklbw xmm1, xmm0 - movdqa XMMWORD Ptr [rdi], xmm1 +filter_block1d16_h6_only_sse2_rowloop: + movq xmm3, MMWORD PTR [rsi - 2] + movq xmm1, MMWORD PTR [rsi + 6] + + movq xmm2, MMWORD PTR [rsi +14] + pslldq xmm2, 8 + + por xmm2, xmm1 + prefetcht2 [rsi+rax-2] + + pslldq xmm1, 8 + por xmm1, xmm3 + movdqa xmm4, xmm1 + movdqa xmm5, xmm1 + + movdqa xmm6, xmm1 + movdqa xmm7, xmm1 + + punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2 + psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1 + + pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1 + punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1 + + psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 + pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2 + + punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00 + psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 + + pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3 + + punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01 + psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 + + pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4 + + punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02 + psrldq xmm1, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 + + pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5 + + punpcklbw xmm1, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03 + pmullw xmm1, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6 + + paddsw xmm4, xmm7 + paddsw xmm4, xmm5 + + paddsw xmm4, xmm3 + paddsw xmm4, xmm6 + + paddsw xmm4, xmm1 + paddsw xmm4, [rd GLOBAL] + + psraw xmm4, 7 + + packuswb xmm4, xmm0 ; lower 8 bytes + + movq QWORD Ptr [rdi], xmm4 ; store the results in the destination + + movdqa xmm3, xmm2 + movdqa xmm4, xmm2 + + movdqa xmm5, xmm2 + movdqa xmm6, xmm2 + + movdqa xmm7, xmm2 + + punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2 + psrldq xmm4, 1 ; xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1 + + pmullw xmm3, XMMWORD PTR [rdx] ; x[-2] * H[-2]; Tap 1 + punpcklbw xmm4, xmm0 ; xx06 xx05 xx04 xx03 xx02 xx01 xx00 xx-1 + + psrldq xmm5, 2 ; xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 + pmullw xmm4, XMMWORD PTR [rdx+16] ; x[-1] * H[-1]; Tap 2 + + punpcklbw xmm5, xmm0 ; xx07 xx06 xx05 xx04 xx03 xx02 xx01 xx00 + psrldq xmm6, 3 ; xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 + + pmullw xmm5, [rdx+32] ; x[ 0] * H[ 0]; Tap 3 + + punpcklbw xmm6, xmm0 ; xx08 xx07 xx06 xx05 xx04 xx03 xx02 xx01 + psrldq xmm7, 4 ; xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 02 + + pmullw xmm6, [rdx+48] ; x[ 1] * h[ 1] ; Tap 4 + + punpcklbw xmm7, xmm0 ; xx09 xx08 xx07 xx06 xx05 xx04 xx03 xx02 + psrldq xmm2, 5 ; xx xx xx xx xx 0d 0c 0b 0a 09 08 07 06 05 04 03 + + pmullw xmm7, [rdx+64] ; x[ 2] * h[ 2] ; Tap 5 + + punpcklbw xmm2, xmm0 ; xx0a xx09 xx08 xx07 xx06 xx05 xx04 xx03 + pmullw xmm2, [rdx+80] ; x[ 3] * h[ 3] ; Tap 6 + + paddsw xmm4, xmm7 + paddsw xmm4, xmm5 + + paddsw xmm4, xmm3 + paddsw xmm4, xmm6 + + paddsw xmm4, xmm2 + paddsw xmm4, [rd GLOBAL] + + psraw xmm4, 7 + + packuswb xmm4, xmm0 ; higher 8 bytes + + movq QWORD Ptr [rdi+8], xmm4 ; store the results in the destination + + lea rsi, [rsi + rax] %if ABI_IS_32BIT - add rdi, DWORD Ptr arg(4) ;[output_width] + add rdi, DWORD Ptr arg(3) ;dst_ptich %else add rdi, r8 %endif + dec rcx - jnz unpack_block1d8_h6_sse2_rowloop ; next row + jnz filter_block1d16_h6_only_sse2_rowloop ; next row ; begin epilog pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret -;void vp8_pack_block1d8_v6_sse2 +;void vp8_filter_block1d8_v6_only_sse2 ;( -; short *src_ptr, +; unsigned char *src_ptr, +; unsigned int src_pixels_per_line, ; unsigned char *output_ptr, ; int dst_ptich, -; unsigned int pixels_per_line, ; unsigned int output_height, -; unsigned int output_width +; const short *vp8_filter ;) -global sym(vp8_pack_block1d8_v6_sse2) -sym(vp8_pack_block1d8_v6_sse2): +; Second-pass filter only when xoffset==0 +global sym(vp8_filter_block1d8_v6_only_sse2) +sym(vp8_filter_block1d8_v6_only_sse2): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 6 + SAVE_XMM GET_GOT rbx push rsi push rdi ; end prolog - movsxd rdx, dword ptr arg(3) ;pixels_per_line - mov rdi, arg(1) ;output_ptr - mov rsi, arg(0) ;src_ptr - movsxd rcx, DWORD PTR arg(4) ;[output_height] + mov rdi, arg(2) ;output_ptr + + movsxd rcx, dword ptr arg(4) ;output_height + movsxd rdx, dword ptr arg(1) ;src_pixels_per_line + + mov rax, arg(5) ;vp8_filter + + pxor xmm0, xmm0 ; clear xmm0 + + movdqa xmm7, XMMWORD PTR [rd GLOBAL] %if ABI_IS_32BIT=0 - movsxd r8, dword ptr arg(5) ;output_width ; Pitch for Source + movsxd r8, dword ptr arg(3) ; dst_ptich %endif -pack_block1d8_v6_sse2_loop: - movdqa xmm0, XMMWORD PTR [rsi] - packuswb xmm0, xmm0 +vp8_filter_block1d8_v6_only_sse2_loop: + movq xmm1, MMWORD PTR [rsi] + movq xmm2, MMWORD PTR [rsi + rdx] + movq xmm3, MMWORD PTR [rsi + rdx * 2] + movq xmm5, MMWORD PTR [rsi + rdx * 4] + add rsi, rdx + movq xmm4, MMWORD PTR [rsi + rdx * 2] + movq xmm6, MMWORD PTR [rsi + rdx * 4] + + punpcklbw xmm1, xmm0 + pmullw xmm1, [rax] + + punpcklbw xmm2, xmm0 + pmullw xmm2, [rax + 16] + + punpcklbw xmm3, xmm0 + pmullw xmm3, [rax + 32] + + punpcklbw xmm5, xmm0 + pmullw xmm5, [rax + 64] + + punpcklbw xmm4, xmm0 + pmullw xmm4, [rax + 48] + + punpcklbw xmm6, xmm0 + pmullw xmm6, [rax + 80] + + paddsw xmm2, xmm5 + paddsw xmm2, xmm3 + + paddsw xmm2, xmm1 + paddsw xmm2, xmm4 + + paddsw xmm2, xmm6 + paddsw xmm2, xmm7 - movq QWORD PTR [rdi], xmm0 ; store the results in the destination - lea rsi, [rsi+rdx] + psraw xmm2, 7 + packuswb xmm2, xmm0 ; pack and saturate + movq QWORD PTR [rdi], xmm2 ; store the results in the destination %if ABI_IS_32BIT - add rdi, DWORD Ptr arg(5) ;[output_width] + add rdi, DWORD PTR arg(3) ;[dst_ptich] %else add rdi, r8 %endif dec rcx ; decrement count - jnz pack_block1d8_v6_sse2_loop ; next row + jnz vp8_filter_block1d8_v6_only_sse2_loop ; next row ; begin epilog pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret -;void vp8_pack_block1d16_v6_sse2 +;void vp8_unpack_block1d16_h6_sse2 ;( -; short *src_ptr, -; unsigned char *output_ptr, -; int dst_ptich, -; unsigned int pixels_per_line, -; unsigned int output_height, -; unsigned int output_width +; unsigned char *src_ptr, +; unsigned short *output_ptr, +; unsigned int src_pixels_per_line, +; unsigned int output_height, +; unsigned int output_width ;) -global sym(vp8_pack_block1d16_v6_sse2) -sym(vp8_pack_block1d16_v6_sse2): +global sym(vp8_unpack_block1d16_h6_sse2) +sym(vp8_unpack_block1d16_h6_sse2): push rbp mov rbp, rsp - SHADOW_ARGS_TO_STACK 6 + SHADOW_ARGS_TO_STACK 5 + ;SAVE_XMM ;xmm6, xmm7 are not used here. GET_GOT rbx push rsi push rdi ; end prolog - movsxd rdx, dword ptr arg(3) ;pixels_per_line + mov rsi, arg(0) ;src_ptr mov rdi, arg(1) ;output_ptr - mov rsi, arg(0) ;src_ptr - movsxd rcx, DWORD PTR arg(4) ;[output_height] + movsxd rcx, dword ptr arg(3) ;output_height + movsxd rax, dword ptr arg(2) ;src_pixels_per_line ; Pitch for Source + + pxor xmm0, xmm0 ; clear xmm0 for unpack %if ABI_IS_32BIT=0 - movsxd r8, dword ptr arg(2) ;dst_pitch + movsxd r8, dword ptr arg(4) ;output_width ; Pitch for Source %endif -pack_block1d16_v6_sse2_loop: - movdqa xmm0, XMMWORD PTR [rsi] - movdqa xmm1, XMMWORD PTR [rsi+16] +unpack_block1d16_h6_sse2_rowloop: + movq xmm1, MMWORD PTR [rsi] ; 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 -1 -2 + movq xmm3, MMWORD PTR [rsi+8] ; make copy of xmm1 + + punpcklbw xmm3, xmm0 ; xx05 xx04 xx03 xx02 xx01 xx01 xx-1 xx-2 + punpcklbw xmm1, xmm0 - packuswb xmm0, xmm1 - movdqa XMMWORD PTR [rdi], xmm0 ; store the results in the destination + movdqa XMMWORD Ptr [rdi], xmm1 + movdqa XMMWORD Ptr [rdi + 16], xmm3 - add rsi, rdx + lea rsi, [rsi + rax] %if ABI_IS_32BIT - add rdi, DWORD Ptr arg(2) ;dst_pitch + add rdi, DWORD Ptr arg(4) ;[output_width] %else add rdi, r8 %endif - dec rcx ; decrement count - jnz pack_block1d16_v6_sse2_loop ; next row + dec rcx + jnz unpack_block1d16_h6_sse2_rowloop ; next row ; begin epilog pop rdi pop rsi RESTORE_GOT + ;RESTORE_XMM UNSHADOW_ARGS pop rbp ret @@ -631,6 +969,7 @@ sym(vp8_bilinear_predict16x16_sse2): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 6 + SAVE_XMM GET_GOT rbx push rsi push rdi @@ -878,6 +1217,7 @@ done: pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret @@ -898,6 +1238,7 @@ sym(vp8_bilinear_predict8x8_sse2): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 6 + SAVE_XMM GET_GOT rbx push rsi push rdi @@ -1021,6 +1362,7 @@ next_row8x8: pop rdi pop rsi RESTORE_GOT + RESTORE_XMM UNSHADOW_ARGS pop rbp ret diff --git a/vp8/common/x86/subpixel_ssse3.asm b/vp8/common/x86/subpixel_ssse3.asm new file mode 100644 index 0000000..f7209cc --- /dev/null +++ b/vp8/common/x86/subpixel_ssse3.asm @@ -0,0 +1,1508 @@ +; +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. +; +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. +; + + +%include "vpx_ports/x86_abi_support.asm" + +%define BLOCK_HEIGHT_WIDTH 4 +%define VP8_FILTER_WEIGHT 128 +%define VP8_FILTER_SHIFT 7 + + +;/************************************************************************************ +; Notes: filter_block1d_h6 applies a 6 tap filter horizontally to the input pixels. The +; input pixel array has output_height rows. This routine assumes that output_height is an +; even number. This function handles 8 pixels in horizontal direction, calculating ONE +; rows each iteration to take advantage of the 128 bits operations. +; +; This is an implementation of some of the SSE optimizations first seen in ffvp8 +; +;*************************************************************************************/ +;void vp8_filter_block1d8_h6_ssse3 +;( +; unsigned char *src_ptr, +; unsigned int src_pixels_per_line, +; unsigned char *output_ptr, +; unsigned int output_pitch, +; unsigned int output_height, +; unsigned int vp8_filter_index +;) +global sym(vp8_filter_block1d8_h6_ssse3) +sym(vp8_filter_block1d8_h6_ssse3): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + GET_GOT rbx + push rsi + push rdi + ; end prolog + + movsxd rdx, DWORD PTR arg(5) ;table index + xor rsi, rsi + shl rdx, 4 + + movdqa xmm7, [rd GLOBAL] + + lea rax, [k0_k5 GLOBAL] + add rax, rdx + mov rdi, arg(2) ;output_ptr + + cmp esi, DWORD PTR [rax] + je vp8_filter_block1d8_h4_ssse3 + + movdqa xmm4, XMMWORD PTR [rax] ;k0_k5 + movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4 + movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3 + + mov rsi, arg(0) ;src_ptr + movsxd rax, dword ptr arg(1) ;src_pixels_per_line + movsxd rcx, dword ptr arg(4) ;output_height + + movsxd rdx, dword ptr arg(3) ;output_pitch + + sub rdi, rdx +;xmm3 free +filter_block1d8_h6_rowloop_ssse3: + movdqu xmm0, XMMWORD PTR [rsi - 2] + + movdqa xmm1, xmm0 + pshufb xmm0, [shuf1b GLOBAL] + + movdqa xmm2, xmm1 + pshufb xmm1, [shuf2b GLOBAL] + pmaddubsw xmm0, xmm4 + pmaddubsw xmm1, xmm5 + + pshufb xmm2, [shuf3b GLOBAL] + add rdi, rdx + pmaddubsw xmm2, xmm6 + + lea rsi, [rsi + rax] + dec rcx + paddsw xmm0, xmm1 + paddsw xmm0, xmm7 + paddsw xmm0, xmm2 + psraw xmm0, 7 + packuswb xmm0, xmm0 + + movq MMWORD Ptr [rdi], xmm0 + jnz filter_block1d8_h6_rowloop_ssse3 + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +vp8_filter_block1d8_h4_ssse3: + movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4 + movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3 + + movdqa xmm3, XMMWORD PTR [shuf2b GLOBAL] + movdqa xmm4, XMMWORD PTR [shuf3b GLOBAL] + + mov rsi, arg(0) ;src_ptr + + movsxd rax, dword ptr arg(1) ;src_pixels_per_line + movsxd rcx, dword ptr arg(4) ;output_height + + movsxd rdx, dword ptr arg(3) ;output_pitch + + sub rdi, rdx +;xmm3 free +filter_block1d8_h4_rowloop_ssse3: + movdqu xmm0, XMMWORD PTR [rsi - 2] + + movdqa xmm2, xmm0 + pshufb xmm0, xmm3 ;[shuf2b GLOBAL] + pshufb xmm2, xmm4 ;[shuf3b GLOBAL] + + pmaddubsw xmm0, xmm5 + add rdi, rdx + pmaddubsw xmm2, xmm6 + + lea rsi, [rsi + rax] + dec rcx + paddsw xmm0, xmm7 + paddsw xmm0, xmm2 + psraw xmm0, 7 + packuswb xmm0, xmm0 + + movq MMWORD Ptr [rdi], xmm0 + + jnz filter_block1d8_h4_rowloop_ssse3 + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret +;void vp8_filter_block1d16_h6_ssse3 +;( +; unsigned char *src_ptr, +; unsigned int src_pixels_per_line, +; unsigned char *output_ptr, +; unsigned int output_pitch, +; unsigned int output_height, +; unsigned int vp8_filter_index +;) +global sym(vp8_filter_block1d16_h6_ssse3) +sym(vp8_filter_block1d16_h6_ssse3): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM + GET_GOT rbx + push rsi + push rdi + ; end prolog + + movsxd rdx, DWORD PTR arg(5) ;table index + xor rsi, rsi + shl rdx, 4 ; + + lea rax, [k0_k5 GLOBAL] + add rax, rdx + + mov rdi, arg(2) ;output_ptr + movdqa xmm7, [rd GLOBAL] + +;; +;; cmp esi, DWORD PTR [rax] +;; je vp8_filter_block1d16_h4_ssse3 + + mov rsi, arg(0) ;src_ptr + + movdqa xmm4, XMMWORD PTR [rax] ;k0_k5 + movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4 + movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3 + + movsxd rax, dword ptr arg(1) ;src_pixels_per_line + movsxd rcx, dword ptr arg(4) ;output_height + movsxd rdx, dword ptr arg(3) ;output_pitch + +filter_block1d16_h6_rowloop_ssse3: + movdqu xmm0, XMMWORD PTR [rsi - 2] + + movdqa xmm1, xmm0 + pshufb xmm0, [shuf1b GLOBAL] + movdqa xmm2, xmm1 + pmaddubsw xmm0, xmm4 + pshufb xmm1, [shuf2b GLOBAL] + pshufb xmm2, [shuf3b GLOBAL] + pmaddubsw xmm1, xmm5 + + movdqu xmm3, XMMWORD PTR [rsi + 6] + + pmaddubsw xmm2, xmm6 + paddsw xmm0, xmm1 + movdqa xmm1, xmm3 + pshufb xmm3, [shuf1b GLOBAL] + paddsw xmm0, xmm7 + pmaddubsw xmm3, xmm4 + paddsw xmm0, xmm2 + movdqa xmm2, xmm1 + pshufb xmm1, [shuf2b GLOBAL] + pshufb xmm2, [shuf3b GLOBAL] + pmaddubsw xmm1, xmm5 + pmaddubsw xmm2, xmm6 + + psraw xmm0, 7 + packuswb xmm0, xmm0 + lea rsi, [rsi + rax] + paddsw xmm3, xmm1 + paddsw xmm3, xmm7 + paddsw xmm3, xmm2 + psraw xmm3, 7 + packuswb xmm3, xmm3 + + punpcklqdq xmm0, xmm3 + + movdqa XMMWORD Ptr [rdi], xmm0 + + add rdi, rdx + dec rcx + jnz filter_block1d16_h6_rowloop_ssse3 + + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +vp8_filter_block1d16_h4_ssse3: + movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4 + movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3 + + mov rsi, arg(0) ;src_ptr + movsxd rax, dword ptr arg(1) ;src_pixels_per_line + movsxd rcx, dword ptr arg(4) ;output_height + movsxd rdx, dword ptr arg(3) ;output_pitch + +filter_block1d16_h4_rowloop_ssse3: + movdqu xmm1, XMMWORD PTR [rsi - 2] + + movdqa xmm2, xmm1 + pshufb xmm1, [shuf2b GLOBAL] + pshufb xmm2, [shuf3b GLOBAL] + pmaddubsw xmm1, xmm5 + + movdqu xmm3, XMMWORD PTR [rsi + 6] + + pmaddubsw xmm2, xmm6 + movdqa xmm0, xmm3 + pshufb xmm3, [shuf3b GLOBAL] + pshufb xmm0, [shuf2b GLOBAL] + + paddsw xmm1, xmm7 + paddsw xmm1, xmm2 + + pmaddubsw xmm0, xmm5 + pmaddubsw xmm3, xmm6 + + psraw xmm1, 7 + packuswb xmm1, xmm1 + lea rsi, [rsi + rax] + paddsw xmm3, xmm0 + paddsw xmm3, xmm7 + psraw xmm3, 7 + packuswb xmm3, xmm3 + + punpcklqdq xmm1, xmm3 + + movdqa XMMWORD Ptr [rdi], xmm1 + + add rdi, rdx + dec rcx + jnz filter_block1d16_h4_rowloop_ssse3 + + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +;void vp8_filter_block1d4_h6_ssse3 +;( +; unsigned char *src_ptr, +; unsigned int src_pixels_per_line, +; unsigned char *output_ptr, +; unsigned int output_pitch, +; unsigned int output_height, +; unsigned int vp8_filter_index +;) +global sym(vp8_filter_block1d4_h6_ssse3) +sym(vp8_filter_block1d4_h6_ssse3): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + GET_GOT rbx + push rsi + push rdi + ; end prolog + + movsxd rdx, DWORD PTR arg(5) ;table index + xor rsi, rsi + shl rdx, 4 ; + + lea rax, [k0_k5 GLOBAL] + add rax, rdx + movdqa xmm7, [rd GLOBAL] + + cmp esi, DWORD PTR [rax] + je vp8_filter_block1d4_h4_ssse3 + + movdqa xmm4, XMMWORD PTR [rax] ;k0_k5 + movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4 + movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3 + + mov rsi, arg(0) ;src_ptr + mov rdi, arg(2) ;output_ptr + movsxd rax, dword ptr arg(1) ;src_pixels_per_line + movsxd rcx, dword ptr arg(4) ;output_height + + movsxd rdx, dword ptr arg(3) ;output_pitch + +;xmm3 free +filter_block1d4_h6_rowloop_ssse3: + movdqu xmm0, XMMWORD PTR [rsi - 2] + + movdqa xmm1, xmm0 + pshufb xmm0, [shuf1b GLOBAL] + + movdqa xmm2, xmm1 + pshufb xmm1, [shuf2b GLOBAL] + pmaddubsw xmm0, xmm4 + pshufb xmm2, [shuf3b GLOBAL] + pmaddubsw xmm1, xmm5 + +;-- + pmaddubsw xmm2, xmm6 + + lea rsi, [rsi + rax] +;-- + paddsw xmm0, xmm1 + paddsw xmm0, xmm7 + pxor xmm1, xmm1 + paddsw xmm0, xmm2 + psraw xmm0, 7 + packuswb xmm0, xmm0 + + movd DWORD PTR [rdi], xmm0 + + add rdi, rdx + dec rcx + jnz filter_block1d4_h6_rowloop_ssse3 + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +vp8_filter_block1d4_h4_ssse3: + movdqa xmm5, XMMWORD PTR [rax+256] ;k2_k4 + movdqa xmm6, XMMWORD PTR [rax+128] ;k1_k3 + movdqa xmm0, XMMWORD PTR [shuf2b GLOBAL] + movdqa xmm3, XMMWORD PTR [shuf3b GLOBAL] + + mov rsi, arg(0) ;src_ptr + mov rdi, arg(2) ;output_ptr + movsxd rax, dword ptr arg(1) ;src_pixels_per_line + movsxd rcx, dword ptr arg(4) ;output_height + + movsxd rdx, dword ptr arg(3) ;output_pitch + +filter_block1d4_h4_rowloop_ssse3: + movdqu xmm1, XMMWORD PTR [rsi - 2] + + movdqa xmm2, xmm1 + pshufb xmm1, xmm0 ;;[shuf2b GLOBAL] + pshufb xmm2, xmm3 ;;[shuf3b GLOBAL] + pmaddubsw xmm1, xmm5 + +;-- + pmaddubsw xmm2, xmm6 + + lea rsi, [rsi + rax] +;-- + paddsw xmm1, xmm7 + paddsw xmm1, xmm2 + psraw xmm1, 7 + packuswb xmm1, xmm1 + + movd DWORD PTR [rdi], xmm1 + + add rdi, rdx + dec rcx + jnz filter_block1d4_h4_rowloop_ssse3 + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + + + +;void vp8_filter_block1d16_v6_ssse3 +;( +; unsigned char *src_ptr, +; unsigned int src_pitch, +; unsigned char *output_ptr, +; unsigned int out_pitch, +; unsigned int output_height, +; unsigned int vp8_filter_index +;) +global sym(vp8_filter_block1d16_v6_ssse3) +sym(vp8_filter_block1d16_v6_ssse3): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + GET_GOT rbx + push rsi + push rdi + ; end prolog + + movsxd rdx, DWORD PTR arg(5) ;table index + xor rsi, rsi + shl rdx, 4 ; + + lea rax, [k0_k5 GLOBAL] + add rax, rdx + + cmp esi, DWORD PTR [rax] + je vp8_filter_block1d16_v4_ssse3 + + movdqa xmm5, XMMWORD PTR [rax] ;k0_k5 + movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4 + movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3 + + mov rsi, arg(0) ;src_ptr + movsxd rdx, DWORD PTR arg(1) ;pixels_per_line + mov rdi, arg(2) ;output_ptr + +%if ABI_IS_32BIT=0 + movsxd r8, DWORD PTR arg(3) ;out_pitch +%endif + mov rax, rsi + movsxd rcx, DWORD PTR arg(4) ;output_height + add rax, rdx + + +vp8_filter_block1d16_v6_ssse3_loop: + movq xmm1, MMWORD PTR [rsi] ;A + movq xmm2, MMWORD PTR [rsi + rdx] ;B + movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C + movq xmm4, MMWORD PTR [rax + rdx * 2] ;D + movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E + + punpcklbw xmm2, xmm4 ;B D + punpcklbw xmm3, xmm0 ;C E + + movq xmm0, MMWORD PTR [rax + rdx * 4] ;F + + pmaddubsw xmm3, xmm6 + punpcklbw xmm1, xmm0 ;A F + pmaddubsw xmm2, xmm7 + pmaddubsw xmm1, xmm5 + + paddsw xmm2, xmm3 + paddsw xmm2, xmm1 + paddsw xmm2, [rd GLOBAL] + psraw xmm2, 7 + packuswb xmm2, xmm2 + + movq MMWORD PTR [rdi], xmm2 ;store the results + + movq xmm1, MMWORD PTR [rsi + 8] ;A + movq xmm2, MMWORD PTR [rsi + rdx + 8] ;B + movq xmm3, MMWORD PTR [rsi + rdx * 2 + 8] ;C + movq xmm4, MMWORD PTR [rax + rdx * 2 + 8] ;D + movq xmm0, MMWORD PTR [rsi + rdx * 4 + 8] ;E + + punpcklbw xmm2, xmm4 ;B D + punpcklbw xmm3, xmm0 ;C E + + movq xmm0, MMWORD PTR [rax + rdx * 4 + 8] ;F + pmaddubsw xmm3, xmm6 + punpcklbw xmm1, xmm0 ;A F + pmaddubsw xmm2, xmm7 + pmaddubsw xmm1, xmm5 + + add rsi, rdx + add rax, rdx +;-- +;-- + paddsw xmm2, xmm3 + paddsw xmm2, xmm1 + paddsw xmm2, [rd GLOBAL] + psraw xmm2, 7 + packuswb xmm2, xmm2 + + movq MMWORD PTR [rdi+8], xmm2 + +%if ABI_IS_32BIT + add rdi, DWORD PTR arg(3) ;out_pitch +%else + add rdi, r8 +%endif + dec rcx + jnz vp8_filter_block1d16_v6_ssse3_loop + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +vp8_filter_block1d16_v4_ssse3: + movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4 + movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3 + + mov rsi, arg(0) ;src_ptr + movsxd rdx, DWORD PTR arg(1) ;pixels_per_line + mov rdi, arg(2) ;output_ptr + +%if ABI_IS_32BIT=0 + movsxd r8, DWORD PTR arg(3) ;out_pitch +%endif + mov rax, rsi + movsxd rcx, DWORD PTR arg(4) ;output_height + add rax, rdx + +vp8_filter_block1d16_v4_ssse3_loop: + movq xmm2, MMWORD PTR [rsi + rdx] ;B + movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C + movq xmm4, MMWORD PTR [rax + rdx * 2] ;D + movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E + + punpcklbw xmm2, xmm4 ;B D + punpcklbw xmm3, xmm0 ;C E + + pmaddubsw xmm3, xmm6 + pmaddubsw xmm2, xmm7 + movq xmm5, MMWORD PTR [rsi + rdx + 8] ;B + movq xmm1, MMWORD PTR [rsi + rdx * 2 + 8] ;C + movq xmm4, MMWORD PTR [rax + rdx * 2 + 8] ;D + movq xmm0, MMWORD PTR [rsi + rdx * 4 + 8] ;E + + paddsw xmm2, [rd GLOBAL] + paddsw xmm2, xmm3 + psraw xmm2, 7 + packuswb xmm2, xmm2 + + punpcklbw xmm5, xmm4 ;B D + punpcklbw xmm1, xmm0 ;C E + + pmaddubsw xmm1, xmm6 + pmaddubsw xmm5, xmm7 + + movdqa xmm4, [rd GLOBAL] + add rsi, rdx + add rax, rdx +;-- +;-- + paddsw xmm5, xmm1 + paddsw xmm5, xmm4 + psraw xmm5, 7 + packuswb xmm5, xmm5 + + punpcklqdq xmm2, xmm5 + + movdqa XMMWORD PTR [rdi], xmm2 + +%if ABI_IS_32BIT + add rdi, DWORD PTR arg(3) ;out_pitch +%else + add rdi, r8 +%endif + dec rcx + jnz vp8_filter_block1d16_v4_ssse3_loop + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +;void vp8_filter_block1d8_v6_ssse3 +;( +; unsigned char *src_ptr, +; unsigned int src_pitch, +; unsigned char *output_ptr, +; unsigned int out_pitch, +; unsigned int output_height, +; unsigned int vp8_filter_index +;) +global sym(vp8_filter_block1d8_v6_ssse3) +sym(vp8_filter_block1d8_v6_ssse3): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + GET_GOT rbx + push rsi + push rdi + ; end prolog + + movsxd rdx, DWORD PTR arg(5) ;table index + xor rsi, rsi + shl rdx, 4 ; + + lea rax, [k0_k5 GLOBAL] + add rax, rdx + + movsxd rdx, DWORD PTR arg(1) ;pixels_per_line + mov rdi, arg(2) ;output_ptr +%if ABI_IS_32BIT=0 + movsxd r8, DWORD PTR arg(3) ; out_pitch +%endif + movsxd rcx, DWORD PTR arg(4) ;[output_height] + + cmp esi, DWORD PTR [rax] + je vp8_filter_block1d8_v4_ssse3 + + movdqa xmm5, XMMWORD PTR [rax] ;k0_k5 + movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4 + movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3 + + mov rsi, arg(0) ;src_ptr + + mov rax, rsi + add rax, rdx + +vp8_filter_block1d8_v6_ssse3_loop: + movq xmm1, MMWORD PTR [rsi] ;A + movq xmm2, MMWORD PTR [rsi + rdx] ;B + movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C + movq xmm4, MMWORD PTR [rax + rdx * 2] ;D + movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E + + punpcklbw xmm2, xmm4 ;B D + punpcklbw xmm3, xmm0 ;C E + + movq xmm0, MMWORD PTR [rax + rdx * 4] ;F + movdqa xmm4, [rd GLOBAL] + + pmaddubsw xmm3, xmm6 + punpcklbw xmm1, xmm0 ;A F + pmaddubsw xmm2, xmm7 + pmaddubsw xmm1, xmm5 + add rsi, rdx + add rax, rdx +;-- +;-- + paddsw xmm2, xmm3 + paddsw xmm2, xmm1 + paddsw xmm2, xmm4 + psraw xmm2, 7 + packuswb xmm2, xmm2 + + movq MMWORD PTR [rdi], xmm2 + +%if ABI_IS_32BIT + add rdi, DWORD PTR arg(3) ;[out_pitch] +%else + add rdi, r8 +%endif + dec rcx + jnz vp8_filter_block1d8_v6_ssse3_loop + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +vp8_filter_block1d8_v4_ssse3: + movdqa xmm6, XMMWORD PTR [rax+256] ;k2_k4 + movdqa xmm7, XMMWORD PTR [rax+128] ;k1_k3 + movdqa xmm5, [rd GLOBAL] + + mov rsi, arg(0) ;src_ptr + + mov rax, rsi + add rax, rdx + +vp8_filter_block1d8_v4_ssse3_loop: + movq xmm2, MMWORD PTR [rsi + rdx] ;B + movq xmm3, MMWORD PTR [rsi + rdx * 2] ;C + movq xmm4, MMWORD PTR [rax + rdx * 2] ;D + movq xmm0, MMWORD PTR [rsi + rdx * 4] ;E + + punpcklbw xmm2, xmm4 ;B D + punpcklbw xmm3, xmm0 ;C E + + pmaddubsw xmm3, xmm6 + pmaddubsw xmm2, xmm7 + add rsi, rdx + add rax, rdx +;-- +;-- + paddsw xmm2, xmm3 + paddsw xmm2, xmm5 + psraw xmm2, 7 + packuswb xmm2, xmm2 + + movq MMWORD PTR [rdi], xmm2 + +%if ABI_IS_32BIT + add rdi, DWORD PTR arg(3) ;[out_pitch] +%else + add rdi, r8 +%endif + dec rcx + jnz vp8_filter_block1d8_v4_ssse3_loop + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret +;void vp8_filter_block1d4_v6_ssse3 +;( +; unsigned char *src_ptr, +; unsigned int src_pitch, +; unsigned char *output_ptr, +; unsigned int out_pitch, +; unsigned int output_height, +; unsigned int vp8_filter_index +;) +global sym(vp8_filter_block1d4_v6_ssse3) +sym(vp8_filter_block1d4_v6_ssse3): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + GET_GOT rbx + push rsi + push rdi + ; end prolog + + movsxd rdx, DWORD PTR arg(5) ;table index + xor rsi, rsi + shl rdx, 4 ; + + lea rax, [k0_k5 GLOBAL] + add rax, rdx + + movsxd rdx, DWORD PTR arg(1) ;pixels_per_line + mov rdi, arg(2) ;output_ptr +%if ABI_IS_32BIT=0 + movsxd r8, DWORD PTR arg(3) ; out_pitch +%endif + movsxd rcx, DWORD PTR arg(4) ;[output_height] + + cmp esi, DWORD PTR [rax] + je vp8_filter_block1d4_v4_ssse3 + + movq mm5, MMWORD PTR [rax] ;k0_k5 + movq mm6, MMWORD PTR [rax+256] ;k2_k4 + movq mm7, MMWORD PTR [rax+128] ;k1_k3 + + mov rsi, arg(0) ;src_ptr + + mov rax, rsi + add rax, rdx + +vp8_filter_block1d4_v6_ssse3_loop: + movd mm1, DWORD PTR [rsi] ;A + movd mm2, DWORD PTR [rsi + rdx] ;B + movd mm3, DWORD PTR [rsi + rdx * 2] ;C + movd mm4, DWORD PTR [rax + rdx * 2] ;D + movd mm0, DWORD PTR [rsi + rdx * 4] ;E + + punpcklbw mm2, mm4 ;B D + punpcklbw mm3, mm0 ;C E + + movd mm0, DWORD PTR [rax + rdx * 4] ;F + + movq mm4, [rd GLOBAL] + + pmaddubsw mm3, mm6 + punpcklbw mm1, mm0 ;A F + pmaddubsw mm2, mm7 + pmaddubsw mm1, mm5 + add rsi, rdx + add rax, rdx +;-- +;-- + paddsw mm2, mm3 + paddsw mm2, mm1 + paddsw mm2, mm4 + psraw mm2, 7 + packuswb mm2, mm2 + + movd DWORD PTR [rdi], mm2 + +%if ABI_IS_32BIT + add rdi, DWORD PTR arg(3) ;[out_pitch] +%else + add rdi, r8 +%endif + dec rcx + jnz vp8_filter_block1d4_v6_ssse3_loop + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +vp8_filter_block1d4_v4_ssse3: + movq mm6, MMWORD PTR [rax+256] ;k2_k4 + movq mm7, MMWORD PTR [rax+128] ;k1_k3 + movq mm5, MMWORD PTR [rd GLOBAL] + + mov rsi, arg(0) ;src_ptr + + mov rax, rsi + add rax, rdx + +vp8_filter_block1d4_v4_ssse3_loop: + movd mm2, DWORD PTR [rsi + rdx] ;B + movd mm3, DWORD PTR [rsi + rdx * 2] ;C + movd mm4, DWORD PTR [rax + rdx * 2] ;D + movd mm0, DWORD PTR [rsi + rdx * 4] ;E + + punpcklbw mm2, mm4 ;B D + punpcklbw mm3, mm0 ;C E + + pmaddubsw mm3, mm6 + pmaddubsw mm2, mm7 + add rsi, rdx + add rax, rdx +;-- +;-- + paddsw mm2, mm3 + paddsw mm2, mm5 + psraw mm2, 7 + packuswb mm2, mm2 + + movd DWORD PTR [rdi], mm2 + +%if ABI_IS_32BIT + add rdi, DWORD PTR arg(3) ;[out_pitch] +%else + add rdi, r8 +%endif + dec rcx + jnz vp8_filter_block1d4_v4_ssse3_loop + + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + UNSHADOW_ARGS + pop rbp + ret + +;void vp8_bilinear_predict16x16_ssse3 +;( +; unsigned char *src_ptr, +; int src_pixels_per_line, +; int xoffset, +; int yoffset, +; unsigned char *dst_ptr, +; int dst_pitch +;) +global sym(vp8_bilinear_predict16x16_ssse3) +sym(vp8_bilinear_predict16x16_ssse3): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM + GET_GOT rbx + push rsi + push rdi + ; end prolog + + lea rcx, [vp8_bilinear_filters_ssse3 GLOBAL] + movsxd rax, dword ptr arg(2) ; xoffset + + cmp rax, 0 ; skip first_pass filter if xoffset=0 + je b16x16_sp_only + + shl rax, 4 + lea rax, [rax + rcx] ; HFilter + + mov rdi, arg(4) ; dst_ptr + mov rsi, arg(0) ; src_ptr + movsxd rdx, dword ptr arg(5) ; dst_pitch + + movdqa xmm1, [rax] + + movsxd rax, dword ptr arg(3) ; yoffset + + cmp rax, 0 ; skip second_pass filter if yoffset=0 + je b16x16_fp_only + + shl rax, 4 + lea rax, [rax + rcx] ; VFilter + + lea rcx, [rdi+rdx*8] + lea rcx, [rcx+rdx*8] + movsxd rdx, dword ptr arg(1) ; src_pixels_per_line + + movdqa xmm2, [rax] + +%if ABI_IS_32BIT=0 + movsxd r8, dword ptr arg(5) ; dst_pitch +%endif + movdqu xmm3, [rsi] ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 + + movdqa xmm4, xmm3 + + movdqu xmm5, [rsi+1] ; 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 + lea rsi, [rsi + rdx] ; next line + + punpcklbw xmm3, xmm5 ; 00 01 01 02 02 03 03 04 04 05 05 06 06 07 07 08 + pmaddubsw xmm3, xmm1 ; 00 02 04 06 08 10 12 14 + + punpckhbw xmm4, xmm5 ; 08 09 09 10 10 11 11 12 12 13 13 14 14 15 15 16 + pmaddubsw xmm4, xmm1 ; 01 03 05 07 09 11 13 15 + + paddw xmm3, [rd GLOBAL] ; xmm3 += round value + psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128 + + paddw xmm4, [rd GLOBAL] ; xmm4 += round value + psraw xmm4, VP8_FILTER_SHIFT ; xmm4 /= 128 + + movdqa xmm7, xmm3 + packuswb xmm7, xmm4 ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 + +.next_row: + movdqu xmm6, [rsi] ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 + + movdqa xmm4, xmm6 + + movdqu xmm5, [rsi+1] ; 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 + lea rsi, [rsi + rdx] ; next line + + punpcklbw xmm6, xmm5 + pmaddubsw xmm6, xmm1 + + punpckhbw xmm4, xmm5 + pmaddubsw xmm4, xmm1 + + paddw xmm6, [rd GLOBAL] ; xmm6 += round value + psraw xmm6, VP8_FILTER_SHIFT ; xmm6 /= 128 + + paddw xmm4, [rd GLOBAL] ; xmm4 += round value + psraw xmm4, VP8_FILTER_SHIFT ; xmm4 /= 128 + + packuswb xmm6, xmm4 + movdqa xmm5, xmm7 + + punpcklbw xmm5, xmm6 + pmaddubsw xmm5, xmm2 + + punpckhbw xmm7, xmm6 + pmaddubsw xmm7, xmm2 + + paddw xmm5, [rd GLOBAL] ; xmm5 += round value + psraw xmm5, VP8_FILTER_SHIFT ; xmm5 /= 128 + + paddw xmm7, [rd GLOBAL] ; xmm7 += round value + psraw xmm7, VP8_FILTER_SHIFT ; xmm7 /= 128 + + packuswb xmm5, xmm7 + movdqa xmm7, xmm6 + + movdqa [rdi], xmm5 ; store the results in the destination +%if ABI_IS_32BIT + add rdi, DWORD PTR arg(5) ; dst_pitch +%else + add rdi, r8 +%endif + + cmp rdi, rcx + jne .next_row + + jmp done + +b16x16_sp_only: + movsxd rax, dword ptr arg(3) ; yoffset + shl rax, 4 + lea rax, [rax + rcx] ; VFilter + + mov rdi, arg(4) ; dst_ptr + mov rsi, arg(0) ; src_ptr + movsxd rdx, dword ptr arg(5) ; dst_pitch + + movdqa xmm1, [rax] ; VFilter + + lea rcx, [rdi+rdx*8] + lea rcx, [rcx+rdx*8] + movsxd rax, dword ptr arg(1) ; src_pixels_per_line + + ; get the first horizontal line done + movdqu xmm2, [rsi] ; load row 0 + + lea rsi, [rsi + rax] ; next line +.next_row: + movdqu xmm3, [rsi] ; load row + 1 + + movdqu xmm4, xmm2 + punpcklbw xmm4, xmm3 + + pmaddubsw xmm4, xmm1 + movdqu xmm7, [rsi + rax] ; load row + 2 + + punpckhbw xmm2, xmm3 + movdqu xmm6, xmm3 + + pmaddubsw xmm2, xmm1 + punpcklbw xmm6, xmm7 + + paddw xmm4, [rd GLOBAL] + pmaddubsw xmm6, xmm1 + + psraw xmm4, VP8_FILTER_SHIFT + punpckhbw xmm3, xmm7 + + paddw xmm2, [rd GLOBAL] + pmaddubsw xmm3, xmm1 + + psraw xmm2, VP8_FILTER_SHIFT + paddw xmm6, [rd GLOBAL] + + packuswb xmm4, xmm2 + psraw xmm6, VP8_FILTER_SHIFT + + movdqa [rdi], xmm4 ; store row 0 + paddw xmm3, [rd GLOBAL] + + psraw xmm3, VP8_FILTER_SHIFT + lea rsi, [rsi + 2*rax] + + packuswb xmm6, xmm3 + movdqa xmm2, xmm7 + + movdqa [rdi + rdx],xmm6 ; store row 1 + lea rdi, [rdi + 2*rdx] + + cmp rdi, rcx + jne .next_row + + jmp done + +b16x16_fp_only: + lea rcx, [rdi+rdx*8] + lea rcx, [rcx+rdx*8] + movsxd rax, dword ptr arg(1) ; src_pixels_per_line + +.next_row: + movdqu xmm2, [rsi] ; row 0 + movdqa xmm3, xmm2 + + movdqu xmm4, [rsi + 1] ; row 0 + 1 + lea rsi, [rsi + rax] ; next line + + punpcklbw xmm2, xmm4 + movdqu xmm5, [rsi] ; row 1 + + pmaddubsw xmm2, xmm1 + movdqa xmm6, xmm5 + + punpckhbw xmm3, xmm4 + movdqu xmm7, [rsi + 1] ; row 1 + 1 + + pmaddubsw xmm3, xmm1 + paddw xmm2, [rd GLOBAL] + + psraw xmm2, VP8_FILTER_SHIFT + punpcklbw xmm5, xmm7 + + paddw xmm3, [rd GLOBAL] + pmaddubsw xmm5, xmm1 + + psraw xmm3, VP8_FILTER_SHIFT + punpckhbw xmm6, xmm7 + + packuswb xmm2, xmm3 + pmaddubsw xmm6, xmm1 + + movdqa [rdi], xmm2 ; store the results in the destination + paddw xmm5, [rd GLOBAL] + + lea rdi, [rdi + rdx] ; dst_pitch + psraw xmm5, VP8_FILTER_SHIFT + + paddw xmm6, [rd GLOBAL] + psraw xmm6, VP8_FILTER_SHIFT + + packuswb xmm5, xmm6 + lea rsi, [rsi + rax] ; next line + + movdqa [rdi], xmm5 ; store the results in the destination + lea rdi, [rdi + rdx] ; dst_pitch + + cmp rdi, rcx + + jne .next_row + +done: + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + RESTORE_XMM + UNSHADOW_ARGS + pop rbp + ret + +;void vp8_bilinear_predict8x8_ssse3 +;( +; unsigned char *src_ptr, +; int src_pixels_per_line, +; int xoffset, +; int yoffset, +; unsigned char *dst_ptr, +; int dst_pitch +;) +global sym(vp8_bilinear_predict8x8_ssse3) +sym(vp8_bilinear_predict8x8_ssse3): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 6 + SAVE_XMM + GET_GOT rbx + push rsi + push rdi + ; end prolog + + ALIGN_STACK 16, rax + sub rsp, 144 ; reserve 144 bytes + + lea rcx, [vp8_bilinear_filters_ssse3 GLOBAL] + + mov rsi, arg(0) ;src_ptr + movsxd rdx, dword ptr arg(1) ;src_pixels_per_line + + ;Read 9-line unaligned data in and put them on stack. This gives a big + ;performance boost. + movdqu xmm0, [rsi] + lea rax, [rdx + rdx*2] + movdqu xmm1, [rsi+rdx] + movdqu xmm2, [rsi+rdx*2] + add rsi, rax + movdqu xmm3, [rsi] + movdqu xmm4, [rsi+rdx] + movdqu xmm5, [rsi+rdx*2] + add rsi, rax + movdqu xmm6, [rsi] + movdqu xmm7, [rsi+rdx] + + movdqa XMMWORD PTR [rsp], xmm0 + + movdqu xmm0, [rsi+rdx*2] + + movdqa XMMWORD PTR [rsp+16], xmm1 + movdqa XMMWORD PTR [rsp+32], xmm2 + movdqa XMMWORD PTR [rsp+48], xmm3 + movdqa XMMWORD PTR [rsp+64], xmm4 + movdqa XMMWORD PTR [rsp+80], xmm5 + movdqa XMMWORD PTR [rsp+96], xmm6 + movdqa XMMWORD PTR [rsp+112], xmm7 + movdqa XMMWORD PTR [rsp+128], xmm0 + + movsxd rax, dword ptr arg(2) ; xoffset + cmp rax, 0 ; skip first_pass filter if xoffset=0 + je b8x8_sp_only + + shl rax, 4 + add rax, rcx ; HFilter + + mov rdi, arg(4) ; dst_ptr + movsxd rdx, dword ptr arg(5) ; dst_pitch + + movdqa xmm0, [rax] + + movsxd rax, dword ptr arg(3) ; yoffset + cmp rax, 0 ; skip second_pass filter if yoffset=0 + je b8x8_fp_only + + shl rax, 4 + lea rax, [rax + rcx] ; VFilter + + lea rcx, [rdi+rdx*8] + + movdqa xmm1, [rax] + + ; get the first horizontal line done + movdqa xmm3, [rsp] ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 + movdqa xmm5, xmm3 ; 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 xx + + psrldq xmm5, 1 + lea rsp, [rsp + 16] ; next line + + punpcklbw xmm3, xmm5 ; 00 01 01 02 02 03 03 04 04 05 05 06 06 07 07 08 + pmaddubsw xmm3, xmm0 ; 00 02 04 06 08 10 12 14 + + paddw xmm3, [rd GLOBAL] ; xmm3 += round value + psraw xmm3, VP8_FILTER_SHIFT ; xmm3 /= 128 + + movdqa xmm7, xmm3 + packuswb xmm7, xmm7 ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 + +.next_row: + movdqa xmm6, [rsp] ; 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 + lea rsp, [rsp + 16] ; next line + + movdqa xmm5, xmm6 + + psrldq xmm5, 1 + + punpcklbw xmm6, xmm5 + pmaddubsw xmm6, xmm0 + + paddw xmm6, [rd GLOBAL] ; xmm6 += round value + psraw xmm6, VP8_FILTER_SHIFT ; xmm6 /= 128 + + packuswb xmm6, xmm6 + + punpcklbw xmm7, xmm6 + pmaddubsw xmm7, xmm1 + + paddw xmm7, [rd GLOBAL] ; xmm7 += round value + psraw xmm7, VP8_FILTER_SHIFT ; xmm7 /= 128 + + packuswb xmm7, xmm7 + + movq [rdi], xmm7 ; store the results in the destination + lea rdi, [rdi + rdx] + + movdqa xmm7, xmm6 + + cmp rdi, rcx + jne .next_row + + jmp done8x8 + +b8x8_sp_only: + movsxd rax, dword ptr arg(3) ; yoffset + shl rax, 4 + lea rax, [rax + rcx] ; VFilter + + mov rdi, arg(4) ;dst_ptr + movsxd rdx, dword ptr arg(5) ; dst_pitch + + movdqa xmm0, [rax] ; VFilter + + movq xmm1, XMMWORD PTR [rsp] + movq xmm2, XMMWORD PTR [rsp+16] + + movq xmm3, XMMWORD PTR [rsp+32] + punpcklbw xmm1, xmm2 + + movq xmm4, XMMWORD PTR [rsp+48] + punpcklbw xmm2, xmm3 + + movq xmm5, XMMWORD PTR [rsp+64] + punpcklbw xmm3, xmm4 + + movq xmm6, XMMWORD PTR [rsp+80] + punpcklbw xmm4, xmm5 + + movq xmm7, XMMWORD PTR [rsp+96] + punpcklbw xmm5, xmm6 + + pmaddubsw xmm1, xmm0 + pmaddubsw xmm2, xmm0 + + pmaddubsw xmm3, xmm0 + pmaddubsw xmm4, xmm0 + + pmaddubsw xmm5, xmm0 + punpcklbw xmm6, xmm7 + + pmaddubsw xmm6, xmm0 + paddw xmm1, [rd GLOBAL] + + paddw xmm2, [rd GLOBAL] + psraw xmm1, VP8_FILTER_SHIFT + + paddw xmm3, [rd GLOBAL] + psraw xmm2, VP8_FILTER_SHIFT + + paddw xmm4, [rd GLOBAL] + psraw xmm3, VP8_FILTER_SHIFT + + paddw xmm5, [rd GLOBAL] + psraw xmm4, VP8_FILTER_SHIFT + + paddw xmm6, [rd GLOBAL] + psraw xmm5, VP8_FILTER_SHIFT + + psraw xmm6, VP8_FILTER_SHIFT + packuswb xmm1, xmm1 + + packuswb xmm2, xmm2 + movq [rdi], xmm1 + + packuswb xmm3, xmm3 + movq [rdi+rdx], xmm2 + + packuswb xmm4, xmm4 + movq xmm1, XMMWORD PTR [rsp+112] + + lea rdi, [rdi + 2*rdx] + movq xmm2, XMMWORD PTR [rsp+128] + + packuswb xmm5, xmm5 + movq [rdi], xmm3 + + packuswb xmm6, xmm6 + movq [rdi+rdx], xmm4 + + lea rdi, [rdi + 2*rdx] + punpcklbw xmm7, xmm1 + + movq [rdi], xmm5 + pmaddubsw xmm7, xmm0 + + movq [rdi+rdx], xmm6 + punpcklbw xmm1, xmm2 + + pmaddubsw xmm1, xmm0 + paddw xmm7, [rd GLOBAL] + + psraw xmm7, VP8_FILTER_SHIFT + paddw xmm1, [rd GLOBAL] + + psraw xmm1, VP8_FILTER_SHIFT + packuswb xmm7, xmm7 + + packuswb xmm1, xmm1 + lea rdi, [rdi + 2*rdx] + + movq [rdi], xmm7 + + movq [rdi+rdx], xmm1 + lea rsp, [rsp + 144] + + jmp done8x8 + +b8x8_fp_only: + lea rcx, [rdi+rdx*8] + +.next_row: + movdqa xmm1, XMMWORD PTR [rsp] + movdqa xmm3, XMMWORD PTR [rsp+16] + + movdqa xmm2, xmm1 + movdqa xmm5, XMMWORD PTR [rsp+32] + + psrldq xmm2, 1 + movdqa xmm7, XMMWORD PTR [rsp+48] + + movdqa xmm4, xmm3 + psrldq xmm4, 1 + + movdqa xmm6, xmm5 + psrldq xmm6, 1 + + punpcklbw xmm1, xmm2 + pmaddubsw xmm1, xmm0 + + punpcklbw xmm3, xmm4 + pmaddubsw xmm3, xmm0 + + punpcklbw xmm5, xmm6 + pmaddubsw xmm5, xmm0 + + movdqa xmm2, xmm7 + psrldq xmm2, 1 + + punpcklbw xmm7, xmm2 + pmaddubsw xmm7, xmm0 + + paddw xmm1, [rd GLOBAL] + psraw xmm1, VP8_FILTER_SHIFT + + paddw xmm3, [rd GLOBAL] + psraw xmm3, VP8_FILTER_SHIFT + + paddw xmm5, [rd GLOBAL] + psraw xmm5, VP8_FILTER_SHIFT + + paddw xmm7, [rd GLOBAL] + psraw xmm7, VP8_FILTER_SHIFT + + packuswb xmm1, xmm1 + packuswb xmm3, xmm3 + + packuswb xmm5, xmm5 + movq [rdi], xmm1 + + packuswb xmm7, xmm7 + movq [rdi+rdx], xmm3 + + lea rdi, [rdi + 2*rdx] + movq [rdi], xmm5 + + lea rsp, [rsp + 4*16] + movq [rdi+rdx], xmm7 + + lea rdi, [rdi + 2*rdx] + cmp rdi, rcx + + jne .next_row + + lea rsp, [rsp + 16] + +done8x8: + ;add rsp, 144 + pop rsp + ; begin epilog + pop rdi + pop rsi + RESTORE_GOT + RESTORE_XMM + UNSHADOW_ARGS + pop rbp + ret + +SECTION_RODATA +align 16 +shuf1b: + db 0, 5, 1, 6, 2, 7, 3, 8, 4, 9, 5, 10, 6, 11, 7, 12 +shuf2b: + db 2, 4, 3, 5, 4, 6, 5, 7, 6, 8, 7, 9, 8, 10, 9, 11 +shuf3b: + db 1, 3, 2, 4, 3, 5, 4, 6, 5, 7, 6, 8, 7, 9, 8, 10 + +align 16 +rd: + times 8 dw 0x40 + +align 16 +k0_k5: + times 8 db 0, 0 ;placeholder + times 8 db 0, 0 + times 8 db 2, 1 + times 8 db 0, 0 + times 8 db 3, 3 + times 8 db 0, 0 + times 8 db 1, 2 + times 8 db 0, 0 +k1_k3: + times 8 db 0, 0 ;placeholder + times 8 db -6, 12 + times 8 db -11, 36 + times 8 db -9, 50 + times 8 db -16, 77 + times 8 db -6, 93 + times 8 db -8, 108 + times 8 db -1, 123 +k2_k4: + times 8 db 128, 0 ;placeholder + times 8 db 123, -1 + times 8 db 108, -8 + times 8 db 93, -6 + times 8 db 77, -16 + times 8 db 50, -9 + times 8 db 36, -11 + times 8 db 12, -6 +align 16 +vp8_bilinear_filters_ssse3: + times 8 db 128, 0 + times 8 db 112, 16 + times 8 db 96, 32 + times 8 db 80, 48 + times 8 db 64, 64 + times 8 db 48, 80 + times 8 db 32, 96 + times 8 db 16, 112 + diff --git a/vp8/common/x86/subpixel_x86.h b/vp8/common/x86/subpixel_x86.h index efa7b2e..75991cc 100644 --- a/vp8/common/x86/subpixel_x86.h +++ b/vp8/common/x86/subpixel_x86.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -85,4 +86,37 @@ extern prototype_subpixel_predict(vp8_bilinear_predict8x8_sse2); #endif #endif +#if HAVE_SSSE3 +extern prototype_subpixel_predict(vp8_sixtap_predict16x16_ssse3); +extern prototype_subpixel_predict(vp8_sixtap_predict8x8_ssse3); +extern prototype_subpixel_predict(vp8_sixtap_predict8x4_ssse3); +extern prototype_subpixel_predict(vp8_sixtap_predict4x4_ssse3); +extern prototype_subpixel_predict(vp8_bilinear_predict16x16_ssse3); +extern prototype_subpixel_predict(vp8_bilinear_predict8x8_ssse3); + +#if !CONFIG_RUNTIME_CPU_DETECT +#undef vp8_subpix_sixtap16x16 +#define vp8_subpix_sixtap16x16 vp8_sixtap_predict16x16_ssse3 + +#undef vp8_subpix_sixtap8x8 +#define vp8_subpix_sixtap8x8 vp8_sixtap_predict8x8_ssse3 + +#undef vp8_subpix_sixtap8x4 +#define vp8_subpix_sixtap8x4 vp8_sixtap_predict8x4_ssse3 + +#undef vp8_subpix_sixtap4x4 +#define vp8_subpix_sixtap4x4 vp8_sixtap_predict4x4_ssse3 + + +#undef vp8_subpix_bilinear16x16 +#define vp8_subpix_bilinear16x16 vp8_bilinear_predict16x16_ssse3 + +#undef vp8_subpix_bilinear8x8 +#define vp8_subpix_bilinear8x8 vp8_bilinear_predict8x8_ssse3 + +#endif +#endif + + + #endif diff --git a/vp8/common/x86/vp8_asm_stubs.c b/vp8/common/x86/vp8_asm_stubs.c index 68454f7..950d962 100644 --- a/vp8/common/x86/vp8_asm_stubs.c +++ b/vp8/common/x86/vp8_asm_stubs.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -67,6 +68,17 @@ extern void vp8_filter_block1d8_v6_sse2 unsigned int output_width, const short *vp8_filter ); +extern void vp8_filter_block1d16_v6_sse2 +( + unsigned short *src_ptr, + unsigned char *output_ptr, + int dst_ptich, + unsigned int pixels_per_line, + unsigned int pixel_step, + unsigned int output_height, + unsigned int output_width, + const short *vp8_filter +); extern void vp8_unpack_block1d16_h6_sse2 ( unsigned char *src_ptr, @@ -75,31 +87,32 @@ extern void vp8_unpack_block1d16_h6_sse2 unsigned int output_height, unsigned int output_width ); -extern void vp8_unpack_block1d8_h6_sse2 +extern void vp8_filter_block1d8_h6_only_sse2 ( unsigned char *src_ptr, - unsigned short *output_ptr, unsigned int src_pixels_per_line, + unsigned char *output_ptr, + int dst_ptich, unsigned int output_height, - unsigned int output_width + const short *vp8_filter ); -extern void vp8_pack_block1d8_v6_sse2 +extern void vp8_filter_block1d16_h6_only_sse2 ( - unsigned short *src_ptr, - unsigned char *output_ptr, + unsigned char *src_ptr, + unsigned int src_pixels_per_line, + unsigned char *output_ptr, int dst_ptich, - unsigned int pixels_per_line, - unsigned int output_height, - unsigned int output_width + unsigned int output_height, + const short *vp8_filter ); -extern void vp8_pack_block1d16_v6_sse2 +extern void vp8_filter_block1d8_v6_only_sse2 ( - unsigned short *src_ptr, + unsigned char *src_ptr, + unsigned int src_pixels_per_line, unsigned char *output_ptr, int dst_ptich, - unsigned int pixels_per_line, - unsigned int output_height, - unsigned int output_width + unsigned int output_height, + const short *vp8_filter ); extern prototype_subpixel_predict(vp8_bilinear_predict8x8_mmx); @@ -246,28 +259,69 @@ void vp8_sixtap_predict16x16_sse2 if (xoffset) { - HFilter = vp8_six_tap_mmx[xoffset]; - vp8_filter_block1d16_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 21, 32, HFilter); + if (yoffset) + { + HFilter = vp8_six_tap_mmx[xoffset]; + vp8_filter_block1d16_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 21, 32, HFilter); + VFilter = vp8_six_tap_mmx[yoffset]; + vp8_filter_block1d16_v6_sse2(FData2 + 32, dst_ptr, dst_pitch, 32, 16 , 16, dst_pitch, VFilter); + } + else + { + // First-pass only + HFilter = vp8_six_tap_mmx[xoffset]; + vp8_filter_block1d16_h6_only_sse2(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 16, HFilter); + } } else { + // Second-pass only + VFilter = vp8_six_tap_mmx[yoffset]; vp8_unpack_block1d16_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 21, 32); + vp8_filter_block1d16_v6_sse2(FData2 + 32, dst_ptr, dst_pitch, 32, 16 , 16, dst_pitch, VFilter); } +} - if (yoffset) + +void vp8_sixtap_predict8x8_sse2 +( + unsigned char *src_ptr, + int src_pixels_per_line, + int xoffset, + int yoffset, + unsigned char *dst_ptr, + int dst_pitch +) +{ + DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 256); // Temp data bufffer used in filtering + const short *HFilter, *VFilter; + + if (xoffset) { - VFilter = vp8_six_tap_mmx[yoffset]; - vp8_filter_block1d8_v6_sse2(FData2 + 32, dst_ptr, dst_pitch, 32, 16 , 16, 16, VFilter); - vp8_filter_block1d8_v6_sse2(FData2 + 40, dst_ptr + 8, dst_pitch, 32, 16 , 16, 16, VFilter); + if (yoffset) + { + HFilter = vp8_six_tap_mmx[xoffset]; + vp8_filter_block1d8_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 13, 16, HFilter); + VFilter = vp8_six_tap_mmx[yoffset]; + vp8_filter_block1d8_v6_sse2(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 8, dst_pitch, VFilter); + } + else + { + // First-pass only + HFilter = vp8_six_tap_mmx[xoffset]; + vp8_filter_block1d8_h6_only_sse2(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 8, HFilter); + } } else { - vp8_pack_block1d16_v6_sse2(FData2 + 32, dst_ptr, dst_pitch, 32, 16, 16); + // Second-pass only + VFilter = vp8_six_tap_mmx[yoffset]; + vp8_filter_block1d8_v6_only_sse2(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 8, VFilter); } } -void vp8_sixtap_predict8x8_sse2 +void vp8_sixtap_predict8x4_sse2 ( unsigned char *src_ptr, int src_pixels_per_line, @@ -282,29 +336,126 @@ void vp8_sixtap_predict8x8_sse2 if (xoffset) { - HFilter = vp8_six_tap_mmx[xoffset]; - vp8_filter_block1d8_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 13, 16, HFilter); + if (yoffset) + { + HFilter = vp8_six_tap_mmx[xoffset]; + vp8_filter_block1d8_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 9, 16, HFilter); + VFilter = vp8_six_tap_mmx[yoffset]; + vp8_filter_block1d8_v6_sse2(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 4, dst_pitch, VFilter); + } + else + { + // First-pass only + HFilter = vp8_six_tap_mmx[xoffset]; + vp8_filter_block1d8_h6_only_sse2(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 4, HFilter); + } } else { - vp8_unpack_block1d8_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 13, 16); + // Second-pass only + VFilter = vp8_six_tap_mmx[yoffset]; + vp8_filter_block1d8_v6_only_sse2(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 4, VFilter); } +} + +#endif + +#if HAVE_SSSE3 + +extern void vp8_filter_block1d8_h6_ssse3 +( + unsigned char *src_ptr, + unsigned int src_pixels_per_line, + unsigned char *output_ptr, + unsigned int output_pitch, + unsigned int output_height, + unsigned int vp8_filter_index +); + +extern void vp8_filter_block1d16_h6_ssse3 +( + unsigned char *src_ptr, + unsigned int src_pixels_per_line, + unsigned char *output_ptr, + unsigned int output_pitch, + unsigned int output_height, + unsigned int vp8_filter_index +); + +extern void vp8_filter_block1d16_v6_ssse3 +( + unsigned char *src_ptr, + unsigned int src_pitch, + unsigned char *output_ptr, + unsigned int out_pitch, + unsigned int output_height, + unsigned int vp8_filter_index +); + +extern void vp8_filter_block1d8_v6_ssse3 +( + unsigned char *src_ptr, + unsigned int src_pitch, + unsigned char *output_ptr, + unsigned int out_pitch, + unsigned int output_height, + unsigned int vp8_filter_index +); + +extern void vp8_filter_block1d4_h6_ssse3 +( + unsigned char *src_ptr, + unsigned int src_pixels_per_line, + unsigned char *output_ptr, + unsigned int output_pitch, + unsigned int output_height, + unsigned int vp8_filter_index +); - if (yoffset) +extern void vp8_filter_block1d4_v6_ssse3 +( + unsigned char *src_ptr, + unsigned int src_pitch, + unsigned char *output_ptr, + unsigned int out_pitch, + unsigned int output_height, + unsigned int vp8_filter_index +); + +void vp8_sixtap_predict16x16_ssse3 +( + unsigned char *src_ptr, + int src_pixels_per_line, + int xoffset, + int yoffset, + unsigned char *dst_ptr, + int dst_pitch + +) +{ + DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 24*24); + + if (xoffset) { - VFilter = vp8_six_tap_mmx[yoffset]; - vp8_filter_block1d8_v6_sse2(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 8, dst_pitch, VFilter); + if (yoffset) + { + vp8_filter_block1d16_h6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, FData2, 16, 21, xoffset); + vp8_filter_block1d16_v6_ssse3(FData2 , 16, dst_ptr, dst_pitch, 16, yoffset); + } + else + { + // First-pass only + vp8_filter_block1d16_h6_ssse3(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 16, xoffset); + } } else { - vp8_pack_block1d8_v6_sse2(FData2 + 16, dst_ptr, dst_pitch, 16, 8, dst_pitch); + // Second-pass only + vp8_filter_block1d16_v6_ssse3(src_ptr - (2 * src_pixels_per_line) , src_pixels_per_line, dst_ptr, dst_pitch, 16, yoffset); } - - } - -void vp8_sixtap_predict8x4_sse2 +void vp8_sixtap_predict8x8_ssse3 ( unsigned char *src_ptr, int src_pixels_per_line, @@ -314,29 +465,89 @@ void vp8_sixtap_predict8x4_sse2 int dst_pitch ) { - DECLARE_ALIGNED_ARRAY(16, unsigned short, FData2, 256); // Temp data bufffer used in filtering - const short *HFilter, *VFilter; + DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 256); if (xoffset) { - HFilter = vp8_six_tap_mmx[xoffset]; - vp8_filter_block1d8_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 1, 9, 16, HFilter); + if (yoffset) + { + vp8_filter_block1d8_h6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, FData2, 8, 13, xoffset); + vp8_filter_block1d8_v6_ssse3(FData2, 8, dst_ptr, dst_pitch, 8, yoffset); + } + else + { + vp8_filter_block1d8_h6_ssse3(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 8, xoffset); + } } else { - vp8_unpack_block1d8_h6_sse2(src_ptr - (2 * src_pixels_per_line), FData2, src_pixels_per_line, 9, 16); + // Second-pass only + vp8_filter_block1d8_v6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 8, yoffset); } +} + + +void vp8_sixtap_predict8x4_ssse3 +( + unsigned char *src_ptr, + int src_pixels_per_line, + int xoffset, + int yoffset, + unsigned char *dst_ptr, + int dst_pitch +) +{ + DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 256); - if (yoffset) + if (xoffset) { - VFilter = vp8_six_tap_mmx[yoffset]; - vp8_filter_block1d8_v6_sse2(FData2 + 16, dst_ptr, dst_pitch, 16, 8 , 4, dst_pitch, VFilter); + if (yoffset) + { + vp8_filter_block1d8_h6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, FData2, 8, 9, xoffset); + vp8_filter_block1d8_v6_ssse3(FData2, 8, dst_ptr, dst_pitch, 4, yoffset); + } + else + { + // First-pass only + vp8_filter_block1d8_h6_ssse3(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 4, xoffset); + } } else { - vp8_pack_block1d8_v6_sse2(FData2 + 16, dst_ptr, dst_pitch, 16, 4, dst_pitch); + // Second-pass only + vp8_filter_block1d8_v6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 4, yoffset); } +} +void vp8_sixtap_predict4x4_ssse3 +( + unsigned char *src_ptr, + int src_pixels_per_line, + int xoffset, + int yoffset, + unsigned char *dst_ptr, + int dst_pitch +) +{ + DECLARE_ALIGNED_ARRAY(16, unsigned char, FData2, 4*9); + + if (xoffset) + { + if (yoffset) + { + vp8_filter_block1d4_h6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, FData2, 4, 9, xoffset); + vp8_filter_block1d4_v6_ssse3(FData2, 4, dst_ptr, dst_pitch, 4, yoffset); + } + else + { + vp8_filter_block1d4_h6_ssse3(src_ptr, src_pixels_per_line, dst_ptr, dst_pitch, 4, xoffset); + } + } + else + { + vp8_filter_block1d4_v6_ssse3(src_ptr - (2 * src_pixels_per_line), src_pixels_per_line, dst_ptr, dst_pitch, 4, yoffset); + } } + #endif diff --git a/vp8/common/x86/x86_systemdependent.c b/vp8/common/x86/x86_systemdependent.c index 5312e06..b983bc2 100644 --- a/vp8/common/x86/x86_systemdependent.c +++ b/vp8/common/x86/x86_systemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -26,6 +27,7 @@ void vp8_arch_x86_common_init(VP8_COMMON *ctx) int mmx_enabled = flags & HAS_MMX; int xmm_enabled = flags & HAS_SSE; int wmt_enabled = flags & HAS_SSE2; + int SSSE3Enabled = flags & HAS_SSSE3; /* Note: * @@ -41,7 +43,7 @@ void vp8_arch_x86_common_init(VP8_COMMON *ctx) { rtcd->idct.idct1 = vp8_short_idct4x4llm_1_mmx; rtcd->idct.idct16 = vp8_short_idct4x4llm_mmx; - rtcd->idct.idct1_scalar = vp8_dc_only_idct_mmx; + rtcd->idct.idct1_scalar_add = vp8_dc_only_idct_add_mmx; rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_mmx; rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_mmx; @@ -113,5 +115,19 @@ void vp8_arch_x86_common_init(VP8_COMMON *ctx) } #endif + +#if HAVE_SSSE3 + + if (SSSE3Enabled) + { + rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_ssse3; + rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_ssse3; + rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_ssse3; + rtcd->subpix.sixtap4x4 = vp8_sixtap_predict4x4_ssse3; + rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_ssse3; + rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_ssse3; + } +#endif + #endif } diff --git a/vp8/decoder/arm/armv5/dequantize_v5.asm b/vp8/decoder/arm/armv5/dequantize_v5.asm index eb3f030..de3648a 100644 --- a/vp8/decoder/arm/armv5/dequantize_v5.asm +++ b/vp8/decoder/arm/armv5/dequantize_v5.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/decoder/arm/armv6/dboolhuff_v6.asm b/vp8/decoder/arm/armv6/dboolhuff_v6.asm index 143e33e..6515804 100644 --- a/vp8/decoder/arm/armv6/dboolhuff_v6.asm +++ b/vp8/decoder/arm/armv6/dboolhuff_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/decoder/arm/armv6/dequant_dc_idct_v6.asm b/vp8/decoder/arm/armv6/dequant_dc_idct_v6.asm new file mode 100644 index 0000000..6bebda2 --- /dev/null +++ b/vp8/decoder/arm/armv6/dequant_dc_idct_v6.asm @@ -0,0 +1,218 @@ +; +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. +; +; Use of this source code is governed by a BSD-style license and patent +; grant that can be found in the LICENSE file in the root of the source +; tree. All contributing project authors may be found in the AUTHORS +; file in the root of the source tree. +; + + + EXPORT |vp8_dequant_dc_idct_add_v6| + + AREA |.text|, CODE, READONLY + +;void vp8_dequant_dc_idct_v6(short *input, short *dq, unsigned char *pred, +; unsigned char *dest, int pitch, int stride, int Dc) +; r0 = input +; r1 = dq +; r2 = pred +; r3 = dest +; sp + 36 = pitch ; +4 = 40 +; sp + 40 = stride ; +4 = 44 +; sp + 44 = Dc ; +4 = 48 + + +|vp8_dequant_dc_idct_add_v6| PROC + stmdb sp!, {r4-r11, lr} + + ldr r6, [sp, #44] + + ldr r4, [r0] ;input + ldr r5, [r1], #4 ;dq + + sub sp, sp, #4 + str r3, [sp] + + smultt r7, r4, r5 + + ldr r4, [r0, #4] ;input + ldr r5, [r1], #4 ;dq + + strh r6, [r0], #2 + strh r7, [r0], #2 + + smulbb r6, r4, r5 + smultt r7, r4, r5 + + ldr r4, [r0, #4] ;input + ldr r5, [r1], #4 ;dq + + strh r6, [r0], #2 + strh r7, [r0], #2 + + mov r12, #3 + +vp8_dequant_dc_add_loop + smulbb r6, r4, r5 + smultt r7, r4, r5 + + ldr r4, [r0, #4] ;input + ldr r5, [r1], #4 ;dq + + strh r6, [r0], #2 + strh r7, [r0], #2 + + smulbb r6, r4, r5 + smultt r7, r4, r5 + + subs r12, r12, #1 + + ldrne r4, [r0, #4] + ldrne r5, [r1], #4 + + strh r6, [r0], #2 + strh r7, [r0], #2 + + bne vp8_dequant_dc_add_loop + + sub r0, r0, #32 + mov r1, r0 + +; short_idct4x4llm_v6_dual + ldr r3, cospi8sqrt2minus1 + ldr r4, sinpi8sqrt2 + ldr r6, [r0, #8] + mov r5, #2 +vp8_dequant_dc_idct_loop1_v6 + ldr r12, [r0, #24] + ldr r14, [r0, #16] + smulwt r9, r3, r6 + smulwb r7, r3, r6 + smulwt r10, r4, r6 + smulwb r8, r4, r6 + pkhbt r7, r7, r9, lsl #16 + smulwt r11, r3, r12 + pkhbt r8, r8, r10, lsl #16 + uadd16 r6, r6, r7 + smulwt r7, r4, r12 + smulwb r9, r3, r12 + smulwb r10, r4, r12 + subs r5, r5, #1 + pkhbt r9, r9, r11, lsl #16 + ldr r11, [r0], #4 + pkhbt r10, r10, r7, lsl #16 + uadd16 r7, r12, r9 + usub16 r7, r8, r7 + uadd16 r6, r6, r10 + uadd16 r10, r11, r14 + usub16 r8, r11, r14 + uadd16 r9, r10, r6 + usub16 r10, r10, r6 + uadd16 r6, r8, r7 + usub16 r7, r8, r7 + str r6, [r1, #8] + ldrne r6, [r0, #8] + str r7, [r1, #16] + str r10, [r1, #24] + str r9, [r1], #4 + bne vp8_dequant_dc_idct_loop1_v6 + + mov r5, #2 + sub r0, r1, #8 +vp8_dequant_dc_idct_loop2_v6 + ldr r6, [r0], #4 + ldr r7, [r0], #4 + ldr r8, [r0], #4 + ldr r9, [r0], #4 + smulwt r1, r3, r6 + smulwt r12, r4, r6 + smulwt lr, r3, r8 + smulwt r10, r4, r8 + pkhbt r11, r8, r6, lsl #16 + pkhbt r1, lr, r1, lsl #16 + pkhbt r12, r10, r12, lsl #16 + pkhtb r6, r6, r8, asr #16 + uadd16 r6, r1, r6 + pkhbt lr, r9, r7, lsl #16 + uadd16 r10, r11, lr + usub16 lr, r11, lr + pkhtb r8, r7, r9, asr #16 + subs r5, r5, #1 + smulwt r1, r3, r8 + smulwb r7, r3, r8 + smulwt r11, r4, r8 + smulwb r9, r4, r8 + pkhbt r1, r7, r1, lsl #16 + uadd16 r8, r1, r8 + pkhbt r11, r9, r11, lsl #16 + usub16 r1, r12, r8 + uadd16 r8, r11, r6 + ldr r9, c0x00040004 + ldr r12, [sp, #40] + uadd16 r6, r10, r8 + usub16 r7, r10, r8 + uadd16 r7, r7, r9 + uadd16 r6, r6, r9 + uadd16 r10, r14, r1 + usub16 r1, r14, r1 + uadd16 r10, r10, r9 + uadd16 r1, r1, r9 + ldr r11, [r2], r12 + mov r8, r7, asr #3 + pkhtb r9, r8, r10, asr #19 + mov r8, r1, asr #3 + pkhtb r8, r8, r6, asr #19 + uxtb16 lr, r11, ror #8 + qadd16 r9, r9, lr + uxtb16 lr, r11 + qadd16 r8, r8, lr + usat16 r9, #8, r9 + usat16 r8, #8, r8 + orr r9, r8, r9, lsl #8 + ldr r11, [r2], r12 + ldr lr, [sp] + ldr r12, [sp, #44] + mov r7, r7, lsl #16 + mov r1, r1, lsl #16 + mov r10, r10, lsl #16 + mov r6, r6, lsl #16 + mov r7, r7, asr #3 + pkhtb r7, r7, r10, asr #19 + mov r1, r1, asr #3 + pkhtb r1, r1, r6, asr #19 + uxtb16 r8, r11, ror #8 + qadd16 r7, r7, r8 + uxtb16 r8, r11 + qadd16 r1, r1, r8 + usat16 r7, #8, r7 + usat16 r1, #8, r1 + orr r1, r1, r7, lsl #8 + str r9, [lr], r12 + str r1, [lr], r12 + str lr, [sp] + bne vp8_dequant_dc_idct_loop2_v6 + +; vpx_memset + sub r0, r0, #32 + add sp, sp, #4 + + mov r12, #0 + str r12, [r0] + str r12, [r0, #4] + str r12, [r0, #8] + str r12, [r0, #12] + str r12, [r0, #16] + str r12, [r0, #20] + str r12, [r0, #24] + str r12, [r0, #28] + + ldmia sp!, {r4 - r11, pc} + ENDP ; |vp8_dequant_dc_idct_add_v6| + +; Constant Pool +cospi8sqrt2minus1 DCD 0x00004E7B +sinpi8sqrt2 DCD 0x00008A8C +c0x00040004 DCD 0x00040004 + + END diff --git a/vp8/decoder/arm/armv6/dequant_idct_v6.asm b/vp8/decoder/arm/armv6/dequant_idct_v6.asm new file mode 100644 index 0000000..47b671c --- /dev/null +++ b/vp8/decoder/arm/armv6/dequant_idct_v6.asm @@ -0,0 +1,196 @@ +; +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. +; +; Use of this source code is governed by a BSD-style license and patent +; grant that can be found in the LICENSE file in the root of the source +; tree. All contributing project authors may be found in the AUTHORS +; file in the root of the source tree. +; + + EXPORT |vp8_dequant_idct_add_v6| + + AREA |.text|, CODE, READONLY +;void vp8_dequant_idct_v6(short *input, short *dq, unsigned char *pred, +; unsigned char *dest, int pitch, int stride) +; r0 = input +; r1 = dq +; r2 = pred +; r3 = dest +; sp + 36 = pitch ; +4 = 40 +; sp + 40 = stride ; +4 = 44 + + +|vp8_dequant_idct_add_v6| PROC + stmdb sp!, {r4-r11, lr} + + ldr r4, [r0] ;input + ldr r5, [r1], #4 ;dq + + sub sp, sp, #4 + str r3, [sp] + + mov r12, #4 + +vp8_dequant_add_loop + smulbb r6, r4, r5 + smultt r7, r4, r5 + + ldr r4, [r0, #4] ;input + ldr r5, [r1], #4 ;dq + + strh r6, [r0], #2 + strh r7, [r0], #2 + + smulbb r6, r4, r5 + smultt r7, r4, r5 + + subs r12, r12, #1 + + ldrne r4, [r0, #4] + ldrne r5, [r1], #4 + + strh r6, [r0], #2 + strh r7, [r0], #2 + + bne vp8_dequant_add_loop + + sub r0, r0, #32 + mov r1, r0 + +; short_idct4x4llm_v6_dual + ldr r3, cospi8sqrt2minus1 + ldr r4, sinpi8sqrt2 + ldr r6, [r0, #8] + mov r5, #2 +vp8_dequant_idct_loop1_v6 + ldr r12, [r0, #24] + ldr r14, [r0, #16] + smulwt r9, r3, r6 + smulwb r7, r3, r6 + smulwt r10, r4, r6 + smulwb r8, r4, r6 + pkhbt r7, r7, r9, lsl #16 + smulwt r11, r3, r12 + pkhbt r8, r8, r10, lsl #16 + uadd16 r6, r6, r7 + smulwt r7, r4, r12 + smulwb r9, r3, r12 + smulwb r10, r4, r12 + subs r5, r5, #1 + pkhbt r9, r9, r11, lsl #16 + ldr r11, [r0], #4 + pkhbt r10, r10, r7, lsl #16 + uadd16 r7, r12, r9 + usub16 r7, r8, r7 + uadd16 r6, r6, r10 + uadd16 r10, r11, r14 + usub16 r8, r11, r14 + uadd16 r9, r10, r6 + usub16 r10, r10, r6 + uadd16 r6, r8, r7 + usub16 r7, r8, r7 + str r6, [r1, #8] + ldrne r6, [r0, #8] + str r7, [r1, #16] + str r10, [r1, #24] + str r9, [r1], #4 + bne vp8_dequant_idct_loop1_v6 + + mov r5, #2 + sub r0, r1, #8 +vp8_dequant_idct_loop2_v6 + ldr r6, [r0], #4 + ldr r7, [r0], #4 + ldr r8, [r0], #4 + ldr r9, [r0], #4 + smulwt r1, r3, r6 + smulwt r12, r4, r6 + smulwt lr, r3, r8 + smulwt r10, r4, r8 + pkhbt r11, r8, r6, lsl #16 + pkhbt r1, lr, r1, lsl #16 + pkhbt r12, r10, r12, lsl #16 + pkhtb r6, r6, r8, asr #16 + uadd16 r6, r1, r6 + pkhbt lr, r9, r7, lsl #16 + uadd16 r10, r11, lr + usub16 lr, r11, lr + pkhtb r8, r7, r9, asr #16 + subs r5, r5, #1 + smulwt r1, r3, r8 + smulwb r7, r3, r8 + smulwt r11, r4, r8 + smulwb r9, r4, r8 + pkhbt r1, r7, r1, lsl #16 + uadd16 r8, r1, r8 + pkhbt r11, r9, r11, lsl #16 + usub16 r1, r12, r8 + uadd16 r8, r11, r6 + ldr r9, c0x00040004 + ldr r12, [sp, #40] + uadd16 r6, r10, r8 + usub16 r7, r10, r8 + uadd16 r7, r7, r9 + uadd16 r6, r6, r9 + uadd16 r10, r14, r1 + usub16 r1, r14, r1 + uadd16 r10, r10, r9 + uadd16 r1, r1, r9 + ldr r11, [r2], r12 + mov r8, r7, asr #3 + pkhtb r9, r8, r10, asr #19 + mov r8, r1, asr #3 + pkhtb r8, r8, r6, asr #19 + uxtb16 lr, r11, ror #8 + qadd16 r9, r9, lr + uxtb16 lr, r11 + qadd16 r8, r8, lr + usat16 r9, #8, r9 + usat16 r8, #8, r8 + orr r9, r8, r9, lsl #8 + ldr r11, [r2], r12 + ldr lr, [sp] + ldr r12, [sp, #44] + mov r7, r7, lsl #16 + mov r1, r1, lsl #16 + mov r10, r10, lsl #16 + mov r6, r6, lsl #16 + mov r7, r7, asr #3 + pkhtb r7, r7, r10, asr #19 + mov r1, r1, asr #3 + pkhtb r1, r1, r6, asr #19 + uxtb16 r8, r11, ror #8 + qadd16 r7, r7, r8 + uxtb16 r8, r11 + qadd16 r1, r1, r8 + usat16 r7, #8, r7 + usat16 r1, #8, r1 + orr r1, r1, r7, lsl #8 + str r9, [lr], r12 + str r1, [lr], r12 + str lr, [sp] + bne vp8_dequant_idct_loop2_v6 + +; vpx_memset + sub r0, r0, #32 + add sp, sp, #4 + + mov r12, #0 + str r12, [r0] + str r12, [r0, #4] + str r12, [r0, #8] + str r12, [r0, #12] + str r12, [r0, #16] + str r12, [r0, #20] + str r12, [r0, #24] + str r12, [r0, #28] + + ldmia sp!, {r4 - r11, pc} + ENDP ; |vp8_dequant_idct_add_v6| + +; Constant Pool +cospi8sqrt2minus1 DCD 0x00004E7B +sinpi8sqrt2 DCD 0x00008A8C +c0x00040004 DCD 0x00040004 + + END diff --git a/vp8/decoder/arm/armv6/dequantdcidct_v6.asm b/vp8/decoder/arm/armv6/dequantdcidct_v6.asm deleted file mode 100644 index 3daa9b3..0000000 --- a/vp8/decoder/arm/armv6/dequantdcidct_v6.asm +++ /dev/null @@ -1,202 +0,0 @@ -; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. -; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. -; - - - EXPORT |vp8_dequant_dc_idct_v6| - ; ARM - ; REQUIRE8 - ; PRESERVE8 - - AREA |.text|, CODE, READONLY ; name this block of code -;void vp8_dequant_dc_idct_v6(short *input, short *dq, short *output, int pitch,int Dc) -|vp8_dequant_dc_idct_v6| PROC - stmdb sp!, {r4-r11, lr} - - ldr r6, [sp, #36] ;load Dc - - ldr r4, [r0] ;input - ldr r5, [r1], #4 ;dq - - sub sp, sp, #4 - str r0, [sp] - - smultt r7, r4, r5 - - ldr r4, [r0, #4] ;input - ldr r5, [r1], #4 ;dq - - strh r6, [r0], #2 - strh r7, [r0], #2 - - smulbb r6, r4, r5 - smultt r7, r4, r5 - - ldr r4, [r0, #4] ;input - ldr r5, [r1], #4 ;dq - - strh r6, [r0], #2 - strh r7, [r0], #2 - - mov r12, #3 - -dequant_dc_idct_loop - smulbb r6, r4, r5 - smultt r7, r4, r5 - - ldr r4, [r0, #4] ;input - ldr r5, [r1], #4 ;dq - - strh r6, [r0], #2 - strh r7, [r0], #2 - - smulbb r6, r4, r5 - smultt r7, r4, r5 - - subs r12, r12, #1 - - ldrne r4, [r0, #4] - ldrne r5, [r1], #4 - - strh r6, [r0], #2 - strh r7, [r0], #2 - - bne dequant_dc_idct_loop - - sub r0, r0, #32 - mov r1, r2 - mov r2, r3 - -; short_idct4x4llm_v6_dual - - mov r3, #0x00004E00 ; cos - orr r3, r3, #0x0000007B ; cospi8sqrt2minus1 - mov r4, #0x00008A00 ; sin - orr r4, r4, #0x0000008C ; sinpi8sqrt2 - mov r5, #0x2 ; i=2 i -loop1_dual_11 - ldr r6, [r0, #(4*2)] ; i5 | i4 5|4 - ldr r12, [r0, #(12*2)] ; i13 | i12 13|12 - ldr r14, [r0, #(8*2)] ; i9 | i8 9|8 - - smulwt r9, r3, r6 ; (ip[5] * cospi8sqrt2minus1) >> 16 5c - smulwb r7, r3, r6 ; (ip[4] * cospi8sqrt2minus1) >> 16 4c - smulwt r10, r4, r6 ; (ip[5] * sinpi8sqrt2) >> 16 5s - smulwb r8, r4, r6 ; (ip[4] * sinpi8sqrt2) >> 16 4s - pkhbt r7, r7, r9, lsl #16 ; 5c | 4c - smulwt r11, r3, r12 ; (ip[13] * cospi8sqrt2minus1) >> 16 13c - pkhbt r8, r8, r10, lsl #16 ; 5s | 4s - uadd16 r6, r6, r7 ; 5c+5 | 4c+4 - smulwt r7, r4, r12 ; (ip[13] * sinpi8sqrt2) >> 16 13s - smulwb r9, r3, r12 ; (ip[12] * cospi8sqrt2minus1) >> 16 12c - smulwb r10, r4, r12 ; (ip[12] * sinpi8sqrt2) >> 16 12s - subs r5, r5, #0x1 ; i-- -- - pkhbt r9, r9, r11, lsl #16 ; 13c | 12c - ldr r11, [r0], #0x4 ; i1 | i0 ++ 1|0 - pkhbt r10, r10, r7, lsl #16 ; 13s | 12s - uadd16 r7, r12, r9 ; 13c+13 | 12c+12 - usub16 r7, r8, r7 ; c c - uadd16 r6, r6, r10 ; d d - uadd16 r10, r11, r14 ; a a - usub16 r8, r11, r14 ; b b - uadd16 r9, r10, r6 ; a+d a+d - usub16 r10, r10, r6 ; a-d a-d - uadd16 r6, r8, r7 ; b+c b+c - usub16 r7, r8, r7 ; b-c b-c - str r6, [r1, r2] ; o5 | o4 - add r6, r2, r2 ; pitch * 2 p2 - str r7, [r1, r6] ; o9 | o8 - add r6, r6, r2 ; pitch * 3 p3 - str r10, [r1, r6] ; o13 | o12 - str r9, [r1], #0x4 ; o1 | o0 ++ - bne loop1_dual_11 ; - mov r5, #0x2 ; i=2 i - sub r0, r1, #8 ; reset input/output i/o -loop2_dual_22 - ldr r6, [r0, r2] ; i5 | i4 5|4 - ldr r1, [r0] ; i1 | i0 1|0 - ldr r12, [r0, #0x4] ; i3 | i2 3|2 - add r14, r2, #0x4 ; pitch + 2 p+2 - ldr r14, [r0, r14] ; i7 | i6 7|6 - smulwt r9, r3, r6 ; (ip[5] * cospi8sqrt2minus1) >> 16 5c - smulwt r7, r3, r1 ; (ip[1] * cospi8sqrt2minus1) >> 16 1c - smulwt r10, r4, r6 ; (ip[5] * sinpi8sqrt2) >> 16 5s - smulwt r8, r4, r1 ; (ip[1] * sinpi8sqrt2) >> 16 1s - pkhbt r11, r6, r1, lsl #16 ; i0 | i4 0|4 - pkhbt r7, r9, r7, lsl #16 ; 1c | 5c - pkhbt r8, r10, r8, lsl #16 ; 1s | 5s = temp1 © tc1 - pkhtb r1, r1, r6, asr #16 ; i1 | i5 1|5 - uadd16 r1, r7, r1 ; 1c+1 | 5c+5 = temp2 (d) td2 - pkhbt r9, r14, r12, lsl #16 ; i2 | i6 2|6 - uadd16 r10, r11, r9 ; a a - usub16 r9, r11, r9 ; b b - pkhtb r6, r12, r14, asr #16 ; i3 | i7 3|7 - subs r5, r5, #0x1 ; i-- -- - smulwt r7, r3, r6 ; (ip[3] * cospi8sqrt2minus1) >> 16 3c - smulwt r11, r4, r6 ; (ip[3] * sinpi8sqrt2) >> 16 3s - smulwb r12, r3, r6 ; (ip[7] * cospi8sqrt2minus1) >> 16 7c - smulwb r14, r4, r6 ; (ip[7] * sinpi8sqrt2) >> 16 7s - - pkhbt r7, r12, r7, lsl #16 ; 3c | 7c - pkhbt r11, r14, r11, lsl #16 ; 3s | 7s = temp1 (d) td1 - uadd16 r6, r7, r6 ; 3c+3 | 7c+7 = temp2 (c) tc2 - usub16 r12, r8, r6 ; c (o1 | o5) c - uadd16 r6, r11, r1 ; d (o3 | o7) d - uadd16 r7, r10, r6 ; a+d a+d - mov r8, #0x4 ; set up 4's 4 - orr r8, r8, #0x40000 ; 4|4 - usub16 r6, r10, r6 ; a-d a-d - uadd16 r6, r6, r8 ; a-d+4 3|7 - uadd16 r7, r7, r8 ; a+d+4 0|4 - uadd16 r10, r9, r12 ; b+c b+c - usub16 r1, r9, r12 ; b-c b-c - uadd16 r10, r10, r8 ; b+c+4 1|5 - uadd16 r1, r1, r8 ; b-c+4 2|6 - mov r8, r10, asr #19 ; o1 >> 3 - strh r8, [r0, #2] ; o1 - mov r8, r1, asr #19 ; o2 >> 3 - strh r8, [r0, #4] ; o2 - mov r8, r6, asr #19 ; o3 >> 3 - strh r8, [r0, #6] ; o3 - mov r8, r7, asr #19 ; o0 >> 3 - strh r8, [r0], r2 ; o0 +p - sxth r10, r10 ; - mov r8, r10, asr #3 ; o5 >> 3 - strh r8, [r0, #2] ; o5 - sxth r1, r1 ; - mov r8, r1, asr #3 ; o6 >> 3 - strh r8, [r0, #4] ; o6 - sxth r6, r6 ; - mov r8, r6, asr #3 ; o7 >> 3 - strh r8, [r0, #6] ; o7 - sxth r7, r7 ; - mov r8, r7, asr #3 ; o4 >> 3 - strh r8, [r0], r2 ; o4 +p -;;;;; subs r5, r5, #0x1 ; i-- -- - bne loop2_dual_22 ; - - -;vpx_memset - ldr r0, [sp] - add sp, sp, #4 - - mov r12, #0 - str r12, [r0] - str r12, [r0, #4] - str r12, [r0, #8] - str r12, [r0, #12] - str r12, [r0, #16] - str r12, [r0, #20] - str r12, [r0, #24] - str r12, [r0, #28] - - ldmia sp!, {r4 - r11, pc} ; replace vars, return restore - - ENDP ;|vp8_dequant_dc_idct_v68| - - END diff --git a/vp8/decoder/arm/armv6/dequantidct_v6.asm b/vp8/decoder/arm/armv6/dequantidct_v6.asm deleted file mode 100644 index 61bb48d..0000000 --- a/vp8/decoder/arm/armv6/dequantidct_v6.asm +++ /dev/null @@ -1,183 +0,0 @@ -; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. -; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. -; - - - EXPORT |vp8_dequant_idct_v6| - ; ARM - ; REQUIRE8 - ; PRESERVE8 - - AREA |.text|, CODE, READONLY ; name this block of code -;void vp8_dequant_idct_v6(short *input, short *dq, short *output, int pitch) -|vp8_dequant_idct_v6| PROC - stmdb sp!, {r4-r11, lr} - - ldr r4, [r0] ;input - ldr r5, [r1], #4 ;dq - - sub sp, sp, #4 - str r0, [sp] - - mov r12, #4 - -dequant_idct_loop - smulbb r6, r4, r5 - smultt r7, r4, r5 - - ldr r4, [r0, #4] ;input - ldr r5, [r1], #4 ;dq - - strh r6, [r0], #2 - strh r7, [r0], #2 - - smulbb r6, r4, r5 - smultt r7, r4, r5 - - subs r12, r12, #1 - - ldrne r4, [r0, #4] - ldrne r5, [r1], #4 - - strh r6, [r0], #2 - strh r7, [r0], #2 - - bne dequant_idct_loop - - sub r0, r0, #32 - mov r1, r2 - mov r2, r3 - -; short_idct4x4llm_v6_dual - - mov r3, #0x00004E00 ; cos - orr r3, r3, #0x0000007B ; cospi8sqrt2minus1 - mov r4, #0x00008A00 ; sin - orr r4, r4, #0x0000008C ; sinpi8sqrt2 - mov r5, #0x2 ; i=2 i -loop1_dual_1 - ldr r6, [r0, #(4*2)] ; i5 | i4 5|4 - ldr r12, [r0, #(12*2)] ; i13 | i12 13|12 - ldr r14, [r0, #(8*2)] ; i9 | i8 9|8 - - smulwt r9, r3, r6 ; (ip[5] * cospi8sqrt2minus1) >> 16 5c - smulwb r7, r3, r6 ; (ip[4] * cospi8sqrt2minus1) >> 16 4c - smulwt r10, r4, r6 ; (ip[5] * sinpi8sqrt2) >> 16 5s - smulwb r8, r4, r6 ; (ip[4] * sinpi8sqrt2) >> 16 4s - pkhbt r7, r7, r9, lsl #16 ; 5c | 4c - smulwt r11, r3, r12 ; (ip[13] * cospi8sqrt2minus1) >> 16 13c - pkhbt r8, r8, r10, lsl #16 ; 5s | 4s - uadd16 r6, r6, r7 ; 5c+5 | 4c+4 - smulwt r7, r4, r12 ; (ip[13] * sinpi8sqrt2) >> 16 13s - smulwb r9, r3, r12 ; (ip[12] * cospi8sqrt2minus1) >> 16 12c - smulwb r10, r4, r12 ; (ip[12] * sinpi8sqrt2) >> 16 12s - subs r5, r5, #0x1 ; i-- -- - pkhbt r9, r9, r11, lsl #16 ; 13c | 12c - ldr r11, [r0], #0x4 ; i1 | i0 ++ 1|0 - pkhbt r10, r10, r7, lsl #16 ; 13s | 12s - uadd16 r7, r12, r9 ; 13c+13 | 12c+12 - usub16 r7, r8, r7 ; c c - uadd16 r6, r6, r10 ; d d - uadd16 r10, r11, r14 ; a a - usub16 r8, r11, r14 ; b b - uadd16 r9, r10, r6 ; a+d a+d - usub16 r10, r10, r6 ; a-d a-d - uadd16 r6, r8, r7 ; b+c b+c - usub16 r7, r8, r7 ; b-c b-c - str r6, [r1, r2] ; o5 | o4 - add r6, r2, r2 ; pitch * 2 p2 - str r7, [r1, r6] ; o9 | o8 - add r6, r6, r2 ; pitch * 3 p3 - str r10, [r1, r6] ; o13 | o12 - str r9, [r1], #0x4 ; o1 | o0 ++ - bne loop1_dual_1 ; - mov r5, #0x2 ; i=2 i - sub r0, r1, #8 ; reset input/output i/o -loop2_dual_2 - ldr r6, [r0, r2] ; i5 | i4 5|4 - ldr r1, [r0] ; i1 | i0 1|0 - ldr r12, [r0, #0x4] ; i3 | i2 3|2 - add r14, r2, #0x4 ; pitch + 2 p+2 - ldr r14, [r0, r14] ; i7 | i6 7|6 - smulwt r9, r3, r6 ; (ip[5] * cospi8sqrt2minus1) >> 16 5c - smulwt r7, r3, r1 ; (ip[1] * cospi8sqrt2minus1) >> 16 1c - smulwt r10, r4, r6 ; (ip[5] * sinpi8sqrt2) >> 16 5s - smulwt r8, r4, r1 ; (ip[1] * sinpi8sqrt2) >> 16 1s - pkhbt r11, r6, r1, lsl #16 ; i0 | i4 0|4 - pkhbt r7, r9, r7, lsl #16 ; 1c | 5c - pkhbt r8, r10, r8, lsl #16 ; 1s | 5s = temp1 © tc1 - pkhtb r1, r1, r6, asr #16 ; i1 | i5 1|5 - uadd16 r1, r7, r1 ; 1c+1 | 5c+5 = temp2 (d) td2 - pkhbt r9, r14, r12, lsl #16 ; i2 | i6 2|6 - uadd16 r10, r11, r9 ; a a - usub16 r9, r11, r9 ; b b - pkhtb r6, r12, r14, asr #16 ; i3 | i7 3|7 - subs r5, r5, #0x1 ; i-- -- - smulwt r7, r3, r6 ; (ip[3] * cospi8sqrt2minus1) >> 16 3c - smulwt r11, r4, r6 ; (ip[3] * sinpi8sqrt2) >> 16 3s - smulwb r12, r3, r6 ; (ip[7] * cospi8sqrt2minus1) >> 16 7c - smulwb r14, r4, r6 ; (ip[7] * sinpi8sqrt2) >> 16 7s - - pkhbt r7, r12, r7, lsl #16 ; 3c | 7c - pkhbt r11, r14, r11, lsl #16 ; 3s | 7s = temp1 (d) td1 - uadd16 r6, r7, r6 ; 3c+3 | 7c+7 = temp2 (c) tc2 - usub16 r12, r8, r6 ; c (o1 | o5) c - uadd16 r6, r11, r1 ; d (o3 | o7) d - uadd16 r7, r10, r6 ; a+d a+d - mov r8, #0x4 ; set up 4's 4 - orr r8, r8, #0x40000 ; 4|4 - usub16 r6, r10, r6 ; a-d a-d - uadd16 r6, r6, r8 ; a-d+4 3|7 - uadd16 r7, r7, r8 ; a+d+4 0|4 - uadd16 r10, r9, r12 ; b+c b+c - usub16 r1, r9, r12 ; b-c b-c - uadd16 r10, r10, r8 ; b+c+4 1|5 - uadd16 r1, r1, r8 ; b-c+4 2|6 - mov r8, r10, asr #19 ; o1 >> 3 - strh r8, [r0, #2] ; o1 - mov r8, r1, asr #19 ; o2 >> 3 - strh r8, [r0, #4] ; o2 - mov r8, r6, asr #19 ; o3 >> 3 - strh r8, [r0, #6] ; o3 - mov r8, r7, asr #19 ; o0 >> 3 - strh r8, [r0], r2 ; o0 +p - sxth r10, r10 ; - mov r8, r10, asr #3 ; o5 >> 3 - strh r8, [r0, #2] ; o5 - sxth r1, r1 ; - mov r8, r1, asr #3 ; o6 >> 3 - strh r8, [r0, #4] ; o6 - sxth r6, r6 ; - mov r8, r6, asr #3 ; o7 >> 3 - strh r8, [r0, #6] ; o7 - sxth r7, r7 ; - mov r8, r7, asr #3 ; o4 >> 3 - strh r8, [r0], r2 ; o4 +p -;;;;; subs r5, r5, #0x1 ; i-- -- - bne loop2_dual_2 ; - ; - -;vpx_memset - ldr r0, [sp] - add sp, sp, #4 - - mov r12, #0 - str r12, [r0] - str r12, [r0, #4] - str r12, [r0, #8] - str r12, [r0, #12] - str r12, [r0, #16] - str r12, [r0, #20] - str r12, [r0, #24] - str r12, [r0, #28] - - ldmia sp!, {r4 - r11, pc} ; replace vars, return restore - - ENDP ;|vp8_dequant_idct_v6| - - END diff --git a/vp8/decoder/arm/armv6/dequantize_v6.asm b/vp8/decoder/arm/armv6/dequantize_v6.asm index 95e3859..72f7e0e 100644 --- a/vp8/decoder/arm/armv6/dequantize_v6.asm +++ b/vp8/decoder/arm/armv6/dequantize_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/decoder/arm/armv6/idct_blk_v6.c b/vp8/decoder/arm/armv6/idct_blk_v6.c new file mode 100644 index 0000000..3c7bc50 --- /dev/null +++ b/vp8/decoder/arm/armv6/idct_blk_v6.c @@ -0,0 +1,151 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include "vpx_ports/config.h" +#include "idct.h" +#include "dequantize.h" + +void vp8_dequant_dc_idct_add_y_block_v6 + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int stride, char *eobs, short *dc) +{ + int i; + + for (i = 0; i < 4; i++) + { + if (eobs[0] > 1) + vp8_dequant_dc_idct_add_v6 (q, dq, pre, dst, 16, stride, dc[0]); + else + vp8_dc_only_idct_add_v6 (dc[0], pre, dst, 16, stride); + + if (eobs[1] > 1) + vp8_dequant_dc_idct_add_v6 (q+16, dq, pre+4, dst+4, 16, stride, dc[1]); + else + vp8_dc_only_idct_add_v6 (dc[1], pre+4, dst+4, 16, stride); + + if (eobs[2] > 1) + vp8_dequant_dc_idct_add_v6 (q+32, dq, pre+8, dst+8, 16, stride, dc[2]); + else + vp8_dc_only_idct_add_v6 (dc[2], pre+8, dst+8, 16, stride); + + if (eobs[3] > 1) + vp8_dequant_dc_idct_add_v6 (q+48, dq, pre+12, dst+12, 16, stride, dc[3]); + else + vp8_dc_only_idct_add_v6 (dc[3], pre+12, dst+12, 16, stride); + + q += 64; + dc += 4; + pre += 64; + dst += 4*stride; + eobs += 4; + } +} + +void vp8_dequant_idct_add_y_block_v6 + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int stride, char *eobs) +{ + int i; + + for (i = 0; i < 4; i++) + { + if (eobs[0] > 1) + vp8_dequant_idct_add_v6 (q, dq, pre, dst, 16, stride); + else + { + vp8_dc_only_idct_add_v6 (q[0]*dq[0], pre, dst, 16, stride); + ((int *)q)[0] = 0; + } + + if (eobs[1] > 1) + vp8_dequant_idct_add_v6 (q+16, dq, pre+4, dst+4, 16, stride); + else + { + vp8_dc_only_idct_add_v6 (q[16]*dq[0], pre+4, dst+4, 16, stride); + ((int *)(q+16))[0] = 0; + } + + if (eobs[2] > 1) + vp8_dequant_idct_add_v6 (q+32, dq, pre+8, dst+8, 16, stride); + else + { + vp8_dc_only_idct_add_v6 (q[32]*dq[0], pre+8, dst+8, 16, stride); + ((int *)(q+32))[0] = 0; + } + + if (eobs[3] > 1) + vp8_dequant_idct_add_v6 (q+48, dq, pre+12, dst+12, 16, stride); + else + { + vp8_dc_only_idct_add_v6 (q[48]*dq[0], pre+12, dst+12, 16, stride); + ((int *)(q+48))[0] = 0; + } + + q += 64; + pre += 64; + dst += 4*stride; + eobs += 4; + } +} + +void vp8_dequant_idct_add_uv_block_v6 + (short *q, short *dq, unsigned char *pre, + unsigned char *dstu, unsigned char *dstv, int stride, char *eobs) +{ + int i; + + for (i = 0; i < 2; i++) + { + if (eobs[0] > 1) + vp8_dequant_idct_add_v6 (q, dq, pre, dstu, 8, stride); + else + { + vp8_dc_only_idct_add_v6 (q[0]*dq[0], pre, dstu, 8, stride); + ((int *)q)[0] = 0; + } + + if (eobs[1] > 1) + vp8_dequant_idct_add_v6 (q+16, dq, pre+4, dstu+4, 8, stride); + else + { + vp8_dc_only_idct_add_v6 (q[16]*dq[0], pre+4, dstu+4, 8, stride); + ((int *)(q+16))[0] = 0; + } + + q += 32; + pre += 32; + dstu += 4*stride; + eobs += 2; + } + + for (i = 0; i < 2; i++) + { + if (eobs[0] > 1) + vp8_dequant_idct_add_v6 (q, dq, pre, dstv, 8, stride); + else + { + vp8_dc_only_idct_add_v6 (q[0]*dq[0], pre, dstv, 8, stride); + ((int *)q)[0] = 0; + } + + if (eobs[1] > 1) + vp8_dequant_idct_add_v6 (q+16, dq, pre+4, dstv+4, 8, stride); + else + { + vp8_dc_only_idct_add_v6 (q[16]*dq[0], pre+4, dstv+4, 8, stride); + ((int *)(q+16))[0] = 0; + } + + q += 32; + pre += 32; + dstv += 4*stride; + eobs += 2; + } +} diff --git a/vp8/decoder/arm/dboolhuff_arm.h b/vp8/decoder/arm/dboolhuff_arm.h index 495004f..d2ebc71 100644 --- a/vp8/decoder/arm/dboolhuff_arm.h +++ b/vp8/decoder/arm/dboolhuff_arm.h @@ -16,9 +16,6 @@ #undef vp8_dbool_start #define vp8_dbool_start vp8dx_start_decode_v6 -#undef vp8_dbool_stop -#define vp8_dbool_stop vp8dx_stop_decode_v6 - #undef vp8_dbool_fill #define vp8_dbool_fill vp8_bool_decoder_fill_v6 @@ -33,9 +30,6 @@ #undef vp8_dbool_start #define vp8_dbool_start vp8dx_start_decode_neon -#undef vp8_dbool_stop -#define vp8_dbool_stop vp8dx_stop_decode_neon - #undef vp8_dbool_fill #define vp8_dbool_fill vp8_bool_decoder_fill_neon diff --git a/vp8/decoder/arm/dequantize_arm.c b/vp8/decoder/arm/dequantize_arm.c index 54006a9..3926587 100644 --- a/vp8/decoder/arm/dequantize_arm.c +++ b/vp8/decoder/arm/dequantize_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/decoder/arm/dequantize_arm.h b/vp8/decoder/arm/dequantize_arm.h index c8a61a4..40151e0 100644 --- a/vp8/decoder/arm/dequantize_arm.h +++ b/vp8/decoder/arm/dequantize_arm.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -13,32 +14,56 @@ #if HAVE_ARMV6 extern prototype_dequant_block(vp8_dequantize_b_v6); -extern prototype_dequant_idct(vp8_dequant_idct_v6); -extern prototype_dequant_idct_dc(vp8_dequant_dc_idct_v6); +extern prototype_dequant_idct_add(vp8_dequant_idct_add_v6); +extern prototype_dequant_dc_idct_add(vp8_dequant_dc_idct_add_v6); +extern prototype_dequant_dc_idct_add_y_block(vp8_dequant_dc_idct_add_y_block_v6); +extern prototype_dequant_idct_add_y_block(vp8_dequant_idct_add_y_block_v6); +extern prototype_dequant_idct_add_uv_block(vp8_dequant_idct_add_uv_block_v6); #undef vp8_dequant_block #define vp8_dequant_block vp8_dequantize_b_v6 -#undef vp8_dequant_idct -#define vp8_dequant_idct vp8_dequant_idct_v6 +#undef vp8_dequant_idct_add +#define vp8_dequant_idct_add vp8_dequant_idct_add_v6 -#undef vp8_dequant_idct_dc -#define vp8_dequant_idct_dc vp8_dequant_dc_idct_v6 +#undef vp8_dequant_dc_idct_add +#define vp8_dequant_dc_idct_add vp8_dequant_dc_idct_add_v6 + +#undef vp8_dequant_dc_idct_add_y_block +#define vp8_dequant_dc_idct_add_y_block vp8_dequant_dc_idct_add_y_block_v6 + +#undef vp8_dequant_idct_add_y_block +#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_v6 + +#undef vp8_dequant_idct_add_uv_block +#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_v6 #endif #if HAVE_ARMV7 extern prototype_dequant_block(vp8_dequantize_b_neon); -extern prototype_dequant_idct(vp8_dequant_idct_neon); -extern prototype_dequant_idct_dc(vp8_dequant_dc_idct_neon); +extern prototype_dequant_idct_add(vp8_dequant_idct_add_neon); +extern prototype_dequant_dc_idct_add(vp8_dequant_dc_idct_add_neon); +extern prototype_dequant_dc_idct_add_y_block(vp8_dequant_dc_idct_add_y_block_neon); +extern prototype_dequant_idct_add_y_block(vp8_dequant_idct_add_y_block_neon); +extern prototype_dequant_idct_add_uv_block(vp8_dequant_idct_add_uv_block_neon); #undef vp8_dequant_block #define vp8_dequant_block vp8_dequantize_b_neon -#undef vp8_dequant_idct -#define vp8_dequant_idct vp8_dequant_idct_neon +#undef vp8_dequant_idct_add +#define vp8_dequant_idct_add vp8_dequant_idct_add_neon + +#undef vp8_dequant_dc_idct_add +#define vp8_dequant_dc_idct_add vp8_dequant_dc_idct_add_neon + +#undef vp8_dequant_dc_idct_add_y_block +#define vp8_dequant_dc_idct_add_y_block vp8_dequant_dc_idct_add_y_block_neon + +#undef vp8_dequant_idct_add_y_block +#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_neon -#undef vp8_dequant_idct_dc -#define vp8_dequant_idct_dc vp8_dequant_dc_idct_neon +#undef vp8_dequant_idct_add_uv_block +#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_neon #endif #endif diff --git a/vp8/decoder/arm/detokenize.asm b/vp8/decoder/arm/detokenize.asm new file mode 100644 index 0000000..45e068a --- /dev/null +++ b/vp8/decoder/arm/detokenize.asm @@ -0,0 +1,320 @@ +; +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. +; +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. +; + + + EXPORT |vp8_decode_mb_tokens_v6| + + AREA |.text|, CODE, READONLY ; name this block of code + + INCLUDE vpx_asm_offsets.asm + +l_qcoeff EQU 0 +l_i EQU 4 +l_type EQU 8 +l_stop EQU 12 +l_c EQU 16 +l_l_ptr EQU 20 +l_a_ptr EQU 24 +l_bc EQU 28 +l_coef_ptr EQU 32 +l_stacksize EQU 64 + + +;; constant offsets -- these should be created at build time +c_block2above_offset EQU 25 +c_entropy_nodes EQU 11 +c_dct_eob_token EQU 11 + +|vp8_decode_mb_tokens_v6| PROC + stmdb sp!, {r4 - r11, lr} + sub sp, sp, #l_stacksize + mov r7, r1 ; type + mov r9, r0 ; detoken + + ldr r1, [r9, #detok_current_bc] + ldr r0, [r9, #detok_qcoeff_start_ptr] + mov r11, #0 ; i + mov r3, #16 ; stop + + cmp r7, #1 ; type ?= 1 + addeq r11, r11, #24 ; i = 24 + addeq r3, r3, #8 ; stop = 24 + addeq r0, r0, #3, 24 ; qcoefptr += 24*16 + + str r0, [sp, #l_qcoeff] + str r11, [sp, #l_i] + str r7, [sp, #l_type] + str r3, [sp, #l_stop] + str r1, [sp, #l_bc] + + add lr, r9, r7, lsl #2 ; detoken + type*4 + + ldr r8, [r1, #bool_decoder_user_buffer] + + ldr r10, [lr, #detok_coef_probs] + ldr r5, [r1, #bool_decoder_count] + ldr r6, [r1, #bool_decoder_range] + ldr r4, [r1, #bool_decoder_value] + + str r10, [sp, #l_coef_ptr] + +BLOCK_LOOP + ldr r3, [r9, #detok_ptr_block2leftabove] + ldr r1, [r9, #detok_L] + ldr r2, [r9, #detok_A] + ldrb r12, [r3, r11]! ; block2left[i] + ldrb r3, [r3, #c_block2above_offset]; block2above[i] + + cmp r7, #0 ; c = !type + moveq r7, #1 + movne r7, #0 + + ldrb r0, [r1, r12]! ; *(L += block2left[i]) + ldrb r3, [r2, r3]! ; *(A += block2above[i]) + mov lr, #c_entropy_nodes ; ENTROPY_NODES = 11 + +; VP8_COMBINEENTROPYCONTETEXTS(t, *a, *l) => t = ((*a) != 0) + ((*l) !=0) + cmp r0, #0 ; *l ?= 0 + movne r0, #1 + cmp r3, #0 ; *a ?= 0 + addne r0, r0, #1 ; t + + str r1, [sp, #l_l_ptr] ; save &l + str r2, [sp, #l_a_ptr] ; save &a + smlabb r0, r0, lr, r10 ; Prob = coef_probs + (t * ENTROPY_NODES) + mov r1, #0 ; t = 0 + str r7, [sp, #l_c] + + ;align 4 +COEFF_LOOP + ldr r3, [r9, #detok_ptr_coef_bands_x] + ldr lr, [r9, #detok_coef_tree_ptr] + ;STALL + ldrb r3, [r3, r7] ; coef_bands_x[c] + ;STALL + ;STALL + add r0, r0, r3 ; Prob += coef_bands_x[c] + +get_token_loop + ldrb r2, [r0, +r1, asr #1] ; Prob[t >> 1] + mov r3, r6, lsl #8 ; range << 8 + sub r3, r3, #256 ; (range << 8) - (1 << 8) + mov r10, #1 ; 1 + + smlawb r2, r3, r2, r10 ; split = 1 + (((range-1) * probability) >> 8) + + ldrb r12, [r8] ; load cx data byte in stall slot : r8 = bufptr + ;++ + + subs r3, r4, r2, lsl #24 ; value-(split<<24): used later to calculate shift for NORMALIZE + addhs r1, r1, #1 ; t += 1 + movhs r4, r3 ; value -= bigsplit (split << 24) + subhs r2, r6, r2 ; range -= split + ; movlo r6, r2 ; range = split + + ldrsb r1, [lr, r1] ; t = onyx_coef_tree_ptr[t] + +; NORMALIZE + clz r3, r2 ; vp8dx_bitreader_norm[range] + 24 + sub r3, r3, #24 ; vp8dx_bitreader_norm[range] + subs r5, r5, r3 ; count -= shift + mov r6, r2, lsl r3 ; range <<= shift + mov r4, r4, lsl r3 ; value <<= shift + +; if count <= 0, += BR_COUNT; value |= *bufptr++ << (BR_COUNT-count); BR_COUNT = 8, but need to upshift values by +16 + addle r5, r5, #8 ; count += 8 + rsble r3, r5, #24 ; 24 - count + addle r8, r8, #1 ; bufptr++ + orrle r4, r4, r12, lsl r3 ; value |= *bufptr << shift + 16 + + cmp r1, #0 ; t ?= 0 + bgt get_token_loop ; while (t > 0) + + cmn r1, #c_dct_eob_token ; if(t == -DCT_EOB_TOKEN) + beq END_OF_BLOCK ; break + + rsb lr, r1, #0 ; v = -t; + + cmp lr, #4 ; if(v > FOUR_TOKEN) + ble SKIP_EXTRABITS + + ldr r3, [r9, #detok_teb_base_ptr] + mov r11, #1 ; 1 in split = 1 + ... nope, v+= 1 << bits_count + add r7, r3, lr, lsl #4 ; detok_teb_base_ptr + (v << 4) + + ldrsh lr, [r7, #tokenextrabits_min_val] ; v = teb_ptr->min_val + ldrsh r0, [r7, #tokenextrabits_length] ; bits_count = teb_ptr->Length + +extrabits_loop + add r3, r0, r7 ; &teb_ptr->Probs[bits_count] + + ldrb r2, [r3, #4] ; probability. why +4? + mov r3, r6, lsl #8 ; range << 8 + sub r3, r3, #256 ; range << 8 + 1 << 8 + + smlawb r2, r3, r2, r11 ; split = 1 + (((range-1) * probability) >> 8) + + ldrb r12, [r8] ; *bufptr + ;++ + + subs r10, r4, r2, lsl #24 ; value - (split<<24) + movhs r4, r10 ; value = value - (split << 24) + subhs r2, r6, r2 ; range = range - split + addhs lr, lr, r11, lsl r0 ; v += ((UINT16)1<<bits_count) + +; NORMALIZE + clz r3, r2 ; shift - leading zeros in split + sub r3, r3, #24 ; don't count first 3 bytes + subs r5, r5, r3 ; count -= shift + mov r6, r2, lsl r3 ; range = range << shift + mov r4, r4, lsl r3 ; value <<= shift + + addle r5, r5, #8 ; count += BR_COUNT + addle r8, r8, #1 ; bufptr++ + rsble r3, r5, #24 ; BR_COUNT - count + orrle r4, r4, r12, lsl r3 ; value |= *bufptr << (BR_COUNT - count) + + subs r0, r0, #1 ; bits_count -- + bpl extrabits_loop + + +SKIP_EXTRABITS + ldr r11, [sp, #l_qcoeff] + ldr r0, [sp, #l_coef_ptr] ; Prob = coef_probs + + cmp r1, #0 ; check for nonzero token - if (t) + beq SKIP_EOB_CHECK ; if t is zero, we will skip the eob table chec + + add r3, r6, #1 ; range + 1 + mov r2, r3, lsr #1 ; split = (range + 1) >> 1 + + subs r3, r4, r2, lsl #24 ; value - (split<<24) + movhs r4, r3 ; value -= (split << 24) + subhs r2, r6, r2 ; range -= split + mvnhs r3, lr ; -v + addhs lr, r3, #1 ; v = (v ^ -1) + 1 + +; NORMALIZE + clz r3, r2 ; leading 0s in split + sub r3, r3, #24 ; shift + subs r5, r5, r3 ; count -= shift + mov r6, r2, lsl r3 ; range <<= shift + mov r4, r4, lsl r3 ; value <<= shift + ldrleb r2, [r8], #1 ; *(bufptr++) + addle r5, r5, #8 ; count += 8 + rsble r3, r5, #24 ; BR_COUNT - count + orrle r4, r4, r2, lsl r3 ; value |= *bufptr << (BR_COUNT - count) + + add r0, r0, #11 ; Prob += ENTROPY_NODES (11) + + cmn r1, #1 ; t < -ONE_TOKEN + + addlt r0, r0, #11 ; Prob += ENTROPY_NODES (11) + + mvn r1, #1 ; t = -1 ???? C is -2 + +SKIP_EOB_CHECK + ldr r7, [sp, #l_c] ; c + ldr r3, [r9, #detok_scan] + add r1, r1, #2 ; t+= 2 + cmp r7, #15 ; c should will be one higher + + ldr r3, [r3, +r7, lsl #2] ; scan[c] this needs pre-inc c value + add r7, r7, #1 ; c++ + add r3, r11, r3, lsl #1 ; qcoeff + scan[c] + + str r7, [sp, #l_c] ; store c + strh lr, [r3] ; qcoef_ptr[scan[c]] = v + + blt COEFF_LOOP + + sub r7, r7, #1 ; if(t != -DCT_EOB_TOKEN) --c + +END_OF_BLOCK + ldr r3, [sp, #l_type] ; type + ldr r10, [sp, #l_coef_ptr] ; coef_ptr + ldr r0, [sp, #l_qcoeff] ; qcoeff + ldr r11, [sp, #l_i] ; i + ldr r12, [sp, #l_stop] ; stop + + cmp r3, #0 ; type ?= 0 + moveq r1, #1 + movne r1, #0 + add r3, r11, r9 ; detok + i + + cmp r7, r1 ; c ?= !type + strb r7, [r3, #detok_eob] ; eob[i] = c + + ldr r7, [sp, #l_l_ptr] ; l + ldr r2, [sp, #l_a_ptr] ; a + movne r3, #1 ; t + moveq r3, #0 + + add r0, r0, #32 ; qcoeff += 32 (16 * 2?) + add r11, r11, #1 ; i++ + strb r3, [r7] ; *l = t + strb r3, [r2] ; *a = t + str r0, [sp, #l_qcoeff] ; qcoeff + str r11, [sp, #l_i] ; i + + cmp r11, r12 ; i < stop + ldr r7, [sp, #l_type] ; type + + blt BLOCK_LOOP + + cmp r11, #25 ; i ?= 25 + bne ln2_decode_mb_to + + ldr r12, [r9, #detok_qcoeff_start_ptr] + ldr r10, [r9, #detok_coef_probs] + mov r7, #0 ; type/i = 0 + mov r3, #16 ; stop = 16 + str r12, [sp, #l_qcoeff] ; qcoeff_ptr = qcoeff_start_ptr + str r7, [sp, #l_i] + str r7, [sp, #l_type] + str r3, [sp, #l_stop] + + str r10, [sp, #l_coef_ptr] ; coef_probs = coef_probs[type=0] + + b BLOCK_LOOP + +ln2_decode_mb_to + cmp r11, #16 ; i ?= 16 + bne ln1_decode_mb_to + + mov r10, #detok_coef_probs + add r10, r10, #2*4 ; coef_probs[type] + ldr r10, [r9, r10] ; detok + detok_coef_probs[type] + + mov r7, #2 ; type = 2 + mov r3, #24 ; stop = 24 + + str r7, [sp, #l_type] + str r3, [sp, #l_stop] + + str r10, [sp, #l_coef_ptr] ; coef_probs = coef_probs[type] + b BLOCK_LOOP + +ln1_decode_mb_to + ldr r2, [sp, #l_bc] + mov r0, #0 + nop + + str r8, [r2, #bool_decoder_user_buffer] + str r5, [r2, #bool_decoder_count] + str r4, [r2, #bool_decoder_value] + str r6, [r2, #bool_decoder_range] + + add sp, sp, #l_stacksize + ldmia sp!, {r4 - r11, pc} + + ENDP ; |vp8_decode_mb_tokens_v6| + + END diff --git a/vp8/decoder/arm/detokenize_arm.h b/vp8/decoder/arm/detokenize_arm.h new file mode 100644 index 0000000..9bb19b6 --- /dev/null +++ b/vp8/decoder/arm/detokenize_arm.h @@ -0,0 +1,22 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + + +#ifndef DETOKENIZE_ARM_H +#define DETOKENIZE_ARM_H + +#if HAVE_ARMV6 +#if CONFIG_ARM_ASM_DETOK +void vp8_init_detokenizer(VP8D_COMP *dx); +void vp8_decode_mb_tokens_v6(DETOK *detoken, int type); +#endif +#endif + +#endif diff --git a/vp8/decoder/arm/detokenizearm_sjl.c b/vp8/decoder/arm/detokenizearm_sjl.c deleted file mode 100644 index c714452..0000000 --- a/vp8/decoder/arm/detokenizearm_sjl.c +++ /dev/null @@ -1,730 +0,0 @@ -/* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. - * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. - */ - - -#include "type_aliases.h" -#include "blockd.h" -#include "onyxd_int.h" -#include "vpx_mem/vpx_mem.h" -#include "vpx_ports/mem.h" - -#define BR_COUNT 8 -#define BOOL_DATA UINT8 - -#define OCB_X PREV_COEF_CONTEXTS * ENTROPY_NODES -//ALIGN16 UINT16 onyx_coef_bands_x[16] = { 0, 1*OCB_X, 2*OCB_X, 3*OCB_X, 6*OCB_X, 4*OCB_X, 5*OCB_X, 6*OCB_X, 6*OCB_X, 6*OCB_X, 6*OCB_X, 6*OCB_X, 6*OCB_X, 6*OCB_X, 6*OCB_X, 7*OCB_X}; -DECLARE_ALIGNED(16, UINT8, vp8_coef_bands_x[16]) = { 0, 1 * OCB_X, 2 * OCB_X, 3 * OCB_X, 6 * OCB_X, 4 * OCB_X, 5 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 7 * OCB_X}; - -#define EOB_CONTEXT_NODE 0 -#define ZERO_CONTEXT_NODE 1 -#define ONE_CONTEXT_NODE 2 -#define LOW_VAL_CONTEXT_NODE 3 -#define TWO_CONTEXT_NODE 4 -#define THREE_CONTEXT_NODE 5 -#define HIGH_LOW_CONTEXT_NODE 6 -#define CAT_ONE_CONTEXT_NODE 7 -#define CAT_THREEFOUR_CONTEXT_NODE 8 -#define CAT_THREE_CONTEXT_NODE 9 -#define CAT_FIVE_CONTEXT_NODE 10 - - - - -DECLARE_ALIGNED(16, static const TOKENEXTRABITS, vp8d_token_extra_bits2[MAX_ENTROPY_TOKENS]) = -{ - { 0, -1, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, //ZERO_TOKEN - { 1, 0, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, //ONE_TOKEN - { 2, 0, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, //TWO_TOKEN - { 3, 0, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, //THREE_TOKEN - { 4, 0, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, //FOUR_TOKEN - { 5, 0, { 159, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, //DCT_VAL_CATEGORY1 - { 7, 1, { 145, 165, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, //DCT_VAL_CATEGORY2 - { 11, 2, { 140, 148, 173, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, //DCT_VAL_CATEGORY3 - { 19, 3, { 135, 140, 155, 176, 0, 0, 0, 0, 0, 0, 0, 0 } }, //DCT_VAL_CATEGORY4 - { 35, 4, { 130, 134, 141, 157, 180, 0, 0, 0, 0, 0, 0, 0 } }, //DCT_VAL_CATEGORY5 - { 67, 10, { 129, 130, 133, 140, 153, 177, 196, 230, 243, 254, 254, 0 } }, //DCT_VAL_CATEGORY6 - { 0, -1, { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 } }, // EOB TOKEN -}; - -/* -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -*/ -DECLARE_ALIGNED(16, const UINT8, vp8_block2context_leftabove[25*3]) = -{ - 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, //end of vp8_block2context - 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 0, 0, 1, 1, 0, 0, 1, 1, 0, //end of vp8_block2left - 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 0, 1, 0, 1, 0, 1, 0 //end of vp8_block2above -}; - -/* -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -*/ - -void vp8_reset_mb_tokens_context(MACROBLOCKD *x) -{ - ENTROPY_CONTEXT **const A = x->above_context; - ENTROPY_CONTEXT(* const L)[4] = x->left_context; - - ENTROPY_CONTEXT *a; - ENTROPY_CONTEXT *l; - int i; - - for (i = 0; i < 24; i++) - { - - a = A[ vp8_block2context[i] ] + vp8_block2above[i]; - l = L[ vp8_block2context[i] ] + vp8_block2left[i]; - - *a = *l = 0; - } - - if (x->mbmi.mode != B_PRED && x->mbmi.mode != SPLITMV) - { - a = A[Y2CONTEXT] + vp8_block2above[24]; - l = L[Y2CONTEXT] + vp8_block2left[24]; - *a = *l = 0; - } - - -} - -#define ONYXBLOCK2CONTEXT_OFFSET 0 -#define ONYXBLOCK2LEFT_OFFSET 25 -#define ONYXBLOCK2ABOVE_OFFSET 50 - -DECLARE_ALIGNED(16, const static unsigned char, norm[128]) = -{ - 0, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, - 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, - 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, - 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 -}; - -/* -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -*/ -void init_detokenizer(VP8D_COMP *dx) -{ - const VP8_COMMON *const oc = & dx->common; - MACROBLOCKD *x = & dx->mb; - - dx->detoken.norm_ptr = (unsigned char *)norm; - dx->detoken.vp8_coef_tree_ptr = (vp8_tree_index *)vp8_coef_tree; - dx->detoken.ptr_onyxblock2context_leftabove = (UINT8 *)vp8_block2context_leftabove; - dx->detoken.ptr_onyx_coef_bands_x = vp8_coef_bands_x; - dx->detoken.scan = (int *)vp8_default_zig_zag1d; - dx->detoken.teb_base_ptr = (TOKENEXTRABITS *)vp8d_token_extra_bits2; - - dx->detoken.qcoeff_start_ptr = &x->qcoeff[0]; - - - dx->detoken.coef_probs[0] = (unsigned char *)(oc->fc.coef_probs [0] [ 0 ] [0]); - dx->detoken.coef_probs[1] = (unsigned char *)(oc->fc.coef_probs [1] [ 0 ] [0]); - dx->detoken.coef_probs[2] = (unsigned char *)(oc->fc.coef_probs [2] [ 0 ] [0]); - dx->detoken.coef_probs[3] = (unsigned char *)(oc->fc.coef_probs [3] [ 0 ] [0]); - -} - -/* -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -*/ - - -//shift = norm[range]; \ -// shift = norm_ptr[range]; \ - -#define NORMALIZE \ - /*if(range < 0x80)*/ \ - { \ - shift = detoken->norm_ptr[range]; \ - range <<= shift; \ - value <<= shift; \ - count -= shift; \ - if(count <= 0) \ - { \ - count += BR_COUNT ; \ - value |= (*bufptr) << (BR_COUNT-count); \ - bufptr++; \ - } \ - } -#if 1 -#define DECODE_AND_APPLYSIGN(value_to_sign) \ - split = (range + 1) >> 1; \ - if ( (value >> 24) < split ) \ - { \ - range = split; \ - v= value_to_sign; \ - } \ - else \ - { \ - range = range-split; \ - value = value-(split<<24); \ - v = -value_to_sign; \ - } \ - range +=range; \ - value +=value; \ - if (!--count) \ - { \ - count = BR_COUNT; \ - value |= *bufptr; \ - bufptr++; \ - } - -#define DECODE_AND_BRANCH_IF_ZERO(probability,branch) \ - { \ - split = 1 + ((( probability*(range-1) ) )>> 8); \ - if ( (value >> 24) < split ) \ - { \ - range = split; \ - NORMALIZE \ - goto branch; \ - } \ - value -= (split<<24); \ - range = range - split; \ - NORMALIZE \ - } - -#define DECODE_AND_LOOP_IF_ZERO(probability,branch) \ - { \ - split = 1 + ((( probability*(range-1) ) ) >> 8); \ - if ( (value >> 24) < split ) \ - { \ - range = split; \ - NORMALIZE \ - Prob = coef_probs; \ - ++c; \ - Prob += vp8_coef_bands_x[c]; \ - goto branch; \ - } \ - value -= (split<<24); \ - range = range - split; \ - NORMALIZE \ - } - -#define DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val) \ - DECODE_AND_APPLYSIGN(val) \ - Prob = coef_probs + (ENTROPY_NODES*2); \ - if(c < 15){\ - qcoeff_ptr [ scan[c] ] = (INT16) v; \ - ++c; \ - goto DO_WHILE; }\ - qcoeff_ptr [ scan[15] ] = (INT16) v; \ - goto BLOCK_FINISHED; - - -#define DECODE_EXTRABIT_AND_ADJUST_VAL(t,bits_count)\ - split = 1 + (((range-1) * vp8d_token_extra_bits2[t].Probs[bits_count]) >> 8); \ - if(value >= (split<<24))\ - {\ - range = range-split;\ - value = value-(split<<24);\ - val += ((UINT16)1<<bits_count);\ - }\ - else\ - {\ - range = split;\ - }\ - NORMALIZE -#endif - -#if 0 -int vp8_decode_mb_tokens(VP8D_COMP *dx, MACROBLOCKD *x) -{ - ENTROPY_CONTEXT **const A = x->above_context; - ENTROPY_CONTEXT(* const L)[4] = x->left_context; - const VP8_COMMON *const oc = & dx->common; - - BOOL_DECODER *bc = x->current_bc; - - ENTROPY_CONTEXT *a; - ENTROPY_CONTEXT *l; - int i; - - int eobtotal = 0; - - register int count; - - BOOL_DATA *bufptr; - register unsigned int range; - register unsigned int value; - const int *scan; - register unsigned int shift; - UINT32 split; - INT16 *qcoeff_ptr; - - UINT8 *coef_probs; - int type; - int stop; - INT16 val, bits_count; - INT16 c; - INT16 t; - INT16 v; - vp8_prob *Prob; - - //int *scan; - type = 3; - i = 0; - stop = 16; - - if (x->mbmi.mode != B_PRED && x->mbmi.mode != SPLITMV) - { - i = 24; - stop = 24; - type = 1; - qcoeff_ptr = &x->qcoeff[24*16]; - scan = vp8_default_zig_zag1d; - eobtotal -= 16; - } - else - { - scan = vp8_default_zig_zag1d; - qcoeff_ptr = &x->qcoeff[0]; - } - - count = bc->count; - range = bc->range; - value = bc->value; - bufptr = &bc->buffer[bc->pos]; - - - coef_probs = (unsigned char *)(oc->fc.coef_probs [type] [ 0 ] [0]); - -BLOCK_LOOP: - a = A[ vp8_block2context[i] ] + vp8_block2above[i]; - l = L[ vp8_block2context[i] ] + vp8_block2left[i]; - c = (INT16)(!type); - - VP8_COMBINEENTROPYCONTEXTS(t, *a, *l); - Prob = coef_probs; - Prob += t * ENTROPY_NODES; - -DO_WHILE: - Prob += vp8_coef_bands_x[c]; - DECODE_AND_BRANCH_IF_ZERO(Prob[EOB_CONTEXT_NODE], BLOCK_FINISHED); - -CHECK_0_: - DECODE_AND_LOOP_IF_ZERO(Prob[ZERO_CONTEXT_NODE], CHECK_0_); - DECODE_AND_BRANCH_IF_ZERO(Prob[ONE_CONTEXT_NODE], ONE_CONTEXT_NODE_0_); - DECODE_AND_BRANCH_IF_ZERO(Prob[LOW_VAL_CONTEXT_NODE], LOW_VAL_CONTEXT_NODE_0_); - DECODE_AND_BRANCH_IF_ZERO(Prob[HIGH_LOW_CONTEXT_NODE], HIGH_LOW_CONTEXT_NODE_0_); - DECODE_AND_BRANCH_IF_ZERO(Prob[CAT_THREEFOUR_CONTEXT_NODE], CAT_THREEFOUR_CONTEXT_NODE_0_); - DECODE_AND_BRANCH_IF_ZERO(Prob[CAT_FIVE_CONTEXT_NODE], CAT_FIVE_CONTEXT_NODE_0_); - val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY6].min_val; - bits_count = vp8d_token_extra_bits2[DCT_VAL_CATEGORY6].Length; - - do - { - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY6, bits_count); - bits_count -- ; - } - while (bits_count >= 0); - - DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val); - -CAT_FIVE_CONTEXT_NODE_0_: - val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY5].min_val; - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY5, 4); - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY5, 3); - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY5, 2); - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY5, 1); - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY5, 0); - DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val); - -CAT_THREEFOUR_CONTEXT_NODE_0_: - DECODE_AND_BRANCH_IF_ZERO(Prob[CAT_THREE_CONTEXT_NODE], CAT_THREE_CONTEXT_NODE_0_); - val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY4].min_val; - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY4, 3); - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY4, 2); - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY4, 1); - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY4, 0); - DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val); - -CAT_THREE_CONTEXT_NODE_0_: - val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY3].min_val; - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY3, 2); - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY3, 1); - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY3, 0); - DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val); - -HIGH_LOW_CONTEXT_NODE_0_: - DECODE_AND_BRANCH_IF_ZERO(Prob[CAT_ONE_CONTEXT_NODE], CAT_ONE_CONTEXT_NODE_0_); - - val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY2].min_val; - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY2, 1); - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY2, 0); - DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val); - -CAT_ONE_CONTEXT_NODE_0_: - val = vp8d_token_extra_bits2[DCT_VAL_CATEGORY1].min_val; - DECODE_EXTRABIT_AND_ADJUST_VAL(DCT_VAL_CATEGORY1, 0); - DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(val); - -LOW_VAL_CONTEXT_NODE_0_: - DECODE_AND_BRANCH_IF_ZERO(Prob[TWO_CONTEXT_NODE], TWO_CONTEXT_NODE_0_); - DECODE_AND_BRANCH_IF_ZERO(Prob[THREE_CONTEXT_NODE], THREE_CONTEXT_NODE_0_); - DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(4); - -THREE_CONTEXT_NODE_0_: - DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(3); - -TWO_CONTEXT_NODE_0_: - DECODE_SIGN_WRITE_COEFF_AND_CHECK_EXIT(2); - -ONE_CONTEXT_NODE_0_: - DECODE_AND_APPLYSIGN(1); - Prob = coef_probs + ENTROPY_NODES; - - if (c < 15) - { - qcoeff_ptr [ scan[c] ] = (INT16) v; - ++c; - goto DO_WHILE; - } - - qcoeff_ptr [ scan[15] ] = (INT16) v; -BLOCK_FINISHED: - t = ((x->Block[i].eob = c) != !type); // any nonzero data? - eobtotal += x->Block[i].eob; - *a = *l = t; - qcoeff_ptr += 16; - - i++; - - if (i < stop) - goto BLOCK_LOOP; - - if (i == 25) - { - scan = vp8_default_zig_zag1d;//x->scan_order1d; - type = 0; - i = 0; - stop = 16; - coef_probs = (unsigned char *)(oc->fc.coef_probs [type] [ 0 ] [0]); - qcoeff_ptr = &x->qcoeff[0]; - goto BLOCK_LOOP; - } - - if (i == 16) - { - type = 2; - coef_probs = (unsigned char *)(oc->fc.coef_probs [type] [ 0 ] [0]); - stop = 24; - goto BLOCK_LOOP; - } - - bc->count = count; - bc->value = value; - bc->range = range; - bc->pos = bufptr - bc->buffer; - return eobtotal; - -} -//#endif -#else -/* -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -*/ - -#if 0 -//uses relative offsets - -const vp8_tree_index vp8_coef_tree_x[ 22] = /* corresponding _CONTEXT_NODEs */ -{ - -DCT_EOB_TOKEN, 1, /* 0 = EOB */ - -ZERO_TOKEN, 1, /* 1 = ZERO */ - -ONE_TOKEN, 1, /* 2 = ONE */ - 2, 5, /* 3 = LOW_VAL */ - -TWO_TOKEN, 1, /* 4 = TWO */ - -THREE_TOKEN, -FOUR_TOKEN, /* 5 = THREE */ - 2, 3, /* 6 = HIGH_LOW */ - -DCT_VAL_CATEGORY1, -DCT_VAL_CATEGORY2, /* 7 = CAT_ONE */ - 2, 3, /* 8 = CAT_THREEFOUR */ - -DCT_VAL_CATEGORY3, -DCT_VAL_CATEGORY4, /* 9 = CAT_THREE */ - -DCT_VAL_CATEGORY5, -DCT_VAL_CATEGORY6 /* 10 = CAT_FIVE */ -}; -#endif - -#define _SCALEDOWN 8 //16 //8 - -int vp8_decode_mb_tokens_v5(DETOK *detoken, int type); - -int vp8_decode_mb_tokens_v5_c(DETOK *detoken, int type) -{ - BOOL_DECODER *bc = detoken->current_bc; - - ENTROPY_CONTEXT *a; - ENTROPY_CONTEXT *l; - int i; - - register int count; - - BOOL_DATA *bufptr; - register unsigned int range; - register unsigned int value; - register unsigned int shift; - UINT32 split; - INT16 *qcoeff_ptr; - - UINT8 *coef_probs; -// int type; - int stop; - INT16 c; - INT16 t; - INT16 v; - vp8_prob *Prob; - - - -// type = 3; - i = 0; - stop = 16; - qcoeff_ptr = detoken->qcoeff_start_ptr; - -// if( detoken->mode != B_PRED && detoken->mode != SPLITMV) - if (type == 1) - { - i += 24; - stop += 8; //24; -// type = 1; - qcoeff_ptr += 24 * 16; -// eobtotal-=16; - } - - count = bc->count; - range = bc->range; - value = bc->value; - bufptr = &bc->buffer[bc->pos]; - - - coef_probs = detoken->coef_probs[type]; //(unsigned char *)( oc->fc.coef_probs [type] [ 0 ] [0]); - -BLOCK_LOOP: - a = detoken->A[ detoken->ptr_onyxblock2context_leftabove[i] ]; - l = detoken->L[ detoken->ptr_onyxblock2context_leftabove[i] ]; - c = !type; - a += detoken->ptr_onyxblock2context_leftabove[i + ONYXBLOCK2ABOVE_OFFSET]; - l += detoken->ptr_onyxblock2context_leftabove[i + ONYXBLOCK2LEFT_OFFSET]; - - //#define ONYX_COMBINEENTROPYCONTEXTS( Dest, A, B) \ - //Dest = ((A)!=0) + ((B)!=0); - - VP8_COMBINEENTROPYCONTEXTS(t, *a, *l); - - Prob = coef_probs; - Prob += t * ENTROPY_NODES; - t = 0; - - do - { - - { -// onyx_tree_index * onyx_coef_tree_ptr = onyx_coef_tree_x; - - Prob += detoken->ptr_onyx_coef_bands_x[c]; - - GET_TOKEN_START: - - do - { - split = 1 + (((range - 1) * (Prob[t>>1])) >> 8); - - if (value >> 24 >= split) - { - range = range - split; - value = value - (split << 24); - t += 1; - - //used to eliminate else branch - split = range; - } - - range = split; - - t = detoken->vp8_coef_tree_ptr[ t ]; - - NORMALIZE - - } - while (t > 0) ; - } - GET_TOKEN_STOP: - - if (t == -DCT_EOB_TOKEN) - { - break; - } - - v = -t; - - if (v > FOUR_TOKEN) - { - INT16 bits_count; - TOKENEXTRABITS *teb_ptr; - -// teb_ptr = &onyxd_token_extra_bits2[t]; -// teb_ptr = &onyxd_token_extra_bits2[v]; - teb_ptr = &detoken->teb_base_ptr[v]; - - - v = teb_ptr->min_val; - bits_count = teb_ptr->Length; - - do - { - split = 1 + (((range - 1) * teb_ptr->Probs[bits_count]) >> _SCALEDOWN); - - if ((value >> 24) >= split) - { - range = range - split; - value = value - (split << 24); - v += ((UINT16)1 << bits_count); - - //used to eliminate else branch - split = range; - } - - range = split; - - NORMALIZE - - bits_count -- ; - } - while (bits_count >= 0); - } - - Prob = coef_probs; - - if (t) - { - split = 1 + (((range - 1) * vp8_prob_half) >> 8); - - if ((value >> 24) >= split) - { - range = range - split; - value = value - (split << 24); - v = (v ^ -1) + 1; /* negate w/out conditionals */ - - //used to eliminate else branch - split = range; - } - - range = split; - - NORMALIZE - Prob += ENTROPY_NODES; - - if (t < -ONE_TOKEN) - Prob += ENTROPY_NODES; - - t = -2; - } - - //if t is zero, we will skip the eob table check - t += 2; - qcoeff_ptr [detoken->scan [c] ] = (INT16) v; - - } - while (++c < 16); - - if (t != -DCT_EOB_TOKEN) - { - --c; - } - - t = ((detoken->eob[i] = c) != !type); // any nonzero data? -// eobtotal += detoken->eob[i]; - *a = *l = t; - qcoeff_ptr += 16; - - i++; - - if (i < stop) - goto BLOCK_LOOP; - - if (i == 25) - { - type = 0; - i = 0; - stop = 16; -// coef_probs = (unsigned char *)(oc->fc.coef_probs [type] [ 0 ] [0]); - coef_probs = detoken->coef_probs[type]; //(unsigned char *)( oc->fc.coef_probs [type] [ 0 ] [0]); - qcoeff_ptr = detoken->qcoeff_start_ptr; - goto BLOCK_LOOP; - } - - if (i == 16) - { - type = 2; -// coef_probs =(unsigned char *)( oc->fc.coef_probs [type] [ 0 ] [0]); - coef_probs = detoken->coef_probs[type]; //(unsigned char *)( oc->fc.coef_probs [type] [ 0 ] [0]); - stop = 24; - goto BLOCK_LOOP; - } - - bc->count = count; - bc->value = value; - bc->range = range; - bc->pos = bufptr - bc->buffer; - return 0; -} -//#if 0 -int vp8_decode_mb_tokens(VP8D_COMP *dx, MACROBLOCKD *x) -{ -// const ONYX_COMMON * const oc = & dx->common; - int eobtotal = 0; - int i, type; - /* - dx->detoken.norm_ptr = norm; - dx->detoken.onyx_coef_tree_ptr = onyx_coef_tree; - dx->detoken.ptr_onyxblock2context_leftabove = ONYXBLOCK2CONTEXT_LEFTABOVE; - dx->detoken.ptr_onyx_coef_bands_x = onyx_coef_bands_x; - dx->detoken.scan = default_zig_zag1d; - dx->detoken.teb_base_ptr = onyxd_token_extra_bits2; - - dx->detoken.qcoeff_start_ptr = &x->qcoeff[0]; - - dx->detoken.A = x->above_context; - dx->detoken.L = x->left_context; - - dx->detoken.coef_probs[0] = (unsigned char *)( oc->fc.coef_probs [0] [ 0 ] [0]); - dx->detoken.coef_probs[1] = (unsigned char *)( oc->fc.coef_probs [1] [ 0 ] [0]); - dx->detoken.coef_probs[2] = (unsigned char *)( oc->fc.coef_probs [2] [ 0 ] [0]); - dx->detoken.coef_probs[3] = (unsigned char *)( oc->fc.coef_probs [3] [ 0 ] [0]); - */ - - dx->detoken.current_bc = x->current_bc; - dx->detoken.A = x->above_context; - dx->detoken.L = x->left_context; - - type = 3; - - if (x->mbmi.mode != B_PRED && x->mbmi.mode != SPLITMV) - { - type = 1; - eobtotal -= 16; - } - - vp8_decode_mb_tokens_v5(&dx->detoken, type); - - for (i = 0; i < 25; i++) - { - x->Block[i].eob = dx->detoken.eob[i]; - eobtotal += dx->detoken.eob[i]; - } - - return eobtotal; -} -#endif diff --git a/vp8/decoder/arm/detokenizearm_v6.asm b/vp8/decoder/arm/detokenizearm_v6.asm deleted file mode 100644 index 4d87ee5..0000000 --- a/vp8/decoder/arm/detokenizearm_v6.asm +++ /dev/null @@ -1,364 +0,0 @@ -; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. -; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. -; - - - EXPORT |vp8_decode_mb_tokens_v5| - - AREA |.text|, CODE, READONLY ; name this block of code - - INCLUDE vpx_asm_offsets.asm - -l_qcoeff EQU 0 -l_i EQU 4 -l_type EQU 8 -l_stop EQU 12 -l_c EQU 16 -l_l_ptr EQU 20 -l_a_ptr EQU 24 -l_bc EQU 28 -l_coef_ptr EQU 32 -l_stacksize EQU 64 - - -;; constant offsets -- these should be created at build time -c_onyxblock2left_offset EQU 25 -c_onyxblock2above_offset EQU 50 -c_entropy_nodes EQU 11 -c_dct_eob_token EQU 11 - -|vp8_decode_mb_tokens_v5| PROC - stmdb sp!, {r4 - r11, lr} - sub sp, sp, #l_stacksize - mov r7, r1 - mov r9, r0 ;DETOK *detoken - - ldr r1, [r9, #detok_current_bc] - ldr r0, [r9, #detok_qcoeff_start_ptr] - mov r11, #0 - mov r3, #0x10 - - cmp r7, #1 - addeq r11, r11, #24 - addeq r3, r3, #8 - addeq r0, r0, #3, 24 - - str r0, [sp, #l_qcoeff] - str r11, [sp, #l_i] - str r7, [sp, #l_type] - str r3, [sp, #l_stop] - str r1, [sp, #l_bc] - - add lr, r9, r7, lsl #2 - - ldr r2, [r1, #bool_decoder_buffer] - ldr r3, [r1, #bool_decoder_pos] - - ldr r10, [lr, #detok_coef_probs] - ldr r5, [r1, #bool_decoder_count] - ldr r6, [r1, #bool_decoder_range] - ldr r4, [r1, #bool_decoder_value] - add r8, r2, r3 - - str r10, [sp, #l_coef_ptr] - - - ;align 4 -BLOCK_LOOP - ldr r3, [r9, #detok_ptr_onyxblock2context_leftabove] - ldr r2, [r9, #DETOK_A] - ldr r1, [r9, #DETOK_L] - ldrb r12, [r3, +r11] ; detoken->ptr_onyxblock2context_leftabove[i] - - cmp r7, #0 ; check type - moveq r7, #1 - movne r7, #0 - - ldr r0, [r2, +r12, lsl #2] ; a - add r1, r1, r12, lsl #4 - add r3, r3, r11 - - ldrb r2, [r3, #c_onyxblock2above_offset] - ldrb r3, [r3, #c_onyxblock2left_offset] - mov lr, #c_entropy_nodes -;; ;++ - - ldr r2, [r0, +r2, lsl #2]! - add r3, r1, r3, lsl #2 - str r3, [sp, #l_l_ptr] - ldr r3, [r3] - - cmp r2, #0 - movne r2, #1 - cmp r3, #0 - addne r2, r2, #1 - - str r0, [sp, #l_a_ptr] - smlabb r0, r2, lr, r10 - mov r1, #0 ; t = 0 - str r7, [sp, #l_c] - - ;align 4 -COEFF_LOOP - ldr r3, [r9, #detok_ptr_onyx_coef_bands_x] - ldr lr, [r9, #detok_onyx_coef_tree_ptr] - -;;the following two lines are used if onyx_coef_bands_x is UINT16 -;; add r3, r3, r7, lsl #1 -;; ldrh r3, [r3] - -;;the following line is used if onyx_coef_bands_x is UINT8 - ldrb r3, [r7, +r3] - - -;; ;++ -;; pld [r8] - ;++ - add r0, r0, r3 - - ;align 4 -get_token_loop - ldrb r2, [r0, +r1, asr #1] - mov r3, r6, lsl #8 - sub r3, r3, #256 ;split = 1 + (((range-1) * probability) >> 8) - mov r10, #1 - - smlawb r2, r3, r2, r10 - ldrb r12, [r8] ;load cx data byte in stall slot - ;++ - - subs r3, r4, r2, lsl #24 ;x = value-(split<<24) - addhs r1, r1, #1 ;t += 1 - movhs r4, r3 ;update value - subhs r2, r6, r2 ;range = range - split - movlo r6, r2 - -;;; ldrsbhs r1, [r1, +lr] - ldrsb r1, [r1, +lr] - - -;; use branch for short pipelines ??? -;; cmp r2, #0x80 -;; bcs |$LN22@decode_mb_to| - - clz r3, r2 - sub r3, r3, #24 - subs r5, r5, r3 - mov r6, r2, lsl r3 - mov r4, r4, lsl r3 - -;; use branch for short pipelines ??? -;; bgt |$LN22@decode_mb_to| - - addle r5, r5, #8 - rsble r3, r5, #8 - addle r8, r8, #1 - orrle r4, r4, r12, lsl r3 - -;;|$LN22@decode_mb_to| - - cmp r1, #0 - bgt get_token_loop - - cmn r1, #c_dct_eob_token ;if(t == -DCT_EOB_TOKEN) - beq END_OF_BLOCK - - rsb lr, r1, #0 ;v = -t; - - cmp lr, #4 ;if(v > FOUR_TOKEN) - ble SKIP_EXTRABITS - - ldr r3, [r9, #detok_teb_base_ptr] - mov r11, #1 - add r7, r3, lr, lsl #4 - - ldrsh lr, [r7, #tokenextrabits_min_val];v = teb_ptr->min_val - ldrsh r0, [r7, #tokenextrabits_length];bits_count = teb_ptr->Length - -extrabits_loop - add r3, r0, r7 - - ldrb r2, [r3, #4] - mov r3, r6, lsl #8 - sub r3, r3, #256 ;split = 1 + (((range-1) * probability) >> 8) - mov r10, #1 - - smlawb r2, r3, r2, r10 - ldrb r12, [r8] - ;++ - - subs r10, r4, r2, lsl #24 ;x = value-(split<<24) - movhs r4, r10 ;update value - subhs r2, r6, r2 ;range = range - split - addhs lr, lr, r11, lsl r0 ;v += ((UINT16)1<<bits_count) - movlo r6, r2 ;range = split - - -;; use branch for short pipelines ??? -;; cmp r2, #0x80 -;; bcs |$LN10@decode_mb_to| - - clz r3, r2 - sub r3, r3, #24 - subs r5, r5, r3 - mov r6, r2, lsl r3 ;range - mov r4, r4, lsl r3 ;value - - addle r5, r5, #8 - addle r8, r8, #1 - rsble r3, r5, #8 - orrle r4, r4, r12, lsl r3 - -;;|$LN10@decode_mb_to| - subs r0, r0, #1 - bpl extrabits_loop - - -SKIP_EXTRABITS - ldr r11, [sp, #l_qcoeff] - ldr r0, [sp, #l_coef_ptr] - - cmp r1, #0 ;check for nonzero token - beq SKIP_EOB_CHECK ;if t is zero, we will skip the eob table chec - - sub r3, r6, #1 ;range - 1 - ;++ - mov r3, r3, lsl #7 ; *= onyx_prob_half (128) - ;++ - mov r3, r3, lsr #8 - add r2, r3, #1 ;split - - subs r3, r4, r2, lsl #24 ;x = value-(split<<24) - movhs r4, r3 ;update value - subhs r2, r6, r2 ;range = range - split - mvnhs r3, lr - addhs lr, r3, #1 ;v = (v ^ -1) + 1 - movlo r6, r2 ;range = split - -;; use branch for short pipelines ??? -;; cmp r2, #0x80 -;; bcs |$LN6@decode_mb_to| - - clz r3, r2 - sub r3, r3, #24 - subs r5, r5, r3 - mov r6, r2, lsl r3 - mov r4, r4, lsl r3 - ldrleb r2, [r8], #1 - addle r5, r5, #8 - rsble r3, r5, #8 - orrle r4, r4, r2, lsl r3 - -;;|$LN6@decode_mb_to| - add r0, r0, #0xB - - cmn r1, #1 - - addlt r0, r0, #0xB - - mvn r1, #1 - -SKIP_EOB_CHECK - ldr r7, [sp, #l_c] - ldr r3, [r9, #detok_scan] - add r1, r1, #2 - cmp r7, #(0x10 - 1) ;assume one less for now.... increment below - - ldr r3, [r3, +r7, lsl #2] - add r7, r7, #1 - add r3, r11, r3, lsl #1 - - str r7, [sp, #l_c] - strh lr, [r3] - - blt COEFF_LOOP - - sub r7, r7, #1 ;if(t != -DCT_EOB_TOKEN) --c - -END_OF_BLOCK - ldr r3, [sp, #l_type] - ldr r10, [sp, #l_coef_ptr] - ldr r0, [sp, #l_qcoeff] - ldr r11, [sp, #l_i] - ldr r12, [sp, #l_stop] - - cmp r3, #0 - moveq r1, #1 - movne r1, #0 - add r3, r11, r9 - - cmp r7, r1 - strb r7, [r3, #detok_eob] - - ldr r7, [sp, #l_l_ptr] - ldr r2, [sp, #l_a_ptr] - movne r3, #1 - moveq r3, #0 - - add r0, r0, #0x20 - add r11, r11, #1 - str r3, [r7] - str r3, [r2] - str r0, [sp, #l_qcoeff] - str r11, [sp, #l_i] - - cmp r11, r12 ;i >= stop ? - ldr r7, [sp, #l_type] - mov lr, #0xB - - blt BLOCK_LOOP - - cmp r11, #0x19 - bne ln2_decode_mb_to - - ldr r12, [r9, #detok_qcoeff_start_ptr] - ldr r10, [r9, #detok_coef_probs] - mov r7, #0 - mov r3, #0x10 - str r12, [sp, #l_qcoeff] - str r7, [sp, #l_i] - str r7, [sp, #l_type] - str r3, [sp, #l_stop] - - str r10, [sp, #l_coef_ptr] - - b BLOCK_LOOP - -ln2_decode_mb_to - cmp r11, #0x10 - bne ln1_decode_mb_to - - ldr r10, [r9, #0x30] - - mov r7, #2 - mov r3, #0x18 - - str r7, [sp, #l_type] - str r3, [sp, #l_stop] - - str r10, [sp, #l_coef_ptr] - b BLOCK_LOOP - -ln1_decode_mb_to - ldr r2, [sp, #l_bc] - mov r0, #0 - nop - - ldr r3, [r2, #bool_decoder_buffer] - str r5, [r2, #bool_decoder_count] - str r4, [r2, #bool_decoder_value] - sub r3, r8, r3 - str r3, [r2, #bool_decoder_pos] - str r6, [r2, #bool_decoder_range] - - add sp, sp, #l_stacksize - ldmia sp!, {r4 - r11, pc} - - ENDP ; |vp8_decode_mb_tokens_v5| - - END diff --git a/vp8/decoder/arm/dsystemdependent.c b/vp8/decoder/arm/dsystemdependent.c index 455c83a..9dcf7b6 100644 --- a/vp8/decoder/arm/dsystemdependent.c +++ b/vp8/decoder/arm/dsystemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -22,20 +23,14 @@ void vp8_dmachine_specific_config(VP8D_COMP *pbi) pbi->mb.rtcd = &pbi->common.rtcd; #if HAVE_ARMV7 pbi->dequant.block = vp8_dequantize_b_neon; - pbi->dequant.idct = vp8_dequant_idct_neon; - pbi->dequant.idct_dc = vp8_dequant_dc_idct_neon; pbi->dboolhuff.start = vp8dx_start_decode_c; - pbi->dboolhuff.stop = vp8dx_stop_decode_c; pbi->dboolhuff.fill = vp8dx_bool_decoder_fill_c; pbi->dboolhuff.debool = vp8dx_decode_bool_c; pbi->dboolhuff.devalue = vp8dx_decode_value_c; #elif HAVE_ARMV6 pbi->dequant.block = vp8_dequantize_b_v6; - pbi->dequant.idct = vp8_dequant_idct_v6; - pbi->dequant.idct_dc = vp8_dequant_dc_idct_v6; pbi->dboolhuff.start = vp8dx_start_decode_c; - pbi->dboolhuff.stop = vp8dx_stop_decode_c; pbi->dboolhuff.fill = vp8dx_bool_decoder_fill_c; pbi->dboolhuff.debool = vp8dx_decode_bool_c; pbi->dboolhuff.devalue = vp8dx_decode_value_c; diff --git a/vp8/decoder/arm/neon/dboolhuff_neon.asm b/vp8/decoder/arm/neon/dboolhuff_neon.asm index 7ec62a3..ff3ffda 100644 --- a/vp8/decoder/arm/neon/dboolhuff_neon.asm +++ b/vp8/decoder/arm/neon/dboolhuff_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/decoder/arm/neon/dequantdcidct_neon.asm b/vp8/decoder/arm/neon/dequant_dc_idct_neon.asm index 3392f2c..f68a780 100644 --- a/vp8/decoder/arm/neon/dequantdcidct_neon.asm +++ b/vp8/decoder/arm/neon/dequant_dc_idct_neon.asm @@ -1,38 +1,51 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; - EXPORT |vp8_dequant_dc_idct_neon| + EXPORT |vp8_dequant_dc_idct_add_neon| ARM REQUIRE8 PRESERVE8 AREA ||.text||, CODE, READONLY, ALIGN=2 -;void vp8_dequant_dc_idct_c(short *input, short *dq, short *output, int pitch, int Dc); +;void vp8_dequant_dc_idct_add_neon(short *input, short *dq, unsigned char *pred, +; unsigned char *dest, int pitch, int stride, +; int Dc); ; r0 short *input, ; r1 short *dq, -; r2 short *output, -; r3 int pitch, -; (stack) int Dc -|vp8_dequant_dc_idct_neon| PROC +; r2 unsigned char *pred +; r3 unsigned char *dest +; sp int pitch +; sp+4 int stride +; sp+8 int Dc +|vp8_dequant_dc_idct_add_neon| PROC vld1.16 {q3, q4}, [r0] vld1.16 {q5, q6}, [r1] - ldr r1, [sp] ;load Dc from stack + ldr r1, [sp, #8] ;load Dc from stack - ldr r12, _dcidct_coeff_ + ldr r12, _CONSTANTS_ vmul.i16 q1, q3, q5 ;input for short_idct4x4llm_neon vmul.i16 q2, q4, q6 vmov.16 d2[0], r1 + ldr r1, [sp] ; pitch + vld1.32 {d14[0]}, [r2], r1 + vld1.32 {d14[1]}, [r2], r1 + vld1.32 {d15[0]}, [r2], r1 + vld1.32 {d15[1]}, [r2] + + ldr r1, [sp, #4] ; stride + ;|short_idct4x4llm_neon| PROC vld1.16 {d0}, [r12] vswp d3, d4 ;q2(vp[4] vp[12]) @@ -46,14 +59,9 @@ vshr.s16 q3, q3, #1 vshr.s16 q4, q4, #1 - vqadd.s16 q3, q3, q2 ;modify since sinpi8sqrt2 > 65536/2 (negtive number) + vqadd.s16 q3, q3, q2 vqadd.s16 q4, q4, q2 - ;d6 - c1:temp1 - ;d7 - d1:temp2 - ;d8 - d1:temp1 - ;d9 - c1:temp2 - vqsub.s16 d10, d6, d9 ;c1 vqadd.s16 d11, d7, d8 ;d1 @@ -82,7 +90,7 @@ vshr.s16 q3, q3, #1 vshr.s16 q4, q4, #1 - vqadd.s16 q3, q3, q2 ;modify since sinpi8sqrt2 > 65536/2 (negtive number) + vqadd.s16 q3, q3, q2 vqadd.s16 q4, q4, q2 vqsub.s16 d10, d6, d9 ;c1 @@ -100,34 +108,29 @@ vrshr.s16 d4, d4, #3 vrshr.s16 d5, d5, #3 - add r1, r2, r3 - add r12, r1, r3 - add r0, r12, r3 - vtrn.32 d2, d4 vtrn.32 d3, d5 vtrn.16 d2, d3 vtrn.16 d4, d5 - vst1.16 {d2}, [r2] - vst1.16 {d3}, [r1] - vst1.16 {d4}, [r12] - vst1.16 {d5}, [r0] + vaddw.u8 q1, q1, d14 + vaddw.u8 q2, q2, d15 - bx lr + vqmovun.s16 d0, q1 + vqmovun.s16 d1, q2 + + vst1.32 {d0[0]}, [r3], r1 + vst1.32 {d0[1]}, [r3], r1 + vst1.32 {d1[0]}, [r3], r1 + vst1.32 {d1[1]}, [r3] - ENDP + bx lr -;----------------- - AREA dcidct4x4_dat, DATA, READWRITE ;read/write by default -;Data section with name data_area is specified. DCD reserves space in memory for 48 data. -;One word each is reserved. Label filter_coeff can be used to access the data. -;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ... -_dcidct_coeff_ - DCD dcidct_coeff -dcidct_coeff - DCD 0x4e7b4e7b, 0x8a8c8a8c + ENDP ; |vp8_dequant_dc_idct_add_neon| -;20091, 20091, 35468, 35468 +; Constant Pool +_CONSTANTS_ DCD cospi8sqrt2minus1 +cospi8sqrt2minus1 DCD 0x4e7b4e7b +sinpi8sqrt2 DCD 0x8a8c8a8c END diff --git a/vp8/decoder/arm/neon/dequantidct_neon.asm b/vp8/decoder/arm/neon/dequant_idct_neon.asm index bba4d5d..1923be4 100644 --- a/vp8/decoder/arm/neon/dequantidct_neon.asm +++ b/vp8/decoder/arm/neon/dequant_idct_neon.asm @@ -1,29 +1,41 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; - EXPORT |vp8_dequant_idct_neon| + EXPORT |vp8_dequant_idct_add_neon| ARM REQUIRE8 PRESERVE8 AREA ||.text||, CODE, READONLY, ALIGN=2 -;void vp8_dequant_idct_c(short *input, short *dq, short *output, int pitch); +;void vp8_dequant_idct_neon(short *input, short *dq, unsigned char *pred, +; unsigned char *dest, int pitch, int stride) ; r0 short *input, ; r1 short *dq, -; r2 short *output, -; r3 int pitch, -|vp8_dequant_idct_neon| PROC +; r2 unsigned char *pred +; r3 unsigned char *dest +; sp int pitch +; sp+4 int stride + +|vp8_dequant_idct_add_neon| PROC vld1.16 {q3, q4}, [r0] vld1.16 {q5, q6}, [r1] + ldr r1, [sp] ; pitch + vld1.32 {d14[0]}, [r2], r1 + vld1.32 {d14[1]}, [r2], r1 + vld1.32 {d15[0]}, [r2], r1 + vld1.32 {d15[1]}, [r2] + + ldr r1, [sp, #4] ; stride - ldr r12, _didct_coeff_ + ldr r12, _CONSTANTS_ vmul.i16 q1, q3, q5 ;input for short_idct4x4llm_neon vmul.i16 q2, q4, q6 @@ -41,14 +53,9 @@ vshr.s16 q3, q3, #1 vshr.s16 q4, q4, #1 - vqadd.s16 q3, q3, q2 ;modify since sinpi8sqrt2 > 65536/2 (negtive number) + vqadd.s16 q3, q3, q2 vqadd.s16 q4, q4, q2 - ;d6 - c1:temp1 - ;d7 - d1:temp2 - ;d8 - d1:temp1 - ;d9 - c1:temp2 - vqsub.s16 d10, d6, d9 ;c1 vqadd.s16 d11, d7, d8 ;d1 @@ -77,7 +84,7 @@ vshr.s16 q3, q3, #1 vshr.s16 q4, q4, #1 - vqadd.s16 q3, q3, q2 ;modify since sinpi8sqrt2 > 65536/2 (negtive number) + vqadd.s16 q3, q3, q2 vqadd.s16 q4, q4, q2 vqsub.s16 d10, d6, d9 ;c1 @@ -95,34 +102,29 @@ vrshr.s16 d4, d4, #3 vrshr.s16 d5, d5, #3 - add r1, r2, r3 - add r12, r1, r3 - add r0, r12, r3 - vtrn.32 d2, d4 vtrn.32 d3, d5 vtrn.16 d2, d3 vtrn.16 d4, d5 - vst1.16 {d2}, [r2] - vst1.16 {d3}, [r1] - vst1.16 {d4}, [r12] - vst1.16 {d5}, [r0] + vaddw.u8 q1, q1, d14 + vaddw.u8 q2, q2, d15 - bx lr + vqmovun.s16 d0, q1 + vqmovun.s16 d1, q2 + + vst1.32 {d0[0]}, [r3], r1 + vst1.32 {d0[1]}, [r3], r1 + vst1.32 {d1[0]}, [r3], r1 + vst1.32 {d1[1]}, [r3] - ENDP + bx lr -;----------------- - AREA didct4x4_dat, DATA, READWRITE ;read/write by default -;Data section with name data_area is specified. DCD reserves space in memory for 48 data. -;One word each is reserved. Label filter_coeff can be used to access the data. -;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ... -_didct_coeff_ - DCD didct_coeff -didct_coeff - DCD 0x4e7b4e7b, 0x8a8c8a8c + ENDP ; |vp8_dequant_idct_add_neon| -;20091, 20091, 35468, 35468 +; Constant Pool +_CONSTANTS_ DCD cospi8sqrt2minus1 +cospi8sqrt2minus1 DCD 0x4e7b4e7b +sinpi8sqrt2 DCD 0x8a8c8a8c END diff --git a/vp8/decoder/arm/neon/dequantizeb_neon.asm b/vp8/decoder/arm/neon/dequantizeb_neon.asm index 1bde946..c8e0c31 100644 --- a/vp8/decoder/arm/neon/dequantizeb_neon.asm +++ b/vp8/decoder/arm/neon/dequantizeb_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/decoder/arm/neon/idct_blk_neon.c b/vp8/decoder/arm/neon/idct_blk_neon.c new file mode 100644 index 0000000..4725e62 --- /dev/null +++ b/vp8/decoder/arm/neon/idct_blk_neon.c @@ -0,0 +1,151 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include "vpx_ports/config.h" +#include "idct.h" +#include "dequantize.h" + +void vp8_dequant_dc_idct_add_y_block_neon + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int stride, char *eobs, short *dc) +{ + int i; + + for (i = 0; i < 4; i++) + { + if (eobs[0] > 1) + vp8_dequant_dc_idct_add_neon (q, dq, pre, dst, 16, stride, dc[0]); + else + vp8_dc_only_idct_add_neon (dc[0], pre, dst, 16, stride); + + if (eobs[1] > 1) + vp8_dequant_dc_idct_add_neon (q+16, dq, pre+4, dst+4, 16, stride, dc[1]); + else + vp8_dc_only_idct_add_neon (dc[1], pre+4, dst+4, 16, stride); + + if (eobs[2] > 1) + vp8_dequant_dc_idct_add_neon (q+32, dq, pre+8, dst+8, 16, stride, dc[2]); + else + vp8_dc_only_idct_add_neon (dc[2], pre+8, dst+8, 16, stride); + + if (eobs[3] > 1) + vp8_dequant_dc_idct_add_neon (q+48, dq, pre+12, dst+12, 16, stride, dc[3]); + else + vp8_dc_only_idct_add_neon (dc[3], pre+12, dst+12, 16, stride); + + q += 64; + dc += 4; + pre += 64; + dst += 4*stride; + eobs += 4; + } +} + +void vp8_dequant_idct_add_y_block_neon + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int stride, char *eobs) +{ + int i; + + for (i = 0; i < 4; i++) + { + if (eobs[0] > 1) + vp8_dequant_idct_add_neon (q, dq, pre, dst, 16, stride); + else + { + vp8_dc_only_idct_add_neon (q[0]*dq[0], pre, dst, 16, stride); + ((int *)q)[0] = 0; + } + + if (eobs[1] > 1) + vp8_dequant_idct_add_neon (q+16, dq, pre+4, dst+4, 16, stride); + else + { + vp8_dc_only_idct_add_neon (q[16]*dq[0], pre+4, dst+4, 16, stride); + ((int *)(q+16))[0] = 0; + } + + if (eobs[2] > 1) + vp8_dequant_idct_add_neon (q+32, dq, pre+8, dst+8, 16, stride); + else + { + vp8_dc_only_idct_add_neon (q[32]*dq[0], pre+8, dst+8, 16, stride); + ((int *)(q+32))[0] = 0; + } + + if (eobs[3] > 1) + vp8_dequant_idct_add_neon (q+48, dq, pre+12, dst+12, 16, stride); + else + { + vp8_dc_only_idct_add_neon (q[48]*dq[0], pre+12, dst+12, 16, stride); + ((int *)(q+48))[0] = 0; + } + + q += 64; + pre += 64; + dst += 4*stride; + eobs += 4; + } +} + +void vp8_dequant_idct_add_uv_block_neon + (short *q, short *dq, unsigned char *pre, + unsigned char *dstu, unsigned char *dstv, int stride, char *eobs) +{ + int i; + + for (i = 0; i < 2; i++) + { + if (eobs[0] > 1) + vp8_dequant_idct_add_neon (q, dq, pre, dstu, 8, stride); + else + { + vp8_dc_only_idct_add_neon (q[0]*dq[0], pre, dstu, 8, stride); + ((int *)q)[0] = 0; + } + + if (eobs[1] > 1) + vp8_dequant_idct_add_neon (q+16, dq, pre+4, dstu+4, 8, stride); + else + { + vp8_dc_only_idct_add_neon (q[16]*dq[0], pre+4, dstu+4, 8, stride); + ((int *)(q+16))[0] = 0; + } + + q += 32; + pre += 32; + dstu += 4*stride; + eobs += 2; + } + + for (i = 0; i < 2; i++) + { + if (eobs[0] > 1) + vp8_dequant_idct_add_neon (q, dq, pre, dstv, 8, stride); + else + { + vp8_dc_only_idct_add_neon (q[0]*dq[0], pre, dstv, 8, stride); + ((int *)q)[0] = 0; + } + + if (eobs[1] > 1) + vp8_dequant_idct_add_neon (q+16, dq, pre+4, dstv+4, 8, stride); + else + { + vp8_dc_only_idct_add_neon (q[16]*dq[0], pre+4, dstv+4, 8, stride); + ((int *)(q+16))[0] = 0; + } + + q += 32; + pre += 32; + dstv += 4*stride; + eobs += 2; + } +} diff --git a/vp8/decoder/dboolhuff.c b/vp8/decoder/dboolhuff.c index 442054e..57cba16 100644 --- a/vp8/decoder/dboolhuff.c +++ b/vp8/decoder/dboolhuff.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -12,7 +13,7 @@ #include "vpx_ports/mem.h" #include "vpx_mem/vpx_mem.h" -DECLARE_ALIGNED(16, const unsigned int, vp8dx_bitreader_norm[256]) = +DECLARE_ALIGNED(16, const unsigned char, vp8dx_bitreader_norm[256]) = { 0, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, @@ -25,86 +26,41 @@ DECLARE_ALIGNED(16, const unsigned int, vp8dx_bitreader_norm[256]) = }; -static void copy_in(BOOL_DECODER *br, unsigned int to_write) -{ - if (to_write > br->user_buffer_sz) - to_write = br->user_buffer_sz; - - memcpy(br->write_ptr, br->user_buffer, to_write); - br->user_buffer += to_write; - br->user_buffer_sz -= to_write; - br->write_ptr = br_ptr_advance(br->write_ptr, to_write); -} - int vp8dx_start_decode_c(BOOL_DECODER *br, const unsigned char *source, unsigned int source_sz) { - br->lowvalue = 0; + br->user_buffer_end = source+source_sz; + br->user_buffer = source; + br->value = 0; + br->count = -8; br->range = 255; - br->count = 0; - br->user_buffer = source; - br->user_buffer_sz = source_sz; if (source_sz && !source) return 1; - /* Allocate the ring buffer backing store with alignment equal to the - * buffer size*2 so that a single pointer can be used for wrapping rather - * than a pointer+offset. - */ - br->decode_buffer = vpx_memalign(VP8_BOOL_DECODER_SZ * 2, - VP8_BOOL_DECODER_SZ); - - if (!br->decode_buffer) - return 1; - /* Populate the buffer */ - br->read_ptr = br->decode_buffer; - br->write_ptr = br->decode_buffer; - copy_in(br, VP8_BOOL_DECODER_SZ); + vp8dx_bool_decoder_fill_c(br); - /* Read the first byte */ - br->value = (*br->read_ptr++) << 8; return 0; } void vp8dx_bool_decoder_fill_c(BOOL_DECODER *br) { - int left, right; - - /* Find available room in the buffer */ - left = 0; - right = br->read_ptr - br->write_ptr; - - if (right < 0) - { - /* Read pointer is behind the write pointer. We can write from the - * write pointer to the end of the buffer. - */ - right = VP8_BOOL_DECODER_SZ - (br->write_ptr - br->decode_buffer); - left = br->read_ptr - br->decode_buffer; - } - - if (right + left < 128) - return; - - if (right) - copy_in(br, right); - - if (left) - { - br->write_ptr = br->decode_buffer; - copy_in(br, left); - } - -} - - -void vp8dx_stop_decode_c(BOOL_DECODER *bc) -{ - vpx_free(bc->decode_buffer); - bc->decode_buffer = 0; + const unsigned char *bufptr; + const unsigned char *bufend; + VP8_BD_VALUE value; + int count; + bufend = br->user_buffer_end; + bufptr = br->user_buffer; + value = br->value; + count = br->count; + + VP8DX_BOOL_DECODER_FILL(count, value, bufptr, bufend); + + br->user_buffer = bufptr; + br->value = value; + br->count = count; } #if 0 @@ -119,13 +75,18 @@ void vp8dx_stop_decode_c(BOOL_DECODER *bc) int vp8dx_decode_bool_c(BOOL_DECODER *br, int probability) { unsigned int bit=0; + VP8_BD_VALUE value; unsigned int split; - unsigned int bigsplit; - register unsigned int range = br->range; - register unsigned int value = br->value; + VP8_BD_VALUE bigsplit; + int count; + unsigned int range; + + value = br->value; + count = br->count; + range = br->range; split = 1 + (((range-1) * probability) >> 8); - bigsplit = (split<<8); + bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8); range = split; if(value >= bigsplit) @@ -143,21 +104,16 @@ int vp8dx_decode_bool_c(BOOL_DECODER *br, int probability) }*/ { - int count = br->count; register unsigned int shift = vp8dx_bitreader_norm[range]; range <<= shift; value <<= shift; count -= shift; - if(count <= 0) - { - value |= (*br->read_ptr) << (-count); - br->read_ptr = br_ptr_advance(br->read_ptr, 1); - count += 8 ; - } - br->count = count; } br->value = value; + br->count = count; br->range = range; + if (count < 0) + vp8dx_bool_decoder_fill_c(br); return bit; } diff --git a/vp8/decoder/dboolhuff.h b/vp8/decoder/dboolhuff.h index f5c9822..c72bc03 100644 --- a/vp8/decoder/dboolhuff.h +++ b/vp8/decoder/dboolhuff.h @@ -1,60 +1,41 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ #ifndef DBOOLHUFF_H #define DBOOLHUFF_H +#include <stddef.h> +#include <limits.h> #include "vpx_ports/config.h" #include "vpx_ports/mem.h" -#include "vpx_ports/vpx_integer.h" +#include "vpx/vpx_integer.h" + +typedef size_t VP8_BD_VALUE; + +# define VP8_BD_VALUE_SIZE ((int)sizeof(VP8_BD_VALUE)*CHAR_BIT) +/*This is meant to be a large, positive constant that can still be efficiently + loaded as an immediate (on platforms like ARM, for example). + Even relatively modest values like 100 would work fine.*/ +# define VP8_LOTS_OF_BITS (0x40000000) + -/* Size of the bool decoder backing storage - * - * This size was chosen to be greater than the worst case encoding of a - * single macroblock. This was calcluated as follows (python): - * - * def max_cost(prob): - * return max(prob_costs[prob], prob_costs[255-prob]) / 256; - * - * tree_nodes_cost = 7 * max_cost(255) - * extra_bits_cost = sum([max_cost(bit) for bit in extra_bits]) - * sign_bit_cost = max_cost(128) - * total_cost = tree_nodes_cost + extra_bits_cost + sign_bit_cost - * - * where the prob_costs table was taken from the C vp8_prob_cost table in - * boolhuff.c and the extra_bits table was taken from the 11 extrabits for - * a category 6 token as defined in vp8d_token_extra_bits2/detokenize.c - * - * This equation produced a maximum of 79 bits per coefficient. Scaling up - * to the macroblock level: - * - * 79 bits/coeff * 16 coeff/block * 25 blocks/macroblock = 31600 b/mb - * - * 4096 bytes = 32768 bits > 31600 - */ -#define VP8_BOOL_DECODER_SZ 4096 -#define VP8_BOOL_DECODER_MASK (VP8_BOOL_DECODER_SZ-1) -#define VP8_BOOL_DECODER_PTR_MASK (~(uintptr_t)(VP8_BOOL_DECODER_SZ)) struct vp8_dboolhuff_rtcd_vtable; typedef struct { - unsigned int lowvalue; - unsigned int range; - unsigned int value; - int count; + const unsigned char *user_buffer_end; const unsigned char *user_buffer; - unsigned int user_buffer_sz; - unsigned char *decode_buffer; - const unsigned char *read_ptr; - unsigned char *write_ptr; + VP8_BD_VALUE value; + int count; + unsigned int range; #if CONFIG_RUNTIME_CPU_DETECT struct vp8_dboolhuff_rtcd_vtable *rtcd; #endif @@ -62,7 +43,6 @@ typedef struct #define prototype_dbool_start(sym) int sym(BOOL_DECODER *br, \ const unsigned char *source, unsigned int source_sz) -#define prototype_dbool_stop(sym) void sym(BOOL_DECODER *bc) #define prototype_dbool_fill(sym) void sym(BOOL_DECODER *br) #define prototype_dbool_debool(sym) int sym(BOOL_DECODER *br, int probability) #define prototype_dbool_devalue(sym) int sym(BOOL_DECODER *br, int bits); @@ -75,10 +55,6 @@ typedef struct #define vp8_dbool_start vp8dx_start_decode_c #endif -#ifndef vp8_dbool_stop -#define vp8_dbool_stop vp8dx_stop_decode_c -#endif - #ifndef vp8_dbool_fill #define vp8_dbool_fill vp8dx_bool_decoder_fill_c #endif @@ -92,20 +68,17 @@ typedef struct #endif extern prototype_dbool_start(vp8_dbool_start); -extern prototype_dbool_stop(vp8_dbool_stop); extern prototype_dbool_fill(vp8_dbool_fill); extern prototype_dbool_debool(vp8_dbool_debool); extern prototype_dbool_devalue(vp8_dbool_devalue); typedef prototype_dbool_start((*vp8_dbool_start_fn_t)); -typedef prototype_dbool_stop((*vp8_dbool_stop_fn_t)); typedef prototype_dbool_fill((*vp8_dbool_fill_fn_t)); typedef prototype_dbool_debool((*vp8_dbool_debool_fn_t)); typedef prototype_dbool_devalue((*vp8_dbool_devalue_fn_t)); typedef struct vp8_dboolhuff_rtcd_vtable { vp8_dbool_start_fn_t start; - vp8_dbool_stop_fn_t stop; vp8_dbool_fill_fn_t fill; vp8_dbool_debool_fn_t debool; vp8_dbool_devalue_fn_t devalue; @@ -122,18 +95,7 @@ typedef struct vp8_dboolhuff_rtcd_vtable { #define IF_RTCD(x) NULL //#endif -static unsigned char *br_ptr_advance(const unsigned char *_ptr, - unsigned int n) -{ - uintptr_t ptr = (uintptr_t)_ptr; - - ptr += n; - ptr &= VP8_BOOL_DECODER_PTR_MASK; - - return (void *)ptr; -} - -DECLARE_ALIGNED(16, extern const unsigned int, vp8dx_bitreader_norm[256]); +DECLARE_ALIGNED(16, extern const unsigned char, vp8dx_bitreader_norm[256]); /* wrapper functions to hide RTCD. static means inline means hopefully no * penalty @@ -146,12 +108,34 @@ static int vp8dx_start_decode(BOOL_DECODER *br, #endif return DBOOLHUFF_INVOKE(rtcd, start)(br, source, source_sz); } -static void vp8dx_stop_decode(BOOL_DECODER *br) { - DBOOLHUFF_INVOKE(br->rtcd, stop)(br); -} static void vp8dx_bool_decoder_fill(BOOL_DECODER *br) { DBOOLHUFF_INVOKE(br->rtcd, fill)(br); } + +/*The refill loop is used in several places, so define it in a macro to make + sure they're all consistent. + An inline function would be cleaner, but has a significant penalty, because + multiple BOOL_DECODER fields must be modified, and the compiler is not smart + enough to eliminate the stores to those fields and the subsequent reloads + from them when inlining the function.*/ +#define VP8DX_BOOL_DECODER_FILL(_count,_value,_bufptr,_bufend) \ + do \ + { \ + int shift; \ + for(shift = VP8_BD_VALUE_SIZE - 8 - ((_count) + 8); shift >= 0; ) \ + { \ + if((_bufptr) >= (_bufend)) { \ + (_count) = VP8_LOTS_OF_BITS; \ + break; \ + } \ + (_count) += 8; \ + (_value) |= (VP8_BD_VALUE)*(_bufptr)++ << shift; \ + shift -= 8; \ + } \ + } \ + while(0) + + static int vp8dx_decode_bool(BOOL_DECODER *br, int probability) { /* * Until optimized versions of this function are available, we @@ -160,13 +144,18 @@ static int vp8dx_decode_bool(BOOL_DECODER *br, int probability) { *return DBOOLHUFF_INVOKE(br->rtcd, debool)(br, probability); */ unsigned int bit = 0; + VP8_BD_VALUE value; unsigned int split; - unsigned int bigsplit; - register unsigned int range = br->range; - register unsigned int value = br->value; + VP8_BD_VALUE bigsplit; + int count; + unsigned int range; + + value = br->value; + count = br->count; + range = br->range; split = 1 + (((range - 1) * probability) >> 8); - bigsplit = (split << 8); + bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8); range = split; @@ -185,23 +174,16 @@ static int vp8dx_decode_bool(BOOL_DECODER *br, int probability) { }*/ { - int count = br->count; register unsigned int shift = vp8dx_bitreader_norm[range]; range <<= shift; value <<= shift; count -= shift; - - if (count <= 0) - { - value |= (*br->read_ptr) << (-count); - br->read_ptr = br_ptr_advance(br->read_ptr, 1); - count += 8 ; - } - - br->count = count; } br->value = value; + br->count = count; br->range = range; + if(count < 0) + vp8dx_bool_decoder_fill(br); return bit; } diff --git a/vp8/decoder/decodemv.c b/vp8/decoder/decodemv.c index 6035f3e..e9281f7 100644..100755 --- a/vp8/decoder/decodemv.c +++ b/vp8/decoder/decodemv.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -13,10 +14,126 @@ #include "entropymode.h" #include "onyxd_int.h" #include "findnearmv.h" -#include "demode.h" + #if CONFIG_DEBUG #include <assert.h> #endif +static int vp8_read_bmode(vp8_reader *bc, const vp8_prob *p) +{ + const int i = vp8_treed_read(bc, vp8_bmode_tree, p); + + return i; +} + + +static int vp8_read_ymode(vp8_reader *bc, const vp8_prob *p) +{ + const int i = vp8_treed_read(bc, vp8_ymode_tree, p); + + return i; +} + +static int vp8_kfread_ymode(vp8_reader *bc, const vp8_prob *p) +{ + const int i = vp8_treed_read(bc, vp8_kf_ymode_tree, p); + + return i; +} + + + +static int vp8_read_uv_mode(vp8_reader *bc, const vp8_prob *p) +{ + const int i = vp8_treed_read(bc, vp8_uv_mode_tree, p); + + return i; +} + +static void vp8_read_mb_features(vp8_reader *r, MB_MODE_INFO *mi, MACROBLOCKD *x) +{ + // Is segmentation enabled + if (x->segmentation_enabled && x->update_mb_segmentation_map) + { + // If so then read the segment id. + if (vp8_read(r, x->mb_segment_tree_probs[0])) + mi->segment_id = (unsigned char)(2 + vp8_read(r, x->mb_segment_tree_probs[2])); + else + mi->segment_id = (unsigned char)(vp8_read(r, x->mb_segment_tree_probs[1])); + } +} + +static void vp8_kfread_modes(VP8D_COMP *pbi, MODE_INFO *m, int mb_row, int mb_col) +{ + vp8_reader *const bc = & pbi->bc; + const int mis = pbi->common.mode_info_stride; + + { + MB_PREDICTION_MODE y_mode; + + // Read the Macroblock segmentation map if it is being updated explicitly this frame (reset to 0 above by default) + // By default on a key frame reset all MBs to segment 0 + m->mbmi.segment_id = 0; + + if (pbi->mb.update_mb_segmentation_map) + vp8_read_mb_features(bc, &m->mbmi, &pbi->mb); + + // Read the macroblock coeff skip flag if this feature is in use, else default to 0 + if (pbi->common.mb_no_coeff_skip) + m->mbmi.mb_skip_coeff = vp8_read(bc, pbi->prob_skip_false); + else + m->mbmi.mb_skip_coeff = 0; + + y_mode = (MB_PREDICTION_MODE) vp8_kfread_ymode(bc, pbi->common.kf_ymode_prob); + + m->mbmi.ref_frame = INTRA_FRAME; + + if ((m->mbmi.mode = y_mode) == B_PRED) + { + int i = 0; + + do + { + const B_PREDICTION_MODE A = vp8_above_bmi(m, i, mis)->mode; + const B_PREDICTION_MODE L = vp8_left_bmi(m, i)->mode; + + m->bmi[i].mode = (B_PREDICTION_MODE) vp8_read_bmode(bc, pbi->common.kf_bmode_prob [A] [L]); + } + while (++i < 16); + } + else + { + int BMode; + int i = 0; + + switch (y_mode) + { + case DC_PRED: + BMode = B_DC_PRED; + break; + case V_PRED: + BMode = B_VE_PRED; + break; + case H_PRED: + BMode = B_HE_PRED; + break; + case TM_PRED: + BMode = B_TM_PRED; + break; + default: + BMode = B_DC_PRED; + break; + } + + do + { + m->bmi[i].mode = (B_PREDICTION_MODE)BMode; + } + while (++i < 16); + } + + m->mbmi.uv_mode = (MB_PREDICTION_MODE)vp8_read_uv_mode(bc, pbi->common.kf_uv_mode_prob); + } +} static int read_mvcomponent(vp8_reader *r, const MV_CONTEXT *mvc) { @@ -98,6 +215,8 @@ static MB_PREDICTION_MODE sub_mv_ref(vp8_reader *bc, const vp8_prob *p) return (MB_PREDICTION_MODE)i; } + +#ifdef VPX_MODE_COUNT unsigned int vp8_mv_cont_count[5][4] = { { 0, 0, 0, 0 }, @@ -106,309 +225,327 @@ unsigned int vp8_mv_cont_count[5][4] = { 0, 0, 0, 0 }, { 0, 0, 0, 0 } }; +#endif -void vp8_decode_mode_mvs(VP8D_COMP *pbi) -{ - const MV Zero = { 0, 0}; - - VP8_COMMON *const pc = & pbi->common; - vp8_reader *const bc = & pbi->bc; +unsigned char vp8_mbsplit_offset[4][16] = { + { 0, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, + { 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, + { 0, 2, 8, 10, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, + { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15} +}; - MODE_INFO *mi = pc->mi, *ms; - const int mis = pc->mode_info_stride; +unsigned char vp8_mbsplit_fill_count[4] = {8, 8, 4, 1}; +unsigned char vp8_mbsplit_fill_offset[4][16] = { + { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15}, + { 0, 1, 4, 5, 8, 9, 12, 13, 2, 3, 6, 7, 10, 11, 14, 15}, + { 0, 1, 4, 5, 2, 3, 6, 7, 8, 9, 12, 13, 10, 11, 14, 15}, + { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15} +}; - MV_CONTEXT *const mvc = pc->fc.mvc; - int mb_row = -1; - vp8_prob prob_intra; - vp8_prob prob_last; - vp8_prob prob_gf; - vp8_prob prob_skip_false = 0; - if (pc->mb_no_coeff_skip) - prob_skip_false = (vp8_prob)vp8_read_literal(bc, 8); - - prob_intra = (vp8_prob)vp8_read_literal(bc, 8); - prob_last = (vp8_prob)vp8_read_literal(bc, 8); - prob_gf = (vp8_prob)vp8_read_literal(bc, 8); +void vp8_mb_mode_mv_init(VP8D_COMP *pbi) +{ + vp8_reader *const bc = & pbi->bc; + MV_CONTEXT *const mvc = pbi->common.fc.mvc; - ms = pc->mi - 1; + pbi->prob_skip_false = 0; + if (pbi->common.mb_no_coeff_skip) + pbi->prob_skip_false = (vp8_prob)vp8_read_literal(bc, 8); - if (vp8_read_bit(bc)) + if(pbi->common.frame_type != KEY_FRAME) { - int i = 0; + pbi->prob_intra = (vp8_prob)vp8_read_literal(bc, 8); + pbi->prob_last = (vp8_prob)vp8_read_literal(bc, 8); + pbi->prob_gf = (vp8_prob)vp8_read_literal(bc, 8); - do + if (vp8_read_bit(bc)) { - pc->fc.ymode_prob[i] = (vp8_prob) vp8_read_literal(bc, 8); - } - while (++i < 4); - } + int i = 0; - if (vp8_read_bit(bc)) - { - int i = 0; + do + { + pbi->common.fc.ymode_prob[i] = (vp8_prob) vp8_read_literal(bc, 8); + } + while (++i < 4); + } - do + if (vp8_read_bit(bc)) { - pc->fc.uv_mode_prob[i] = (vp8_prob) vp8_read_literal(bc, 8); + int i = 0; + + do + { + pbi->common.fc.uv_mode_prob[i] = (vp8_prob) vp8_read_literal(bc, 8); + } + while (++i < 3); } - while (++i < 3); - } - read_mvcontexts(bc, mvc); + read_mvcontexts(bc, mvc); + } +} - while (++mb_row < pc->mb_rows) +void vp8_read_mb_modes_mv(VP8D_COMP *pbi, MODE_INFO *mi, MB_MODE_INFO *mbmi, + int mb_row, int mb_col) +{ + const MV Zero = { 0, 0}; + vp8_reader *const bc = & pbi->bc; + MV_CONTEXT *const mvc = pbi->common.fc.mvc; + const int mis = pbi->common.mode_info_stride; + + MV *const mv = & mbmi->mv.as_mv; + int mb_to_left_edge; + int mb_to_right_edge; + int mb_to_top_edge; + int mb_to_bottom_edge; + + mb_to_top_edge = pbi->mb.mb_to_top_edge; + mb_to_bottom_edge = pbi->mb.mb_to_bottom_edge; + mb_to_top_edge -= LEFT_TOP_MARGIN; + mb_to_bottom_edge += RIGHT_BOTTOM_MARGIN; + + mbmi->need_to_clamp_mvs = 0; + // Distance of Mb to the various image edges. + // These specified to 8th pel as they are always compared to MV values that are in 1/8th pel units + pbi->mb.mb_to_left_edge = + mb_to_left_edge = -((mb_col * 16) << 3); + mb_to_left_edge -= LEFT_TOP_MARGIN; + + pbi->mb.mb_to_right_edge = + mb_to_right_edge = ((pbi->common.mb_cols - 1 - mb_col) * 16) << 3; + mb_to_right_edge += RIGHT_BOTTOM_MARGIN; + + // If required read in new segmentation data for this MB + if (pbi->mb.update_mb_segmentation_map) + vp8_read_mb_features(bc, mbmi, &pbi->mb); + + // Read the macroblock coeff skip flag if this feature is in use, else default to 0 + if (pbi->common.mb_no_coeff_skip) + mbmi->mb_skip_coeff = vp8_read(bc, pbi->prob_skip_false); + else + mbmi->mb_skip_coeff = 0; + + if ((mbmi->ref_frame = (MV_REFERENCE_FRAME) vp8_read(bc, pbi->prob_intra))) /* inter MB */ { - int mb_col = -1; + int rct[4]; + vp8_prob mv_ref_p [VP8_MVREFS-1]; + MV nearest, nearby, best_mv; - while (++mb_col < pc->mb_cols) + if (vp8_read(bc, pbi->prob_last)) { - MB_MODE_INFO *const mbmi = & mi->mbmi; - MV *const mv = & mbmi->mv.as_mv; - VP8_COMMON *const pc = &pbi->common; - MACROBLOCKD *xd = &pbi->mb; + mbmi->ref_frame = (MV_REFERENCE_FRAME)((int)mbmi->ref_frame + (int)(1 + vp8_read(bc, pbi->prob_gf))); + } - vp8dx_bool_decoder_fill(bc); + vp8_find_near_mvs(&pbi->mb, mi, &nearest, &nearby, &best_mv, rct, mbmi->ref_frame, pbi->common.ref_frame_sign_bias); - // Distance of Mb to the various image edges. - // These specified to 8th pel as they are always compared to MV values that are in 1/8th pel units - xd->mb_to_left_edge = -((mb_col * 16) << 3); - xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3; - xd->mb_to_top_edge = -((mb_row * 16)) << 3; - xd->mb_to_bottom_edge = ((pc->mb_rows - 1 - mb_row) * 16) << 3; + vp8_mv_ref_probs(mv_ref_p, rct); - // If required read in new segmentation data for this MB - if (pbi->mb.update_mb_segmentation_map) - vp8_read_mb_features(bc, mbmi, &pbi->mb); + mbmi->uv_mode = DC_PRED; + switch (mbmi->mode = read_mv_ref(bc, mv_ref_p)) + { + case SPLITMV: + { + const int s = mbmi->partitioning = + vp8_treed_read(bc, vp8_mbsplit_tree, vp8_mbsplit_probs); + const int num_p = vp8_mbsplit_count [s]; + int j = 0; - // Read the macroblock coeff skip flag if this feature is in use, else default to 0 - if (pc->mb_no_coeff_skip) - mbmi->mb_skip_coeff = vp8_read(bc, prob_skip_false); - else - mbmi->mb_skip_coeff = 0; + do /* for each subset j */ + { + B_MODE_INFO bmi; + MV *const mv = & bmi.mv.as_mv; - mbmi->uv_mode = DC_PRED; + int k; /* first block in subset j */ + int mv_contz; + k = vp8_mbsplit_offset[s][j]; - if ((mbmi->ref_frame = (MV_REFERENCE_FRAME) vp8_read(bc, prob_intra))) /* inter MB */ - { - int rct[4]; - vp8_prob mv_ref_p [VP8_MVREFS-1]; - MV nearest, nearby, best_mv; + mv_contz = vp8_mv_cont(&(vp8_left_bmi(mi, k)->mv.as_mv), &(vp8_above_bmi(mi, k, mis)->mv.as_mv)); - if (vp8_read(bc, prob_last)) + switch (bmi.mode = (B_PREDICTION_MODE) sub_mv_ref(bc, vp8_sub_mv_ref_prob2 [mv_contz])) //pc->fc.sub_mv_ref_prob)) { - mbmi->ref_frame = (MV_REFERENCE_FRAME)((int)mbmi->ref_frame + (int)(1 + vp8_read(bc, prob_gf))); + case NEW4X4: + read_mv(bc, mv, (const MV_CONTEXT *) mvc); + mv->row += best_mv.row; + mv->col += best_mv.col; + #ifdef VPX_MODE_COUNT + vp8_mv_cont_count[mv_contz][3]++; + #endif + break; + case LEFT4X4: + *mv = vp8_left_bmi(mi, k)->mv.as_mv; + #ifdef VPX_MODE_COUNT + vp8_mv_cont_count[mv_contz][0]++; + #endif + break; + case ABOVE4X4: + *mv = vp8_above_bmi(mi, k, mis)->mv.as_mv; + #ifdef VPX_MODE_COUNT + vp8_mv_cont_count[mv_contz][1]++; + #endif + break; + case ZERO4X4: + *mv = Zero; + #ifdef VPX_MODE_COUNT + vp8_mv_cont_count[mv_contz][2]++; + #endif + break; + default: + break; } - vp8_find_near_mvs(xd, mi, &nearest, &nearby, &best_mv, rct, mbmi->ref_frame, pbi->common.ref_frame_sign_bias); - - vp8_mv_ref_probs(mv_ref_p, rct); + mbmi->need_to_clamp_mvs = (mv->col < mb_to_left_edge) ? 1 : 0; + mbmi->need_to_clamp_mvs |= (mv->col > mb_to_right_edge) ? 1 : 0; + mbmi->need_to_clamp_mvs |= (mv->row < mb_to_top_edge) ? 1 : 0; + mbmi->need_to_clamp_mvs |= (mv->row > mb_to_bottom_edge) ? 1 : 0; - switch (mbmi->mode = read_mv_ref(bc, mv_ref_p)) { - case SPLITMV: - { - const int s = mbmi->partitioning = vp8_treed_read( - bc, vp8_mbsplit_tree, vp8_mbsplit_probs - ); - const int num_p = vp8_mbsplit_count [s]; - const int *const L = vp8_mbsplits [s]; - int j = 0; - - do /* for each subset j */ - { - B_MODE_INFO *const bmi = mbmi->partition_bmi + j; - MV *const mv = & bmi->mv.as_mv; - - int k = -1; /* first block in subset j */ - int mv_contz; - - while (j != L[++k]) - if (k >= 16) -#if CONFIG_DEBUG - assert(0); + /* Fill (uniform) modes, mvs of jth subset. + Must do it here because ensuing subsets can + refer back to us via "left" or "above". */ + unsigned char *fill_offset; + unsigned int fill_count = vp8_mbsplit_fill_count[s]; -#else - ; -#endif + fill_offset = &vp8_mbsplit_fill_offset[s][(unsigned char)j * vp8_mbsplit_fill_count[s]]; - mv_contz = vp8_mv_cont(&(vp8_left_bmi(mi, k)->mv.as_mv), &(vp8_above_bmi(mi, k, mis)->mv.as_mv)); + do { + mi->bmi[ *fill_offset] = bmi; + fill_offset++; - switch (bmi->mode = (B_PREDICTION_MODE) sub_mv_ref(bc, vp8_sub_mv_ref_prob2 [mv_contz])) //pc->fc.sub_mv_ref_prob)) - { - case NEW4X4: - read_mv(bc, mv, (const MV_CONTEXT *) mvc); - mv->row += best_mv.row; - mv->col += best_mv.col; -#ifdef VPX_MODE_COUNT - vp8_mv_cont_count[mv_contz][3]++; -#endif - break; - case LEFT4X4: - *mv = vp8_left_bmi(mi, k)->mv.as_mv; -#ifdef VPX_MODE_COUNT - vp8_mv_cont_count[mv_contz][0]++; -#endif - break; - case ABOVE4X4: - *mv = vp8_above_bmi(mi, k, mis)->mv.as_mv; -#ifdef VPX_MODE_COUNT - vp8_mv_cont_count[mv_contz][1]++; -#endif - break; - case ZERO4X4: - *mv = Zero; -#ifdef VPX_MODE_COUNT - vp8_mv_cont_count[mv_contz][2]++; -#endif - break; - default: - break; - } - - /* Fill (uniform) modes, mvs of jth subset. - Must do it here because ensuing subsets can - refer back to us via "left" or "above". */ - do - if (j == L[k]) - mi->bmi[k] = *bmi; - - while (++k < 16); - } - while (++j < num_p); + }while (--fill_count); } - *mv = mi->bmi[15].mv.as_mv; - - break; /* done with SPLITMV */ - - case NEARMV: - *mv = nearby; - - // Clip "next_nearest" so that it does not extend to far out of image - if (mv->col < (xd->mb_to_left_edge - LEFT_TOP_MARGIN)) - mv->col = xd->mb_to_left_edge - LEFT_TOP_MARGIN; - else if (mv->col > xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN) - mv->col = xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN; - - if (mv->row < (xd->mb_to_top_edge - LEFT_TOP_MARGIN)) - mv->row = xd->mb_to_top_edge - LEFT_TOP_MARGIN; - else if (mv->row > xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN) - mv->row = xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN; - - goto propagate_mv; + } + while (++j < num_p); + } - case NEARESTMV: - *mv = nearest; + *mv = mi->bmi[15].mv.as_mv; + + break; /* done with SPLITMV */ + + case NEARMV: + *mv = nearby; + // Clip "next_nearest" so that it does not extend to far out of image + mv->col = (mv->col < mb_to_left_edge) ? mb_to_left_edge : mv->col; + mv->col = (mv->col > mb_to_right_edge) ? mb_to_right_edge : mv->col; + mv->row = (mv->row < mb_to_top_edge) ? mb_to_top_edge : mv->row; + mv->row = (mv->row > mb_to_bottom_edge) ? mb_to_bottom_edge : mv->row; + goto propagate_mv; + + case NEARESTMV: + *mv = nearest; + // Clip "next_nearest" so that it does not extend to far out of image + mv->col = (mv->col < mb_to_left_edge) ? mb_to_left_edge : mv->col; + mv->col = (mv->col > mb_to_right_edge) ? mb_to_right_edge : mv->col; + mv->row = (mv->row < mb_to_top_edge) ? mb_to_top_edge : mv->row; + mv->row = (mv->row > mb_to_bottom_edge) ? mb_to_bottom_edge : mv->row; + goto propagate_mv; + + case ZEROMV: + *mv = Zero; + goto propagate_mv; + + case NEWMV: + read_mv(bc, mv, (const MV_CONTEXT *) mvc); + mv->row += best_mv.row; + mv->col += best_mv.col; + + /* Don't need to check this on NEARMV and NEARESTMV modes + * since those modes clamp the MV. The NEWMV mode does not, + * so signal to the prediction stage whether special + * handling may be required. + */ + mbmi->need_to_clamp_mvs = (mv->col < mb_to_left_edge) ? 1 : 0; + mbmi->need_to_clamp_mvs |= (mv->col > mb_to_right_edge) ? 1 : 0; + mbmi->need_to_clamp_mvs |= (mv->row < mb_to_top_edge) ? 1 : 0; + mbmi->need_to_clamp_mvs |= (mv->row > mb_to_bottom_edge) ? 1 : 0; + + propagate_mv: /* same MV throughout */ + { + //int i=0; + //do + //{ + // mi->bmi[i].mv.as_mv = *mv; + //} + //while( ++i < 16); + + mi->bmi[0].mv.as_mv = *mv; + mi->bmi[1].mv.as_mv = *mv; + mi->bmi[2].mv.as_mv = *mv; + mi->bmi[3].mv.as_mv = *mv; + mi->bmi[4].mv.as_mv = *mv; + mi->bmi[5].mv.as_mv = *mv; + mi->bmi[6].mv.as_mv = *mv; + mi->bmi[7].mv.as_mv = *mv; + mi->bmi[8].mv.as_mv = *mv; + mi->bmi[9].mv.as_mv = *mv; + mi->bmi[10].mv.as_mv = *mv; + mi->bmi[11].mv.as_mv = *mv; + mi->bmi[12].mv.as_mv = *mv; + mi->bmi[13].mv.as_mv = *mv; + mi->bmi[14].mv.as_mv = *mv; + mi->bmi[15].mv.as_mv = *mv; + } + break; + default:; + #if CONFIG_DEBUG + assert(0); + #endif + } + } + else + { + /* MB is intra coded */ + int j = 0; + do + { + mi->bmi[j].mv.as_mv = Zero; + } + while (++j < 16); - // Clip "next_nearest" so that it does not extend to far out of image - if (mv->col < (xd->mb_to_left_edge - LEFT_TOP_MARGIN)) - mv->col = xd->mb_to_left_edge - LEFT_TOP_MARGIN; - else if (mv->col > xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN) - mv->col = xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN; + if ((mbmi->mode = (MB_PREDICTION_MODE) vp8_read_ymode(bc, pbi->common.fc.ymode_prob)) == B_PRED) + { + j = 0; + do + { + mi->bmi[j].mode = (B_PREDICTION_MODE)vp8_read_bmode(bc, pbi->common.fc.bmode_prob); + } + while (++j < 16); + } - if (mv->row < (xd->mb_to_top_edge - LEFT_TOP_MARGIN)) - mv->row = xd->mb_to_top_edge - LEFT_TOP_MARGIN; - else if (mv->row > xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN) - mv->row = xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN; + mbmi->uv_mode = (MB_PREDICTION_MODE)vp8_read_uv_mode(bc, pbi->common.fc.uv_mode_prob); + } - goto propagate_mv; +} - case ZEROMV: - *mv = Zero; - goto propagate_mv; +void vp8_decode_mode_mvs(VP8D_COMP *pbi) +{ + MODE_INFO *mi = pbi->common.mi; + int mb_row = -1; - case NEWMV: - read_mv(bc, mv, (const MV_CONTEXT *) mvc); - mv->row += best_mv.row; - mv->col += best_mv.col; - /* Encoder should not produce invalid motion vectors, but since - * arbitrary length MVs can be parsed from the bitstream, we - * need to clamp them here in case we're reading bad data to - * avoid a crash. - */ -#if CONFIG_DEBUG - assert(mv->col >= (xd->mb_to_left_edge - LEFT_TOP_MARGIN)); - assert(mv->col <= (xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN)); - assert(mv->row >= (xd->mb_to_top_edge - LEFT_TOP_MARGIN)); - assert(mv->row <= (xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN)); -#endif + vp8_mb_mode_mv_init(pbi); - if (mv->col < (xd->mb_to_left_edge - LEFT_TOP_MARGIN)) - mv->col = xd->mb_to_left_edge - LEFT_TOP_MARGIN; - else if (mv->col > xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN) - mv->col = xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN; - - if (mv->row < (xd->mb_to_top_edge - LEFT_TOP_MARGIN)) - mv->row = xd->mb_to_top_edge - LEFT_TOP_MARGIN; - else if (mv->row > xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN) - mv->row = xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN; - - propagate_mv: /* same MV throughout */ - { - //int i=0; - //do - //{ - // mi->bmi[i].mv.as_mv = *mv; - //} - //while( ++i < 16); - - mi->bmi[0].mv.as_mv = *mv; - mi->bmi[1].mv.as_mv = *mv; - mi->bmi[2].mv.as_mv = *mv; - mi->bmi[3].mv.as_mv = *mv; - mi->bmi[4].mv.as_mv = *mv; - mi->bmi[5].mv.as_mv = *mv; - mi->bmi[6].mv.as_mv = *mv; - mi->bmi[7].mv.as_mv = *mv; - mi->bmi[8].mv.as_mv = *mv; - mi->bmi[9].mv.as_mv = *mv; - mi->bmi[10].mv.as_mv = *mv; - mi->bmi[11].mv.as_mv = *mv; - mi->bmi[12].mv.as_mv = *mv; - mi->bmi[13].mv.as_mv = *mv; - mi->bmi[14].mv.as_mv = *mv; - mi->bmi[15].mv.as_mv = *mv; - } + while (++mb_row < pbi->common.mb_rows) + { + int mb_col = -1; + int mb_to_top_edge; + int mb_to_bottom_edge; - break; + pbi->mb.mb_to_top_edge = + mb_to_top_edge = -((mb_row * 16)) << 3; + mb_to_top_edge -= LEFT_TOP_MARGIN; - default:; -#if CONFIG_DEBUG - assert(0); -#endif - } + pbi->mb.mb_to_bottom_edge = + mb_to_bottom_edge = ((pbi->common.mb_rows - 1 - mb_row) * 16) << 3; + mb_to_bottom_edge += RIGHT_BOTTOM_MARGIN; - } + while (++mb_col < pbi->common.mb_cols) + { +// vp8_read_mb_modes_mv(pbi, xd->mode_info_context, &xd->mode_info_context->mbmi, mb_row, mb_col); + if(pbi->common.frame_type == KEY_FRAME) + vp8_kfread_modes(pbi, mi, mb_row, mb_col); else - { - /* MB is intra coded */ - - int j = 0; - - do - { - mi->bmi[j].mv.as_mv = Zero; - } - while (++j < 16); - - *mv = Zero; - - if ((mbmi->mode = (MB_PREDICTION_MODE) vp8_read_ymode(bc, pc->fc.ymode_prob)) == B_PRED) - { - int j = 0; - - do - { - mi->bmi[j].mode = (B_PREDICTION_MODE)vp8_read_bmode(bc, pc->fc.bmode_prob); - } - while (++j < 16); - } - - mbmi->uv_mode = (MB_PREDICTION_MODE)vp8_read_uv_mode(bc, pc->fc.uv_mode_prob); - } + vp8_read_mb_modes_mv(pbi, mi, &mi->mbmi, mb_row, mb_col); mi++; // next macroblock } @@ -416,3 +553,4 @@ void vp8_decode_mode_mvs(VP8D_COMP *pbi) mi++; // skip left predictor each row } } + diff --git a/vp8/decoder/decodemv.h b/vp8/decoder/decodemv.h index 4030071..9403424 100644 --- a/vp8/decoder/decodemv.h +++ b/vp8/decoder/decodemv.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/decoder/decoderthreading.h b/vp8/decoder/decoderthreading.h index ebc5c27..c8e3f02 100644 --- a/vp8/decoder/decoderthreading.h +++ b/vp8/decoder/decoderthreading.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -17,6 +18,7 @@ extern void vp8_mtdecode_mb_rows(VP8D_COMP *pbi, MACROBLOCKD *xd); +extern void vp8_mt_loop_filter_frame(VP8D_COMP *pbi); extern void vp8_stop_lfthread(VP8D_COMP *pbi); extern void vp8_start_lfthread(VP8D_COMP *pbi); extern void vp8_decoder_remove_threads(VP8D_COMP *pbi); diff --git a/vp8/decoder/decodframe.c b/vp8/decoder/decodframe.c index 4edf4f6..4f5b7c7 100644 --- a/vp8/decoder/decodframe.c +++ b/vp8/decoder/decodframe.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -20,9 +21,9 @@ #include "alloccommon.h" #include "entropymode.h" #include "quant_common.h" -#include "segmentation_common.h" + #include "setupintrarecon.h" -#include "demode.h" + #include "decodemv.h" #include "extend.h" #include "vpx_mem/vpx_mem.h" @@ -112,7 +113,7 @@ static void mb_init_dequantizer(VP8D_COMP *pbi, MACROBLOCKD *xd) // to dst buffer, we can write the result directly to dst buffer. This eliminates unnecessary copy. static void skip_recon_mb(VP8D_COMP *pbi, MACROBLOCKD *xd) { - if (xd->frame_type == KEY_FRAME || xd->mbmi.ref_frame == INTRA_FRAME) + if (xd->frame_type == KEY_FRAME || xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) { vp8_build_intra_predictors_mbuv_s(xd); @@ -125,42 +126,114 @@ static void skip_recon_mb(VP8D_COMP *pbi, MACROBLOCKD *xd) } } -static void reconstruct_mb(VP8D_COMP *pbi, MACROBLOCKD *xd) +static void clamp_mv_to_umv_border(MV *mv, const MACROBLOCKD *xd) +{ + /* If the MV points so far into the UMV border that no visible pixels + * are used for reconstruction, the subpel part of the MV can be + * discarded and the MV limited to 16 pixels with equivalent results. + * + * This limit kicks in at 19 pixels for the top and left edges, for + * the 16 pixels plus 3 taps right of the central pixel when subpel + * filtering. The bottom and right edges use 16 pixels plus 2 pixels + * left of the central pixel when filtering. + */ + if (mv->col < (xd->mb_to_left_edge - (19 << 3))) + mv->col = xd->mb_to_left_edge - (16 << 3); + else if (mv->col > xd->mb_to_right_edge + (18 << 3)) + mv->col = xd->mb_to_right_edge + (16 << 3); + + if (mv->row < (xd->mb_to_top_edge - (19 << 3))) + mv->row = xd->mb_to_top_edge - (16 << 3); + else if (mv->row > xd->mb_to_bottom_edge + (18 << 3)) + mv->row = xd->mb_to_bottom_edge + (16 << 3); +} + +/* A version of the above function for chroma block MVs.*/ +static void clamp_uvmv_to_umv_border(MV *mv, const MACROBLOCKD *xd) +{ + mv->col = (2*mv->col < (xd->mb_to_left_edge - (19 << 3))) ? (xd->mb_to_left_edge - (16 << 3)) >> 1 : mv->col; + mv->col = (2*mv->col > xd->mb_to_right_edge + (18 << 3)) ? (xd->mb_to_right_edge + (16 << 3)) >> 1 : mv->col; + + mv->row = (2*mv->row < (xd->mb_to_top_edge - (19 << 3))) ? (xd->mb_to_top_edge - (16 << 3)) >> 1 : mv->row; + mv->row = (2*mv->row > xd->mb_to_bottom_edge + (18 << 3)) ? (xd->mb_to_bottom_edge + (16 << 3)) >> 1 : mv->row; +} + +static void clamp_mvs(MACROBLOCKD *xd) +{ + if (xd->mode_info_context->mbmi.mode == SPLITMV) + { + int i; + + for (i=0; i<16; i++) + clamp_mv_to_umv_border(&xd->block[i].bmi.mv.as_mv, xd); + for (i=16; i<24; i++) + clamp_uvmv_to_umv_border(&xd->block[i].bmi.mv.as_mv, xd); + } + else + { + clamp_mv_to_umv_border(&xd->mode_info_context->mbmi.mv.as_mv, xd); + clamp_uvmv_to_umv_border(&xd->block[16].bmi.mv.as_mv, xd); + } + +} + +void vp8_decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd) { - if (xd->frame_type == KEY_FRAME || xd->mbmi.ref_frame == INTRA_FRAME) + int eobtotal = 0; + int i, do_clamp = xd->mode_info_context->mbmi.need_to_clamp_mvs; + + if (xd->mode_info_context->mbmi.mb_skip_coeff) + { + vp8_reset_mb_tokens_context(xd); + } + else + { + eobtotal = vp8_decode_mb_tokens(pbi, xd); + } + + /* Perform temporary clamping of the MV to be used for prediction */ + if (do_clamp) + { + clamp_mvs(xd); + } + + xd->mode_info_context->mbmi.dc_diff = 1; + + if (xd->mode_info_context->mbmi.mode != B_PRED && xd->mode_info_context->mbmi.mode != SPLITMV && eobtotal == 0) + { + xd->mode_info_context->mbmi.dc_diff = 0; + skip_recon_mb(pbi, xd); + return; + } + + if (xd->segmentation_enabled) + mb_init_dequantizer(pbi, xd); + + // do prediction + if (xd->frame_type == KEY_FRAME || xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) { vp8_build_intra_predictors_mbuv(xd); - if (xd->mbmi.mode != B_PRED) + if (xd->mode_info_context->mbmi.mode != B_PRED) { vp8_build_intra_predictors_mby_ptr(xd); - vp8_recon16x16mb(RTCD_VTABLE(recon), xd); - } - else - { - vp8_recon_intra4x4mb(RTCD_VTABLE(recon), xd); + } else { + vp8_intra_prediction_down_copy(xd); } } else { vp8_build_inter_predictors_mb(xd); - vp8_recon16x16mb(RTCD_VTABLE(recon), xd); } -} - - -static void de_quantand_idct(VP8D_COMP *pbi, MACROBLOCKD *xd) -{ - int i; - BLOCKD *b = &xd->block[24]; - - if (xd->mbmi.mode != B_PRED && xd->mbmi.mode != SPLITMV) + // dequantization and idct + if (xd->mode_info_context->mbmi.mode != B_PRED && xd->mode_info_context->mbmi.mode != SPLITMV) { + BLOCKD *b = &xd->block[24]; DEQUANT_INVOKE(&pbi->dequant, block)(b); // do 2nd order transform on the dc block - if (b->eob > 1) + if (xd->eobs[24] > 1) { IDCT_INVOKE(RTCD_VTABLE(idct), iwalsh16)(&b->dqcoeff[0], b->diff); ((int *)b->qcoeff)[0] = 0; @@ -178,84 +251,47 @@ static void de_quantand_idct(VP8D_COMP *pbi, MACROBLOCKD *xd) ((int *)b->qcoeff)[0] = 0; } - - for (i = 0; i < 16; i++) - { - - b = &xd->block[i]; - - if (b->eob > 1) - { - DEQUANT_INVOKE(&pbi->dequant, idct_dc)(b->qcoeff, &b->dequant[0][0], b->diff, 32, xd->block[24].diff[i]); - } - else - { - IDCT_INVOKE(RTCD_VTABLE(idct), idct1_scalar)(xd->block[24].diff[i], b->diff, 32); - } - } - - for (i = 16; i < 24; i++) - { - b = &xd->block[i]; - - if (b->eob > 1) - { - DEQUANT_INVOKE(&pbi->dequant, idct)(b->qcoeff, &b->dequant[0][0], b->diff, 16); - } - else - { - IDCT_INVOKE(RTCD_VTABLE(idct), idct1_scalar)(b->qcoeff[0] * b->dequant[0][0], b->diff, 16); - ((int *)b->qcoeff)[0] = 0; - } - } + DEQUANT_INVOKE (&pbi->dequant, dc_idct_add_y_block) + (xd->qcoeff, &xd->block[0].dequant[0][0], + xd->predictor, xd->dst.y_buffer, + xd->dst.y_stride, xd->eobs, xd->block[24].diff); } - else + else if ((xd->frame_type == KEY_FRAME || xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) && xd->mode_info_context->mbmi.mode == B_PRED) { - for (i = 0; i < 24; i++) + for (i = 0; i < 16; i++) { - b = &xd->block[i]; + BLOCKD *b = &xd->block[i]; + vp8_predict_intra4x4(b, b->bmi.mode, b->predictor); - if (b->eob > 1) + if (xd->eobs[i] > 1) { - DEQUANT_INVOKE(&pbi->dequant, idct)(b->qcoeff, &b->dequant[0][0], b->diff, (32 - (i & 16))); + DEQUANT_INVOKE(&pbi->dequant, idct_add) + (b->qcoeff, &b->dequant[0][0], b->predictor, + *(b->base_dst) + b->dst, 16, b->dst_stride); } else { - IDCT_INVOKE(RTCD_VTABLE(idct), idct1_scalar)(b->qcoeff[0] * b->dequant[0][0], b->diff, (32 - (i & 16))); + IDCT_INVOKE(RTCD_VTABLE(idct), idct1_scalar_add) + (b->qcoeff[0] * b->dequant[0][0], b->predictor, + *(b->base_dst) + b->dst, 16, b->dst_stride); ((int *)b->qcoeff)[0] = 0; } } - } -} - -void vp8_decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd) -{ - int eobtotal = 0; - if (xd->mbmi.mb_skip_coeff) - { - vp8_reset_mb_tokens_context(xd); } else { - eobtotal = vp8_decode_mb_tokens(pbi, xd); - } - - xd->mode_info_context->mbmi.dc_diff = 1; - - if (xd->mbmi.mode != B_PRED && xd->mbmi.mode != SPLITMV && eobtotal == 0) - { - xd->mode_info_context->mbmi.dc_diff = 0; - skip_recon_mb(pbi, xd); - return; + DEQUANT_INVOKE (&pbi->dequant, idct_add_y_block) + (xd->qcoeff, &xd->block[0].dequant[0][0], + xd->predictor, xd->dst.y_buffer, + xd->dst.y_stride, xd->eobs); } - if (xd->segmentation_enabled) - mb_init_dequantizer(pbi, xd); - - de_quantand_idct(pbi, xd); - reconstruct_mb(pbi, xd); + DEQUANT_INVOKE (&pbi->dequant, idct_add_uv_block) + (xd->qcoeff+16*16, &xd->block[16].dequant[0][0], + xd->predictor+16*16, xd->dst.u_buffer, xd->dst.v_buffer, + xd->dst.uv_stride, xd->eobs+16); } static int get_delta_q(vp8_reader *bc, int prev, int *q_update) @@ -293,18 +329,17 @@ void vp8_decode_mb_row(VP8D_COMP *pbi, int i; int recon_yoffset, recon_uvoffset; int mb_col; - int recon_y_stride = pc->last_frame.y_stride; - int recon_uv_stride = pc->last_frame.uv_stride; + int ref_fb_idx = pc->lst_fb_idx; + int dst_fb_idx = pc->new_fb_idx; + int recon_y_stride = pc->yv12_fb[ref_fb_idx].y_stride; + int recon_uv_stride = pc->yv12_fb[ref_fb_idx].uv_stride; - vpx_memset(pc->left_context, 0, sizeof(pc->left_context)); + vpx_memset(&pc->left_context, 0, sizeof(pc->left_context)); recon_yoffset = mb_row * recon_y_stride * 16; recon_uvoffset = mb_row * recon_uv_stride * 8; // reset above block coeffs - xd->above_context[Y1CONTEXT] = pc->above_context[Y1CONTEXT]; - xd->above_context[UCONTEXT ] = pc->above_context[UCONTEXT]; - xd->above_context[VCONTEXT ] = pc->above_context[VCONTEXT]; - xd->above_context[Y2CONTEXT] = pc->above_context[Y2CONTEXT]; + xd->above_context = pc->above_context; xd->up_available = (mb_row != 0); xd->mb_to_top_edge = -((mb_row * 16)) << 3; @@ -312,10 +347,8 @@ void vp8_decode_mb_row(VP8D_COMP *pbi, for (mb_col = 0; mb_col < pc->mb_cols; mb_col++) { - // Take a copy of the mode and Mv information for this macroblock into the xd->mbmi - vpx_memcpy(&xd->mbmi, &xd->mode_info_context->mbmi, 32); //sizeof(MB_MODE_INFO) ); - if (xd->mbmi.mode == SPLITMV || xd->mbmi.mode == B_PRED) + if (xd->mode_info_context->mbmi.mode == SPLITMV || xd->mode_info_context->mbmi.mode == B_PRED) { for (i = 0; i < 16; i++) { @@ -329,43 +362,32 @@ void vp8_decode_mb_row(VP8D_COMP *pbi, xd->mb_to_left_edge = -((mb_col * 16) << 3); xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3; - xd->dst.y_buffer = pc->new_frame.y_buffer + recon_yoffset; - xd->dst.u_buffer = pc->new_frame.u_buffer + recon_uvoffset; - xd->dst.v_buffer = pc->new_frame.v_buffer + recon_uvoffset; + xd->dst.y_buffer = pc->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset; + xd->dst.u_buffer = pc->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset; + xd->dst.v_buffer = pc->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset; xd->left_available = (mb_col != 0); // Select the appropriate reference frame for this MB - if (xd->mbmi.ref_frame == LAST_FRAME) - { - xd->pre.y_buffer = pc->last_frame.y_buffer + recon_yoffset; - xd->pre.u_buffer = pc->last_frame.u_buffer + recon_uvoffset; - xd->pre.v_buffer = pc->last_frame.v_buffer + recon_uvoffset; - } - else if (xd->mbmi.ref_frame == GOLDEN_FRAME) - { - // Golden frame reconstruction buffer - xd->pre.y_buffer = pc->golden_frame.y_buffer + recon_yoffset; - xd->pre.u_buffer = pc->golden_frame.u_buffer + recon_uvoffset; - xd->pre.v_buffer = pc->golden_frame.v_buffer + recon_uvoffset; - } + if (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME) + ref_fb_idx = pc->lst_fb_idx; + else if (xd->mode_info_context->mbmi.ref_frame == GOLDEN_FRAME) + ref_fb_idx = pc->gld_fb_idx; else - { - // Alternate reference frame reconstruction buffer - xd->pre.y_buffer = pc->alt_ref_frame.y_buffer + recon_yoffset; - xd->pre.u_buffer = pc->alt_ref_frame.u_buffer + recon_uvoffset; - xd->pre.v_buffer = pc->alt_ref_frame.v_buffer + recon_uvoffset; - } + ref_fb_idx = pc->alt_fb_idx; + + xd->pre.y_buffer = pc->yv12_fb[ref_fb_idx].y_buffer + recon_yoffset; + xd->pre.u_buffer = pc->yv12_fb[ref_fb_idx].u_buffer + recon_uvoffset; + xd->pre.v_buffer = pc->yv12_fb[ref_fb_idx].v_buffer + recon_uvoffset; vp8_build_uvmvs(xd, pc->full_pixel); /* - if(pbi->common.current_video_frame==0 &&mb_col==1 && mb_row==0) + if(pc->current_video_frame==0 &&mb_col==1 && mb_row==0) pbi->debugoutput =1; else pbi->debugoutput =0; */ - vp8dx_bool_decoder_fill(xd->current_bc); vp8_decode_macroblock(pbi, xd); @@ -374,19 +396,14 @@ void vp8_decode_mb_row(VP8D_COMP *pbi, ++xd->mode_info_context; /* next mb */ - xd->gf_active_ptr++; // GF useage flag for next MB - - xd->above_context[Y1CONTEXT] += 4; - xd->above_context[UCONTEXT ] += 2; - xd->above_context[VCONTEXT ] += 2; - xd->above_context[Y2CONTEXT] ++; + xd->above_context++; pbi->current_mb_col_main = mb_col; } // adjust to the next row of mbs vp8_extend_mb_row( - &pc->new_frame, + &pc->yv12_fb[dst_fb_idx], xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8 ); @@ -473,18 +490,7 @@ static void stop_token_decoder(VP8D_COMP *pbi) VP8_COMMON *pc = &pbi->common; if (pc->multi_token_partition != ONE_PARTITION) - { - int num_part = (1 << pc->multi_token_partition); - - for (i = 0; i < num_part; i++) - { - vp8dx_stop_decode(&pbi->mbc[i]); - } - vpx_free(pbi->mbc); - } - else - vp8dx_stop_decode(& pbi->bc2); } static void init_frame(VP8D_COMP *pbi) @@ -545,10 +551,10 @@ static void init_frame(VP8D_COMP *pbi) } } - xd->left_context = pc->left_context; + xd->left_context = &pc->left_context; xd->mode_info_context = pc->mi; xd->frame_type = pc->frame_type; - xd->mbmi.mode = DC_PRED; + xd->mode_info_context->mbmi.mode = DC_PRED; xd->mode_info_stride = pc->mode_info_stride; } @@ -609,7 +615,7 @@ int vp8_decode_frame(VP8D_COMP *pbi) "Invalid frame height"); } - if (vp8_alloc_frame_buffers(&pbi->common, pc->Width, pc->Height)) + if (vp8_alloc_frame_buffers(pc, pc->Width, pc->Height)) vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR, "Failed to allocate frame buffers"); } @@ -793,7 +799,6 @@ int vp8_decode_frame(VP8D_COMP *pbi) } - vp8dx_bool_decoder_fill(bc); { // read coef probability tree @@ -813,11 +818,11 @@ int vp8_decode_frame(VP8D_COMP *pbi) } } - vpx_memcpy(&xd->pre, &pc->last_frame, sizeof(YV12_BUFFER_CONFIG)); - vpx_memcpy(&xd->dst, &pc->new_frame, sizeof(YV12_BUFFER_CONFIG)); + vpx_memcpy(&xd->pre, &pc->yv12_fb[pc->lst_fb_idx], sizeof(YV12_BUFFER_CONFIG)); + vpx_memcpy(&xd->dst, &pc->yv12_fb[pc->new_fb_idx], sizeof(YV12_BUFFER_CONFIG)); // set up frame new frame for intra coded blocks - vp8_setup_intra_recon(&pc->new_frame); + vp8_setup_intra_recon(&pc->yv12_fb[pc->new_fb_idx]); vp8_setup_block_dptrs(xd); @@ -829,34 +834,25 @@ int vp8_decode_frame(VP8D_COMP *pbi) // Read the mb_no_coeff_skip flag pc->mb_no_coeff_skip = (int)vp8_read_bit(bc); - if (pc->frame_type == KEY_FRAME) - vp8_kfread_modes(pbi); - else - vp8_decode_mode_mvs(pbi); - - // reset since these guys are used as iterators - vpx_memset(pc->above_context[Y1CONTEXT], 0, sizeof(ENTROPY_CONTEXT) * pc->mb_cols * 4); - vpx_memset(pc->above_context[UCONTEXT ], 0, sizeof(ENTROPY_CONTEXT) * pc->mb_cols * 2); - vpx_memset(pc->above_context[VCONTEXT ], 0, sizeof(ENTROPY_CONTEXT) * pc->mb_cols * 2); - vpx_memset(pc->above_context[Y2CONTEXT], 0, sizeof(ENTROPY_CONTEXT) * pc->mb_cols); - xd->gf_active_ptr = (signed char *)pc->gf_active_flags; // Point to base of GF active flags data structure + vp8_decode_mode_mvs(pbi); + vpx_memset(pc->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES) * pc->mb_cols); vpx_memcpy(&xd->block[0].bmi, &xd->mode_info_context->bmi[0], sizeof(B_MODE_INFO)); - if (pbi->b_multithreaded_lf && pbi->common.filter_level != 0) + if (pbi->b_multithreaded_lf && pc->filter_level != 0) vp8_start_lfthread(pbi); - if (pbi->b_multithreaded_rd && pbi->common.multi_token_partition != ONE_PARTITION) + if (pbi->b_multithreaded_rd && pc->multi_token_partition != ONE_PARTITION) { vp8_mtdecode_mb_rows(pbi, xd); } else { int ibc = 0; - int num_part = 1 << pbi->common.multi_token_partition; + int num_part = 1 << pc->multi_token_partition; // Decode the individual macro block for (mb_row = 0; mb_row < pc->mb_rows; mb_row++) @@ -880,13 +876,14 @@ int vp8_decode_frame(VP8D_COMP *pbi) stop_token_decoder(pbi); - vp8dx_stop_decode(bc); - // vpx_log("Decoder: Frame Decoded, Size Roughly:%d bytes \n",bc->pos+pbi->bc2.pos); // If this was a kf or Gf note the Q used - if ((pc->frame_type == KEY_FRAME) || (pc->refresh_golden_frame) || pbi->common.refresh_alt_ref_frame) + if ((pc->frame_type == KEY_FRAME) || + pc->refresh_golden_frame || pc->refresh_alt_ref_frame) + { pc->last_kf_gf_q = pc->base_qindex; + } if (pc->refresh_entropy_probs == 0) { diff --git a/vp8/decoder/demode.c b/vp8/decoder/demode.c deleted file mode 100644 index fd05e6d..0000000 --- a/vp8/decoder/demode.c +++ /dev/null @@ -1,149 +0,0 @@ -/* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. - * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. - */ - - -#include "onyxd_int.h" -#include "entropymode.h" -#include "findnearmv.h" - - -int vp8_read_bmode(vp8_reader *bc, const vp8_prob *p) -{ - const int i = vp8_treed_read(bc, vp8_bmode_tree, p); - - return i; -} - - -int vp8_read_ymode(vp8_reader *bc, const vp8_prob *p) -{ - const int i = vp8_treed_read(bc, vp8_ymode_tree, p); - - return i; -} - -int vp8_kfread_ymode(vp8_reader *bc, const vp8_prob *p) -{ - const int i = vp8_treed_read(bc, vp8_kf_ymode_tree, p); - - return i; -} - - - -int vp8_read_uv_mode(vp8_reader *bc, const vp8_prob *p) -{ - const int i = vp8_treed_read(bc, vp8_uv_mode_tree, p); - - return i; -} - -void vp8_read_mb_features(vp8_reader *r, MB_MODE_INFO *mi, MACROBLOCKD *x) -{ - // Is segmentation enabled - if (x->segmentation_enabled && x->update_mb_segmentation_map) - { - // If so then read the segment id. - if (vp8_read(r, x->mb_segment_tree_probs[0])) - mi->segment_id = (unsigned char)(2 + vp8_read(r, x->mb_segment_tree_probs[2])); - else - mi->segment_id = (unsigned char)(vp8_read(r, x->mb_segment_tree_probs[1])); - } -} - -void vp8_kfread_modes(VP8D_COMP *pbi) -{ - VP8_COMMON *const cp = & pbi->common; - vp8_reader *const bc = & pbi->bc; - - MODE_INFO *m = cp->mi; - const int ms = cp->mode_info_stride; - - int mb_row = -1; - vp8_prob prob_skip_false = 0; - - if (cp->mb_no_coeff_skip) - prob_skip_false = (vp8_prob)(vp8_read_literal(bc, 8)); - - while (++mb_row < cp->mb_rows) - { - int mb_col = -1; - - while (++mb_col < cp->mb_cols) - { - MB_PREDICTION_MODE y_mode; - - vp8dx_bool_decoder_fill(bc); - // Read the Macroblock segmentation map if it is being updated explicitly this frame (reset to 0 above by default) - // By default on a key frame reset all MBs to segment 0 - m->mbmi.segment_id = 0; - - if (pbi->mb.update_mb_segmentation_map) - vp8_read_mb_features(bc, &m->mbmi, &pbi->mb); - - // Read the macroblock coeff skip flag if this feature is in use, else default to 0 - if (cp->mb_no_coeff_skip) - m->mbmi.mb_skip_coeff = vp8_read(bc, prob_skip_false); - else - m->mbmi.mb_skip_coeff = 0; - - y_mode = (MB_PREDICTION_MODE) vp8_kfread_ymode(bc, cp->kf_ymode_prob); - - m->mbmi.ref_frame = INTRA_FRAME; - - if ((m->mbmi.mode = y_mode) == B_PRED) - { - int i = 0; - - do - { - const B_PREDICTION_MODE A = vp8_above_bmi(m, i, ms)->mode; - const B_PREDICTION_MODE L = vp8_left_bmi(m, i)->mode; - - m->bmi[i].mode = (B_PREDICTION_MODE) vp8_read_bmode(bc, cp->kf_bmode_prob [A] [L]); - } - while (++i < 16); - } - else - { - int BMode; - int i = 0; - - switch (y_mode) - { - case DC_PRED: - BMode = B_DC_PRED; - break; - case V_PRED: - BMode = B_VE_PRED; - break; - case H_PRED: - BMode = B_HE_PRED; - break; - case TM_PRED: - BMode = B_TM_PRED; - break; - default: - BMode = B_DC_PRED; - break; - } - - do - { - m->bmi[i].mode = (B_PREDICTION_MODE)BMode; - } - while (++i < 16); - } - - (m++)->mbmi.uv_mode = (MB_PREDICTION_MODE)vp8_read_uv_mode(bc, cp->kf_uv_mode_prob); - } - - m++; // skip the border - } -} diff --git a/vp8/decoder/demode.h b/vp8/decoder/demode.h deleted file mode 100644 index 51bbc5e..0000000 --- a/vp8/decoder/demode.h +++ /dev/null @@ -1,32 +0,0 @@ -/* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. - * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. - */ - - -#include "onyxd_int.h" - -/* Read (intra) modes for all blocks in a keyframe */ - -void vp8_kfread_modes(VP8D_COMP *pbi); - -/* Intra mode for a Y subblock */ - -int vp8_read_bmode(vp8_reader *, const vp8_prob *); - -/* MB intra Y mode trees differ for key and inter frames. */ - -int vp8_read_ymode(vp8_reader *, const vp8_prob *); -int vp8_kfread_ymode(vp8_reader *, const vp8_prob *); - -/* MB intra UV mode trees are the same for key and inter frames. */ - -int vp8_read_uv_mode(vp8_reader *, const vp8_prob *); - -/* Read any macroblock-level features that may be present. */ - -void vp8_read_mb_features(vp8_reader *, MB_MODE_INFO *, MACROBLOCKD *); diff --git a/vp8/decoder/dequantize.c b/vp8/decoder/dequantize.c index 14798d9..8cfa2a3 100644 --- a/vp8/decoder/dequantize.c +++ b/vp8/decoder/dequantize.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -31,8 +32,12 @@ void vp8_dequantize_b_c(BLOCKD *d) } } -void vp8_dequant_idct_c(short *input, short *dq, short *output, int pitch) +void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *pred, + unsigned char *dest, int pitch, int stride) { + short output[16]; + short *diff_ptr = output; + int r, c; int i; for (i = 0; i < 16; i++) @@ -40,13 +45,40 @@ void vp8_dequant_idct_c(short *input, short *dq, short *output, int pitch) input[i] = dq[i] * input[i]; } - vp8_short_idct4x4llm_c(input, output, pitch); + // the idct halves ( >> 1) the pitch + vp8_short_idct4x4llm_c(input, output, 4 << 1); + vpx_memset(input, 0, 32); + + for (r = 0; r < 4; r++) + { + for (c = 0; c < 4; c++) + { + int a = diff_ptr[c] + pred[c]; + + if (a < 0) + a = 0; + + if (a > 255) + a = 255; + + dest[c] = (unsigned char) a; + } + + dest += stride; + diff_ptr += 4; + pred += pitch; + } } -void vp8_dequant_dc_idct_c(short *input, short *dq, short *output, int pitch, int Dc) +void vp8_dequant_dc_idct_add_c(short *input, short *dq, unsigned char *pred, + unsigned char *dest, int pitch, int stride, + int Dc) { int i; + short output[16]; + short *diff_ptr = output; + int r, c; input[0] = (short)Dc; @@ -55,6 +87,28 @@ void vp8_dequant_dc_idct_c(short *input, short *dq, short *output, int pitch, in input[i] = dq[i] * input[i]; } - vp8_short_idct4x4llm_c(input, output, pitch); + // the idct halves ( >> 1) the pitch + vp8_short_idct4x4llm_c(input, output, 4 << 1); + vpx_memset(input, 0, 32); + + for (r = 0; r < 4; r++) + { + for (c = 0; c < 4; c++) + { + int a = diff_ptr[c] + pred[c]; + + if (a < 0) + a = 0; + + if (a > 255) + a = 255; + + dest[c] = (unsigned char) a; + } + + dest += stride; + diff_ptr += 4; + pred += pitch; + } } diff --git a/vp8/decoder/dequantize.h b/vp8/decoder/dequantize.h index d16b02e..b78e39c 100644 --- a/vp8/decoder/dequantize.h +++ b/vp8/decoder/dequantize.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -15,11 +16,31 @@ #define prototype_dequant_block(sym) \ void sym(BLOCKD *x) -#define prototype_dequant_idct(sym) \ - void sym(short *input, short *dq, short *output, int pitch) +#define prototype_dequant_idct_add(sym) \ + void sym(short *input, short *dq, \ + unsigned char *pred, unsigned char *output, \ + int pitch, int stride) -#define prototype_dequant_idct_dc(sym) \ - void sym(short *input, short *dq, short *output, int pitch, int dc) +#define prototype_dequant_dc_idct_add(sym) \ + void sym(short *input, short *dq, \ + unsigned char *pred, unsigned char *output, \ + int pitch, int stride, \ + int dc) + +#define prototype_dequant_dc_idct_add_y_block(sym) \ + void sym(short *q, short *dq, \ + unsigned char *pre, unsigned char *dst, \ + int stride, char *eobs, short *dc) + +#define prototype_dequant_idct_add_y_block(sym) \ + void sym(short *q, short *dq, \ + unsigned char *pre, unsigned char *dst, \ + int stride, char *eobs) + +#define prototype_dequant_idct_add_uv_block(sym) \ + void sym(short *q, short *dq, \ + unsigned char *pre, unsigned char *dst_u, \ + unsigned char *dst_v, int stride, char *eobs) #if ARCH_X86 || ARCH_X86_64 #include "x86/dequantize_x86.h" @@ -34,25 +55,52 @@ #endif extern prototype_dequant_block(vp8_dequant_block); -#ifndef vp8_dequant_idct -#define vp8_dequant_idct vp8_dequant_idct_c +#ifndef vp8_dequant_idct_add +#define vp8_dequant_idct_add vp8_dequant_idct_add_c +#endif +extern prototype_dequant_idct_add(vp8_dequant_idct_add); + +#ifndef vp8_dequant_dc_idct_add +#define vp8_dequant_dc_idct_add vp8_dequant_dc_idct_add_c +#endif +extern prototype_dequant_dc_idct_add(vp8_dequant_dc_idct_add); + +#ifndef vp8_dequant_dc_idct_add_y_block +#define vp8_dequant_dc_idct_add_y_block vp8_dequant_dc_idct_add_y_block_c #endif -extern prototype_dequant_idct(vp8_dequant_idct); +extern prototype_dequant_dc_idct_add_y_block(vp8_dequant_dc_idct_add_y_block); -#ifndef vp8_dequant_idct_dc -#define vp8_dequant_idct_dc vp8_dequant_dc_idct_c +#ifndef vp8_dequant_idct_add_y_block +#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_c #endif -extern prototype_dequant_idct_dc(vp8_dequant_idct_dc); +extern prototype_dequant_idct_add_y_block(vp8_dequant_idct_add_y_block); + +#ifndef vp8_dequant_idct_add_uv_block +#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_c +#endif +extern prototype_dequant_idct_add_uv_block(vp8_dequant_idct_add_uv_block); typedef prototype_dequant_block((*vp8_dequant_block_fn_t)); -typedef prototype_dequant_idct((*vp8_dequant_idct_fn_t)); -typedef prototype_dequant_idct_dc((*vp8_dequant_idct_dc_fn_t)); + +typedef prototype_dequant_idct_add((*vp8_dequant_idct_add_fn_t)); + +typedef prototype_dequant_dc_idct_add((*vp8_dequant_dc_idct_add_fn_t)); + +typedef prototype_dequant_dc_idct_add_y_block((*vp8_dequant_dc_idct_add_y_block_fn_t)); + +typedef prototype_dequant_idct_add_y_block((*vp8_dequant_idct_add_y_block_fn_t)); + +typedef prototype_dequant_idct_add_uv_block((*vp8_dequant_idct_add_uv_block_fn_t)); + typedef struct { - vp8_dequant_block_fn_t block; - vp8_dequant_idct_fn_t idct; - vp8_dequant_idct_dc_fn_t idct_dc; + vp8_dequant_block_fn_t block; + vp8_dequant_idct_add_fn_t idct_add; + vp8_dequant_dc_idct_add_fn_t dc_idct_add; + vp8_dequant_dc_idct_add_y_block_fn_t dc_idct_add_y_block; + vp8_dequant_idct_add_y_block_fn_t idct_add_y_block; + vp8_dequant_idct_add_uv_block_fn_t idct_add_uv_block; } vp8_dequant_rtcd_vtable_t; #if CONFIG_RUNTIME_CPU_DETECT diff --git a/vp8/decoder/detokenize.c b/vp8/decoder/detokenize.c index a42f18d..65c7d53 100644 --- a/vp8/decoder/detokenize.c +++ b/vp8/decoder/detokenize.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -13,12 +14,12 @@ #include "onyxd_int.h" #include "vpx_mem/vpx_mem.h" #include "vpx_ports/mem.h" +#include "detokenize.h" -#define BR_COUNT 8 #define BOOL_DATA UINT8 #define OCB_X PREV_COEF_CONTEXTS * ENTROPY_NODES -DECLARE_ALIGNED(16, UINT16, vp8_coef_bands_x[16]) = { 0, 1 * OCB_X, 2 * OCB_X, 3 * OCB_X, 6 * OCB_X, 4 * OCB_X, 5 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 7 * OCB_X}; +DECLARE_ALIGNED(16, UINT8, vp8_coef_bands_x[16]) = { 0, 1 * OCB_X, 2 * OCB_X, 3 * OCB_X, 6 * OCB_X, 4 * OCB_X, 5 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 6 * OCB_X, 7 * OCB_X}; #define EOB_CONTEXT_NODE 0 #define ZERO_CONTEXT_NODE 1 #define ONE_CONTEXT_NODE 2 @@ -60,32 +61,54 @@ DECLARE_ALIGNED(16, static const TOKENEXTRABITS, vp8d_token_extra_bits2[MAX_ENTR void vp8_reset_mb_tokens_context(MACROBLOCKD *x) { - ENTROPY_CONTEXT **const A = x->above_context; - ENTROPY_CONTEXT(* const L)[4] = x->left_context; - - ENTROPY_CONTEXT *a; - ENTROPY_CONTEXT *l; - int i; - - for (i = 0; i < 24; i++) + /* Clear entropy contexts for Y2 blocks */ + if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV) { - - a = A[ vp8_block2context[i] ] + vp8_block2above[i]; - l = L[ vp8_block2context[i] ] + vp8_block2left[i]; - - *a = *l = 0; + vpx_memset(x->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memset(x->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)); } - - if (x->mbmi.mode != B_PRED && x->mbmi.mode != SPLITMV) + else { - a = A[Y2CONTEXT] + vp8_block2above[24]; - l = L[Y2CONTEXT] + vp8_block2left[24]; - *a = *l = 0; + vpx_memset(x->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1); + vpx_memset(x->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1); } +} +#if CONFIG_ARM_ASM_DETOK +// mashup of vp8_block2left and vp8_block2above so we only need one pointer +// for the assembly version. +DECLARE_ALIGNED(16, const UINT8, vp8_block2leftabove[25*2]) = +{ + //vp8_block2left + 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, + //vp8_block2above + 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 0, 1, 2, 3, 4, 5, 4, 5, 6, 7, 6, 7, 8 +}; +void vp8_init_detokenizer(VP8D_COMP *dx) +{ + const VP8_COMMON *const oc = & dx->common; + MACROBLOCKD *x = & dx->mb; + + dx->detoken.vp8_coef_tree_ptr = vp8_coef_tree; + dx->detoken.ptr_block2leftabove = vp8_block2leftabove; + dx->detoken.ptr_coef_bands_x = vp8_coef_bands_x; + dx->detoken.scan = vp8_default_zig_zag1d; + dx->detoken.teb_base_ptr = vp8d_token_extra_bits2; + dx->detoken.qcoeff_start_ptr = &x->qcoeff[0]; + + dx->detoken.coef_probs[0] = (oc->fc.coef_probs [0] [ 0 ] [0]); + dx->detoken.coef_probs[1] = (oc->fc.coef_probs [1] [ 0 ] [0]); + dx->detoken.coef_probs[2] = (oc->fc.coef_probs [2] [ 0 ] [0]); + dx->detoken.coef_probs[3] = (oc->fc.coef_probs [3] [ 0 ] [0]); } -DECLARE_ALIGNED(16, extern const unsigned int, vp8dx_bitreader_norm[256]); +#endif + +DECLARE_ALIGNED(16, extern const unsigned char, vp8dx_bitreader_norm[256]); +#define FILL \ + if(count < 0) \ + VP8DX_BOOL_DECODER_FILL(count, value, bufptr, bufend); + #define NORMALIZE \ /*if(range < 0x80)*/ \ { \ @@ -93,17 +116,13 @@ DECLARE_ALIGNED(16, extern const unsigned int, vp8dx_bitreader_norm[256]); range <<= shift; \ value <<= shift; \ count -= shift; \ - if(count <= 0) \ - { \ - count += BR_COUNT ; \ - value |= (*bufptr) << (BR_COUNT-count); \ - bufptr = br_ptr_advance(bufptr, 1); \ - } \ } #define DECODE_AND_APPLYSIGN(value_to_sign) \ split = (range + 1) >> 1; \ - if ( (value >> 8) < split ) \ + bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8); \ + FILL \ + if ( value < bigsplit ) \ { \ range = split; \ v= value_to_sign; \ @@ -111,28 +130,25 @@ DECLARE_ALIGNED(16, extern const unsigned int, vp8dx_bitreader_norm[256]); else \ { \ range = range-split; \ - value = value-(split<<8); \ + value = value-bigsplit; \ v = -value_to_sign; \ } \ range +=range; \ value +=value; \ - if (!--count) \ - { \ - count = BR_COUNT; \ - value |= *bufptr; \ - bufptr = br_ptr_advance(bufptr, 1); \ - } + count--; #define DECODE_AND_BRANCH_IF_ZERO(probability,branch) \ { \ split = 1 + ((( probability*(range-1) ) )>> 8); \ - if ( (value >> 8) < split ) \ + bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8); \ + FILL \ + if ( value < bigsplit ) \ { \ range = split; \ NORMALIZE \ goto branch; \ } \ - value -= (split<<8); \ + value -= bigsplit; \ range = range - split; \ NORMALIZE \ } @@ -140,7 +156,9 @@ DECLARE_ALIGNED(16, extern const unsigned int, vp8dx_bitreader_norm[256]); #define DECODE_AND_LOOP_IF_ZERO(probability,branch) \ { \ split = 1 + ((( probability*(range-1) ) ) >> 8); \ - if ( (value >> 8) < split ) \ + bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8); \ + FILL \ + if ( value < bigsplit ) \ { \ range = split; \ NORMALIZE \ @@ -151,7 +169,7 @@ DECLARE_ALIGNED(16, extern const unsigned int, vp8dx_bitreader_norm[256]); goto branch; \ } goto BLOCK_FINISHED; /*for malformed input */\ } \ - value -= (split<<8); \ + value -= bigsplit; \ range = range - split; \ NORMALIZE \ } @@ -169,10 +187,12 @@ DECLARE_ALIGNED(16, extern const unsigned int, vp8dx_bitreader_norm[256]); #define DECODE_EXTRABIT_AND_ADJUST_VAL(t,bits_count)\ split = 1 + (((range-1) * vp8d_token_extra_bits2[t].Probs[bits_count]) >> 8); \ - if(value >= (split<<8))\ + bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8); \ + FILL \ + if(value >= bigsplit)\ {\ range = range-split;\ - value = value-(split<<8);\ + value = value-bigsplit;\ val += ((UINT16)1<<bits_count);\ }\ else\ @@ -181,14 +201,45 @@ DECLARE_ALIGNED(16, extern const unsigned int, vp8dx_bitreader_norm[256]); }\ NORMALIZE +#if CONFIG_ARM_ASM_DETOK +int vp8_decode_mb_tokens(VP8D_COMP *dx, MACROBLOCKD *x) +{ + int eobtotal = 0; + int i, type; + + dx->detoken.current_bc = x->current_bc; + dx->detoken.A = x->above_context; + dx->detoken.L = x->left_context; + + type = 3; + + if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV) + { + type = 1; + eobtotal -= 16; + } + + vp8_decode_mb_tokens_v6(&dx->detoken, type); + + for (i = 0; i < 25; i++) + { + x->eobs[i] = dx->detoken.eob[i]; + eobtotal += dx->detoken.eob[i]; + } + + return eobtotal; +} +#else int vp8_decode_mb_tokens(VP8D_COMP *dx, MACROBLOCKD *x) { - ENTROPY_CONTEXT **const A = x->above_context; - ENTROPY_CONTEXT(* const L)[4] = x->left_context; + ENTROPY_CONTEXT *A = (ENTROPY_CONTEXT *)x->above_context; + ENTROPY_CONTEXT *L = (ENTROPY_CONTEXT *)x->left_context; const VP8_COMMON *const oc = & dx->common; BOOL_DECODER *bc = x->current_bc; + char *eobs = x->eobs; + ENTROPY_CONTEXT *a; ENTROPY_CONTEXT *l; int i; @@ -198,11 +249,13 @@ int vp8_decode_mb_tokens(VP8D_COMP *dx, MACROBLOCKD *x) register int count; const BOOL_DATA *bufptr; + const BOOL_DATA *bufend; register unsigned int range; - register unsigned int value; + VP8_BD_VALUE value; const int *scan; register unsigned int shift; UINT32 split; + VP8_BD_VALUE bigsplit; INT16 *qcoeff_ptr; const vp8_prob *coef_probs; @@ -210,46 +263,44 @@ int vp8_decode_mb_tokens(VP8D_COMP *dx, MACROBLOCKD *x) int stop; INT16 val, bits_count; INT16 c; - INT16 t; INT16 v; const vp8_prob *Prob; - //int *scan; type = 3; i = 0; stop = 16; - if (x->mbmi.mode != B_PRED && x->mbmi.mode != SPLITMV) + scan = vp8_default_zig_zag1d; + qcoeff_ptr = &x->qcoeff[0]; + + if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV) { i = 24; stop = 24; type = 1; - qcoeff_ptr = &x->qcoeff[24*16]; - scan = vp8_default_zig_zag1d; + qcoeff_ptr += 24*16; eobtotal -= 16; } - else - { - scan = vp8_default_zig_zag1d; - qcoeff_ptr = &x->qcoeff[0]; - } + bufend = bc->user_buffer_end; + bufptr = bc->user_buffer; + value = bc->value; count = bc->count; range = bc->range; - value = bc->value; - bufptr = bc->read_ptr; coef_probs = oc->fc.coef_probs [type] [ 0 ] [0]; BLOCK_LOOP: - a = A[ vp8_block2context[i] ] + vp8_block2above[i]; - l = L[ vp8_block2context[i] ] + vp8_block2left[i]; + a = A + vp8_block2above[i]; + l = L + vp8_block2left[i]; + c = (INT16)(!type); - VP8_COMBINEENTROPYCONTEXTS(t, *a, *l); +// Dest = ((A)!=0) + ((B)!=0); + VP8_COMBINEENTROPYCONTEXTS(v, *a, *l); Prob = coef_probs; - Prob += t * ENTROPY_NODES; + Prob += v * ENTROPY_NODES; DO_WHILE: Prob += vp8_coef_bands_x[c]; @@ -336,9 +387,8 @@ ONE_CONTEXT_NODE_0_: qcoeff_ptr [ scan[15] ] = (INT16) v; BLOCK_FINISHED: - t = ((x->block[i].eob = c) != !type); // any nonzero data? - eobtotal += x->block[i].eob; - *a = *l = t; + *a = *l = ((eobs[i] = c) != !type); // any nonzero data? + eobtotal += c; qcoeff_ptr += 16; i++; @@ -348,12 +398,11 @@ BLOCK_FINISHED: if (i == 25) { - scan = vp8_default_zig_zag1d;//x->scan_order1d; type = 0; i = 0; stop = 16; coef_probs = oc->fc.coef_probs [type] [ 0 ] [0]; - qcoeff_ptr = &x->qcoeff[0]; + qcoeff_ptr -= (24*16 + 16); goto BLOCK_LOOP; } @@ -365,10 +414,12 @@ BLOCK_FINISHED: goto BLOCK_LOOP; } - bc->count = count; + FILL + bc->user_buffer = bufptr; bc->value = value; + bc->count = count; bc->range = range; - bc->read_ptr = bufptr; return eobtotal; } +#endif //!CONFIG_ASM_DETOK diff --git a/vp8/decoder/detokenize.h b/vp8/decoder/detokenize.h index 6a9a476..294a4a5 100644 --- a/vp8/decoder/detokenize.h +++ b/vp8/decoder/detokenize.h @@ -1,19 +1,24 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ -#ifndef detokenize_h -#define detokenize_h 1 +#ifndef DETOKENIZE_H +#define DETOKENIZE_H #include "onyxd_int.h" +#if ARCH_ARM +#include "arm/detokenize_arm.h" +#endif + void vp8_reset_mb_tokens_context(MACROBLOCKD *x); int vp8_decode_mb_tokens(VP8D_COMP *, MACROBLOCKD *); -#endif /* detokenize_h */ +#endif /* DETOKENIZE_H */ diff --git a/vp8/decoder/generic/dsystemdependent.c b/vp8/decoder/generic/dsystemdependent.c index 302b64b..60f2af5 100644 --- a/vp8/decoder/generic/dsystemdependent.c +++ b/vp8/decoder/generic/dsystemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -18,13 +19,15 @@ void vp8_dmachine_specific_config(VP8D_COMP *pbi) { // Pure C: #if CONFIG_RUNTIME_CPU_DETECT - pbi->mb.rtcd = &pbi->common.rtcd; - pbi->dequant.block = vp8_dequantize_b_c; - pbi->dequant.idct = vp8_dequant_idct_c; - pbi->dequant.idct_dc = vp8_dequant_dc_idct_c; - pbi->dboolhuff.start = vp8dx_start_decode_c; - pbi->dboolhuff.stop = vp8dx_stop_decode_c; - pbi->dboolhuff.fill = vp8dx_bool_decoder_fill_c; + pbi->mb.rtcd = &pbi->common.rtcd; + pbi->dequant.block = vp8_dequantize_b_c; + pbi->dequant.idct_add = vp8_dequant_idct_add_c; + pbi->dequant.dc_idct_add = vp8_dequant_dc_idct_add_c; + pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_c; + pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_c; + pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_c; + pbi->dboolhuff.start = vp8dx_start_decode_c; + pbi->dboolhuff.fill = vp8dx_bool_decoder_fill_c; #if 0 //For use with RTCD, when implemented pbi->dboolhuff.debool = vp8dx_decode_bool_c; pbi->dboolhuff.devalue = vp8dx_decode_value_c; diff --git a/vp8/decoder/idct_blk.c b/vp8/decoder/idct_blk.c new file mode 100644 index 0000000..c6a4257 --- /dev/null +++ b/vp8/decoder/idct_blk.c @@ -0,0 +1,116 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include "vpx_ports/config.h" +#include "idct.h" +#include "dequantize.h" + +void vp8_dequant_dc_idct_add_y_block_c + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int stride, char *eobs, short *dc) +{ + int i, j; + + for (i = 0; i < 4; i++) + { + for (j = 0; j < 4; j++) + { + if (*eobs++ > 1) + vp8_dequant_dc_idct_add_c (q, dq, pre, dst, 16, stride, dc[0]); + else + vp8_dc_only_idct_add_c (dc[0], pre, dst, 16, stride); + + q += 16; + pre += 4; + dst += 4; + dc ++; + } + + pre += 64 - 16; + dst += 4*stride - 16; + } +} + +void vp8_dequant_idct_add_y_block_c + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int stride, char *eobs) +{ + int i, j; + + for (i = 0; i < 4; i++) + { + for (j = 0; j < 4; j++) + { + if (*eobs++ > 1) + vp8_dequant_idct_add_c (q, dq, pre, dst, 16, stride); + else + { + vp8_dc_only_idct_add_c (q[0]*dq[0], pre, dst, 16, stride); + ((int *)q)[0] = 0; + } + + q += 16; + pre += 4; + dst += 4; + } + + pre += 64 - 16; + dst += 4*stride - 16; + } +} + +void vp8_dequant_idct_add_uv_block_c + (short *q, short *dq, unsigned char *pre, + unsigned char *dstu, unsigned char *dstv, int stride, char *eobs) +{ + int i, j; + + for (i = 0; i < 2; i++) + { + for (j = 0; j < 2; j++) + { + if (*eobs++ > 1) + vp8_dequant_idct_add_c (q, dq, pre, dstu, 8, stride); + else + { + vp8_dc_only_idct_add_c (q[0]*dq[0], pre, dstu, 8, stride); + ((int *)q)[0] = 0; + } + + q += 16; + pre += 4; + dstu += 4; + } + + pre += 32 - 8; + dstu += 4*stride - 8; + } + + for (i = 0; i < 2; i++) + { + for (j = 0; j < 2; j++) + { + if (*eobs++ > 1) + vp8_dequant_idct_add_c (q, dq, pre, dstv, 8, stride); + else + { + vp8_dc_only_idct_add_c (q[0]*dq[0], pre, dstv, 8, stride); + ((int *)q)[0] = 0; + } + + q += 16; + pre += 4; + dstv += 4; + } + + pre += 32 - 8; + dstv += 4*stride - 8; + } +} diff --git a/vp8/decoder/onyxd_if.c b/vp8/decoder/onyxd_if.c index 6875585..1651784 100644 --- a/vp8/decoder/onyxd_if.c +++ b/vp8/decoder/onyxd_if.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -23,18 +24,16 @@ #include "threading.h" #include "decoderthreading.h" #include <stdio.h> -#include "segmentation_common.h" + #include "quant_common.h" #include "vpx_scale/vpxscale.h" #include "systemdependent.h" #include "vpx_ports/vpx_timer.h" - +#include "detokenize.h" extern void vp8_init_loop_filter(VP8_COMMON *cm); - extern void vp8cx_init_de_quantizer(VP8D_COMP *pbi); -// DEBUG code #if CONFIG_DEBUG void vp8_recon_write_yuv_frame(unsigned char *name, YV12_BUFFER_CONFIG *s) { @@ -128,6 +127,9 @@ VP8D_PTR vp8dx_create_decompressor(VP8D_CONFIG *oxcf) cm->last_sharpness_level = cm->sharpness_level; } +#if CONFIG_ARM_ASM_DETOK + vp8_init_detokenizer(pbi); +#endif pbi->common.error.setjmp = 0; return (VP8D_PTR) pbi; } @@ -179,38 +181,38 @@ int vp8dx_get_reference(VP8D_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_C { VP8D_COMP *pbi = (VP8D_COMP *) ptr; VP8_COMMON *cm = &pbi->common; + int ref_fb_idx; if (ref_frame_flag == VP8_LAST_FLAG) - vp8_yv12_copy_frame_ptr(&cm->last_frame, sd); - + ref_fb_idx = cm->lst_fb_idx; else if (ref_frame_flag == VP8_GOLD_FLAG) - vp8_yv12_copy_frame_ptr(&cm->golden_frame, sd); - + ref_fb_idx = cm->gld_fb_idx; else if (ref_frame_flag == VP8_ALT_FLAG) - vp8_yv12_copy_frame_ptr(&cm->alt_ref_frame, sd); - + ref_fb_idx = cm->alt_fb_idx; else return -1; + vp8_yv12_copy_frame_ptr(&cm->yv12_fb[ref_fb_idx], sd); + return 0; } int vp8dx_set_reference(VP8D_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd) { VP8D_COMP *pbi = (VP8D_COMP *) ptr; VP8_COMMON *cm = &pbi->common; + int ref_fb_idx; if (ref_frame_flag == VP8_LAST_FLAG) - vp8_yv12_copy_frame_ptr(sd, &cm->last_frame); - + ref_fb_idx = cm->lst_fb_idx; else if (ref_frame_flag == VP8_GOLD_FLAG) - vp8_yv12_copy_frame_ptr(sd, &cm->golden_frame); - + ref_fb_idx = cm->gld_fb_idx; else if (ref_frame_flag == VP8_ALT_FLAG) - vp8_yv12_copy_frame_ptr(sd, &cm->alt_ref_frame); - + ref_fb_idx = cm->alt_fb_idx; else return -1; + vp8_yv12_copy_frame_ptr(sd, &cm->yv12_fb[ref_fb_idx]); + return 0; } @@ -220,12 +222,95 @@ extern void vp8_push_neon(INT64 *store); extern void vp8_pop_neon(INT64 *store); static INT64 dx_store_reg[8]; #endif + +static int get_free_fb (VP8_COMMON *cm) +{ + int i; + for (i = 0; i < NUM_YV12_BUFFERS; i++) + if (cm->fb_idx_ref_cnt[i] == 0) + break; + + cm->fb_idx_ref_cnt[i] = 1; + return i; +} + +static void ref_cnt_fb (int *buf, int *idx, int new_idx) +{ + if (buf[*idx] > 0) + buf[*idx]--; + + *idx = new_idx; + + buf[new_idx]++; +} + +// If any buffer copy / swapping is signalled it should be done here. +static int swap_frame_buffers (VP8_COMMON *cm) +{ + int fb_to_update_with, err = 0; + + if (cm->refresh_last_frame) + fb_to_update_with = cm->lst_fb_idx; + else + fb_to_update_with = cm->new_fb_idx; + + // The alternate reference frame or golden frame can be updated + // using the new, last, or golden/alt ref frame. If it + // is updated using the newly decoded frame it is a refresh. + // An update using the last or golden/alt ref frame is a copy. + if (cm->copy_buffer_to_arf) + { + int new_fb = 0; + + if (cm->copy_buffer_to_arf == 1) + new_fb = fb_to_update_with; + else if (cm->copy_buffer_to_arf == 2) + new_fb = cm->gld_fb_idx; + else + err = -1; + + ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->alt_fb_idx, new_fb); + } + + if (cm->copy_buffer_to_gf) + { + int new_fb = 0; + + if (cm->copy_buffer_to_gf == 1) + new_fb = fb_to_update_with; + else if (cm->copy_buffer_to_gf == 2) + new_fb = cm->alt_fb_idx; + else + err = -1; + + ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->gld_fb_idx, new_fb); + } + + if (cm->refresh_golden_frame) + ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->gld_fb_idx, cm->new_fb_idx); + + if (cm->refresh_alt_ref_frame) + ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->alt_fb_idx, cm->new_fb_idx); + + if (cm->refresh_last_frame) + { + ref_cnt_fb (cm->fb_idx_ref_cnt, &cm->lst_fb_idx, cm->new_fb_idx); + + cm->frame_to_show = &cm->yv12_fb[cm->lst_fb_idx]; + } + else + cm->frame_to_show = &cm->yv12_fb[cm->new_fb_idx]; + + cm->fb_idx_ref_cnt[cm->new_fb_idx]--; + + return err; +} + int vp8dx_receive_compressed_data(VP8D_PTR ptr, unsigned long size, const unsigned char *source, INT64 time_stamp) { VP8D_COMP *pbi = (VP8D_COMP *) ptr; VP8_COMMON *cm = &pbi->common; int retcode = 0; - struct vpx_usec_timer timer; // if(pbi->ready_for_new_data == 0) @@ -256,6 +341,8 @@ int vp8dx_receive_compressed_data(VP8D_PTR ptr, unsigned long size, const unsign pbi->Source = source; pbi->source_sz = size; + cm->new_fb_idx = get_free_fb (cm); + retcode = vp8_decode_frame(pbi); if (retcode < 0) @@ -268,23 +355,17 @@ int vp8dx_receive_compressed_data(VP8D_PTR ptr, unsigned long size, const unsign return retcode; } - // Update the GF useage maps. - vp8_update_gf_useage_maps(cm, &pbi->mb); - if (pbi->b_multithreaded_lf && pbi->common.filter_level != 0) vp8_stop_lfthread(pbi); - if (cm->refresh_last_frame) - { - vp8_swap_yv12_buffer(&cm->last_frame, &cm->new_frame); - - cm->frame_to_show = &cm->last_frame; - } - else + if (swap_frame_buffers (cm)) { - cm->frame_to_show = &cm->new_frame; + pbi->common.error.error_code = VPX_CODEC_ERROR; + pbi->common.error.setjmp = 0; + return -1; } +/* if (!pbi->b_multithreaded_lf) { struct vpx_usec_timer lpftimer; @@ -292,16 +373,45 @@ int vp8dx_receive_compressed_data(VP8D_PTR ptr, unsigned long size, const unsign // Apply the loop filter if appropriate. if (cm->filter_level > 0) - { vp8_loop_filter_frame(cm, &pbi->mb, cm->filter_level); - cm->last_frame_type = cm->frame_type; - cm->last_filter_type = cm->filter_type; - cm->last_sharpness_level = cm->sharpness_level; - } + vpx_usec_timer_mark(&lpftimer); + pbi->time_loop_filtering += vpx_usec_timer_elapsed(&lpftimer); + }else{ + struct vpx_usec_timer lpftimer; + vpx_usec_timer_start(&lpftimer); + // Apply the loop filter if appropriate. + + if (cm->filter_level > 0) + vp8_mt_loop_filter_frame(cm, &pbi->mb, cm->filter_level); + + vpx_usec_timer_mark(&lpftimer); + pbi->time_loop_filtering += vpx_usec_timer_elapsed(&lpftimer); + } + if (cm->filter_level > 0) { + cm->last_frame_type = cm->frame_type; + cm->last_filter_type = cm->filter_type; + cm->last_sharpness_level = cm->sharpness_level; + } +*/ + + if(pbi->common.filter_level) + { + struct vpx_usec_timer lpftimer; + vpx_usec_timer_start(&lpftimer); + // Apply the loop filter if appropriate. + + if (pbi->b_multithreaded_lf && cm->multi_token_partition != ONE_PARTITION) + vp8_mt_loop_filter_frame(pbi); //cm, &pbi->mb, cm->filter_level); + else + vp8_loop_filter_frame(cm, &pbi->mb, cm->filter_level); vpx_usec_timer_mark(&lpftimer); pbi->time_loop_filtering += vpx_usec_timer_elapsed(&lpftimer); + + cm->last_frame_type = cm->frame_type; + cm->last_filter_type = cm->filter_type; + cm->last_sharpness_level = cm->sharpness_level; } vp8_yv12_extend_frame_borders_ptr(cm->frame_to_show); @@ -313,49 +423,6 @@ int vp8dx_receive_compressed_data(VP8D_PTR ptr, unsigned long size, const unsign write_dx_frame_to_file(cm->frame_to_show, cm->current_video_frame); #endif - // If any buffer copy / swaping is signalled it should be done here. - if (cm->copy_buffer_to_arf) - { - if (cm->copy_buffer_to_arf == 1) - { - if (cm->refresh_last_frame) - vp8_yv12_copy_frame_ptr(&cm->new_frame, &cm->alt_ref_frame); - else - vp8_yv12_copy_frame_ptr(&cm->last_frame, &cm->alt_ref_frame); - } - else if (cm->copy_buffer_to_arf == 2) - vp8_yv12_copy_frame_ptr(&cm->golden_frame, &cm->alt_ref_frame); - } - - if (cm->copy_buffer_to_gf) - { - if (cm->copy_buffer_to_gf == 1) - { - if (cm->refresh_last_frame) - vp8_yv12_copy_frame_ptr(&cm->new_frame, &cm->golden_frame); - else - vp8_yv12_copy_frame_ptr(&cm->last_frame, &cm->golden_frame); - } - else if (cm->copy_buffer_to_gf == 2) - vp8_yv12_copy_frame_ptr(&cm->alt_ref_frame, &cm->golden_frame); - } - - // Should the golden or alternate reference frame be refreshed? - if (cm->refresh_golden_frame || cm->refresh_alt_ref_frame) - { - if (cm->refresh_golden_frame) - vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->golden_frame); - - if (cm->refresh_alt_ref_frame) - vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->alt_ref_frame); - - //vpx_log("Decoder: recovery frame received \n"); - - // Update data structures that monitors GF useage - vpx_memset(cm->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols)); - cm->gf_active_count = cm->mb_rows * cm->mb_cols; - } - vp8_clear_system_state(); vpx_usec_timer_mark(&timer); diff --git a/vp8/decoder/onyxd_if_sjl.c b/vp8/decoder/onyxd_if_sjl.c deleted file mode 100644 index 363ad5d..0000000 --- a/vp8/decoder/onyxd_if_sjl.c +++ /dev/null @@ -1,398 +0,0 @@ -/* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. - * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. - */ - - -#include "onyxc_int.h" -#include "postproc.h" -#include "onyxd.h" -#include "onyxd_int.h" -#include "vpx_mem/vpx_mem.h" -#include "alloccommon.h" -#include "vpx_scale/yv12extend.h" -#include "loopfilter.h" -#include "swapyv12buffer.h" -#include "g_common.h" -#include "threading.h" -#include "decoderthreading.h" -#include <stdio.h> -#include "segmentation_common.h" -#include "quant_common.h" -#include "vpx_scale/vpxscale.h" -#include "systemdependent.h" -#include "vpx_ports/vpx_timer.h" - - -#ifndef VPX_NO_GLOBALS -static int init_ct = 0; -#else -# include "vpx_global_handling.h" -# define init_ct ((int)vpxglobalm(onyxd,init_ct)) -#endif - -extern void vp8_init_loop_filter(VP8_COMMON *cm); - -extern void vp8cx_init_de_quantizer(VP8D_COMP *pbi); -extern void init_detokenizer(VP8D_COMP *dx); - -// DEBUG code -void vp8_recon_write_yuv_frame(unsigned char *name, YV12_BUFFER_CONFIG *s) -{ - FILE *yuv_file = fopen((char *)name, "ab"); - unsigned char *src = s->y_buffer; - int h = s->y_height; - - do - { - fwrite(src, s->y_width, 1, yuv_file); - src += s->y_stride; - } - while (--h); - - src = s->u_buffer; - h = s->uv_height; - - do - { - fwrite(src, s->uv_width, 1, yuv_file); - src += s->uv_stride; - } - while (--h); - - src = s->v_buffer; - h = s->uv_height; - - do - { - fwrite(src, s->uv_width, 1, yuv_file); - src += s->uv_stride; - } - while (--h); - - fclose(yuv_file); -} - -void vp8dx_initialize() -{ - if (!init_ct++) - { - vp8_initialize_common(); - vp8_scale_machine_specific_config(); - } -} - -void vp8dx_shutdown() -{ - if (!--init_ct) - { - vp8_shutdown_common(); - } -} - - -VP8D_PTR vp8dx_create_decompressor(VP8D_CONFIG *oxcf) -{ - VP8D_COMP *pbi = vpx_memalign(32, sizeof(VP8D_COMP)); - - if (!pbi) - return NULL; - - vpx_memset(pbi, 0, sizeof(VP8D_COMP)); - - vp8dx_initialize(); - - vp8_create_common(&pbi->common); - vp8_dmachine_specific_config(pbi); - - pbi->common.current_video_frame = 0; - pbi->ready_for_new_data = 1; - - pbi->CPUFreq = 0; //vp8_get_processor_freq(); - pbi->max_threads = oxcf->max_threads; - vp8_decoder_create_threads(pbi); - - //vp8cx_init_de_quantizer() is first called here. Add check in frame_init_dequantizer() to avoid - // unnecessary calling of vp8cx_init_de_quantizer() for every frame. - vp8cx_init_de_quantizer(pbi); - - { - VP8_COMMON *cm = &pbi->common; - - vp8_init_loop_filter(cm); - cm->last_frame_type = KEY_FRAME; - cm->last_filter_type = cm->filter_type; - cm->last_sharpness_level = cm->sharpness_level; - } - - init_detokenizer(pbi); - - return (VP8D_PTR) pbi; -} -void vp8dx_remove_decompressor(VP8D_PTR ptr) -{ - VP8D_COMP *pbi = (VP8D_COMP *) ptr; - - if (!pbi) - return; - - vp8_decoder_remove_threads(pbi); - vp8_remove_common(&pbi->common); - vpx_free(pbi); - vp8dx_shutdown(); - -} - -void vp8dx_set_setting(VP8D_PTR comp, VP8D_SETTING oxst, int x) -{ - VP8D_COMP *pbi = (VP8D_COMP *) comp; - - (void) pbi; - (void) x; - - switch (oxst) - { - case VP8D_OK: - break; - } -} - -int vp8dx_get_setting(VP8D_PTR comp, VP8D_SETTING oxst) -{ - VP8D_COMP *pbi = (VP8D_COMP *) comp; - - (void) pbi; - - switch (oxst) - { - case VP8D_OK: - break; - } - - return -1; -} - -int vp8dx_get_reference(VP8D_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd) -{ - VP8D_COMP *pbi = (VP8D_COMP *) ptr; - VP8_COMMON *cm = &pbi->common; - - if (ref_frame_flag == VP8_LAST_FLAG) - vp8_yv12_copy_frame_ptr(&cm->last_frame, sd); - - else if (ref_frame_flag == VP8_GOLD_FLAG) - vp8_yv12_copy_frame_ptr(&cm->golden_frame, sd); - - else if (ref_frame_flag == VP8_ALT_FLAG) - vp8_yv12_copy_frame_ptr(&cm->alt_ref_frame, sd); - - else - return -1; - - return 0; -} -int vp8dx_set_reference(VP8D_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd) -{ - VP8D_COMP *pbi = (VP8D_COMP *) ptr; - VP8_COMMON *cm = &pbi->common; - - if (ref_frame_flag == VP8_LAST_FLAG) - vp8_yv12_copy_frame_ptr(sd, &cm->last_frame); - - else if (ref_frame_flag == VP8_GOLD_FLAG) - vp8_yv12_copy_frame_ptr(sd, &cm->golden_frame); - - else if (ref_frame_flag == VP8_ALT_FLAG) - vp8_yv12_copy_frame_ptr(sd, &cm->alt_ref_frame); - - else - return -1; - - return 0; -} -int vp8dx_receive_compressed_data(VP8D_PTR ptr, unsigned long size, char *source, INT64 time_stamp) -{ - VP8D_COMP *pbi = (VP8D_COMP *) ptr; - VP8_COMMON *cm = &pbi->common; - int retcode = 0; - - struct vpx_usec_timer timer; - (void) size; - -// if(pbi->ready_for_new_data == 0) -// return -1; - - vpx_usec_timer_start(&timer); - - if (ptr == 0) - { - return -1; - } - - //cm->current_video_frame++; - pbi->Source = source; - - retcode = vp8_decode_frame(pbi); - - if (retcode < 0) - return retcode; - - // Update the GF useage maps. - vp8_update_gf_useage_maps(cm, &pbi->mb); - - if (pbi->b_multithreaded) - vp8_stop_lfthread(pbi); - - if (cm->refresh_last_frame) - { - vp8_swap_yv12_buffer(&cm->last_frame, &cm->new_frame); - - cm->frame_to_show = &cm->last_frame; - } - else - { - cm->frame_to_show = &cm->new_frame; - } - - if (!pbi->b_multithreaded) - { - struct vpx_usec_timer lpftimer; - vpx_usec_timer_start(&lpftimer); - // Apply the loop filter if appropriate. - - if (cm->filter_level > 0) - { - vp8_loop_filter_frame(cm, &pbi->mb, cm->filter_level); - cm->last_frame_type = cm->frame_type; - cm->last_filter_type = cm->filter_type; - cm->last_sharpness_level = cm->sharpness_level; - - } - - vpx_usec_timer_mark(&lpftimer); - pbi->time_loop_filtering += vpx_usec_timer_elapsed(&lpftimer); - } - - vp8_yv12_extend_frame_borders_ptr(cm->frame_to_show); - -#if 0 - // DEBUG code - //vp8_recon_write_yuv_frame("recon.yuv", cm->frame_to_show); - if (cm->current_video_frame <= 5) - write_dx_frame_to_file(cm->frame_to_show, cm->current_video_frame); -#endif - - // If any buffer copy / swaping is signalled it should be done here. - if (cm->copy_buffer_to_arf) - { - if (cm->copy_buffer_to_arf == 1) - { - if (cm->refresh_last_frame) - vp8_yv12_copy_frame_ptr(&cm->new_frame, &cm->alt_ref_frame); - else - vp8_yv12_copy_frame_ptr(&cm->last_frame, &cm->alt_ref_frame); - } - else if (cm->copy_buffer_to_arf == 2) - vp8_yv12_copy_frame_ptr(&cm->golden_frame, &cm->alt_ref_frame); - } - - if (cm->copy_buffer_to_gf) - { - if (cm->copy_buffer_to_gf == 1) - { - if (cm->refresh_last_frame) - vp8_yv12_copy_frame_ptr(&cm->new_frame, &cm->golden_frame); - else - vp8_yv12_copy_frame_ptr(&cm->last_frame, &cm->golden_frame); - } - else if (cm->copy_buffer_to_gf == 2) - vp8_yv12_copy_frame_ptr(&cm->alt_ref_frame, &cm->golden_frame); - } - - // Should the golden or alternate reference frame be refreshed? - if (cm->refresh_golden_frame || cm->refresh_alt_ref_frame) - { - if (cm->refresh_golden_frame) - vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->golden_frame); - - if (cm->refresh_alt_ref_frame) - vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->alt_ref_frame); - - //vpx_log("Decoder: recovery frame received \n"); - - // Update data structures that monitors GF useage - vpx_memset(cm->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols)); - cm->gf_active_count = cm->mb_rows * cm->mb_cols; - } - - vp8_clear_system_state(); - - vpx_usec_timer_mark(&timer); - pbi->decode_microseconds = vpx_usec_timer_elapsed(&timer); - - pbi->time_decoding += pbi->decode_microseconds; - -// vp8_print_modes_and_motion_vectors( cm->mi, cm->mb_rows,cm->mb_cols, cm->current_video_frame); - - cm->current_video_frame++; - pbi->ready_for_new_data = 0; - pbi->last_time_stamp = time_stamp; - - { - int i; - INT64 earliest_time = pbi->dr[0].time_stamp; - INT64 latest_time = pbi->dr[0].time_stamp; - INT64 time_diff = 0; - int bytes = 0; - - pbi->dr[pbi->common.current_video_frame&0xf].size = pbi->bc.pos + pbi->bc2.pos + 4;; - pbi->dr[pbi->common.current_video_frame&0xf].time_stamp = time_stamp; - - for (i = 0; i < 16; i++) - { - - bytes += pbi->dr[i].size; - - if (pbi->dr[i].time_stamp < earliest_time) - earliest_time = pbi->dr[i].time_stamp; - - if (pbi->dr[i].time_stamp > latest_time) - latest_time = pbi->dr[i].time_stamp; - } - - time_diff = latest_time - earliest_time; - - if (time_diff > 0) - { - pbi->common.bitrate = 80000.00 * bytes / time_diff ; - pbi->common.framerate = 160000000.00 / time_diff ; - } - - } - return retcode; -} -int vp8dx_get_raw_frame(VP8D_PTR ptr, YV12_BUFFER_CONFIG *sd, INT64 *time_stamp, INT64 *time_end_stamp, int deblock_level, int noise_level, int flags) -{ - int ret = -1; - VP8D_COMP *pbi = (VP8D_COMP *) ptr; - - if (pbi->ready_for_new_data == 1) - return ret; - - // ie no raw frame to show!!! - if (pbi->common.show_frame == 0) - return ret; - - pbi->ready_for_new_data = 1; - *time_stamp = pbi->last_time_stamp; - *time_end_stamp = 0; - - sd->clrtype = pbi->common.clr_type; - ret = vp8_post_proc_frame(&pbi->common, sd, deblock_level, noise_level, flags); - vp8_clear_system_state(); - return ret; -} diff --git a/vp8/decoder/onyxd_int.h b/vp8/decoder/onyxd_int.h index fa4fa48..ad21ae3 100644 --- a/vp8/decoder/onyxd_int.h +++ b/vp8/decoder/onyxd_int.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -47,21 +48,20 @@ typedef struct typedef struct { - int *scan; - UINT8 *ptr_onyxblock2context_leftabove; - vp8_tree_index *vp8_coef_tree_ptr; //onyx_coef_tree_ptr; ??? - TOKENEXTRABITS *teb_base_ptr; + int const *scan; + UINT8 const *ptr_block2leftabove; + vp8_tree_index const *vp8_coef_tree_ptr; + TOKENEXTRABITS const *teb_base_ptr; unsigned char *norm_ptr; -// UINT16 *ptr_onyx_coef_bands_x; - UINT8 *ptr_onyx_coef_bands_x; + UINT8 *ptr_coef_bands_x; - ENTROPY_CONTEXT **A; - ENTROPY_CONTEXT(*L)[4]; + ENTROPY_CONTEXT_PLANES *A; + ENTROPY_CONTEXT_PLANES *L; INT16 *qcoeff_start_ptr; BOOL_DECODER *current_bc; - UINT8 *coef_probs[4]; + vp8_prob const *coef_probs[4]; UINT8 eob[25]; @@ -94,20 +94,22 @@ typedef struct VP8Decompressor int current_mb_col_main; int decoding_thread_count; int allocated_decoding_thread_count; + int *current_mb_col; //Each row remembers its already decoded column. + int mt_baseline_filter_level[MAX_MB_SEGMENTS]; // variable for threading DECLARE_ALIGNED(16, MACROBLOCKD, lpfmb); #if CONFIG_MULTITHREAD - pthread_t h_thread_lpf; // thread for postprocessing - sem_t h_event_lpf; // Event for post_proc completed - sem_t h_event_start_lpf; + //pthread_t h_thread_lpf; // thread for postprocessing + sem_t h_event_end_lpf; // Event for post_proc completed + sem_t *h_event_start_lpf; #endif MB_ROW_DEC *mb_row_di; DECODETHREAD_DATA *de_thread_data; #if CONFIG_MULTITHREAD pthread_t *h_decoding_thread; - sem_t *h_event_mbrdecoding; - sem_t h_event_main; + sem_t *h_event_start_decoding; + sem_t h_event_end_decoding; // end of threading data #endif vp8_reader *mbc; @@ -123,6 +125,12 @@ typedef struct VP8Decompressor struct vp8_dboolhuff_rtcd_vtable dboolhuff; #endif + + vp8_prob prob_intra; + vp8_prob prob_last; + vp8_prob prob_gf; + vp8_prob prob_skip_false; + } VP8D_COMP; int vp8_decode_frame(VP8D_COMP *cpi); diff --git a/vp8/decoder/threading.c b/vp8/decoder/threading.c index e35d175..a77552c 100644 --- a/vp8/decoder/threading.c +++ b/vp8/decoder/threading.c @@ -1,16 +1,20 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ #ifndef WIN32 # include <unistd.h> #endif +#ifdef __APPLE__ +#include <mach/mach_init.h> +#endif #include "onyxd_int.h" #include "vpx_mem/vpx_mem.h" #include "threading.h" @@ -19,6 +23,8 @@ #include "extend.h" #include "vpx_ports/vpx_timer.h" +#define MAX_ROWS 256 + extern void vp8_decode_mb_row(VP8D_COMP *pbi, VP8_COMMON *pc, int mb_row, @@ -27,11 +33,10 @@ extern void vp8_decode_mb_row(VP8D_COMP *pbi, extern void vp8_build_uvmvs(MACROBLOCKD *x, int fullpixel); extern void vp8_decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd); +void vp8_thread_loop_filter(VP8D_COMP *pbi, MB_ROW_DEC *mbrd, int ithread); + void vp8_setup_decoding_thread_data(VP8D_COMP *pbi, MACROBLOCKD *xd, MB_ROW_DEC *mbrd, int count) { - - - #if CONFIG_MULTITHREAD VP8_COMMON *const pc = & pbi->common; int i, j; @@ -42,15 +47,11 @@ void vp8_setup_decoding_thread_data(VP8D_COMP *pbi, MACROBLOCKD *xd, MB_ROW_DEC #if CONFIG_RUNTIME_CPU_DETECT mbd->rtcd = xd->rtcd; #endif - - mbd->subpixel_predict = xd->subpixel_predict; mbd->subpixel_predict8x4 = xd->subpixel_predict8x4; mbd->subpixel_predict8x8 = xd->subpixel_predict8x8; mbd->subpixel_predict16x16 = xd->subpixel_predict16x16; - mbd->gf_active_ptr = xd->gf_active_ptr; - mbd->mode_info = pc->mi - 1; mbd->mode_info_context = pc->mi + pc->mode_info_stride * (i + 1); mbd->mode_info_stride = pc->mode_info_stride; @@ -58,11 +59,8 @@ void vp8_setup_decoding_thread_data(VP8D_COMP *pbi, MACROBLOCKD *xd, MB_ROW_DEC mbd->frames_since_golden = pc->frames_since_golden; mbd->frames_till_alt_ref_frame = pc->frames_till_alt_ref_frame; - mbd->pre = pc->last_frame; - mbd->dst = pc->new_frame; - - - + mbd->pre = pc->yv12_fb[pc->lst_fb_idx]; + mbd->dst = pc->yv12_fb[pc->new_fb_idx]; vp8_setup_block_dptrs(mbd); vp8_build_block_doffsets(mbd); @@ -70,9 +68,6 @@ void vp8_setup_decoding_thread_data(VP8D_COMP *pbi, MACROBLOCKD *xd, MB_ROW_DEC mbd->mb_segement_abs_delta = xd->mb_segement_abs_delta; vpx_memcpy(mbd->segment_feature_data, xd->segment_feature_data, sizeof(xd->segment_feature_data)); - mbd->mbmi.mode = DC_PRED; - mbd->mbmi.uv_mode = DC_PRED; - mbd->current_bc = &pbi->bc2; for (j = 0; j < 25; j++) @@ -81,6 +76,8 @@ void vp8_setup_decoding_thread_data(VP8D_COMP *pbi, MACROBLOCKD *xd, MB_ROW_DEC } } + for (i=0; i< pc->mb_rows; i++) + pbi->current_mb_col[i]=-1; #else (void) pbi; (void) xd; @@ -89,6 +86,68 @@ void vp8_setup_decoding_thread_data(VP8D_COMP *pbi, MACROBLOCKD *xd, MB_ROW_DEC #endif } +void vp8_setup_loop_filter_thread_data(VP8D_COMP *pbi, MACROBLOCKD *xd, MB_ROW_DEC *mbrd, int count) +{ +#if CONFIG_MULTITHREAD + VP8_COMMON *const pc = & pbi->common; + int i, j; + + for (i = 0; i < count; i++) + { + MACROBLOCKD *mbd = &mbrd[i].mbd; +//#if CONFIG_RUNTIME_CPU_DETECT +// mbd->rtcd = xd->rtcd; +//#endif + + //mbd->subpixel_predict = xd->subpixel_predict; + //mbd->subpixel_predict8x4 = xd->subpixel_predict8x4; + //mbd->subpixel_predict8x8 = xd->subpixel_predict8x8; + //mbd->subpixel_predict16x16 = xd->subpixel_predict16x16; + + mbd->mode_info_context = pc->mi + pc->mode_info_stride * (i + 1); + mbd->mode_info_stride = pc->mode_info_stride; + + //mbd->frame_type = pc->frame_type; + //mbd->frames_since_golden = pc->frames_since_golden; + //mbd->frames_till_alt_ref_frame = pc->frames_till_alt_ref_frame; + + //mbd->pre = pc->yv12_fb[pc->lst_fb_idx]; + //mbd->dst = pc->yv12_fb[pc->new_fb_idx]; + + //vp8_setup_block_dptrs(mbd); + //vp8_build_block_doffsets(mbd); + mbd->segmentation_enabled = xd->segmentation_enabled; // + mbd->mb_segement_abs_delta = xd->mb_segement_abs_delta; // + vpx_memcpy(mbd->segment_feature_data, xd->segment_feature_data, sizeof(xd->segment_feature_data)); // + + //signed char ref_lf_deltas[MAX_REF_LF_DELTAS]; + vpx_memcpy(mbd->ref_lf_deltas, xd->ref_lf_deltas, sizeof(xd->ref_lf_deltas)); + //signed char mode_lf_deltas[MAX_MODE_LF_DELTAS]; + vpx_memcpy(mbd->mode_lf_deltas, xd->mode_lf_deltas, sizeof(xd->mode_lf_deltas)); + //unsigned char mode_ref_lf_delta_enabled; + //unsigned char mode_ref_lf_delta_update; + mbd->mode_ref_lf_delta_enabled = xd->mode_ref_lf_delta_enabled; + mbd->mode_ref_lf_delta_update = xd->mode_ref_lf_delta_update; + + //mbd->mbmi.mode = DC_PRED; + //mbd->mbmi.uv_mode = DC_PRED; + //mbd->current_bc = &pbi->bc2; + + //for (j = 0; j < 25; j++) + //{ + // mbd->block[j].dequant = xd->block[j].dequant; + //} + } + + for (i=0; i< pc->mb_rows; i++) + pbi->current_mb_col[i]=-1; +#else + (void) pbi; + (void) xd; + (void) mbrd; + (void) count; +#endif +} THREAD_FUNCTION vp8_thread_decoding_proc(void *p_data) { @@ -96,156 +155,146 @@ THREAD_FUNCTION vp8_thread_decoding_proc(void *p_data) int ithread = ((DECODETHREAD_DATA *)p_data)->ithread; VP8D_COMP *pbi = (VP8D_COMP *)(((DECODETHREAD_DATA *)p_data)->ptr1); MB_ROW_DEC *mbrd = (MB_ROW_DEC *)(((DECODETHREAD_DATA *)p_data)->ptr2); - ENTROPY_CONTEXT mb_row_left_context[4][4]; + ENTROPY_CONTEXT_PLANES mb_row_left_context; while (1) { + int current_filter_level = 0; + if (pbi->b_multithreaded_rd == 0) break; - //if(WaitForSingleObject(pbi->h_event_mbrdecoding[ithread], INFINITE) == WAIT_OBJECT_0) - if (sem_wait(&pbi->h_event_mbrdecoding[ithread]) == 0) + //if(WaitForSingleObject(pbi->h_event_start_decoding[ithread], INFINITE) == WAIT_OBJECT_0) + if (sem_wait(&pbi->h_event_start_decoding[ithread]) == 0) { if (pbi->b_multithreaded_rd == 0) break; else { VP8_COMMON *pc = &pbi->common; - int mb_row = mbrd->mb_row; MACROBLOCKD *xd = &mbrd->mbd; - //printf("ithread:%d mb_row %d\n", ithread, mb_row); - int i; - int recon_yoffset, recon_uvoffset; - int mb_col; - int recon_y_stride = pc->last_frame.y_stride; - int recon_uv_stride = pc->last_frame.uv_stride; - + int mb_row; + int num_part = 1 << pbi->common.multi_token_partition; volatile int *last_row_current_mb_col; - if (ithread > 0) - last_row_current_mb_col = &pbi->mb_row_di[ithread-1].current_mb_col; - else - last_row_current_mb_col = &pbi->current_mb_col_main; + for (mb_row = ithread+1; mb_row < pc->mb_rows; mb_row += (pbi->decoding_thread_count + 1)) + { + int i; + int recon_yoffset, recon_uvoffset; + int mb_col; + int ref_fb_idx = pc->lst_fb_idx; + int dst_fb_idx = pc->new_fb_idx; + int recon_y_stride = pc->yv12_fb[ref_fb_idx].y_stride; + int recon_uv_stride = pc->yv12_fb[ref_fb_idx].uv_stride; - recon_yoffset = mb_row * recon_y_stride * 16; - recon_uvoffset = mb_row * recon_uv_stride * 8; - // reset above block coeffs + pbi->mb_row_di[ithread].mb_row = mb_row; + pbi->mb_row_di[ithread].mbd.current_bc = &pbi->mbc[mb_row%num_part]; - xd->above_context[Y1CONTEXT] = pc->above_context[Y1CONTEXT]; - xd->above_context[UCONTEXT ] = pc->above_context[UCONTEXT]; - xd->above_context[VCONTEXT ] = pc->above_context[VCONTEXT]; - xd->above_context[Y2CONTEXT] = pc->above_context[Y2CONTEXT]; - xd->left_context = mb_row_left_context; - vpx_memset(mb_row_left_context, 0, sizeof(mb_row_left_context)); - xd->up_available = (mb_row != 0); + last_row_current_mb_col = &pbi->current_mb_col[mb_row -1]; - xd->mb_to_top_edge = -((mb_row * 16)) << 3; - xd->mb_to_bottom_edge = ((pc->mb_rows - 1 - mb_row) * 16) << 3; + recon_yoffset = mb_row * recon_y_stride * 16; + recon_uvoffset = mb_row * recon_uv_stride * 8; + // reset above block coeffs - for (mb_col = 0; mb_col < pc->mb_cols; mb_col++) - { + xd->above_context = pc->above_context; + xd->left_context = &mb_row_left_context; + vpx_memset(&mb_row_left_context, 0, sizeof(mb_row_left_context)); + xd->up_available = (mb_row != 0); - while (mb_col > (*last_row_current_mb_col - 1) && *last_row_current_mb_col != pc->mb_cols - 1) - { - x86_pause_hint(); - thread_sleep(0); - } + xd->mb_to_top_edge = -((mb_row * 16)) << 3; + xd->mb_to_bottom_edge = ((pc->mb_rows - 1 - mb_row) * 16) << 3; - // Take a copy of the mode and Mv information for this macroblock into the xd->mbmi - vpx_memcpy(&xd->mbmi, &xd->mode_info_context->mbmi, 32); //sizeof(MB_MODE_INFO) ); - - if (xd->mbmi.mode == SPLITMV || xd->mbmi.mode == B_PRED) + for (mb_col = 0; mb_col < pc->mb_cols; mb_col++) { - for (i = 0; i < 16; i++) + if ((mb_col & 7) == 0) { - BLOCKD *d = &xd->block[i]; - vpx_memcpy(&d->bmi, &xd->mode_info_context->bmi[i], sizeof(B_MODE_INFO)); + while (mb_col > (*last_row_current_mb_col - 8) && *last_row_current_mb_col != pc->mb_cols - 1) + { + x86_pause_hint(); + thread_sleep(0); + } } - } - // Distance of Mb to the various image edges. - // These specified to 8th pel as they are always compared to values that are in 1/8th pel units - xd->mb_to_left_edge = -((mb_col * 16) << 3); - xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3; + if (xd->mode_info_context->mbmi.mode == SPLITMV || xd->mode_info_context->mbmi.mode == B_PRED) + { + for (i = 0; i < 16; i++) + { + BLOCKD *d = &xd->block[i]; + vpx_memcpy(&d->bmi, &xd->mode_info_context->bmi[i], sizeof(B_MODE_INFO)); + } + } - xd->dst.y_buffer = pc->new_frame.y_buffer + recon_yoffset; - xd->dst.u_buffer = pc->new_frame.u_buffer + recon_uvoffset; - xd->dst.v_buffer = pc->new_frame.v_buffer + recon_uvoffset; + // Distance of Mb to the various image edges. + // These specified to 8th pel as they are always compared to values that are in 1/8th pel units + xd->mb_to_left_edge = -((mb_col * 16) << 3); + xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3; - xd->left_available = (mb_col != 0); + xd->dst.y_buffer = pc->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset; + xd->dst.u_buffer = pc->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset; + xd->dst.v_buffer = pc->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset; - // Select the appropriate reference frame for this MB - if (xd->mbmi.ref_frame == LAST_FRAME) - { - xd->pre.y_buffer = pc->last_frame.y_buffer + recon_yoffset; - xd->pre.u_buffer = pc->last_frame.u_buffer + recon_uvoffset; - xd->pre.v_buffer = pc->last_frame.v_buffer + recon_uvoffset; - } - else if (xd->mbmi.ref_frame == GOLDEN_FRAME) - { - // Golden frame reconstruction buffer - xd->pre.y_buffer = pc->golden_frame.y_buffer + recon_yoffset; - xd->pre.u_buffer = pc->golden_frame.u_buffer + recon_uvoffset; - xd->pre.v_buffer = pc->golden_frame.v_buffer + recon_uvoffset; - } - else - { - // Alternate reference frame reconstruction buffer - xd->pre.y_buffer = pc->alt_ref_frame.y_buffer + recon_yoffset; - xd->pre.u_buffer = pc->alt_ref_frame.u_buffer + recon_uvoffset; - xd->pre.v_buffer = pc->alt_ref_frame.v_buffer + recon_uvoffset; - } + xd->left_available = (mb_col != 0); - vp8_build_uvmvs(xd, pc->full_pixel); + // Select the appropriate reference frame for this MB + if (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME) + ref_fb_idx = pc->lst_fb_idx; + else if (xd->mode_info_context->mbmi.ref_frame == GOLDEN_FRAME) + ref_fb_idx = pc->gld_fb_idx; + else + ref_fb_idx = pc->alt_fb_idx; - vp8dx_bool_decoder_fill(xd->current_bc); - vp8_decode_macroblock(pbi, xd); + xd->pre.y_buffer = pc->yv12_fb[ref_fb_idx].y_buffer + recon_yoffset; + xd->pre.u_buffer = pc->yv12_fb[ref_fb_idx].u_buffer + recon_uvoffset; + xd->pre.v_buffer = pc->yv12_fb[ref_fb_idx].v_buffer + recon_uvoffset; + vp8_build_uvmvs(xd, pc->full_pixel); - recon_yoffset += 16; - recon_uvoffset += 8; + vp8_decode_macroblock(pbi, xd); - ++xd->mode_info_context; /* next mb */ + recon_yoffset += 16; + recon_uvoffset += 8; - xd->gf_active_ptr++; // GF useage flag for next MB + ++xd->mode_info_context; /* next mb */ - xd->above_context[Y1CONTEXT] += 4; - xd->above_context[UCONTEXT ] += 2; - xd->above_context[VCONTEXT ] += 2; - xd->above_context[Y2CONTEXT] ++; - pbi->mb_row_di[ithread].current_mb_col = mb_col; + xd->above_context++; - } + //pbi->mb_row_di[ithread].current_mb_col = mb_col; + pbi->current_mb_col[mb_row] = mb_col; + } - // adjust to the next row of mbs - vp8_extend_mb_row( - &pc->new_frame, + // adjust to the next row of mbs + vp8_extend_mb_row( + &pc->yv12_fb[dst_fb_idx], xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8 - ); + ); - ++xd->mode_info_context; /* skip prediction column */ + ++xd->mode_info_context; /* skip prediction column */ - // since we have multithread - xd->mode_info_context += xd->mode_info_stride * pbi->decoding_thread_count; + // since we have multithread + xd->mode_info_context += xd->mode_info_stride * pbi->decoding_thread_count; - //memcpy(&pbi->lpfmb, &pbi->mb, sizeof(pbi->mb)); - if ((mb_row & 1) == 1) - { pbi->last_mb_row_decoded = mb_row; - //printf("S%d", pbi->last_mb_row_decoded); - } - - if (ithread == (pbi->decoding_thread_count - 1) || mb_row == pc->mb_rows - 1) - { - //SetEvent(pbi->h_event_main); - sem_post(&pbi->h_event_main); } } } - } + // If |pbi->common.filter_level| is 0 the value can change in-between + // the sem_post and the check to call vp8_thread_loop_filter. + current_filter_level = pbi->common.filter_level; + + // add this to each frame + if ((mbrd->mb_row == pbi->common.mb_rows-1) || ((mbrd->mb_row == pbi->common.mb_rows-2) && (pbi->common.mb_rows % (pbi->decoding_thread_count+1))==1)) + { + //SetEvent(pbi->h_event_end_decoding); + sem_post(&pbi->h_event_end_decoding); + } + + if ((pbi->b_multithreaded_lf) && (current_filter_level)) + vp8_thread_loop_filter(pbi, mbrd, ithread); + + } #else (void) p_data; #endif @@ -253,93 +302,60 @@ THREAD_FUNCTION vp8_thread_decoding_proc(void *p_data) return 0 ; } -THREAD_FUNCTION vp8_thread_loop_filter(void *p_data) + +void vp8_thread_loop_filter(VP8D_COMP *pbi, MB_ROW_DEC *mbrd, int ithread) { #if CONFIG_MULTITHREAD - VP8D_COMP *pbi = (VP8D_COMP *)p_data; - while (1) - { - if (pbi->b_multithreaded_lf == 0) - break; - - //printf("before waiting for start_lpf\n"); - - //if(WaitForSingleObject(pbi->h_event_start_lpf, INFINITE) == WAIT_OBJECT_0) - if (sem_wait(&pbi->h_event_start_lpf) == 0) + if (sem_wait(&pbi->h_event_start_lpf[ithread]) == 0) { - if (pbi->b_multithreaded_lf == 0) // we're shutting down - break; - else + // if (pbi->b_multithreaded_lf == 0) // we're shutting down ???? + // break; + // else { - VP8_COMMON *cm = &pbi->common; - MACROBLOCKD *mbd = &pbi->lpfmb; + MACROBLOCKD *mbd = &mbrd->mbd; int default_filt_lvl = pbi->common.filter_level; - YV12_BUFFER_CONFIG *post = &cm->new_frame; + YV12_BUFFER_CONFIG *post = cm->frame_to_show; loop_filter_info *lfi = cm->lf_info; + //int frame_type = cm->frame_type; int mb_row; int mb_col; - - int baseline_filter_level[MAX_MB_SEGMENTS]; int filter_level; int alt_flt_enabled = mbd->segmentation_enabled; int i; unsigned char *y_ptr, *u_ptr, *v_ptr; - volatile int *last_mb_row_decoded = &pbi->last_mb_row_decoded; - - //MODE_INFO * this_mb_mode_info = cm->mi; - mbd->mode_info_context = cm->mi; // Point at base of Mb MODE_INFO list - - // Note the baseline filter values for each segment - if (alt_flt_enabled) - { - for (i = 0; i < MAX_MB_SEGMENTS; i++) - { - if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA) - baseline_filter_level[i] = mbd->segment_feature_data[MB_LVL_ALT_LF][i]; - else - { - baseline_filter_level[i] = default_filt_lvl + mbd->segment_feature_data[MB_LVL_ALT_LF][i]; - baseline_filter_level[i] = (baseline_filter_level[i] >= 0) ? ((baseline_filter_level[i] <= MAX_LOOP_FILTER) ? baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; // Clamp to valid range - } - } - } - else - { - for (i = 0; i < MAX_MB_SEGMENTS; i++) - baseline_filter_level[i] = default_filt_lvl; - } - - // Initialize the loop filter for this frame. - vp8_init_loop_filter(cm); + volatile int *last_row_current_mb_col; // Set up the buffer pointers - y_ptr = post->y_buffer; - u_ptr = post->u_buffer; - v_ptr = post->v_buffer; + y_ptr = post->y_buffer + post->y_stride * 16 * (ithread +1); + u_ptr = post->u_buffer + post->uv_stride * 8 * (ithread +1); + v_ptr = post->v_buffer + post->uv_stride * 8 * (ithread +1); // vp8_filter each macro block - for (mb_row = 0; mb_row < cm->mb_rows; mb_row++) + for (mb_row = ithread+1; mb_row < cm->mb_rows; mb_row+= (pbi->decoding_thread_count + 1)) { + last_row_current_mb_col = &pbi->current_mb_col[mb_row -1]; - while (mb_row >= *last_mb_row_decoded) - { - x86_pause_hint(); - thread_sleep(0); - } - - //printf("R%d", mb_row); for (mb_col = 0; mb_col < cm->mb_cols; mb_col++) { int Segment = (alt_flt_enabled) ? mbd->mode_info_context->mbmi.segment_id : 0; - filter_level = baseline_filter_level[Segment]; + if ((mb_col & 7) == 0) + { + while (mb_col > (*last_row_current_mb_col-8) && *last_row_current_mb_col != cm->mb_cols - 1) + { + x86_pause_hint(); + thread_sleep(0); + } + } + + filter_level = pbi->mt_baseline_filter_level[Segment]; // Apply any context driven MB level adjustment vp8_adjust_mb_lf_value(mbd, &filter_level); @@ -365,29 +381,28 @@ THREAD_FUNCTION vp8_thread_loop_filter(void *p_data) v_ptr += 8; mbd->mode_info_context++; // step to next MB - + pbi->current_mb_col[mb_row] = mb_col; } - y_ptr += post->y_stride * 16 - post->y_width; - u_ptr += post->uv_stride * 8 - post->uv_width; - v_ptr += post->uv_stride * 8 - post->uv_width; - mbd->mode_info_context++; // Skip border mb - } - //printf("R%d\n", mb_row); - // When done, signal main thread that ME is finished - //SetEvent(pbi->h_event_lpf); - sem_post(&pbi->h_event_lpf); - } + y_ptr += post->y_stride * 16 * (pbi->decoding_thread_count + 1) - post->y_width; + u_ptr += post->uv_stride * 8 * (pbi->decoding_thread_count + 1) - post->uv_width; + v_ptr += post->uv_stride * 8 * (pbi->decoding_thread_count + 1) - post->uv_width; + mbd->mode_info_context += pbi->decoding_thread_count * mbd->mode_info_stride; // Skip border mb + } + } } - } + // add this to each frame + if ((mbrd->mb_row == pbi->common.mb_rows-1) || ((mbrd->mb_row == pbi->common.mb_rows-2) && (pbi->common.mb_rows % (pbi->decoding_thread_count+1))==1)) + { + sem_post(&pbi->h_event_end_lpf); + } #else - (void) p_data; + (void) pbi; #endif - return 0; } void vp8_decoder_create_threads(VP8D_COMP *pbi) @@ -399,39 +414,38 @@ void vp8_decoder_create_threads(VP8D_COMP *pbi) pbi->b_multithreaded_rd = 0; pbi->b_multithreaded_lf = 0; pbi->allocated_decoding_thread_count = 0; - core_count = (pbi->max_threads > 16) ? 16 : pbi->max_threads; //vp8_get_proc_core_count(); - if (core_count > 1) - { - sem_init(&pbi->h_event_lpf, 0, 0); - sem_init(&pbi->h_event_start_lpf, 0, 0); - pbi->b_multithreaded_lf = 1; - pthread_create(&pbi->h_thread_lpf, 0, vp8_thread_loop_filter, (pbi)); - } + core_count = (pbi->max_threads > 16) ? 16 : pbi->max_threads; if (core_count > 1) { pbi->b_multithreaded_rd = 1; - pbi->decoding_thread_count = core_count - 1; + pbi->b_multithreaded_lf = 1; // this can be merged with pbi->b_multithreaded_rd ? + pbi->decoding_thread_count = core_count -1; CHECK_MEM_ERROR(pbi->h_decoding_thread, vpx_malloc(sizeof(pthread_t) * pbi->decoding_thread_count)); - CHECK_MEM_ERROR(pbi->h_event_mbrdecoding, vpx_malloc(sizeof(sem_t) * pbi->decoding_thread_count)); + CHECK_MEM_ERROR(pbi->h_event_start_decoding, vpx_malloc(sizeof(sem_t) * pbi->decoding_thread_count)); CHECK_MEM_ERROR(pbi->mb_row_di, vpx_memalign(32, sizeof(MB_ROW_DEC) * pbi->decoding_thread_count)); vpx_memset(pbi->mb_row_di, 0, sizeof(MB_ROW_DEC) * pbi->decoding_thread_count); CHECK_MEM_ERROR(pbi->de_thread_data, vpx_malloc(sizeof(DECODETHREAD_DATA) * pbi->decoding_thread_count)); + CHECK_MEM_ERROR(pbi->current_mb_col, vpx_malloc(sizeof(int) * MAX_ROWS)); // pc->mb_rows)); + CHECK_MEM_ERROR(pbi->h_event_start_lpf, vpx_malloc(sizeof(sem_t) * pbi->decoding_thread_count)); + for (ithread = 0; ithread < pbi->decoding_thread_count; ithread++) { - sem_init(&pbi->h_event_mbrdecoding[ithread], 0, 0); + sem_init(&pbi->h_event_start_decoding[ithread], 0, 0); + sem_init(&pbi->h_event_start_lpf[ithread], 0, 0); pbi->de_thread_data[ithread].ithread = ithread; pbi->de_thread_data[ithread].ptr1 = (void *)pbi; pbi->de_thread_data[ithread].ptr2 = (void *) &pbi->mb_row_di[ithread]; pthread_create(&pbi->h_decoding_thread[ithread], 0, vp8_thread_decoding_proc, (&pbi->de_thread_data[ithread])); - } - sem_init(&pbi->h_event_main, 0, 0); + sem_init(&pbi->h_event_end_decoding, 0, 0); + sem_init(&pbi->h_event_end_lpf, 0, 0); + pbi->allocated_decoding_thread_count = pbi->decoding_thread_count; } @@ -446,39 +460,35 @@ void vp8_decoder_remove_threads(VP8D_COMP *pbi) if (pbi->b_multithreaded_lf) { + int i; pbi->b_multithreaded_lf = 0; - sem_post(&pbi->h_event_start_lpf); - pthread_join(pbi->h_thread_lpf, 0); - sem_destroy(&pbi->h_event_start_lpf); + + for (i = 0; i < pbi->allocated_decoding_thread_count; i++) + sem_destroy(&pbi->h_event_start_lpf[i]); + + sem_destroy(&pbi->h_event_end_lpf); } //shutdown MB Decoding thread; if (pbi->b_multithreaded_rd) { + int i; + pbi->b_multithreaded_rd = 0; + // allow all threads to exit + for (i = 0; i < pbi->allocated_decoding_thread_count; i++) { - int i; - - for (i = 0; i < pbi->allocated_decoding_thread_count; i++) - { - - sem_post(&pbi->h_event_mbrdecoding[i]); - pthread_join(pbi->h_decoding_thread[i], NULL); - } + sem_post(&pbi->h_event_start_decoding[i]); + pthread_join(pbi->h_decoding_thread[i], NULL); } - { - - int i; - for (i = 0; i < pbi->allocated_decoding_thread_count; i++) - { - sem_destroy(&pbi->h_event_mbrdecoding[i]); - } - + for (i = 0; i < pbi->allocated_decoding_thread_count; i++) + { + sem_destroy(&pbi->h_event_start_decoding[i]); } - sem_destroy(&pbi->h_event_main); + sem_destroy(&pbi->h_event_end_decoding); if (pbi->h_decoding_thread) { @@ -486,10 +496,16 @@ void vp8_decoder_remove_threads(VP8D_COMP *pbi) pbi->h_decoding_thread = NULL; } - if (pbi->h_event_mbrdecoding) + if (pbi->h_event_start_decoding) { - vpx_free(pbi->h_event_mbrdecoding); - pbi->h_event_mbrdecoding = NULL; + vpx_free(pbi->h_event_start_decoding); + pbi->h_event_start_decoding = NULL; + } + + if (pbi->h_event_start_lpf) + { + vpx_free(pbi->h_event_start_lpf); + pbi->h_event_start_lpf = NULL; } if (pbi->mb_row_di) @@ -503,8 +519,13 @@ void vp8_decoder_remove_threads(VP8D_COMP *pbi) vpx_free(pbi->de_thread_data); pbi->de_thread_data = NULL; } - } + if (pbi->current_mb_col) + { + vpx_free(pbi->current_mb_col); + pbi->current_mb_col = NULL ; + } + } #else (void) pbi; #endif @@ -514,9 +535,12 @@ void vp8_decoder_remove_threads(VP8D_COMP *pbi) void vp8_start_lfthread(VP8D_COMP *pbi) { #if CONFIG_MULTITHREAD + /* memcpy(&pbi->lpfmb, &pbi->mb, sizeof(pbi->mb)); pbi->last_mb_row_decoded = 0; sem_post(&pbi->h_event_start_lpf); + */ + (void) pbi; #else (void) pbi; #endif @@ -525,14 +549,17 @@ void vp8_start_lfthread(VP8D_COMP *pbi) void vp8_stop_lfthread(VP8D_COMP *pbi) { #if CONFIG_MULTITHREAD + /* struct vpx_usec_timer timer; vpx_usec_timer_start(&timer); - sem_wait(&pbi->h_event_lpf); + sem_wait(&pbi->h_event_end_lpf); vpx_usec_timer_mark(&timer); pbi->time_loop_filtering += vpx_usec_timer_elapsed(&timer); + */ + (void) pbi; #else (void) pbi; #endif @@ -548,49 +575,246 @@ void vp8_mtdecode_mb_rows(VP8D_COMP *pbi, int ibc = 0; int num_part = 1 << pbi->common.multi_token_partition; + int i; + volatile int *last_row_current_mb_col = NULL; vp8_setup_decoding_thread_data(pbi, xd, pbi->mb_row_di, pbi->decoding_thread_count); + for (i = 0; i < pbi->decoding_thread_count; i++) + sem_post(&pbi->h_event_start_decoding[i]); + for (mb_row = 0; mb_row < pc->mb_rows; mb_row += (pbi->decoding_thread_count + 1)) { int i; - pbi->current_mb_col_main = -1; - - xd->current_bc = &pbi->mbc[ibc]; - ibc++ ; - if (ibc == num_part) - ibc = 0; + xd->current_bc = &pbi->mbc[mb_row%num_part]; - for (i = 0; i < pbi->decoding_thread_count; i++) + //vp8_decode_mb_row(pbi, pc, mb_row, xd); { - if ((mb_row + i + 1) >= pc->mb_rows) - break; + int i; + int recon_yoffset, recon_uvoffset; + int mb_col; + int ref_fb_idx = pc->lst_fb_idx; + int dst_fb_idx = pc->new_fb_idx; + int recon_y_stride = pc->yv12_fb[ref_fb_idx].y_stride; + int recon_uv_stride = pc->yv12_fb[ref_fb_idx].uv_stride; - pbi->mb_row_di[i].mb_row = mb_row + i + 1; - pbi->mb_row_di[i].mbd.current_bc = &pbi->mbc[ibc]; - ibc++; + // volatile int *last_row_current_mb_col = NULL; + if (mb_row > 0) + last_row_current_mb_col = &pbi->current_mb_col[mb_row -1]; - if (ibc == num_part) - ibc = 0; + vpx_memset(&pc->left_context, 0, sizeof(pc->left_context)); + recon_yoffset = mb_row * recon_y_stride * 16; + recon_uvoffset = mb_row * recon_uv_stride * 8; + // reset above block coeffs - pbi->mb_row_di[i].current_mb_col = -1; - sem_post(&pbi->h_event_mbrdecoding[i]); - } + xd->above_context = pc->above_context; + xd->up_available = (mb_row != 0); + + xd->mb_to_top_edge = -((mb_row * 16)) << 3; + xd->mb_to_bottom_edge = ((pc->mb_rows - 1 - mb_row) * 16) << 3; + + for (mb_col = 0; mb_col < pc->mb_cols; mb_col++) + { + if ( mb_row > 0 && (mb_col & 7) == 0){ + while (mb_col > (*last_row_current_mb_col - 8) && *last_row_current_mb_col != pc->mb_cols - 1) + { + x86_pause_hint(); + thread_sleep(0); + } + } + + if (xd->mode_info_context->mbmi.mode == SPLITMV || xd->mode_info_context->mbmi.mode == B_PRED) + { + for (i = 0; i < 16; i++) + { + BLOCKD *d = &xd->block[i]; + vpx_memcpy(&d->bmi, &xd->mode_info_context->bmi[i], sizeof(B_MODE_INFO)); + } + } + + // Distance of Mb to the various image edges. + // These specified to 8th pel as they are always compared to values that are in 1/8th pel units + xd->mb_to_left_edge = -((mb_col * 16) << 3); + xd->mb_to_right_edge = ((pc->mb_cols - 1 - mb_col) * 16) << 3; + + xd->dst.y_buffer = pc->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset; + xd->dst.u_buffer = pc->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset; + xd->dst.v_buffer = pc->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset; + + xd->left_available = (mb_col != 0); + + // Select the appropriate reference frame for this MB + if (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME) + ref_fb_idx = pc->lst_fb_idx; + else if (xd->mode_info_context->mbmi.ref_frame == GOLDEN_FRAME) + ref_fb_idx = pc->gld_fb_idx; + else + ref_fb_idx = pc->alt_fb_idx; + + xd->pre.y_buffer = pc->yv12_fb[ref_fb_idx].y_buffer + recon_yoffset; + xd->pre.u_buffer = pc->yv12_fb[ref_fb_idx].u_buffer + recon_uvoffset; + xd->pre.v_buffer = pc->yv12_fb[ref_fb_idx].v_buffer + recon_uvoffset; + + vp8_build_uvmvs(xd, pc->full_pixel); + + vp8_decode_macroblock(pbi, xd); - vp8_decode_mb_row(pbi, pc, mb_row, xd); + recon_yoffset += 16; + recon_uvoffset += 8; + ++xd->mode_info_context; /* next mb */ + + xd->above_context++; + + //pbi->current_mb_col_main = mb_col; + pbi->current_mb_col[mb_row] = mb_col; + } + + // adjust to the next row of mbs + vp8_extend_mb_row( + &pc->yv12_fb[dst_fb_idx], + xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8 + ); + + ++xd->mode_info_context; /* skip prediction column */ + + pbi->last_mb_row_decoded = mb_row; + } xd->mode_info_context += xd->mode_info_stride * pbi->decoding_thread_count; + } + + sem_wait(&pbi->h_event_end_decoding); // add back for each frame +#else + (void) pbi; + (void) xd; +#endif +} + - if (mb_row < pc->mb_rows - 1) +void vp8_mt_loop_filter_frame( VP8D_COMP *pbi) +{ +#if CONFIG_MULTITHREAD + VP8_COMMON *cm = &pbi->common; + MACROBLOCKD *mbd = &pbi->mb; + int default_filt_lvl = pbi->common.filter_level; + + YV12_BUFFER_CONFIG *post = cm->frame_to_show; + loop_filter_info *lfi = cm->lf_info; + int frame_type = cm->frame_type; + + int mb_row; + int mb_col; + + int filter_level; + int alt_flt_enabled = mbd->segmentation_enabled; + + int i; + unsigned char *y_ptr, *u_ptr, *v_ptr; + + volatile int *last_row_current_mb_col=NULL; + + vp8_setup_loop_filter_thread_data(pbi, mbd, pbi->mb_row_di, pbi->decoding_thread_count); + + mbd->mode_info_context = cm->mi; // Point at base of Mb MODE_INFO list + + // Note the baseline filter values for each segment + if (alt_flt_enabled) + { + for (i = 0; i < MAX_MB_SEGMENTS; i++) { - sem_wait(&pbi->h_event_main); + // Abs value + if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA) + pbi->mt_baseline_filter_level[i] = mbd->segment_feature_data[MB_LVL_ALT_LF][i]; + // Delta Value + else + { + pbi->mt_baseline_filter_level[i] = default_filt_lvl + mbd->segment_feature_data[MB_LVL_ALT_LF][i]; + pbi->mt_baseline_filter_level[i] = (pbi->mt_baseline_filter_level[i] >= 0) ? ((pbi->mt_baseline_filter_level[i] <= MAX_LOOP_FILTER) ? pbi->mt_baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; // Clamp to valid range + } + } + } + else + { + for (i = 0; i < MAX_MB_SEGMENTS; i++) + pbi->mt_baseline_filter_level[i] = default_filt_lvl; + } + + // Initialize the loop filter for this frame. + if ((cm->last_filter_type != cm->filter_type) || (cm->last_sharpness_level != cm->sharpness_level)) + vp8_init_loop_filter(cm); + else if (frame_type != cm->last_frame_type) + vp8_frame_init_loop_filter(lfi, frame_type); + + for (i = 0; i < pbi->decoding_thread_count; i++) + sem_post(&pbi->h_event_start_lpf[i]); + // sem_post(&pbi->h_event_start_lpf); + + // Set up the buffer pointers + y_ptr = post->y_buffer; + u_ptr = post->u_buffer; + v_ptr = post->v_buffer; + + // vp8_filter each macro block + for (mb_row = 0; mb_row < cm->mb_rows; mb_row+= (pbi->decoding_thread_count + 1)) + { + if (mb_row > 0) + last_row_current_mb_col = &pbi->current_mb_col[mb_row -1]; + + for (mb_col = 0; mb_col < cm->mb_cols; mb_col++) + { + int Segment = (alt_flt_enabled) ? mbd->mode_info_context->mbmi.segment_id : 0; + + if ( mb_row > 0 && (mb_col & 7) == 0){ + // if ( mb_row > 0 ){ + while (mb_col > (*last_row_current_mb_col-8) && *last_row_current_mb_col != cm->mb_cols - 1) + { + x86_pause_hint(); + thread_sleep(0); + } + } + + filter_level = pbi->mt_baseline_filter_level[Segment]; + + // Distance of Mb to the various image edges. + // These specified to 8th pel as they are always compared to values that are in 1/8th pel units + // Apply any context driven MB level adjustment + vp8_adjust_mb_lf_value(mbd, &filter_level); + + if (filter_level) + { + if (mb_col > 0) + cm->lf_mbv(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf); + + if (mbd->mode_info_context->mbmi.dc_diff > 0) + cm->lf_bv(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf); + + // don't apply across umv border + if (mb_row > 0) + cm->lf_mbh(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf); + + if (mbd->mode_info_context->mbmi.dc_diff > 0) + cm->lf_bh(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf); + } + + y_ptr += 16; + u_ptr += 8; + v_ptr += 8; + + mbd->mode_info_context++; // step to next MB + pbi->current_mb_col[mb_row] = mb_col; } + mbd->mode_info_context++; // Skip border mb + + //update for multi-thread + y_ptr += post->y_stride * 16 * (pbi->decoding_thread_count + 1) - post->y_width; + u_ptr += post->uv_stride * 8 * (pbi->decoding_thread_count + 1) - post->uv_width; + v_ptr += post->uv_stride * 8 * (pbi->decoding_thread_count + 1) - post->uv_width; + mbd->mode_info_context += pbi->decoding_thread_count * mbd->mode_info_stride; } - pbi->last_mb_row_decoded = mb_row; + sem_wait(&pbi->h_event_end_lpf); #else (void) pbi; - (void) xd; #endif } diff --git a/vp8/decoder/treereader.h b/vp8/decoder/treereader.h index eb10e24..2778428 100644 --- a/vp8/decoder/treereader.h +++ b/vp8/decoder/treereader.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/decoder/x86/dequantize_mmx.asm b/vp8/decoder/x86/dequantize_mmx.asm index 02be487..150d090 100644 --- a/vp8/decoder/x86/dequantize_mmx.asm +++ b/vp8/decoder/x86/dequantize_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -49,12 +50,12 @@ sym(vp8_dequantize_b_impl_mmx): ret -;void dequant_idct_mmx(short *input, short *dq, short *output, int pitch) -global sym(vp8_dequant_idct_mmx) -sym(vp8_dequant_idct_mmx): +;void dequant_idct_add_mmx(short *input, short *dq, unsigned char *pred, unsigned char *dest, int pitch, int stride) +global sym(vp8_dequant_idct_add_mmx) +sym(vp8_dequant_idct_add_mmx): push rbp mov rbp, rsp - SHADOW_ARGS_TO_STACK 4 + SHADOW_ARGS_TO_STACK 6 GET_GOT rbx push rsi push rdi @@ -76,7 +77,8 @@ sym(vp8_dequant_idct_mmx): movq mm3, [rax+24] pmullw mm3, [rdx+24] - mov rdx, arg(2) ;output + mov rdx, arg(3) ;dest + mov rsi, arg(2) ;pred pxor mm7, mm7 @@ -87,7 +89,8 @@ sym(vp8_dequant_idct_mmx): movq [rax+24],mm7 - movsxd rax, dword ptr arg(3) ;pitch + movsxd rax, dword ptr arg(4) ;pitch + movsxd rdi, dword ptr arg(5) ;stride psubw mm0, mm2 ; b1= 0-2 paddw mm2, mm2 ; @@ -206,13 +209,34 @@ sym(vp8_dequant_idct_mmx): punpckldq mm2, mm4 ; 32 22 12 02 punpckhdq mm5, mm4 ; 33 23 13 03 - movq [rdx], mm0 + pxor mm7, mm7 - movq [rdx+rax], mm1 - movq [rdx+rax*2], mm2 + movd mm4, [rsi] + punpcklbw mm4, mm7 + paddsw mm0, mm4 + packuswb mm0, mm7 + movd [rdx], mm0 - add rdx, rax - movq [rdx+rax*2], mm5 + movd mm4, [rsi+rax] + punpcklbw mm4, mm7 + paddsw mm1, mm4 + packuswb mm1, mm7 + movd [rdx+rdi], mm1 + + movd mm4, [rsi+2*rax] + punpcklbw mm4, mm7 + paddsw mm2, mm4 + packuswb mm2, mm7 + movd [rdx+rdi*2], mm2 + + add rdx, rdi + add rsi, rax + + movd mm4, [rsi+2*rax] + punpcklbw mm4, mm7 + paddsw mm5, mm4 + packuswb mm5, mm7 + movd [rdx+rdi*2], mm5 ; begin epilog pop rdi @@ -223,12 +247,12 @@ sym(vp8_dequant_idct_mmx): ret -;void dequant_dc_idct_mmx(short *input, short *dq, short *output, int pitch, int Dc) -global sym(vp8_dequant_dc_idct_mmx) -sym(vp8_dequant_dc_idct_mmx): +;void dequant_dc_idct_add_mmx(short *input, short *dq, unsigned char *pred, unsigned char *dest, int pitch, int stride, int Dc) +global sym(vp8_dequant_dc_idct_add_mmx) +sym(vp8_dequant_dc_idct_add_mmx): push rbp mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 + SHADOW_ARGS_TO_STACK 7 GET_GOT rbx push rsi push rdi @@ -237,8 +261,6 @@ sym(vp8_dequant_dc_idct_mmx): mov rax, arg(0) ;input mov rdx, arg(1) ;dq - movsxd rcx, dword ptr arg(4) ;Dc - movq mm0, [rax ] pmullw mm0, [rdx] @@ -251,7 +273,8 @@ sym(vp8_dequant_dc_idct_mmx): movq mm3, [rax+24] pmullw mm3, [rdx+24] - mov rdx, arg(2) ;output + mov rdx, arg(3) ;dest + mov rsi, arg(2) ;pred pxor mm7, mm7 @@ -261,8 +284,15 @@ sym(vp8_dequant_dc_idct_mmx): movq [rax+16],mm7 movq [rax+24],mm7 - pinsrw mm0, rcx, 0 - movsxd rax, dword ptr arg(3) ;pitch + ; move lower word of Dc to lower word of mm0 + psrlq mm0, 16 + movzx rcx, word ptr arg(6) ;Dc + psllq mm0, 16 + movd mm7, rcx + por mm0, mm7 + + movsxd rax, dword ptr arg(4) ;pitch + movsxd rdi, dword ptr arg(5) ;stride psubw mm0, mm2 ; b1= 0-2 paddw mm2, mm2 ; @@ -381,13 +411,34 @@ sym(vp8_dequant_dc_idct_mmx): punpckldq mm2, mm4 ; 32 22 12 02 punpckhdq mm5, mm4 ; 33 23 13 03 - movq [rdx], mm0 - - movq [rdx+rax], mm1 - movq [rdx+rax*2], mm2 - - add rdx, rax - movq [rdx+rax*2], mm5 + pxor mm7, mm7 + + movd mm4, [rsi] + punpcklbw mm4, mm7 + paddsw mm0, mm4 + packuswb mm0, mm7 + movd [rdx], mm0 + + movd mm4, [rsi+rax] + punpcklbw mm4, mm7 + paddsw mm1, mm4 + packuswb mm1, mm7 + movd [rdx+rdi], mm1 + + movd mm4, [rsi+2*rax] + punpcklbw mm4, mm7 + paddsw mm2, mm4 + packuswb mm2, mm7 + movd [rdx+rdi*2], mm2 + + add rdx, rdi + add rsi, rax + + movd mm4, [rsi+2*rax] + punpcklbw mm4, mm7 + paddsw mm5, mm4 + packuswb mm5, mm7 + movd [rdx+rdi*2], mm5 ; begin epilog pop rdi diff --git a/vp8/decoder/x86/dequantize_x86.h b/vp8/decoder/x86/dequantize_x86.h index 5def406..dc68daa 100644 --- a/vp8/decoder/x86/dequantize_x86.h +++ b/vp8/decoder/x86/dequantize_x86.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -20,19 +21,48 @@ */ #if HAVE_MMX extern prototype_dequant_block(vp8_dequantize_b_mmx); -extern prototype_dequant_idct(vp8_dequant_idct_mmx); -extern prototype_dequant_idct_dc(vp8_dequant_dc_idct_mmx); - +extern prototype_dequant_idct_add(vp8_dequant_idct_add_mmx); +extern prototype_dequant_dc_idct_add(vp8_dequant_dc_idct_add_mmx); +extern prototype_dequant_dc_idct_add_y_block(vp8_dequant_dc_idct_add_y_block_mmx); +extern prototype_dequant_idct_add_y_block(vp8_dequant_idct_add_y_block_mmx); +extern prototype_dequant_idct_add_uv_block(vp8_dequant_idct_add_uv_block_mmx); #if !CONFIG_RUNTIME_CPU_DETECT #undef vp8_dequant_block #define vp8_dequant_block vp8_dequantize_b_mmx -#undef vp8_dequant_idct -#define vp8_dequant_idct vp8_dequant_idct_mmx +#undef vp8_dequant_idct_add +#define vp8_dequant_idct_add vp8_dequant_idct_add_mmx + +#undef vp8_dequant_dc_idct_add +#define vp8_dequant_dc_idct_add vp8_dequant_dc_idct_add_mmx + +#undef vp8_dequant_dc_idct_add_y_block +#define vp8_dequant_dc_idct_add_y_block vp8_dequant_dc_idct_add_y_block_mmx + +#undef vp8_dequant_idct_add_y_block +#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_mmx + +#undef vp8_dequant_idct_add_uv_block +#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_mmx + +#endif +#endif + +#if HAVE_SSE2 +extern prototype_dequant_dc_idct_add_y_block(vp8_dequant_dc_idct_add_y_block_sse2); +extern prototype_dequant_idct_add_y_block(vp8_dequant_idct_add_y_block_sse2); +extern prototype_dequant_idct_add_uv_block(vp8_dequant_idct_add_uv_block_sse2); + +#if !CONFIG_RUNTIME_CPU_DETECT +#undef vp8_dequant_dc_idct_add_y_block +#define vp8_dequant_dc_idct_add_y_block vp8_dequant_dc_idct_add_y_block_sse2 + +#undef vp8_dequant_idct_add_y_block +#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_sse2 -#undef vp8_dequant_idct_dc -#define vp8_dequant_idct_dc vp8_dequant_dc_idct_mmx +#undef vp8_dequant_idct_add_uv_block +#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_sse2 #endif #endif diff --git a/vp8/decoder/x86/idct_blk_mmx.c b/vp8/decoder/x86/idct_blk_mmx.c new file mode 100644 index 0000000..78c91d3 --- /dev/null +++ b/vp8/decoder/x86/idct_blk_mmx.c @@ -0,0 +1,151 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include "vpx_ports/config.h" +#include "idct.h" +#include "dequantize.h" + +void vp8_dequant_dc_idct_add_y_block_mmx + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int stride, char *eobs, short *dc) +{ + int i; + + for (i = 0; i < 4; i++) + { + if (eobs[0] > 1) + vp8_dequant_dc_idct_add_mmx (q, dq, pre, dst, 16, stride, dc[0]); + else + vp8_dc_only_idct_add_mmx (dc[0], pre, dst, 16, stride); + + if (eobs[1] > 1) + vp8_dequant_dc_idct_add_mmx (q+16, dq, pre+4, dst+4, 16, stride, dc[1]); + else + vp8_dc_only_idct_add_mmx (dc[1], pre+4, dst+4, 16, stride); + + if (eobs[2] > 1) + vp8_dequant_dc_idct_add_mmx (q+32, dq, pre+8, dst+8, 16, stride, dc[2]); + else + vp8_dc_only_idct_add_mmx (dc[2], pre+8, dst+8, 16, stride); + + if (eobs[3] > 1) + vp8_dequant_dc_idct_add_mmx (q+48, dq, pre+12, dst+12, 16, stride, dc[3]); + else + vp8_dc_only_idct_add_mmx (dc[3], pre+12, dst+12, 16, stride); + + q += 64; + dc += 4; + pre += 64; + dst += 4*stride; + eobs += 4; + } +} + +void vp8_dequant_idct_add_y_block_mmx + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int stride, char *eobs) +{ + int i; + + for (i = 0; i < 4; i++) + { + if (eobs[0] > 1) + vp8_dequant_idct_add_mmx (q, dq, pre, dst, 16, stride); + else + { + vp8_dc_only_idct_add_mmx (q[0]*dq[0], pre, dst, 16, stride); + ((int *)q)[0] = 0; + } + + if (eobs[1] > 1) + vp8_dequant_idct_add_mmx (q+16, dq, pre+4, dst+4, 16, stride); + else + { + vp8_dc_only_idct_add_mmx (q[16]*dq[0], pre+4, dst+4, 16, stride); + ((int *)(q+16))[0] = 0; + } + + if (eobs[2] > 1) + vp8_dequant_idct_add_mmx (q+32, dq, pre+8, dst+8, 16, stride); + else + { + vp8_dc_only_idct_add_mmx (q[32]*dq[0], pre+8, dst+8, 16, stride); + ((int *)(q+32))[0] = 0; + } + + if (eobs[3] > 1) + vp8_dequant_idct_add_mmx (q+48, dq, pre+12, dst+12, 16, stride); + else + { + vp8_dc_only_idct_add_mmx (q[48]*dq[0], pre+12, dst+12, 16, stride); + ((int *)(q+48))[0] = 0; + } + + q += 64; + pre += 64; + dst += 4*stride; + eobs += 4; + } +} + +void vp8_dequant_idct_add_uv_block_mmx + (short *q, short *dq, unsigned char *pre, + unsigned char *dstu, unsigned char *dstv, int stride, char *eobs) +{ + int i; + + for (i = 0; i < 2; i++) + { + if (eobs[0] > 1) + vp8_dequant_idct_add_mmx (q, dq, pre, dstu, 8, stride); + else + { + vp8_dc_only_idct_add_mmx (q[0]*dq[0], pre, dstu, 8, stride); + ((int *)q)[0] = 0; + } + + if (eobs[1] > 1) + vp8_dequant_idct_add_mmx (q+16, dq, pre+4, dstu+4, 8, stride); + else + { + vp8_dc_only_idct_add_mmx (q[16]*dq[0], pre+4, dstu+4, 8, stride); + ((int *)(q+16))[0] = 0; + } + + q += 32; + pre += 32; + dstu += 4*stride; + eobs += 2; + } + + for (i = 0; i < 2; i++) + { + if (eobs[0] > 1) + vp8_dequant_idct_add_mmx (q, dq, pre, dstv, 8, stride); + else + { + vp8_dc_only_idct_add_mmx (q[0]*dq[0], pre, dstv, 8, stride); + ((int *)q)[0] = 0; + } + + if (eobs[1] > 1) + vp8_dequant_idct_add_mmx (q+16, dq, pre+4, dstv+4, 8, stride); + else + { + vp8_dc_only_idct_add_mmx (q[16]*dq[0], pre+4, dstv+4, 8, stride); + ((int *)(q+16))[0] = 0; + } + + q += 32; + pre += 32; + dstv += 4*stride; + eobs += 2; + } +} diff --git a/vp8/decoder/x86/idct_blk_sse2.c b/vp8/decoder/x86/idct_blk_sse2.c new file mode 100644 index 0000000..0273d6e --- /dev/null +++ b/vp8/decoder/x86/idct_blk_sse2.c @@ -0,0 +1,114 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + +#include "vpx_ports/config.h" +#include "idct.h" +#include "dequantize.h" + +void idct_dequant_dc_0_2x_sse2 + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int dst_stride, short *dc); +void idct_dequant_dc_full_2x_sse2 + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int dst_stride, short *dc); + +void idct_dequant_0_2x_sse2 + (short *q, short *dq ,unsigned char *pre, + unsigned char *dst, int dst_stride, int blk_stride); +void idct_dequant_full_2x_sse2 + (short *q, short *dq ,unsigned char *pre, + unsigned char *dst, int dst_stride, int blk_stride); + +void vp8_dequant_dc_idct_add_y_block_sse2 + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int stride, char *eobs, short *dc) +{ + int i; + + for (i = 0; i < 4; i++) + { + if (((short *)(eobs))[0] & 0xfefe) + idct_dequant_dc_full_2x_sse2 (q, dq, pre, dst, stride, dc); + else + idct_dequant_dc_0_2x_sse2 (q, dq, pre, dst, stride, dc); + + if (((short *)(eobs))[1] & 0xfefe) + idct_dequant_dc_full_2x_sse2 (q+32, dq, pre+8, dst+8, stride, dc+2); + else + idct_dequant_dc_0_2x_sse2 (q+32, dq, pre+8, dst+8, stride, dc+2); + + q += 64; + dc += 4; + pre += 64; + dst += stride*4; + eobs += 4; + } +} + +void vp8_dequant_idct_add_y_block_sse2 + (short *q, short *dq, unsigned char *pre, + unsigned char *dst, int stride, char *eobs) +{ + int i; + + for (i = 0; i < 4; i++) + { + if (((short *)(eobs))[0] & 0xfefe) + idct_dequant_full_2x_sse2 (q, dq, pre, dst, stride, 16); + else + idct_dequant_0_2x_sse2 (q, dq, pre, dst, stride, 16); + + if (((short *)(eobs))[1] & 0xfefe) + idct_dequant_full_2x_sse2 (q+32, dq, pre+8, dst+8, stride, 16); + else + idct_dequant_0_2x_sse2 (q+32, dq, pre+8, dst+8, stride, 16); + + q += 64; + pre += 64; + dst += stride*4; + eobs += 4; + } +} + +void vp8_dequant_idct_add_uv_block_sse2 + (short *q, short *dq, unsigned char *pre, + unsigned char *dstu, unsigned char *dstv, int stride, char *eobs) +{ + if (((short *)(eobs))[0] & 0xfefe) + idct_dequant_full_2x_sse2 (q, dq, pre, dstu, stride, 8); + else + idct_dequant_0_2x_sse2 (q, dq, pre, dstu, stride, 8); + + q += 32; + pre += 32; + dstu += stride*4; + + if (((short *)(eobs))[1] & 0xfefe) + idct_dequant_full_2x_sse2 (q, dq, pre, dstu, stride, 8); + else + idct_dequant_0_2x_sse2 (q, dq, pre, dstu, stride, 8); + + q += 32; + pre += 32; + + if (((short *)(eobs))[2] & 0xfefe) + idct_dequant_full_2x_sse2 (q, dq, pre, dstv, stride, 8); + else + idct_dequant_0_2x_sse2 (q, dq, pre, dstv, stride, 8); + + q += 32; + pre += 32; + dstv += stride*4; + + if (((short *)(eobs))[3] & 0xfefe) + idct_dequant_full_2x_sse2 (q, dq, pre, dstv, stride, 8); + else + idct_dequant_0_2x_sse2 (q, dq, pre, dstv, stride, 8); +} diff --git a/vp8/decoder/x86/onyxdxv.c b/vp8/decoder/x86/onyxdxv.c index 75a676a..50293c7 100644 --- a/vp8/decoder/x86/onyxdxv.c +++ b/vp8/decoder/x86/onyxdxv.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/decoder/x86/x86_dsystemdependent.c b/vp8/decoder/x86/x86_dsystemdependent.c index 6d7cc36..47e346d 100644 --- a/vp8/decoder/x86/x86_dsystemdependent.c +++ b/vp8/decoder/x86/x86_dsystemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -38,14 +39,24 @@ void vp8_arch_x86_decode_init(VP8D_COMP *pbi) #if CONFIG_RUNTIME_CPU_DETECT /* Override default functions with fastest ones for this CPU. */ #if HAVE_MMX - if (flags & HAS_MMX) { - pbi->dequant.block = vp8_dequantize_b_mmx; - pbi->dequant.idct = vp8_dequant_idct_mmx; - pbi->dequant.idct_dc = vp8_dequant_dc_idct_mmx; + pbi->dequant.block = vp8_dequantize_b_mmx; + pbi->dequant.idct_add = vp8_dequant_idct_add_mmx; + pbi->dequant.dc_idct_add = vp8_dequant_dc_idct_add_mmx; + pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_mmx; + pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_mmx; + pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_mmx; + } +#endif +#if HAVE_SSE2 + if (flags & HAS_SSE2) + { + pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_sse2; + pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_sse2; + pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_sse2; } - #endif + #endif } diff --git a/vp8/decoder/xprintf.c b/vp8/decoder/xprintf.c index cb2221c..e3b953e 100644 --- a/vp8/decoder/xprintf.c +++ b/vp8/decoder/xprintf.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/decoder/xprintf.h b/vp8/decoder/xprintf.h index 2f175e9..f83dd39 100644 --- a/vp8/decoder/xprintf.h +++ b/vp8/decoder/xprintf.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/arm/armv6/walsh_v6.asm b/vp8/encoder/arm/armv6/walsh_v6.asm index 608c9ae..61ffdb3 100644 --- a/vp8/encoder/arm/armv6/walsh_v6.asm +++ b/vp8/encoder/arm/armv6/walsh_v6.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; EXPORT |vp8_short_walsh4x4_armv6| diff --git a/vp8/encoder/arm/boolhuff_arm.c b/vp8/encoder/arm/boolhuff_arm.c index e70b3ad..fe8e70c 100644 --- a/vp8/encoder/arm/boolhuff_arm.c +++ b/vp8/encoder/arm/boolhuff_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/arm/csystemdependent.c b/vp8/encoder/arm/csystemdependent.c index 0039796..8d70d63 100644 --- a/vp8/encoder/arm/csystemdependent.c +++ b/vp8/encoder/arm/csystemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -62,7 +63,11 @@ void vp8_cmachine_specific_config(VP8_COMP *cpi) cpi->rtcd.encodemb.submbuv = vp8_subtract_mbuv_neon; cpi->rtcd.quantize.quantb = vp8_regular_quantize_b; - cpi->rtcd.quantize.fastquantb = vp8_fast_quantize_b_neon; + cpi->rtcd.quantize.fastquantb = vp8_fast_quantize_b_c; + /* The neon quantizer has not been updated to match the new exact + * quantizer introduced in commit e04e2935 + */ + /*cpi->rtcd.quantize.fastquantb = vp8_fast_quantize_b_neon;*/ #elif HAVE_ARMV6 cpi->rtcd.variance.sad16x16 = vp8_sad16x16_c; cpi->rtcd.variance.sad16x8 = vp8_sad16x8_c; diff --git a/vp8/encoder/arm/dct_arm.h b/vp8/encoder/arm/dct_arm.h index a671862..774599b 100644 --- a/vp8/encoder/arm/dct_arm.h +++ b/vp8/encoder/arm/dct_arm.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/arm/encodemb_arm.c b/vp8/encoder/arm/encodemb_arm.c index 3f1d053..cc9e014 100644 --- a/vp8/encoder/arm/encodemb_arm.c +++ b/vp8/encoder/arm/encodemb_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/arm/encodemb_arm.h b/vp8/encoder/arm/encodemb_arm.h index 28f9e5c..eb69943 100644 --- a/vp8/encoder/arm/encodemb_arm.h +++ b/vp8/encoder/arm/encodemb_arm.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/arm/mcomp_arm.c b/vp8/encoder/arm/mcomp_arm.c index 07f2186..4e95c47 100644 --- a/vp8/encoder/arm/mcomp_arm.c +++ b/vp8/encoder/arm/mcomp_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/arm/neon/boolhuff_armv7.asm b/vp8/encoder/arm/neon/boolhuff_armv7.asm index 9a5f366..9c4823c 100644 --- a/vp8/encoder/arm/neon/boolhuff_armv7.asm +++ b/vp8/encoder/arm/neon/boolhuff_armv7.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/fastfdct4x4_neon.asm b/vp8/encoder/arm/neon/fastfdct4x4_neon.asm index d5dec44..8c191a7 100644 --- a/vp8/encoder/arm/neon/fastfdct4x4_neon.asm +++ b/vp8/encoder/arm/neon/fastfdct4x4_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/fastfdct8x4_neon.asm b/vp8/encoder/arm/neon/fastfdct8x4_neon.asm index de1c254..ca351a1 100644 --- a/vp8/encoder/arm/neon/fastfdct8x4_neon.asm +++ b/vp8/encoder/arm/neon/fastfdct8x4_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/fastquantizeb_neon.asm b/vp8/encoder/arm/neon/fastquantizeb_neon.asm index 1107037..ca1ea9c 100644 --- a/vp8/encoder/arm/neon/fastquantizeb_neon.asm +++ b/vp8/encoder/arm/neon/fastquantizeb_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/sad16_neon.asm b/vp8/encoder/arm/neon/sad16_neon.asm index 6169f10..d7c590e 100644 --- a/vp8/encoder/arm/neon/sad16_neon.asm +++ b/vp8/encoder/arm/neon/sad16_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/sad8_neon.asm b/vp8/encoder/arm/neon/sad8_neon.asm index 28604dd..23ba6df 100644 --- a/vp8/encoder/arm/neon/sad8_neon.asm +++ b/vp8/encoder/arm/neon/sad8_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/shortfdct_neon.asm b/vp8/encoder/arm/neon/shortfdct_neon.asm index 26bc0d0..5af5cb8 100644 --- a/vp8/encoder/arm/neon/shortfdct_neon.asm +++ b/vp8/encoder/arm/neon/shortfdct_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/subtract_neon.asm b/vp8/encoder/arm/neon/subtract_neon.asm index 8781ca0..3ea00f8 100644 --- a/vp8/encoder/arm/neon/subtract_neon.asm +++ b/vp8/encoder/arm/neon/subtract_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/variance_neon.asm b/vp8/encoder/arm/neon/variance_neon.asm index 64b83ca..e1a4686 100644 --- a/vp8/encoder/arm/neon/variance_neon.asm +++ b/vp8/encoder/arm/neon/variance_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/vp8_memcpy_neon.asm b/vp8/encoder/arm/neon/vp8_memcpy_neon.asm index f26b4d7..b0450e5 100644 --- a/vp8/encoder/arm/neon/vp8_memcpy_neon.asm +++ b/vp8/encoder/arm/neon/vp8_memcpy_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/vp8_mse16x16_neon.asm b/vp8/encoder/arm/neon/vp8_mse16x16_neon.asm index f535967..6af4e87 100644 --- a/vp8/encoder/arm/neon/vp8_mse16x16_neon.asm +++ b/vp8/encoder/arm/neon/vp8_mse16x16_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/vp8_packtokens_armv7.asm b/vp8/encoder/arm/neon/vp8_packtokens_armv7.asm index 9c52c52..c19ac82 100644 --- a/vp8/encoder/arm/neon/vp8_packtokens_armv7.asm +++ b/vp8/encoder/arm/neon/vp8_packtokens_armv7.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/vp8_packtokens_mbrow_armv7.asm b/vp8/encoder/arm/neon/vp8_packtokens_mbrow_armv7.asm index 92b0989..0756455 100644 --- a/vp8/encoder/arm/neon/vp8_packtokens_mbrow_armv7.asm +++ b/vp8/encoder/arm/neon/vp8_packtokens_mbrow_armv7.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/vp8_packtokens_partitions_armv7.asm b/vp8/encoder/arm/neon/vp8_packtokens_partitions_armv7.asm index 6d5f882..10a3d98 100644 --- a/vp8/encoder/arm/neon/vp8_packtokens_partitions_armv7.asm +++ b/vp8/encoder/arm/neon/vp8_packtokens_partitions_armv7.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.asm b/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.asm index 5269c0a..ba3decf 100644 --- a/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.asm +++ b/vp8/encoder/arm/neon/vp8_shortwalsh4x4_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/vp8_subpixelvariance16x16_neon.asm b/vp8/encoder/arm/neon/vp8_subpixelvariance16x16_neon.asm index aec716e..1b09cfe 100644 --- a/vp8/encoder/arm/neon/vp8_subpixelvariance16x16_neon.asm +++ b/vp8/encoder/arm/neon/vp8_subpixelvariance16x16_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/vp8_subpixelvariance16x16s_neon.asm b/vp8/encoder/arm/neon/vp8_subpixelvariance16x16s_neon.asm index 3d02d7c..1c1441c 100644 --- a/vp8/encoder/arm/neon/vp8_subpixelvariance16x16s_neon.asm +++ b/vp8/encoder/arm/neon/vp8_subpixelvariance16x16s_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/neon/vp8_subpixelvariance8x8_neon.asm b/vp8/encoder/arm/neon/vp8_subpixelvariance8x8_neon.asm index bd56761..cf4da62 100644 --- a/vp8/encoder/arm/neon/vp8_subpixelvariance8x8_neon.asm +++ b/vp8/encoder/arm/neon/vp8_subpixelvariance8x8_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/arm/picklpf_arm.c b/vp8/encoder/arm/picklpf_arm.c index 0586e55..b2d8f2b 100644 --- a/vp8/encoder/arm/picklpf_arm.c +++ b/vp8/encoder/arm/picklpf_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/arm/quantize_arm.c b/vp8/encoder/arm/quantize_arm.c index 46906d3..50f58bf 100644 --- a/vp8/encoder/arm/quantize_arm.c +++ b/vp8/encoder/arm/quantize_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/arm/quantize_arm.h b/vp8/encoder/arm/quantize_arm.h index e93f0fe..5f9155e 100644 --- a/vp8/encoder/arm/quantize_arm.h +++ b/vp8/encoder/arm/quantize_arm.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -14,8 +15,11 @@ #if HAVE_ARMV7 extern prototype_quantize_block(vp8_fast_quantize_b_neon); -#undef vp8_quantize_fastquantb -#define vp8_quantize_fastquantb vp8_fast_quantize_b_neon +/* The neon quantizer has not been updated to match the new exact + * quantizer introduced in commit e04e2935 + */ +//#undef vp8_quantize_fastquantb +//#define vp8_quantize_fastquantb vp8_fast_quantize_b_neon #endif diff --git a/vp8/encoder/arm/variance_arm.h b/vp8/encoder/arm/variance_arm.h index d9fc9b3..859e43f 100644 --- a/vp8/encoder/arm/variance_arm.h +++ b/vp8/encoder/arm/variance_arm.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/arm/vpx_vp8_enc_asm_offsets.c b/vp8/encoder/arm/vpx_vp8_enc_asm_offsets.c index 8cdf079..c595ca3 100644 --- a/vp8/encoder/arm/vpx_vp8_enc_asm_offsets.c +++ b/vp8/encoder/arm/vpx_vp8_enc_asm_offsets.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/bitstream.c b/vp8/encoder/bitstream.c index 31ad56a..929c178 100644 --- a/vp8/encoder/bitstream.c +++ b/vp8/encoder/bitstream.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -791,7 +792,8 @@ static void write_mv_ref assert(NEARESTMV <= m && m <= SPLITMV); - vp8_write_token(w, vp8_mv_ref_tree, p, VP8_MVREFENCODINGS + m); + vp8_write_token(w, vp8_mv_ref_tree, p, + vp8_mv_ref_encoding_array - NEARESTMV + m); } static void write_sub_mv_ref @@ -801,7 +803,8 @@ static void write_sub_mv_ref { assert(LEFT4X4 <= m && m <= NEW4X4); - vp8_write_token(w, vp8_sub_mv_ref_tree, p, VP8_SUBMVREFENCODINGS + m); + vp8_write_token(w, vp8_sub_mv_ref_tree, p, + vp8_sub_mv_ref_encoding_array - LEFT4X4 + m); } static void write_mv @@ -869,6 +872,8 @@ static void pack_inter_mode_mvs(VP8_COMP *const cpi) int prob_skip_false = 0; ms = pc->mi - 1; + cpi->mb.partition_info = cpi->mb.pi; + // Calculate the probabilities to be used to code the reference frame based on actual useage this frame if (!(cpi->prob_intra_coded = rf_intra * 255 / (rf_intra + rf_inter))) cpi->prob_intra_coded = 1; @@ -1017,7 +1022,7 @@ static void pack_inter_mode_mvs(VP8_COMP *const cpi) do { - const B_MODE_INFO *const b = mi->partition_bmi + j; + const B_MODE_INFO *const b = cpi->mb.partition_info->bmi + j; const int *const L = vp8_mbsplits [mi->partitioning]; int k = -1; /* first block in subset j */ int mv_contz; @@ -1039,7 +1044,7 @@ static void pack_inter_mode_mvs(VP8_COMP *const cpi) write_mv(w, &b->mv.as_mv, &best_mv, (const MV_CONTEXT *) mvc); } } - while (++j < mi->partition_count); + while (++j < cpi->mb.partition_info->count); } break; default: @@ -1048,9 +1053,11 @@ static void pack_inter_mode_mvs(VP8_COMP *const cpi) } ++m; + cpi->mb.partition_info++; } ++m; /* skip L prediction border */ + cpi->mb.partition_info++; } } @@ -1385,8 +1392,6 @@ void vp8_pack_bitstream(VP8_COMP *cpi, unsigned char *dest, unsigned long *size) // every keyframe send startcode, width, height, scale factor, clamp and color type if (oh.type == KEY_FRAME) { - int w, h, hs, vs; - // Start / synch code cx_data[0] = 0x9D; cx_data[1] = 0x01; diff --git a/vp8/encoder/bitstream.h b/vp8/encoder/bitstream.h index ee69f66..5596313 100644 --- a/vp8/encoder/bitstream.h +++ b/vp8/encoder/bitstream.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/block.h b/vp8/encoder/block.h index cc4cbe0..ffb8890 100644 --- a/vp8/encoder/block.h +++ b/vp8/encoder/block.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -32,6 +33,7 @@ typedef struct // 16 Y blocks, 4 U blocks, 4 V blocks each with 16 entries short(*quant)[4]; + short(*quant_shift)[4]; short(*zbin)[4]; short(*zrun_zbin_boost); short(*round)[4]; @@ -50,6 +52,12 @@ typedef struct typedef struct { + int count; + B_MODE_INFO bmi[16]; +} PARTITION_INFO; + +typedef struct +{ DECLARE_ALIGNED(16, short, src_diff[400]); // 16x16 Y 8x8 U 8x8 V 4x4 2nd Y DECLARE_ALIGNED(16, short, coeff[400]); // 16x16 Y 8x8 U 8x8 V 4x4 2nd Y @@ -59,6 +67,9 @@ typedef struct YV12_BUFFER_CONFIG src; MACROBLOCKD e_mbd; + PARTITION_INFO *partition_info; /* work pointer */ + PARTITION_INFO *pi; /* Corresponds to upper left visible macroblock */ + PARTITION_INFO *pip; /* Base of allocated array */ search_site *ss; int ss_count; @@ -91,6 +102,9 @@ typedef struct int encode_breakout; + //char * gf_active_ptr; + signed char *gf_active_ptr; + unsigned char *active_ptr; MV_CONTEXT *mvc; @@ -99,15 +113,8 @@ typedef struct void (*vp8_short_fdct4x4)(short *input, short *output, int pitch); void (*vp8_short_fdct8x4)(short *input, short *output, int pitch); - void (*short_fdct4x4rd)(short *input, short *output, int pitch); - void (*short_fdct8x4rd)(short *input, short *output, int pitch); - void (*vp8_short_fdct4x4_ptr)(short *input, short *output, int pitch); void (*short_walsh4x4)(short *input, short *output, int pitch); - void (*quantize_b)(BLOCK *b, BLOCKD *d); - void (*quantize_brd)(BLOCK *b, BLOCKD *d); - - } MACROBLOCK; diff --git a/vp8/encoder/boolhuff.c b/vp8/encoder/boolhuff.c index c101384..82006b1 100644 --- a/vp8/encoder/boolhuff.c +++ b/vp8/encoder/boolhuff.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/boolhuff.h b/vp8/encoder/boolhuff.h index 0d929f0..f723da3 100644 --- a/vp8/encoder/boolhuff.h +++ b/vp8/encoder/boolhuff.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/dct.c b/vp8/encoder/dct.c index 5207e39..b5a11ae 100644 --- a/vp8/encoder/dct.c +++ b/vp8/encoder/dct.c @@ -1,172 +1,64 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ #include <math.h> - -static const short dct_matrix2[4][4] = -{ - { 23170, 30274, 23170, 12540 }, - { 23170, 12540, -23170, -30274 }, - { 23170, -12540, -23170, 30274 }, - { 23170, -30274, 23170, -12540 } -}; - -static const short dct_matrix1[4][4] = -{ - { 23170, 23170, 23170, 23170 }, - { 30274, 12540, -12540, -30274 }, - { 23170, -23170, -23170, 23170 }, - { 12540, -30274, 30274, -12540 } -}; - - -#define _1STSTAGESHIFT 14 -#define _1STSTAGEROUNDING (1<<( _1STSTAGESHIFT-1)) -#define _2NDSTAGESHIFT 16 -#define _2NDSTAGEROUNDING (1<<( _2NDSTAGESHIFT-1)) - -// using matrix multiply void vp8_short_fdct4x4_c(short *input, short *output, int pitch) { - int i, j, k; - short temp[4][4]; - int sumtemp; - pitch >>= 1; - - for (i = 0; i < 4; i++) - { - for (j = 0; j < 4; j++) - { - sumtemp = 0; - - for (k = 0; k < 4; k++) - { - sumtemp += input[i*pitch+k] * dct_matrix2[k][j]; - - } - - temp[i][j] = (short)((sumtemp + _1STSTAGEROUNDING) >> _1STSTAGESHIFT); - } - } - - - for (i = 0; i < 4; i++) - { - for (j = 0; j < 4; j++) - { - sumtemp = 0; - - for (k = 0; k < 4; k++) - { - sumtemp += dct_matrix1[i][ k] * temp[k][ j]; - } - - output[i*4+j] = (short)((sumtemp + _2NDSTAGEROUNDING) >> _2NDSTAGESHIFT); - } - } - -} - - -void vp8_short_fdct8x4_c(short *input, short *output, int pitch) -{ - vp8_short_fdct4x4_c(input, output, pitch); - vp8_short_fdct4x4_c(input + 4, output + 16, pitch); -} - - -static const signed short x_c1 = 60547; -static const signed short x_c2 = 46341; -static const signed short x_c3 = 25080; - -void vp8_fast_fdct4x4_c(short *input, short *output, int pitch) -{ int i; int a1, b1, c1, d1; - int a2, b2, c2, d2; short *ip = input; - short *op = output; - int temp1, temp2; for (i = 0; i < 4; i++) { - a1 = (ip[0] + ip[3]) * 2; - b1 = (ip[1] + ip[2]) * 2; - c1 = (ip[1] - ip[2]) * 2; - d1 = (ip[0] - ip[3]) * 2; - - temp1 = a1 + b1; - temp2 = a1 - b1; - - op[0] = ((temp1 * x_c2) >> 16) + temp1; - op[2] = ((temp2 * x_c2) >> 16) + temp2; - - temp1 = (c1 * x_c3) >> 16; - temp2 = ((d1 * x_c1) >> 16) + d1; + a1 = ((ip[0] + ip[3])<<3); + b1 = ((ip[1] + ip[2])<<3); + c1 = ((ip[1] - ip[2])<<3); + d1 = ((ip[0] - ip[3])<<3); - op[1] = temp1 + temp2; - - temp1 = (d1 * x_c3) >> 16; - temp2 = ((c1 * x_c1) >> 16) + c1; + op[0] = a1 + b1; + op[2] = a1 - b1; - op[3] = temp1 - temp2; + op[1] = (c1 * 2217 + d1 * 5352 + 14500)>>12; + op[3] = (d1 * 2217 - c1 * 5352 + 7500)>>12; ip += pitch / 2; op += 4; - } + } ip = output; op = output; - for (i = 0; i < 4; i++) { - a1 = ip[0] + ip[12]; b1 = ip[4] + ip[8]; c1 = ip[4] - ip[8]; d1 = ip[0] - ip[12]; + op[0] = ( a1 + b1 + 7)>>4; + op[8] = ( a1 - b1 + 7)>>4; - temp1 = a1 + b1; - temp2 = a1 - b1; - - a2 = ((temp1 * x_c2) >> 16) + temp1; - c2 = ((temp2 * x_c2) >> 16) + temp2; - - temp1 = (c1 * x_c3) >> 16; - temp2 = ((d1 * x_c1) >> 16) + d1; - - b2 = temp1 + temp2; - - temp1 = (d1 * x_c3) >> 16; - temp2 = ((c1 * x_c1) >> 16) + c1; - - d2 = temp1 - temp2; - - - op[0] = (a2 + 1) >> 1; - op[4] = (b2 + 1) >> 1; - op[8] = (c2 + 1) >> 1; - op[12] = (d2 + 1) >> 1; + op[4] =((c1 * 2217 + d1 * 5352 + 12000)>>16) + (d1!=0); + op[12] = (d1 * 2217 - c1 * 5352 + 51000)>>16; ip++; op++; } } -void vp8_fast_fdct8x4_c(short *input, short *output, int pitch) +void vp8_short_fdct8x4_c(short *input, short *output, int pitch) { - vp8_fast_fdct4x4_c(input, output, pitch); - vp8_fast_fdct4x4_c(input + 4, output + 16, pitch); + vp8_short_fdct4x4_c(input, output, pitch); + vp8_short_fdct4x4_c(input + 4, output + 16, pitch); } void vp8_short_walsh4x4_c(short *input, short *output, int pitch) @@ -177,17 +69,18 @@ void vp8_short_walsh4x4_c(short *input, short *output, int pitch) short *ip = input; short *op = output; + for (i = 0; i < 4; i++) { - a1 = ip[0] + ip[3]; - b1 = ip[1] + ip[2]; - c1 = ip[1] - ip[2]; - d1 = ip[0] - ip[3]; - - op[0] = a1 + b1; - op[1] = c1 + d1; - op[2] = a1 - b1; - op[3] = d1 - c1; + a1 = ((ip[0] + ip[2])<<2); + d1 = ((ip[1] + ip[3])<<2); + c1 = ((ip[1] - ip[3])<<2); + b1 = ((ip[0] - ip[2])<<2); + + op[0] = a1 + d1 + (a1!=0); + op[1] = b1 + c1; + op[2] = b1 - c1; + op[3] = a1 - d1; ip += pitch / 2; op += 4; } @@ -197,25 +90,25 @@ void vp8_short_walsh4x4_c(short *input, short *output, int pitch) for (i = 0; i < 4; i++) { - a1 = ip[0] + ip[12]; - b1 = ip[4] + ip[8]; - c1 = ip[4] - ip[8]; - d1 = ip[0] - ip[12]; - - a2 = a1 + b1; - b2 = c1 + d1; - c2 = a1 - b1; - d2 = d1 - c1; - - a2 += (a2 > 0); - b2 += (b2 > 0); - c2 += (c2 > 0); - d2 += (d2 > 0); - - op[0] = (a2) >> 1; - op[4] = (b2) >> 1; - op[8] = (c2) >> 1; - op[12] = (d2) >> 1; + a1 = ip[0] + ip[8]; + d1 = ip[4] + ip[12]; + c1 = ip[4] - ip[12]; + b1 = ip[0] - ip[8]; + + a2 = a1 + d1; + b2 = b1 + c1; + c2 = b1 - c1; + d2 = a1 - d1; + + a2 += a2<0; + b2 += b2<0; + c2 += c2<0; + d2 += d2<0; + + op[0] = (a2+3) >> 3; + op[4] = (b2+3) >> 3; + op[8] = (c2+3) >> 3; + op[12]= (d2+3) >> 3; ip++; op++; diff --git a/vp8/encoder/dct.h b/vp8/encoder/dct.h index fb307cf..fec3b4c 100644 --- a/vp8/encoder/dct.h +++ b/vp8/encoder/dct.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -31,15 +32,14 @@ extern prototype_fdct(vp8_fdct_short4x4); #endif extern prototype_fdct(vp8_fdct_short8x4); +// There is no fast4x4 (for now) #ifndef vp8_fdct_fast4x4 -#define vp8_fdct_fast4x4 vp8_fast_fdct4x4_c +#define vp8_fdct_fast4x4 vp8_short_fdct4x4_c #endif -extern prototype_fdct(vp8_fdct_fast4x4); #ifndef vp8_fdct_fast8x4 -#define vp8_fdct_fast8x4 vp8_fast_fdct8x4_c +#define vp8_fdct_fast8x4 vp8_short_fdct8x4_c #endif -extern prototype_fdct(vp8_fdct_fast8x4); #ifndef vp8_fdct_walsh_short4x4 #define vp8_fdct_walsh_short4x4 vp8_short_walsh4x4_c diff --git a/vp8/encoder/encodeframe.c b/vp8/encoder/encodeframe.c index a4e3772..d8a76d5 100644 --- a/vp8/encoder/encodeframe.c +++ b/vp8/encoder/encodeframe.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -16,7 +17,7 @@ #include "extend.h" #include "entropymode.h" #include "quant_common.h" -#include "segmentation_common.h" +#include "segmentation.h" #include "setupintrarecon.h" #include "encodeintra.h" #include "reconinter.h" @@ -59,10 +60,9 @@ unsigned int uv_modes[4] = {0, 0, 0, 0}; unsigned int b_modes[14] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}; #endif -// The first four entries are dummy values static const int qrounding_factors[129] = { - 56, 56, 56, 56, 56, 56, 56, 56, + 56, 56, 56, 56, 48, 48, 56, 56, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, 48, @@ -83,7 +83,7 @@ static const int qrounding_factors[129] = static const int qzbin_factors[129] = { - 64, 64, 64, 64, 80, 80, 80, 80, + 72, 72, 72, 72, 80, 80, 72, 72, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, 80, @@ -102,6 +102,132 @@ static const int qzbin_factors[129] = 80, }; +static const int qrounding_factors_y2[129] = +{ + 56, 56, 56, 56, 48, 48, 56, 56, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, 48, 48, 48, 48, 48, 48, 48, + 48, +}; + +static const int qzbin_factors_y2[129] = +{ + 72, 72, 72, 72, 80, 80, 72, 72, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, 80, 80, 80, 80, 80, 80, 80, + 80, +}; + +//#define EXACT_QUANT +#ifdef EXACT_QUANT +static void vp8cx_invert_quant(short *quant, short *shift, short d) +{ + unsigned t; + int l; + t = d; + for(l = 0; t > 1; l++) + t>>=1; + t = 1 + (1<<(16+l))/d; + *quant = (short)(t - (1<<16)); + *shift = l; +} + +void vp8cx_init_quantizer(VP8_COMP *cpi) +{ + int r, c; + int i; + int quant_val; + int Q; + + int zbin_boost[16] = {0, 0, 8, 10, 12, 14, 16, 20, 24, 28, 32, 36, 40, 44, 44, 44}; + + for (Q = 0; Q < QINDEX_RANGE; Q++) + { + // dc values + quant_val = vp8_dc_quant(Q, cpi->common.y1dc_delta_q); + vp8cx_invert_quant(cpi->Y1quant[Q][0] + 0, + cpi->Y1quant_shift[Q][0] + 0, quant_val); + cpi->Y1zbin[Q][0][0] = ((qzbin_factors[Q] * quant_val) + 64) >> 7; + cpi->Y1round[Q][0][0] = (qrounding_factors[Q] * quant_val) >> 7; + cpi->common.Y1dequant[Q][0][0] = quant_val; + cpi->zrun_zbin_boost_y1[Q][0] = (quant_val * zbin_boost[0]) >> 7; + + quant_val = vp8_dc2quant(Q, cpi->common.y2dc_delta_q); + vp8cx_invert_quant(cpi->Y2quant[Q][0] + 0, + cpi->Y2quant_shift[Q][0] + 0, quant_val); + cpi->Y2zbin[Q][0][0] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7; + cpi->Y2round[Q][0][0] = (qrounding_factors_y2[Q] * quant_val) >> 7; + cpi->common.Y2dequant[Q][0][0] = quant_val; + cpi->zrun_zbin_boost_y2[Q][0] = (quant_val * zbin_boost[0]) >> 7; + + quant_val = vp8_dc_uv_quant(Q, cpi->common.uvdc_delta_q); + vp8cx_invert_quant(cpi->UVquant[Q][0] + 0, + cpi->UVquant_shift[Q][0] + 0, quant_val); + cpi->UVzbin[Q][0][0] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;; + cpi->UVround[Q][0][0] = (qrounding_factors[Q] * quant_val) >> 7; + cpi->common.UVdequant[Q][0][0] = quant_val; + cpi->zrun_zbin_boost_uv[Q][0] = (quant_val * zbin_boost[0]) >> 7; + + // all the ac values = ; + for (i = 1; i < 16; i++) + { + int rc = vp8_default_zig_zag1d[i]; + r = (rc >> 2); + c = (rc & 3); + + quant_val = vp8_ac_yquant(Q); + vp8cx_invert_quant(cpi->Y1quant[Q][r] + c, + cpi->Y1quant_shift[Q][r] + c, quant_val); + cpi->Y1zbin[Q][r][c] = ((qzbin_factors[Q] * quant_val) + 64) >> 7; + cpi->Y1round[Q][r][c] = (qrounding_factors[Q] * quant_val) >> 7; + cpi->common.Y1dequant[Q][r][c] = quant_val; + cpi->zrun_zbin_boost_y1[Q][i] = (quant_val * zbin_boost[i]) >> 7; + + quant_val = vp8_ac2quant(Q, cpi->common.y2ac_delta_q); + vp8cx_invert_quant(cpi->Y2quant[Q][r] + c, + cpi->Y2quant_shift[Q][r] + c, quant_val); + cpi->Y2zbin[Q][r][c] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7; + cpi->Y2round[Q][r][c] = (qrounding_factors_y2[Q] * quant_val) >> 7; + cpi->common.Y2dequant[Q][r][c] = quant_val; + cpi->zrun_zbin_boost_y2[Q][i] = (quant_val * zbin_boost[i]) >> 7; + + quant_val = vp8_ac_uv_quant(Q, cpi->common.uvac_delta_q); + vp8cx_invert_quant(cpi->UVquant[Q][r] + c, + cpi->UVquant_shift[Q][r] + c, quant_val); + cpi->UVzbin[Q][r][c] = ((qzbin_factors[Q] * quant_val) + 64) >> 7; + cpi->UVround[Q][r][c] = (qrounding_factors[Q] * quant_val) >> 7; + cpi->common.UVdequant[Q][r][c] = quant_val; + cpi->zrun_zbin_boost_uv[Q][i] = (quant_val * zbin_boost[i]) >> 7; + } + } +} +#else void vp8cx_init_quantizer(VP8_COMP *cpi) { int r, c; @@ -123,8 +249,8 @@ void vp8cx_init_quantizer(VP8_COMP *cpi) quant_val = vp8_dc2quant(Q, cpi->common.y2dc_delta_q); cpi->Y2quant[Q][0][0] = (1 << 16) / quant_val; - cpi->Y2zbin[Q][0][0] = ((qzbin_factors[Q] * quant_val) + 64) >> 7; - cpi->Y2round[Q][0][0] = (qrounding_factors[Q] * quant_val) >> 7; + cpi->Y2zbin[Q][0][0] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7; + cpi->Y2round[Q][0][0] = (qrounding_factors_y2[Q] * quant_val) >> 7; cpi->common.Y2dequant[Q][0][0] = quant_val; cpi->zrun_zbin_boost_y2[Q][0] = (quant_val * zbin_boost[0]) >> 7; @@ -151,8 +277,8 @@ void vp8cx_init_quantizer(VP8_COMP *cpi) quant_val = vp8_ac2quant(Q, cpi->common.y2ac_delta_q); cpi->Y2quant[Q][r][c] = (1 << 16) / quant_val; - cpi->Y2zbin[Q][r][c] = ((qzbin_factors[Q] * quant_val) + 64) >> 7; - cpi->Y2round[Q][r][c] = (qrounding_factors[Q] * quant_val) >> 7; + cpi->Y2zbin[Q][r][c] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7; + cpi->Y2round[Q][r][c] = (qrounding_factors_y2[Q] * quant_val) >> 7; cpi->common.Y2dequant[Q][r][c] = quant_val; cpi->zrun_zbin_boost_y2[Q][i] = (quant_val * zbin_boost[i]) >> 7; @@ -165,13 +291,12 @@ void vp8cx_init_quantizer(VP8_COMP *cpi) } } } - +#endif void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x) { int i; int QIndex; MACROBLOCKD *xd = &x->e_mbd; - MB_MODE_INFO *mbmi = &xd->mbmi; int zbin_extra; // Select the baseline MB Q index. @@ -179,12 +304,12 @@ void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x) { // Abs Value if (xd->mb_segement_abs_delta == SEGMENT_ABSDATA) - QIndex = xd->segment_feature_data[MB_LVL_ALT_Q][mbmi->segment_id]; + QIndex = xd->segment_feature_data[MB_LVL_ALT_Q][xd->mode_info_context->mbmi.segment_id]; // Delta Value else { - QIndex = cpi->common.base_qindex + xd->segment_feature_data[MB_LVL_ALT_Q][mbmi->segment_id]; + QIndex = cpi->common.base_qindex + xd->segment_feature_data[MB_LVL_ALT_Q][xd->mode_info_context->mbmi.segment_id]; QIndex = (QIndex >= 0) ? ((QIndex <= MAXQ) ? QIndex : MAXQ) : 0; // Clamp to valid range } } @@ -197,6 +322,7 @@ void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x) for (i = 0; i < 16; i++) { x->block[i].quant = cpi->Y1quant[QIndex]; + x->block[i].quant_shift = cpi->Y1quant_shift[QIndex]; x->block[i].zbin = cpi->Y1zbin[QIndex]; x->block[i].round = cpi->Y1round[QIndex]; x->e_mbd.block[i].dequant = cpi->common.Y1dequant[QIndex]; @@ -210,6 +336,7 @@ void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x) for (i = 16; i < 24; i++) { x->block[i].quant = cpi->UVquant[QIndex]; + x->block[i].quant_shift = cpi->UVquant_shift[QIndex]; x->block[i].zbin = cpi->UVzbin[QIndex]; x->block[i].round = cpi->UVround[QIndex]; x->e_mbd.block[i].dequant = cpi->common.UVdequant[QIndex]; @@ -220,6 +347,7 @@ void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x) // Y2 zbin_extra = (cpi->common.Y2dequant[QIndex][0][1] * ((cpi->zbin_over_quant / 2) + cpi->zbin_mode_boost)) >> 7; x->block[24].quant = cpi->Y2quant[QIndex]; + x->block[24].quant_shift = cpi->Y2quant_shift[QIndex]; x->block[24].zbin = cpi->Y2zbin[QIndex]; x->block[24].round = cpi->Y2round[QIndex]; x->e_mbd.block[24].dequant = cpi->common.Y2dequant[QIndex]; @@ -255,16 +383,15 @@ void encode_mb_row(VP8_COMP *cpi, int i; int recon_yoffset, recon_uvoffset; int mb_col; - int recon_y_stride = cm->last_frame.y_stride; - int recon_uv_stride = cm->last_frame.uv_stride; + int ref_fb_idx = cm->lst_fb_idx; + int dst_fb_idx = cm->new_fb_idx; + int recon_y_stride = cm->yv12_fb[ref_fb_idx].y_stride; + int recon_uv_stride = cm->yv12_fb[ref_fb_idx].uv_stride; int seg_map_index = (mb_row * cpi->common.mb_cols); // reset above block coeffs - xd->above_context[Y1CONTEXT] = cm->above_context[Y1CONTEXT]; - xd->above_context[UCONTEXT ] = cm->above_context[UCONTEXT ]; - xd->above_context[VCONTEXT ] = cm->above_context[VCONTEXT ]; - xd->above_context[Y2CONTEXT] = cm->above_context[Y2CONTEXT]; + xd->above_context = cm->above_context; xd->up_available = (mb_row != 0); recon_yoffset = (mb_row * recon_y_stride * 16); @@ -289,9 +416,9 @@ void encode_mb_row(VP8_COMP *cpi, x->mv_row_min = -((mb_row * 16) + (VP8BORDERINPIXELS - 16)); x->mv_row_max = ((cm->mb_rows - 1 - mb_row) * 16) + (VP8BORDERINPIXELS - 16); - xd->dst.y_buffer = cm->new_frame.y_buffer + recon_yoffset; - xd->dst.u_buffer = cm->new_frame.u_buffer + recon_uvoffset; - xd->dst.v_buffer = cm->new_frame.v_buffer + recon_uvoffset; + xd->dst.y_buffer = cm->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset; + xd->dst.u_buffer = cm->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset; + xd->dst.v_buffer = cm->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset; xd->left_available = (mb_col != 0); // Is segmentation enabled @@ -300,14 +427,14 @@ void encode_mb_row(VP8_COMP *cpi, { // Code to set segment id in xd->mbmi.segment_id for current MB (with range checking) if (cpi->segmentation_map[seg_map_index+mb_col] <= 3) - xd->mbmi.segment_id = cpi->segmentation_map[seg_map_index+mb_col]; + xd->mode_info_context->mbmi.segment_id = cpi->segmentation_map[seg_map_index+mb_col]; else - xd->mbmi.segment_id = 0; + xd->mode_info_context->mbmi.segment_id = 0; vp8cx_mb_init_quantizer(cpi, x); } else - xd->mbmi.segment_id = 0; // Set to Segment 0 by default + xd->mode_info_context->mbmi.segment_id = 0; // Set to Segment 0 by default x->active_ptr = cpi->active_map + seg_map_index + mb_col; @@ -331,14 +458,14 @@ void encode_mb_row(VP8_COMP *cpi, for (b = 0; b < xd->mbmi.partition_count; b++) { - inter_b_modes[xd->mbmi.partition_bmi[b].mode] ++; + inter_b_modes[x->partition->bmi[b].mode] ++; } } #endif // Count of last ref frame 0,0 useage - if ((xd->mbmi.mode == ZEROMV) && (xd->mbmi.ref_frame == LAST_FRAME)) + if ((xd->mode_info_context->mbmi.mode == ZEROMV) && (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME)) cpi->inter_zz_count ++; // Special case code for cyclic refresh @@ -346,14 +473,14 @@ void encode_mb_row(VP8_COMP *cpi, // during vp8cx_encode_inter_macroblock()) back into the global sgmentation map if (cpi->cyclic_refresh_mode_enabled && xd->segmentation_enabled) { - cpi->segmentation_map[seg_map_index+mb_col] = xd->mbmi.segment_id; + cpi->segmentation_map[seg_map_index+mb_col] = xd->mode_info_context->mbmi.segment_id; // If the block has been refreshed mark it as clean (the magnitude of the -ve influences how long it will be before we consider another refresh): // Else if it was coded (last frame 0,0) and has not already been refreshed then mark it as a candidate for cleanup next time (marked 0) // else mark it as dirty (1). - if (xd->mbmi.segment_id) + if (xd->mode_info_context->mbmi.segment_id) cpi->cyclic_refresh_map[seg_map_index+mb_col] = -1; - else if ((xd->mbmi.mode == ZEROMV) && (xd->mbmi.ref_frame == LAST_FRAME)) + else if ((xd->mode_info_context->mbmi.mode == ZEROMV) && (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME)) { if (cpi->cyclic_refresh_map[seg_map_index+mb_col] == 1) cpi->cyclic_refresh_map[seg_map_index+mb_col] = 0; @@ -366,10 +493,7 @@ void encode_mb_row(VP8_COMP *cpi, cpi->tplist[mb_row].stop = *tp; - xd->gf_active_ptr++; // Increment pointer into gf useage flags structure for next mb - - // store macroblock mode info into context array - vpx_memcpy(&xd->mode_info_context->mbmi, &xd->mbmi, sizeof(xd->mbmi)); + x->gf_active_ptr++; // Increment pointer into gf useage flags structure for next mb for (i = 0; i < 16; i++) vpx_memcpy(&xd->mode_info_context->bmi[i], &xd->block[i].bmi, sizeof(xd->block[i].bmi)); @@ -383,27 +507,26 @@ void encode_mb_row(VP8_COMP *cpi, recon_uvoffset += 8; // Keep track of segment useage - segment_counts[xd->mbmi.segment_id] ++; + segment_counts[xd->mode_info_context->mbmi.segment_id] ++; // skip to next mb xd->mode_info_context++; + x->partition_info++; - xd->above_context[Y1CONTEXT] += 4; - xd->above_context[UCONTEXT ] += 2; - xd->above_context[VCONTEXT ] += 2; - xd->above_context[Y2CONTEXT] ++; + xd->above_context++; cpi->current_mb_col_main = mb_col; } //extend the recon for intra prediction vp8_extend_mb_row( - &cm->new_frame, + &cm->yv12_fb[dst_fb_idx], xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8); // this is to account for the border xd->mode_info_context++; + x->partition_info++; } @@ -448,7 +571,7 @@ void vp8_encode_frame(VP8_COMP *cpi) //} - xd->gf_active_ptr = (signed char *)cm->gf_active_flags; // Point to base of GF active flags data structure + x->gf_active_ptr = (signed char *)cpi->gf_active_flags; // Point to base of GF active flags data structure x->vector_range = 32; @@ -467,13 +590,13 @@ void vp8_encode_frame(VP8_COMP *cpi) #if 0 // Experimental code - cpi->frame_distortion = 0; + cpi->frame_distortion = 0; cpi->last_mb_distortion = 0; #endif totalrate = 0; - xd->mode_info = cm->mi - 1; + x->partition_info = x->pi; xd->mode_info_context = cm->mi; xd->mode_info_stride = cm->mode_info_stride; @@ -509,12 +632,12 @@ void vp8_encode_frame(VP8_COMP *cpi) // Copy data over into macro block data sturctures. x->src = * cpi->Source; - xd->pre = cm->last_frame; - xd->dst = cm->new_frame; + xd->pre = cm->yv12_fb[cm->lst_fb_idx]; + xd->dst = cm->yv12_fb[cm->new_fb_idx]; // set up frame new frame for intra coded blocks - vp8_setup_intra_recon(&cm->new_frame); + vp8_setup_intra_recon(&cm->yv12_fb[cm->new_fb_idx]); vp8_build_block_offsets(x); @@ -539,10 +662,10 @@ void vp8_encode_frame(VP8_COMP *cpi) //x->rdmult = (int)(cpi->RDMULT * pow( (cpi->rate_correction_factor * 2.0), 0.75 )); #endif - xd->mbmi.mode = DC_PRED; - xd->mbmi.uv_mode = DC_PRED; + xd->mode_info_context->mbmi.mode = DC_PRED; + xd->mode_info_context->mbmi.uv_mode = DC_PRED; - xd->left_context = cm->left_context; + xd->left_context = &cm->left_context; vp8_zero(cpi->count_mb_ref_frame_usage) vp8_zero(cpi->ymode_count) @@ -550,17 +673,7 @@ void vp8_encode_frame(VP8_COMP *cpi) x->mvc = cm->fc.mvc; - // vp8_zero( entropy_stats) - { - ENTROPY_CONTEXT **p = cm->above_context; - const size_t L = cm->mb_cols; - - vp8_zero_array(p [Y1CONTEXT], L * 4) - vp8_zero_array(p [ UCONTEXT], L * 2) - vp8_zero_array(p [ VCONTEXT], L * 2) - vp8_zero_array(p [Y2CONTEXT], L) - } - + vpx_memset(cm->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES) * cm->mb_cols); { struct vpx_usec_timer emr_timer; @@ -619,6 +732,7 @@ void vp8_encode_frame(VP8_COMP *cpi) x->src.v_buffer += 8 * x->src.uv_stride * (cpi->encoding_thread_count + 1) - 8 * cm->mb_cols; xd->mode_info_context += xd->mode_info_stride * cpi->encoding_thread_count; + x->partition_info += xd->mode_info_stride * cpi->encoding_thread_count; if (mb_row < cm->mb_rows - 1) //WaitForSingleObject(cpi->h_event_main, INFINITE); @@ -894,8 +1008,8 @@ void vp8_build_block_offsets(MACROBLOCK *x) static void sum_intra_stats(VP8_COMP *cpi, MACROBLOCK *x) { const MACROBLOCKD *xd = & x->e_mbd; - const MB_PREDICTION_MODE m = xd->mbmi.mode; - const MB_PREDICTION_MODE uvm = xd->mbmi.uv_mode; + const MB_PREDICTION_MODE m = xd->mode_info_context->mbmi.mode; + const MB_PREDICTION_MODE uvm = xd->mode_info_context->mbmi.uv_mode; #ifdef MODE_STATS const int is_key = cpi->common.frame_type == KEY_FRAME; @@ -933,7 +1047,7 @@ int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t) int rateuv_tokenonly = 0; int i; - x->e_mbd.mbmi.ref_frame = INTRA_FRAME; + x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME; #if !(CONFIG_REALTIME_ONLY) @@ -949,7 +1063,7 @@ int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t) error_uv = vp8_rd_pick_intra_mbuv_mode(cpi, x, &rateuv, &rateuv_tokenonly, &distuv); - x->e_mbd.mbmi.mb_skip_coeff = (cpi->common.mb_no_coeff_skip) ? 1 : 0; + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff = (cpi->common.mb_no_coeff_skip) ? 1 : 0; vp8_encode_intra16x16mbuv(IF_RTCD(&cpi->rtcd), x); rate += rateuv; @@ -957,7 +1071,7 @@ int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t) if (Error4x4 < Error16x16) { rate += rate4x4; - x->e_mbd.mbmi.mode = B_PRED; + x->e_mbd.mode_info_context->mbmi.mode = B_PRED; // get back the intra block modes for (i = 0; i < 16; i++) @@ -997,7 +1111,7 @@ int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t) for (mode = DC_PRED; mode <= TM_PRED; mode ++) { - x->e_mbd.mbmi.mode = mode; + x->e_mbd.mode_info_context->mbmi.mode = mode; vp8_build_intra_predictors_mby_ptr(&x->e_mbd); distortion2 = VARIANCE_INVOKE(&cpi->rtcd.variance, get16x16prederror)(x->src.y_buffer, x->src.y_stride, x->e_mbd.predictor, 16, 0x7fffffff); rate2 = x->mbmode_cost[x->e_mbd.frame_type][mode]; @@ -1017,17 +1131,17 @@ int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t) else Error4x4 = RD_ESTIMATE(x->rdmult, x->rddiv, rate2, distortion2); - x->e_mbd.mbmi.mb_skip_coeff = (cpi->common.mb_no_coeff_skip) ? 1 : 0; + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff = (cpi->common.mb_no_coeff_skip) ? 1 : 0; if (Error4x4 < Error16x16) { - x->e_mbd.mbmi.mode = B_PRED; + x->e_mbd.mode_info_context->mbmi.mode = B_PRED; vp8_encode_intra4x4mby(IF_RTCD(&cpi->rtcd), x); cpi->prediction_error += Error4x4; } else { - x->e_mbd.mbmi.mode = best_mode; + x->e_mbd.mode_info_context->mbmi.mode = best_mode; vp8_encode_intra16x16mby(IF_RTCD(&cpi->rtcd), x); cpi->prediction_error += Error16x16; } @@ -1044,7 +1158,7 @@ int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t) extern int cnt_pm; #endif -extern void vp8_fix_contexts(VP8_COMP *cpi, MACROBLOCKD *x); +extern void vp8_fix_contexts(MACROBLOCKD *x); int vp8cx_encode_inter_macroblock ( @@ -1061,7 +1175,7 @@ int vp8cx_encode_inter_macroblock x->skip = 0; if (xd->segmentation_enabled) - x->encode_breakout = cpi->segment_encode_breakout[xd->mbmi.segment_id]; + x->encode_breakout = cpi->segment_encode_breakout[xd->mode_info_context->mbmi.segment_id]; else x->encode_breakout = cpi->oxcf.encode_breakout; @@ -1092,17 +1206,17 @@ int vp8cx_encode_inter_macroblock if (cpi->cyclic_refresh_mode_enabled) { // Clear segment_id back to 0 if not coded (last frame 0,0) - if ((xd->mbmi.segment_id == 1) && - ((xd->mbmi.ref_frame != LAST_FRAME) || (xd->mbmi.mode != ZEROMV))) + if ((xd->mode_info_context->mbmi.segment_id == 1) && + ((xd->mode_info_context->mbmi.ref_frame != LAST_FRAME) || (xd->mode_info_context->mbmi.mode != ZEROMV))) { - xd->mbmi.segment_id = 0; + xd->mode_info_context->mbmi.segment_id = 0; } } // Experimental code. Special case for gf and arf zeromv modes. Increase zbin size to supress noise if (cpi->zbin_mode_boost_enabled) { - if ((xd->mbmi.mode == ZEROMV) && (xd->mbmi.ref_frame != LAST_FRAME)) + if ((xd->mode_info_context->mbmi.mode == ZEROMV) && (xd->mode_info_context->mbmi.ref_frame != LAST_FRAME)) cpi->zbin_mode_boost = GF_ZEROMV_ZBIN_BOOST; else cpi->zbin_mode_boost = 0; @@ -1111,15 +1225,15 @@ int vp8cx_encode_inter_macroblock vp8cx_mb_init_quantizer(cpi, x); } - cpi->count_mb_ref_frame_usage[xd->mbmi.ref_frame] ++; + cpi->count_mb_ref_frame_usage[xd->mode_info_context->mbmi.ref_frame] ++; - if (xd->mbmi.ref_frame == INTRA_FRAME) + if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) { - x->e_mbd.mbmi.mb_skip_coeff = (cpi->common.mb_no_coeff_skip) ? 1 : 0; + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff = (cpi->common.mb_no_coeff_skip) ? 1 : 0; vp8_encode_intra16x16mbuv(IF_RTCD(&cpi->rtcd), x); - if (xd->mbmi.mode == B_PRED) + if (xd->mode_info_context->mbmi.mode == B_PRED) { vp8_encode_intra4x4mby(IF_RTCD(&cpi->rtcd), x); } @@ -1135,36 +1249,25 @@ int vp8cx_encode_inter_macroblock MV best_ref_mv; MV nearest, nearby; int mdcounts[4]; + int ref_fb_idx; vp8_find_near_mvs(xd, xd->mode_info_context, - &nearest, &nearby, &best_ref_mv, mdcounts, xd->mbmi.ref_frame, cpi->common.ref_frame_sign_bias); + &nearest, &nearby, &best_ref_mv, mdcounts, xd->mode_info_context->mbmi.ref_frame, cpi->common.ref_frame_sign_bias); vp8_build_uvmvs(xd, cpi->common.full_pixel); - // store motion vectors in our motion vector list - if (xd->mbmi.ref_frame == LAST_FRAME) - { - // Set up pointers for this macro block into the previous frame recon buffer - xd->pre.y_buffer = cpi->common.last_frame.y_buffer + recon_yoffset; - xd->pre.u_buffer = cpi->common.last_frame.u_buffer + recon_uvoffset; - xd->pre.v_buffer = cpi->common.last_frame.v_buffer + recon_uvoffset; - } - else if (xd->mbmi.ref_frame == GOLDEN_FRAME) - { - // Set up pointers for this macro block into the golden frame recon buffer - xd->pre.y_buffer = cpi->common.golden_frame.y_buffer + recon_yoffset; - xd->pre.u_buffer = cpi->common.golden_frame.u_buffer + recon_uvoffset; - xd->pre.v_buffer = cpi->common.golden_frame.v_buffer + recon_uvoffset; - } + if (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME) + ref_fb_idx = cpi->common.lst_fb_idx; + else if (xd->mode_info_context->mbmi.ref_frame == GOLDEN_FRAME) + ref_fb_idx = cpi->common.gld_fb_idx; else - { - // Set up pointers for this macro block into the alternate reference frame recon buffer - xd->pre.y_buffer = cpi->common.alt_ref_frame.y_buffer + recon_yoffset; - xd->pre.u_buffer = cpi->common.alt_ref_frame.u_buffer + recon_uvoffset; - xd->pre.v_buffer = cpi->common.alt_ref_frame.v_buffer + recon_uvoffset; - } + ref_fb_idx = cpi->common.alt_fb_idx; + + xd->pre.y_buffer = cpi->common.yv12_fb[ref_fb_idx].y_buffer + recon_yoffset; + xd->pre.u_buffer = cpi->common.yv12_fb[ref_fb_idx].u_buffer + recon_uvoffset; + xd->pre.v_buffer = cpi->common.yv12_fb[ref_fb_idx].v_buffer + recon_uvoffset; - if (xd->mbmi.mode == SPLITMV) + if (xd->mode_info_context->mbmi.mode == SPLITMV) { int i; @@ -1177,19 +1280,19 @@ int vp8cx_encode_inter_macroblock } } } - else if (xd->mbmi.mode == NEWMV) + else if (xd->mode_info_context->mbmi.mode == NEWMV) { cpi->MVcount[0][mv_max+((xd->block[0].bmi.mv.as_mv.row - best_ref_mv.row) >> 1)]++; cpi->MVcount[1][mv_max+((xd->block[0].bmi.mv.as_mv.col - best_ref_mv.col) >> 1)]++; } - if (!x->skip && !x->e_mbd.mbmi.force_no_skip) + if (!x->skip && !x->e_mbd.mode_info_context->mbmi.force_no_skip) { vp8_encode_inter16x16(IF_RTCD(&cpi->rtcd), x); // Clear mb_skip_coeff if mb_no_coeff_skip is not set if (!cpi->common.mb_no_coeff_skip) - xd->mbmi.mb_skip_coeff = 0; + xd->mode_info_context->mbmi.mb_skip_coeff = 0; } else @@ -1202,19 +1305,19 @@ int vp8cx_encode_inter_macroblock { if (cpi->common.mb_no_coeff_skip) { - if (xd->mbmi.mode != B_PRED && xd->mbmi.mode != SPLITMV) - xd->mbmi.dc_diff = 0; + if (xd->mode_info_context->mbmi.mode != B_PRED && xd->mode_info_context->mbmi.mode != SPLITMV) + xd->mode_info_context->mbmi.dc_diff = 0; else - xd->mbmi.dc_diff = 1; + xd->mode_info_context->mbmi.dc_diff = 1; - xd->mbmi.mb_skip_coeff = 1; + xd->mode_info_context->mbmi.mb_skip_coeff = 1; cpi->skip_true_count ++; - vp8_fix_contexts(cpi, xd); + vp8_fix_contexts(xd); } else { vp8_stuff_mb(cpi, xd, t); - xd->mbmi.mb_skip_coeff = 0; + xd->mode_info_context->mbmi.mb_skip_coeff = 0; cpi->skip_false_count ++; } } diff --git a/vp8/encoder/encodeintra.c b/vp8/encoder/encodeintra.c index 403d020..af80857 100644 --- a/vp8/encoder/encodeintra.c +++ b/vp8/encoder/encodeintra.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -52,7 +53,7 @@ void vp8_encode_intra4x4block(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x, BLOCK x->quantize_b(be, b); - x->e_mbd.mbmi.mb_skip_coeff &= (!b->eob); + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff &= (!b->eob); vp8_inverse_transform_b(IF_RTCD(&rtcd->common->idct), b, 32); @@ -65,11 +66,11 @@ void vp8_encode_intra4x4block_rd(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x, BL ENCODEMB_INVOKE(&rtcd->encodemb, subb)(be, b, 16); - x->short_fdct4x4rd(be->src_diff, be->coeff, 32); + x->vp8_short_fdct4x4(be->src_diff, be->coeff, 32); - x->quantize_brd(be, b); + x->quantize_b(be, b); - x->e_mbd.mbmi.mb_skip_coeff &= (!b->eob); + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff &= (!b->eob); IDCT_INVOKE(&rtcd->common->idct, idct16)(b->dqcoeff, b->diff, 32); @@ -108,7 +109,6 @@ void vp8_encode_intra16x16mby(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x) #if !(CONFIG_REALTIME_ONLY) #if 1 - if (x->optimize && x->rddiv > 1) vp8_optimize_mby(x, rtcd); @@ -124,7 +124,7 @@ void vp8_encode_intra16x16mby(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x) { BLOCKD *d = &x->e_mbd.block[b]; - switch (x->e_mbd.mbmi.mode) + switch (x->e_mbd.mode_info_context->mbmi.mode) { case DC_PRED: @@ -155,12 +155,11 @@ void vp8_encode_intra16x16mbyrd(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x) ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, x->src.y_buffer, x->e_mbd.predictor, x->src.y_stride); - vp8_transform_intra_mbyrd(x); - - x->e_mbd.mbmi.mb_skip_coeff = 1; + vp8_transform_intra_mby(x); - vp8_quantize_mbyrd(x); + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff = 1; + vp8_quantize_mby(x); vp8_inverse_transform_mby(IF_RTCD(&rtcd->common->idct), &x->e_mbd); @@ -171,7 +170,7 @@ void vp8_encode_intra16x16mbyrd(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x) { BLOCKD *d = &x->e_mbd.block[b]; - switch (x->e_mbd.mbmi.mode) + switch (x->e_mbd.mode_info_context->mbmi.mode) { case DC_PRED: @@ -224,11 +223,9 @@ void vp8_encode_intra16x16mbuvrd(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x) ENCODEMB_INVOKE(&rtcd->encodemb, submbuv)(x->src_diff, x->src.u_buffer, x->src.v_buffer, x->e_mbd.predictor, x->src.uv_stride); - vp8_transform_mbuvrd(x); - - vp8_quantize_mbuvrd(x); - + vp8_transform_mbuv(x); + vp8_quantize_mbuv(x); vp8_inverse_transform_mbuv(IF_RTCD(&rtcd->common->idct), &x->e_mbd); diff --git a/vp8/encoder/encodeintra.h b/vp8/encoder/encodeintra.h index 4a43ab2..5be23d1 100644 --- a/vp8/encoder/encodeintra.h +++ b/vp8/encoder/encodeintra.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/encodemb.c b/vp8/encoder/encodemb.c index d825133..e10b515 100644 --- a/vp8/encoder/encodemb.c +++ b/vp8/encoder/encodemb.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -12,6 +13,7 @@ #include "encodemb.h" #include "reconinter.h" #include "quantize.h" +#include "tokenize.h" #include "invtrans.h" #include "recon.h" #include "reconintra.h" @@ -119,19 +121,11 @@ void vp8_transform_mbuv(MACROBLOCK *x) for (i = 16; i < 24; i += 2) { - x->vp8_short_fdct8x4(&x->block[i].src_diff[0], &x->block[i].coeff[0], 16); + x->vp8_short_fdct8x4(&x->block[i].src_diff[0], + &x->block[i].coeff[0], 16); } } -void vp8_transform_mbuvrd(MACROBLOCK *x) -{ - int i; - - for (i = 16; i < 24; i += 2) - { - x->short_fdct8x4rd(&x->block[i].src_diff[0], &x->block[i].coeff[0], 16); - } -} void vp8_transform_intra_mby(MACROBLOCK *x) { @@ -139,32 +133,19 @@ void vp8_transform_intra_mby(MACROBLOCK *x) for (i = 0; i < 16; i += 2) { - x->vp8_short_fdct8x4(&x->block[i].src_diff[0], &x->block[i].coeff[0], 32); + x->vp8_short_fdct8x4(&x->block[i].src_diff[0], + &x->block[i].coeff[0], 32); } // build dc block from 16 y dc values vp8_build_dcblock(x); // do 2nd order transform on the dc block - x->short_walsh4x4(&x->block[24].src_diff[0], &x->block[24].coeff[0], 8); + x->short_walsh4x4(&x->block[24].src_diff[0], + &x->block[24].coeff[0], 8); } -void vp8_transform_intra_mbyrd(MACROBLOCK *x) -{ - int i; - - for (i = 0; i < 16; i += 2) - { - x->short_fdct8x4rd(&x->block[i].src_diff[0], &x->block[i].coeff[0], 32); - } - - // build dc block from 16 y dc values - vp8_build_dcblock(x); - - // do 2nd order transform on the dc block - x->short_walsh4x4(&x->block[24].src_diff[0], &x->block[24].coeff[0], 8); -} void vp8_transform_mb(MACROBLOCK *x) { @@ -172,21 +153,24 @@ void vp8_transform_mb(MACROBLOCK *x) for (i = 0; i < 16; i += 2) { - x->vp8_short_fdct8x4(&x->block[i].src_diff[0], &x->block[i].coeff[0], 32); + x->vp8_short_fdct8x4(&x->block[i].src_diff[0], + &x->block[i].coeff[0], 32); } // build dc block from 16 y dc values - if (x->e_mbd.mbmi.mode != SPLITMV) + if (x->e_mbd.mode_info_context->mbmi.mode != SPLITMV) vp8_build_dcblock(x); for (i = 16; i < 24; i += 2) { - x->vp8_short_fdct8x4(&x->block[i].src_diff[0], &x->block[i].coeff[0], 16); + x->vp8_short_fdct8x4(&x->block[i].src_diff[0], + &x->block[i].coeff[0], 16); } // do 2nd order transform on the dc block - if (x->e_mbd.mbmi.mode != SPLITMV) - x->short_walsh4x4(&x->block[24].src_diff[0], &x->block[24].coeff[0], 8); + if (x->e_mbd.mode_info_context->mbmi.mode != SPLITMV) + x->short_walsh4x4(&x->block[24].src_diff[0], + &x->block[24].coeff[0], 8); } @@ -196,39 +180,19 @@ void vp8_transform_mby(MACROBLOCK *x) for (i = 0; i < 16; i += 2) { - x->vp8_short_fdct8x4(&x->block[i].src_diff[0], &x->block[i].coeff[0], 32); + x->vp8_short_fdct8x4(&x->block[i].src_diff[0], + &x->block[i].coeff[0], 32); } // build dc block from 16 y dc values - if (x->e_mbd.mbmi.mode != SPLITMV) + if (x->e_mbd.mode_info_context->mbmi.mode != SPLITMV) { vp8_build_dcblock(x); - x->short_walsh4x4(&x->block[24].src_diff[0], &x->block[24].coeff[0], 8); + x->short_walsh4x4(&x->block[24].src_diff[0], + &x->block[24].coeff[0], 8); } } -void vp8_transform_mbrd(MACROBLOCK *x) -{ - int i; - - for (i = 0; i < 16; i += 2) - { - x->short_fdct8x4rd(&x->block[i].src_diff[0], &x->block[i].coeff[0], 32); - } - - // build dc block from 16 y dc values - if (x->e_mbd.mbmi.mode != SPLITMV) - vp8_build_dcblock(x); - - for (i = 16; i < 24; i += 2) - { - x->short_fdct8x4rd(&x->block[i].src_diff[0], &x->block[i].coeff[0], 16); - } - - // do 2nd order transform on the dc block - if (x->e_mbd.mbmi.mode != SPLITMV) - x->short_walsh4x4(&x->block[24].src_diff[0], &x->block[24].coeff[0], 8); -} void vp8_stuff_inter16x16(MACROBLOCK *x) { @@ -265,792 +229,411 @@ void vp8_stuff_inter16x16(MACROBLOCK *x) } #if !(CONFIG_REALTIME_ONLY) -extern const TOKENEXTRA vp8_dct_value_tokens[DCT_MAX_VALUE*2]; -extern const TOKENEXTRA *vp8_dct_value_tokens_ptr; -extern int vp8_dct_value_cost[DCT_MAX_VALUE*2]; -extern int *vp8_dct_value_cost_ptr; - -static int cost_coeffs(MACROBLOCK *mb, BLOCKD *b, int type, ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l) -{ - int c = !type; /* start at coef 0, unless Y with Y2 */ - int eob = b->eob; - int pt ; /* surrounding block/prev coef predictor */ - int cost = 0; - short *qcoeff_ptr = b->qcoeff; - - VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l); - -# define QC( I) ( qcoeff_ptr [vp8_default_zig_zag1d[I]] ) - - for (; c < eob; c++) - { - int v = QC(c); - int t = vp8_dct_value_tokens_ptr[v].Token; - cost += mb->token_costs [type] [vp8_coef_bands[c]] [pt] [t]; - cost += vp8_dct_value_cost_ptr[v]; - pt = vp8_prev_token_class[t]; - } - -# undef QC - - if (c < 16) - cost += mb->token_costs [type] [vp8_coef_bands[c]] [pt] [DCT_EOB_TOKEN]; - - return cost; -} - -static int mbycost_coeffs(MACROBLOCK *mb) +#define RDCOST(RM,DM,R,D) ( ((128+(R)*(RM)) >> 8) + (DM)*(D) ) +#define RDTRUNC(RM,DM,R,D) ( (128+(R)*(RM)) & 0xFF ) + +typedef struct vp8_token_state vp8_token_state; + +struct vp8_token_state{ + int rate; + int error; + signed char next; + signed char token; + short qc; +}; + +void vp8_optimize_b(MACROBLOCK *mb, int i, int type, + ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l, + const VP8_ENCODER_RTCD *rtcd) { - int cost = 0; - int b; - TEMP_CONTEXT t; - int type = 0; - - MACROBLOCKD *x = &mb->e_mbd; - - vp8_setup_temp_context(&t, x->above_context[Y1CONTEXT], x->left_context[Y1CONTEXT], 4); - - if (x->mbmi.mode == SPLITMV) - type = 3; - - for (b = 0; b < 16; b++) - cost += cost_coeffs(mb, x->block + b, type, - t.a + vp8_block2above[b], t.l + vp8_block2left[b]); - - return cost; -} - -#define RDFUNC(RM,DM,R,D,target_rd) ( ((128+(R)*(RM)) >> 8) + (DM)*(D) ) - -void vp8_optimize_b(MACROBLOCK *x, int i, int type, ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l, const VP8_ENCODER_RTCD *rtcd) -{ - BLOCK *b = &x->block[i]; - BLOCKD *bd = &x->e_mbd.block[i]; - short *dequant_ptr = &bd->dequant[0][0]; - int nzpos[16] = {0}; - short saved_qcoefs[16]; - short saved_dqcoefs[16]; - int baserate, baseerror, baserd; - int rate, error, thisrd; - int k; - int nzcoefcount = 0; - int nc, bestnc = 0; - int besteob; - - // count potential coefficient to be optimized - for (k = !type; k < 16; k++) - { - int qcoef = abs(bd->qcoeff[k]); - int coef = abs(b->coeff[k]); - int dq = dequant_ptr[k]; - - if (qcoef && (qcoef * dq > coef) && (qcoef * dq < coef + dq)) - { - nzpos[nzcoefcount] = k; - nzcoefcount++; - } - } - - // if nothing here, do nothing for this block. - if (!nzcoefcount) - { - *a = *l = (bd->eob != !type); - return; - } - - // save a copy of quantized coefficients - vpx_memcpy(saved_qcoefs, bd->qcoeff, 32); - vpx_memcpy(saved_dqcoefs, bd->dqcoeff, 32); - - besteob = bd->eob; - baserate = cost_coeffs(x, bd, type, a, l); - baseerror = ENCODEMB_INVOKE(&rtcd->encodemb, berr)(b->coeff, bd->dqcoeff) >> 2; - baserd = RDFUNC(x->rdmult, x->rddiv, baserate, baseerror, 100); - - for (nc = 1; nc < (1 << nzcoefcount); nc++) - { - //reset coefficients - vpx_memcpy(bd->qcoeff, saved_qcoefs, 32); - vpx_memcpy(bd->dqcoeff, saved_dqcoefs, 32); - - for (k = 0; k < nzcoefcount; k++) - { - int pos = nzpos[k]; + BLOCK *b; + BLOCKD *d; + vp8_token_state tokens[17][2]; + unsigned best_mask[2]; + const short *dequant_ptr; + const short *coeff_ptr; + short *qcoeff_ptr; + short *dqcoeff_ptr; + int eob; + int i0; + int rc; + int x; + int sz; + int next; + int path; + int rdmult; + int rddiv; + int final_eob; + int rd_cost0; + int rd_cost1; + int rate0; + int rate1; + int error0; + int error1; + int t0; + int t1; + int best; + int band; + int pt; + + b = &mb->block[i]; + d = &mb->e_mbd.block[i]; + + /* Enable this to test the effect of RDO as a replacement for the dynamic + * zero bin instead of an augmentation of it. + */ +#if 0 + vp8_strict_quantize_b(b, d); +#endif - if ((nc & (1 << k))) + dequant_ptr = &d->dequant[0][0]; + coeff_ptr = &b->coeff[0]; + qcoeff_ptr = d->qcoeff; + dqcoeff_ptr = d->dqcoeff; + i0 = !type; + eob = d->eob; + + /* Now set up a Viterbi trellis to evaluate alternative roundings. */ + /* TODO: These should vary with the block type, since the quantizer does. */ + rdmult = mb->rdmult << 2; + rddiv = mb->rddiv; + best_mask[0] = best_mask[1] = 0; + /* Initialize the sentinel node of the trellis. */ + tokens[eob][0].rate = 0; + tokens[eob][0].error = 0; + tokens[eob][0].next = 16; + tokens[eob][0].token = DCT_EOB_TOKEN; + tokens[eob][0].qc = 0; + *(tokens[eob] + 1) = *(tokens[eob] + 0); + next = eob; + for (i = eob; i-- > i0;) + { + int base_bits; + int d2; + int dx; + + rc = vp8_default_zig_zag1d[i]; + x = qcoeff_ptr[rc]; + /* Only add a trellis state for non-zero coefficients. */ + if (x) + { + int shortcut=0; + error0 = tokens[next][0].error; + error1 = tokens[next][1].error; + /* Evaluate the first possibility for this state. */ + rate0 = tokens[next][0].rate; + rate1 = tokens[next][1].rate; + t0 = (vp8_dct_value_tokens_ptr + x)->Token; + /* Consider both possible successor states. */ + if (next < 16) { - int cur_qcoef = bd->qcoeff[pos]; - - if (cur_qcoef < 0) - { - bd->qcoeff[pos]++; - bd->dqcoeff[pos] = bd->qcoeff[pos] * dequant_ptr[pos]; - } - else - { - bd->qcoeff[pos]--; - bd->dqcoeff[pos] = bd->qcoeff[pos] * dequant_ptr[pos]; - } + band = vp8_coef_bands[i + 1]; + pt = vp8_prev_token_class[t0]; + rate0 += + mb->token_costs[type][band][pt][tokens[next][0].token]; + rate1 += + mb->token_costs[type][band][pt][tokens[next][1].token]; } - } - - { - int eob = -1; - int rc; - int m; - - for (m = 0; m < 16; m++) + rd_cost0 = RDCOST(rdmult, rddiv, rate0, error0); + rd_cost1 = RDCOST(rdmult, rddiv, rate1, error1); + if (rd_cost0 == rd_cost1) { - rc = vp8_default_zig_zag1d[m]; - - if (bd->qcoeff[rc]) - eob = m; + rd_cost0 = RDTRUNC(rdmult, rddiv, rate0, error0); + rd_cost1 = RDTRUNC(rdmult, rddiv, rate1, error1); } - - bd->eob = eob + 1; - } - - rate = cost_coeffs(x, bd, type, a, l); - error = ENCODEMB_INVOKE(&rtcd->encodemb, berr)(b->coeff, bd->dqcoeff) >> 2; - thisrd = RDFUNC(x->rdmult, x->rddiv, rate, error, 100); - - if (thisrd < baserd) - { - baserd = thisrd; - bestnc = nc; - besteob = bd->eob; - } - } - - //reset coefficients - vpx_memcpy(bd->qcoeff, saved_qcoefs, 32); - vpx_memcpy(bd->dqcoeff, saved_dqcoefs, 32); - - if (bestnc) - { - for (k = 0; k < nzcoefcount; k++) - { - int pos = nzpos[k]; - - if (bestnc & (1 << k)) + /* And pick the best. */ + best = rd_cost1 < rd_cost0; + base_bits = *(vp8_dct_value_cost_ptr + x); + dx = dqcoeff_ptr[rc] - coeff_ptr[rc]; + d2 = dx*dx; + tokens[i][0].rate = base_bits + (best ? rate1 : rate0); + tokens[i][0].error = d2 + (best ? error1 : error0); + tokens[i][0].next = next; + tokens[i][0].token = t0; + tokens[i][0].qc = x; + best_mask[0] |= best << i; + /* Evaluate the second possibility for this state. */ + rate0 = tokens[next][0].rate; + rate1 = tokens[next][1].rate; + + if((abs(x)*dequant_ptr[rc]>abs(coeff_ptr[rc])) && + (abs(x)*dequant_ptr[rc]<abs(coeff_ptr[rc])+dequant_ptr[rc])) + shortcut = 1; + else + shortcut = 0; + + if(shortcut) { - int cur_qcoef = bd->qcoeff[pos]; - - if (cur_qcoef < 0) - { - bd->qcoeff[pos]++; - bd->dqcoeff[pos] = bd->qcoeff[pos] * dequant_ptr[pos]; - } - else - { - bd->qcoeff[pos]--; - bd->dqcoeff[pos] = bd->qcoeff[pos] * dequant_ptr[pos]; - } + sz = -(x < 0); + x -= 2*sz + 1; } - } -#if 0 - { - int eob = -1; - int rc; - int m; - - for (m = 0; m < 16; m++) + /* Consider both possible successor states. */ + if (!x) { - rc = vp8_default_zig_zag1d[m]; - - if (bd->qcoeff[rc]) - eob = m; + /* If we reduced this coefficient to zero, check to see if + * we need to move the EOB back here. + */ + t0 = tokens[next][0].token == DCT_EOB_TOKEN ? + DCT_EOB_TOKEN : ZERO_TOKEN; + t1 = tokens[next][1].token == DCT_EOB_TOKEN ? + DCT_EOB_TOKEN : ZERO_TOKEN; } - - bd->eob = eob + 1; - } -#endif - } - -#if 1 - bd->eob = besteob; -#endif -#if 0 - { - int eob = -1; - int rc; - int m; - - for (m = 0; m < 16; m++) - { - rc = vp8_default_zig_zag1d[m]; - - if (bd->qcoeff[rc]) - eob = m; - } - - bd->eob = eob + 1; - } - -#endif - *a = *l = (bd->eob != !type); - return; -} - -void vp8_optimize_bplus(MACROBLOCK *x, int i, int type, ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l, const VP8_ENCODER_RTCD *rtcd) -{ - BLOCK *b = &x->block[i]; - BLOCKD *bd = &x->e_mbd.block[i]; - short *dequant_ptr = &bd->dequant[0][0]; - int nzpos[16] = {0}; - short saved_qcoefs[16]; - short saved_dqcoefs[16]; - int baserate, baseerror, baserd; - int rate, error, thisrd; - int k; - int nzcoefcount = 0; - int nc, bestnc = 0; - int besteob; - - // count potential coefficient to be optimized - for (k = !type; k < 16; k++) - { - int qcoef = abs(bd->qcoeff[k]); - int coef = abs(b->coeff[k]); - int dq = dequant_ptr[k]; - - if (qcoef && (qcoef * dq < coef) && (coef < (qcoef * dq + dq))) - { - nzpos[nzcoefcount] = k; - nzcoefcount++; - } - } - - // if nothing here, do nothing for this block. - if (!nzcoefcount) - { - //do not update context, we need do the other half. - //*a = *l = (bd->eob != !type); - return; - } - - // save a copy of quantized coefficients - vpx_memcpy(saved_qcoefs, bd->qcoeff, 32); - vpx_memcpy(saved_dqcoefs, bd->dqcoeff, 32); - - besteob = bd->eob; - baserate = cost_coeffs(x, bd, type, a, l); - baseerror = ENCODEMB_INVOKE(&rtcd->encodemb, berr)(b->coeff, bd->dqcoeff) >> 2; - baserd = RDFUNC(x->rdmult, x->rddiv, baserate, baseerror, 100); - - for (nc = 1; nc < (1 << nzcoefcount); nc++) - { - //reset coefficients - vpx_memcpy(bd->qcoeff, saved_qcoefs, 32); - vpx_memcpy(bd->dqcoeff, saved_dqcoefs, 32); - - for (k = 0; k < nzcoefcount; k++) - { - int pos = nzpos[k]; - - if ((nc & (1 << k))) + else { - int cur_qcoef = bd->qcoeff[pos]; - - if (cur_qcoef < 0) + t0=t1 = (vp8_dct_value_tokens_ptr + x)->Token; + } + if (next < 16) + { + band = vp8_coef_bands[i + 1]; + if(t0!=DCT_EOB_TOKEN) { - bd->qcoeff[pos]--; - bd->dqcoeff[pos] = bd->qcoeff[pos] * dequant_ptr[pos]; + pt = vp8_prev_token_class[t0]; + rate0 += mb->token_costs[type][band][pt][ + tokens[next][0].token]; } - else + if(t1!=DCT_EOB_TOKEN) { - bd->qcoeff[pos]++; - bd->dqcoeff[pos] = bd->qcoeff[pos] * dequant_ptr[pos]; + pt = vp8_prev_token_class[t1]; + rate1 += mb->token_costs[type][band][pt][ + tokens[next][1].token]; } } - } - - { - int eob = -1; - int rc; - int m; - for (m = 0; m < 16; m++) + rd_cost0 = RDCOST(rdmult, rddiv, rate0, error0); + rd_cost1 = RDCOST(rdmult, rddiv, rate1, error1); + if (rd_cost0 == rd_cost1) { - rc = vp8_default_zig_zag1d[m]; - - if (bd->qcoeff[rc]) - eob = m; + rd_cost0 = RDTRUNC(rdmult, rddiv, rate0, error0); + rd_cost1 = RDTRUNC(rdmult, rddiv, rate1, error1); } + /* And pick the best. */ + best = rd_cost1 < rd_cost0; + base_bits = *(vp8_dct_value_cost_ptr + x); - bd->eob = eob + 1; - } - - rate = cost_coeffs(x, bd, type, a, l); - error = ENCODEMB_INVOKE(&rtcd->encodemb, berr)(b->coeff, bd->dqcoeff) >> 2; - thisrd = RDFUNC(x->rdmult, x->rddiv, rate, error, 100); - - if (thisrd < baserd) - { - baserd = thisrd; - bestnc = nc; - besteob = bd->eob; - } - } - - //reset coefficients - vpx_memcpy(bd->qcoeff, saved_qcoefs, 32); - vpx_memcpy(bd->dqcoeff, saved_dqcoefs, 32); - - if (bestnc) - { - for (k = 0; k < nzcoefcount; k++) - { - int pos = nzpos[k]; - - if (bestnc & (1 << k)) + if(shortcut) { - int cur_qcoef = bd->qcoeff[pos]; - - if (cur_qcoef < 0) - { - bd->qcoeff[pos]++; - bd->dqcoeff[pos] = bd->qcoeff[pos] * dequant_ptr[pos]; - } - else - { - bd->qcoeff[pos]--; - bd->dqcoeff[pos] = bd->qcoeff[pos] * dequant_ptr[pos]; - } + dx -= (dequant_ptr[rc] + sz) ^ sz; + d2 = dx*dx; } - } - } - - bd->eob = besteob; - //do not update context, we need do the other half. - //*a = *l = (bd->eob != !type); - return; -} - -void vp8_optimize_y2b(MACROBLOCK *x, int i, int type, ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l, const VP8_ENCODER_RTCD *rtcd) -{ - - BLOCK *b = &x->block[i]; - BLOCKD *bd = &x->e_mbd.block[i]; - short *dequant_ptr = &bd->dequant[0][0]; - - int baserate, baseerror, baserd; - int rate, error, thisrd; - int k; - - if (bd->eob == 0) - return; - - baserate = cost_coeffs(x, bd, type, a, l); - baseerror = ENCODEMB_INVOKE(&rtcd->encodemb, berr)(b->coeff, bd->dqcoeff) >> 4; - baserd = RDFUNC(x->rdmult, x->rddiv, baserate, baseerror, 100); - - for (k = 0; k < 16; k++) - { - int cur_qcoef = bd->qcoeff[k]; - - if (!cur_qcoef) - continue; - - if (cur_qcoef < 0) - { - bd->qcoeff[k]++; - bd->dqcoeff[k] = bd->qcoeff[k] * dequant_ptr[k]; - } + tokens[i][1].rate = base_bits + (best ? rate1 : rate0); + tokens[i][1].error = d2 + (best ? error1 : error0); + tokens[i][1].next = next; + tokens[i][1].token =best?t1:t0; + tokens[i][1].qc = x; + best_mask[1] |= best << i; + /* Finally, make this the new head of the trellis. */ + next = i; + } + /* There's no choice to make for a zero coefficient, so we don't + * add a new trellis node, but we do need to update the costs. + */ else { - bd->qcoeff[k]--; - bd->dqcoeff[k] = bd->qcoeff[k] * dequant_ptr[k]; - } - - if (bd->qcoeff[k] == 0) - { - int eob = -1; - int rc; - int l; - - for (l = 0; l < 16; l++) + band = vp8_coef_bands[i + 1]; + t0 = tokens[next][0].token; + t1 = tokens[next][1].token; + /* Update the cost of each path if we're past the EOB token. */ + if (t0 != DCT_EOB_TOKEN) { - rc = vp8_default_zig_zag1d[l]; - - if (bd->qcoeff[rc]) - eob = l; + tokens[next][0].rate += mb->token_costs[type][band][0][t0]; + tokens[next][0].token = ZERO_TOKEN; } - - bd->eob = eob + 1; - } - - rate = cost_coeffs(x, bd, type, a, l); - error = ENCODEMB_INVOKE(&rtcd->encodemb, berr)(b->coeff, bd->dqcoeff) >> 4; - thisrd = RDFUNC(x->rdmult, x->rddiv, rate, error, 100); - - if (thisrd > baserd) - { - bd->qcoeff[k] = cur_qcoef; - bd->dqcoeff[k] = cur_qcoef * dequant_ptr[k]; - } - else - { - baserd = thisrd; - } - - } - - { - int eob = -1; - int rc; - - for (k = 0; k < 16; k++) - { - rc = vp8_default_zig_zag1d[k]; - - if (bd->qcoeff[rc]) - eob = k; + if (t1 != DCT_EOB_TOKEN) + { + tokens[next][1].rate += mb->token_costs[type][band][0][t1]; + tokens[next][1].token = ZERO_TOKEN; + } + /* Don't update next, because we didn't add a new node. */ } - - bd->eob = eob + 1; } - return; + /* Now pick the best path through the whole trellis. */ + band = vp8_coef_bands[i + 1]; + VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l); + rate0 = tokens[next][0].rate; + rate1 = tokens[next][1].rate; + error0 = tokens[next][0].error; + error1 = tokens[next][1].error; + t0 = tokens[next][0].token; + t1 = tokens[next][1].token; + rate0 += mb->token_costs[type][band][pt][t0]; + rate1 += mb->token_costs[type][band][pt][t1]; + rd_cost0 = RDCOST(rdmult, rddiv, rate0, error0); + rd_cost1 = RDCOST(rdmult, rddiv, rate1, error1); + if (rd_cost0 == rd_cost1) + { + rd_cost0 = RDTRUNC(rdmult, rddiv, rate0, error0); + rd_cost1 = RDTRUNC(rdmult, rddiv, rate1, error1); + } + best = rd_cost1 < rd_cost0; + final_eob = i0 - 1; + for (i = next; i < eob; i = next) + { + x = tokens[i][best].qc; + if (x) + final_eob = i; + rc = vp8_default_zig_zag1d[i]; + qcoeff_ptr[rc] = x; + dqcoeff_ptr[rc] = x * dequant_ptr[rc]; + next = tokens[i][best].next; + best = (best_mask[best] >> i) & 1; + } + final_eob++; + + d->eob = final_eob; + *a = *l = (d->eob != !type); } - void vp8_optimize_mb(MACROBLOCK *x, const VP8_ENCODER_RTCD *rtcd) { - int cost = 0; int b; - TEMP_CONTEXT t, t2; - int type = 0; + int type; + int has_2nd_order; + ENTROPY_CONTEXT_PLANES t_above, t_left; + ENTROPY_CONTEXT *ta; + ENTROPY_CONTEXT *tl; - vp8_setup_temp_context(&t, x->e_mbd.above_context[Y1CONTEXT], x->e_mbd.left_context[Y1CONTEXT], 4); + vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES)); - if (x->e_mbd.mbmi.mode == SPLITMV || x->e_mbd.mbmi.mode == B_PRED) - type = 3; + ta = (ENTROPY_CONTEXT *)&t_above; + tl = (ENTROPY_CONTEXT *)&t_left; + + has_2nd_order = (x->e_mbd.mode_info_context->mbmi.mode != B_PRED + && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV); + type = has_2nd_order ? 0 : 3; for (b = 0; b < 16; b++) { - //vp8_optimize_bplus(x, b, type, t.a + vp8_block2above[b], t.l + vp8_block2left[b]); - vp8_optimize_b(x, b, type, t.a + vp8_block2above[b], t.l + vp8_block2left[b], rtcd); + vp8_optimize_b(x, b, type, + ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd); } - vp8_setup_temp_context(&t, x->e_mbd.above_context[UCONTEXT], x->e_mbd.left_context[UCONTEXT], 2); - vp8_setup_temp_context(&t2, x->e_mbd.above_context[VCONTEXT], x->e_mbd.left_context[VCONTEXT], 2); - for (b = 16; b < 20; b++) { - //vp8_optimize_bplus(x, b, vp8_block2type[b], t.a + vp8_block2above[b], t.l + vp8_block2left[b]); - vp8_optimize_b(x, b, vp8_block2type[b], t.a + vp8_block2above[b], t.l + vp8_block2left[b], rtcd); + vp8_optimize_b(x, b, vp8_block2type[b], + ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd); } for (b = 20; b < 24; b++) { - //vp8_optimize_bplus(x, b, vp8_block2type[b], t2.a + vp8_block2above[b], t2.l + vp8_block2left[b]); - vp8_optimize_b(x, b, vp8_block2type[b], t2.a + vp8_block2above[b], t2.l + vp8_block2left[b], rtcd); - } -} - - - -void vp8_super_slow_yquant_optimization(MACROBLOCK *x, int type, const VP8_ENCODER_RTCD *rtcd) -{ - BLOCK *b = &x->block[0]; - BLOCKD *bd = &x->e_mbd.block[0]; - short *dequant_ptr = &bd->dequant[0][0]; - struct - { - int block; - int pos; - } nzpos[256]; - short saved_qcoefs[256]; - short saved_dqcoefs[256]; - short *coef_ptr = x->coeff; - short *qcoef_ptr = x->e_mbd.qcoeff; - short *dqcoef_ptr = x->e_mbd.dqcoeff; - - int baserate, baseerror, baserd; - int rate, error, thisrd; - int i, k; - int nzcoefcount = 0; - int nc, bestnc = 0; - int besteob; - - //this code has assumption in macroblock coeff buffer layout - for (i = 0; i < 16; i++) - { - // count potential coefficient to be optimized - for (k = !type; k < 16; k++) - { - int qcoef = abs(qcoef_ptr[i*16 + k]); - int coef = abs(coef_ptr[i*16 + k]); - int dq = dequant_ptr[k]; - - if (qcoef && (qcoef * dq > coef) && (qcoef * dq < coef + dq)) - { - nzpos[nzcoefcount].block = i; - nzpos[nzcoefcount].pos = k; - nzcoefcount++; - } - } - } - - // if nothing here, do nothing for this macro_block. - if (!nzcoefcount || nzcoefcount > 15) - { - return; - } - - /****************************************************************************** - looking from each coeffient's perspective, each identifed coefficent above could - have 2 values:roundeddown(x) and roundedup(x). Therefore the total number of - different states is less than 2**nzcoefcount. - ******************************************************************************/ - // save the qunatized coefficents and dequantized coefficicents - vpx_memcpy(saved_qcoefs, x->e_mbd.qcoeff, 256); - vpx_memcpy(saved_dqcoefs, x->e_mbd.dqcoeff, 256); - - baserate = mbycost_coeffs(x); - baseerror = ENCODEMB_INVOKE(&rtcd->encodemb, mberr)(x, !type); - baserd = RDFUNC(x->rdmult, x->rddiv, baserate, baseerror, 100); - - for (nc = 1; nc < (1 << nzcoefcount); nc++) - { - //reset coefficients - vpx_memcpy(x->e_mbd.qcoeff, saved_qcoefs, 256); - vpx_memcpy(x->e_mbd.dqcoeff, saved_dqcoefs, 256); - - for (k = 0; k < nzcoefcount; k++) - { - int bk = nzpos[k].block; - int pos = nzpos[k].pos; - int mbkpos = bk * 16 + pos; - - if ((nc & (1 << k))) - { - int cur_qcoef = x->e_mbd.qcoeff[mbkpos]; - - if (cur_qcoef < 0) - { - x->e_mbd.qcoeff[mbkpos]++; - x->e_mbd.dqcoeff[mbkpos] = x->e_mbd.qcoeff[mbkpos] * dequant_ptr[pos]; - } - else - { - x->e_mbd.qcoeff[mbkpos]--; - x->e_mbd.dqcoeff[mbkpos] = x->e_mbd.qcoeff[mbkpos] * dequant_ptr[pos]; - } - } - } - - for (i = 0; i < 16; i++) - { - BLOCKD *bd = &x->e_mbd.block[i]; - { - int eob = -1; - int rc; - int l; - - for (l = 0; l < 16; l++) - { - rc = vp8_default_zig_zag1d[l]; - - if (bd->qcoeff[rc]) - eob = l; - } - - bd->eob = eob + 1; - } - } - - rate = mbycost_coeffs(x); - error = ENCODEMB_INVOKE(&rtcd->encodemb, mberr)(x, !type);; - thisrd = RDFUNC(x->rdmult, x->rddiv, rate, error, 100); - - if (thisrd < baserd) - { - baserd = thisrd; - bestnc = nc; - besteob = bd->eob; - } + vp8_optimize_b(x, b, vp8_block2type[b], + ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd); } - //reset coefficients - vpx_memcpy(x->e_mbd.qcoeff, saved_qcoefs, 256); - vpx_memcpy(x->e_mbd.dqcoeff, saved_dqcoefs, 256); - if (bestnc) + /* + if (has_2nd_order) { - for (k = 0; k < nzcoefcount; k++) - { - int bk = nzpos[k].block; - int pos = nzpos[k].pos; - int mbkpos = bk * 16 + pos; - - if ((nc & (1 << k))) - { - int cur_qcoef = x->e_mbd.qcoeff[mbkpos]; - - if (cur_qcoef < 0) - { - x->e_mbd.qcoeff[mbkpos]++; - x->e_mbd.dqcoeff[mbkpos] = x->e_mbd.qcoeff[mbkpos] * dequant_ptr[pos]; - } - else - { - x->e_mbd.qcoeff[mbkpos]--; - x->e_mbd.dqcoeff[mbkpos] = x->e_mbd.qcoeff[mbkpos] * dequant_ptr[pos]; - } - } - } + vp8_setup_temp_context(&t, x->e_mbd.above_context[Y2CONTEXT], + x->e_mbd.left_context[Y2CONTEXT], 1); + vp8_optimize_b(x, 24, 1, t.a, t.l, rtcd); } + */ +} - for (i = 0; i < 16; i++) - { - BLOCKD *bd = &x->e_mbd.block[i]; - { - int eob = -1; - int rc; - int l; - - for (l = 0; l < 16; l++) - { - rc = vp8_default_zig_zag1d[l]; - - if (bd->qcoeff[rc]) - eob = l; - } - bd->eob = eob + 1; - } - } - - return; -} static void vp8_find_mb_skip_coef(MACROBLOCK *x) { int i; - x->e_mbd.mbmi.mb_skip_coeff = 1; + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff = 1; - if (x->e_mbd.mbmi.mode != B_PRED && x->e_mbd.mbmi.mode != SPLITMV) + if (x->e_mbd.mode_info_context->mbmi.mode != B_PRED && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV) { for (i = 0; i < 16; i++) { - x->e_mbd.mbmi.mb_skip_coeff &= (x->e_mbd.block[i].eob < 2); + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff &= (x->e_mbd.block[i].eob < 2); } for (i = 16; i < 25; i++) { - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); } } else { for (i = 0; i < 24; i++) { - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); } } } -void vp8_optimize_mb_slow(MACROBLOCK *x, const VP8_ENCODER_RTCD *rtcd) -{ - int cost = 0; - int b; - TEMP_CONTEXT t, t2; - int type = 0; - - - vp8_setup_temp_context(&t, x->e_mbd.above_context[Y1CONTEXT], x->e_mbd.left_context[Y1CONTEXT], 4); - - if (x->e_mbd.mbmi.mode == SPLITMV || x->e_mbd.mbmi.mode == B_PRED) - type = 3; - - vp8_super_slow_yquant_optimization(x, type, rtcd); - /* - for(b=0;b<16;b++) - { - vp8_optimize_b(x, b, type, t.a + vp8_block2above[b], t.l + vp8_block2left[b]); - } - */ - - vp8_setup_temp_context(&t, x->e_mbd.above_context[UCONTEXT], x->e_mbd.left_context[UCONTEXT], 2); - - for (b = 16; b < 20; b++) - { - vp8_optimize_b(x, b, vp8_block2type[b], t.a + vp8_block2above[b], t.l + vp8_block2left[b], rtcd); - } - - vp8_setup_temp_context(&t2, x->e_mbd.above_context[VCONTEXT], x->e_mbd.left_context[VCONTEXT], 2); - - for (b = 20; b < 24; b++) - { - vp8_optimize_b(x, b, vp8_block2type[b], t2.a + vp8_block2above[b], t2.l + vp8_block2left[b], rtcd); - } -} - - void vp8_optimize_mby(MACROBLOCK *x, const VP8_ENCODER_RTCD *rtcd) { - int cost = 0; int b; - TEMP_CONTEXT t; - int type = 0; + int type; + int has_2nd_order; + + ENTROPY_CONTEXT_PLANES t_above, t_left; + ENTROPY_CONTEXT *ta; + ENTROPY_CONTEXT *tl; - if (!x->e_mbd.above_context[Y1CONTEXT]) + if (!x->e_mbd.above_context) return; - if (!x->e_mbd.left_context[Y1CONTEXT]) + if (!x->e_mbd.left_context) return; - vp8_setup_temp_context(&t, x->e_mbd.above_context[Y1CONTEXT], x->e_mbd.left_context[Y1CONTEXT], 4); + vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES)); - if (x->e_mbd.mbmi.mode == SPLITMV || x->e_mbd.mbmi.mode == B_PRED) - type = 3; + ta = (ENTROPY_CONTEXT *)&t_above; + tl = (ENTROPY_CONTEXT *)&t_left; + + has_2nd_order = (x->e_mbd.mode_info_context->mbmi.mode != B_PRED + && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV); + type = has_2nd_order ? 0 : 3; for (b = 0; b < 16; b++) { - vp8_optimize_b(x, b, type, t.a + vp8_block2above[b], t.l + vp8_block2left[b], rtcd); + vp8_optimize_b(x, b, type, + ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd); } + /* + if (has_2nd_order) + { + vp8_setup_temp_context(&t, x->e_mbd.above_context[Y2CONTEXT], + x->e_mbd.left_context[Y2CONTEXT], 1); + vp8_optimize_b(x, 24, 1, t.a, t.l, rtcd); + } + */ } void vp8_optimize_mbuv(MACROBLOCK *x, const VP8_ENCODER_RTCD *rtcd) { - int cost = 0; int b; - TEMP_CONTEXT t, t2; - int type = 0; + ENTROPY_CONTEXT_PLANES t_above, t_left; + ENTROPY_CONTEXT *ta; + ENTROPY_CONTEXT *tl; - if (!x->e_mbd.above_context[UCONTEXT]) + if (!x->e_mbd.above_context) return; - if (!x->e_mbd.left_context[UCONTEXT]) + if (!x->e_mbd.left_context) return; - if (!x->e_mbd.above_context[VCONTEXT]) - return; - - if (!x->e_mbd.left_context[VCONTEXT]) - return; + vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES)); - - vp8_setup_temp_context(&t, x->e_mbd.above_context[UCONTEXT], x->e_mbd.left_context[UCONTEXT], 2); - vp8_setup_temp_context(&t2, x->e_mbd.above_context[VCONTEXT], x->e_mbd.left_context[VCONTEXT], 2); + ta = (ENTROPY_CONTEXT *)&t_above; + tl = (ENTROPY_CONTEXT *)&t_left; for (b = 16; b < 20; b++) { vp8_optimize_b(x, b, vp8_block2type[b], - t.a + vp8_block2above[b], t.l + vp8_block2left[b], rtcd); - + ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd); } for (b = 20; b < 24; b++) { vp8_optimize_b(x, b, vp8_block2type[b], - t2.a + vp8_block2above[b], t2.l + vp8_block2left[b], rtcd); + ta + vp8_block2above[b], tl + vp8_block2left[b], rtcd); } } @@ -1067,15 +650,11 @@ void vp8_encode_inter16x16(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x) vp8_quantize_mb(x); #if !(CONFIG_REALTIME_ONLY) -#if 1 - if (x->optimize && x->rddiv > 1) { vp8_optimize_mb(x, rtcd); vp8_find_mb_skip_coef(x); } - -#endif #endif vp8_inverse_transform_mb(IF_RTCD(&rtcd->common->idct), &x->e_mbd); @@ -1122,8 +701,8 @@ void vp8_encode_inter16x16uvrd(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x) vp8_build_inter_predictors_mbuv(&x->e_mbd); ENCODEMB_INVOKE(&rtcd->encodemb, submbuv)(x->src_diff, x->src.u_buffer, x->src.v_buffer, x->e_mbd.predictor, x->src.uv_stride); - vp8_transform_mbuvrd(x); + vp8_transform_mbuv(x); - vp8_quantize_mbuvrd(x); + vp8_quantize_mbuv(x); } diff --git a/vp8/encoder/encodemb.h b/vp8/encoder/encodemb.h index 91ca8f5..08f75c3 100644 --- a/vp8/encoder/encodemb.h +++ b/vp8/encoder/encodemb.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -99,9 +100,7 @@ extern void vp8_stuff_inter16x16(MACROBLOCK *x); void vp8_build_dcblock(MACROBLOCK *b); void vp8_transform_mb(MACROBLOCK *mb); void vp8_transform_mbuv(MACROBLOCK *x); -void vp8_transform_mbuvrd(MACROBLOCK *x); void vp8_transform_intra_mby(MACROBLOCK *x); -void vp8_transform_intra_mbyrd(MACROBLOCK *x); void Encode16x16Y(MACROBLOCK *x); void Encode16x16UV(MACROBLOCK *x); void vp8_encode_inter16x16uv(const struct VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x); diff --git a/vp8/encoder/encodemv.c b/vp8/encoder/encodemv.c index f287edc..cce7530 100644 --- a/vp8/encoder/encodemv.c +++ b/vp8/encoder/encodemv.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -252,7 +253,7 @@ static void write_component_probs( vp8_writer *const w, struct mv_context *cur_mvc, const struct mv_context *default_mvc_, - const struct mv_context *update_mvc, + const struct mv_context *update_mvc, const unsigned int events [MVvals], unsigned int rc, int *updated @@ -282,8 +283,6 @@ static void write_component_probs( //j=0 { - int j = 0; - const int c = events [mv_max]; is_short_ct [0] += c; // Short vector diff --git a/vp8/encoder/encodemv.h b/vp8/encoder/encodemv.h index 1c1f450..e4481bf 100644 --- a/vp8/encoder/encodemv.h +++ b/vp8/encoder/encodemv.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/ethreading.c b/vp8/encoder/ethreading.c index a0b50d2..962e741 100644 --- a/vp8/encoder/ethreading.c +++ b/vp8/encoder/ethreading.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -27,7 +28,7 @@ THREAD_FUNCTION thread_encoding_proc(void *p_data) int ithread = ((ENCODETHREAD_DATA *)p_data)->ithread; VP8_COMP *cpi = (VP8_COMP *)(((ENCODETHREAD_DATA *)p_data)->ptr1); MB_ROW_COMP *mbri = (MB_ROW_COMP *)(((ENCODETHREAD_DATA *)p_data)->ptr2); - ENTROPY_CONTEXT mb_row_left_context[4][4]; + ENTROPY_CONTEXT_PLANES mb_row_left_context; //printf("Started thread %d\n", ithread); @@ -55,8 +56,10 @@ THREAD_FUNCTION thread_encoding_proc(void *p_data) int i; int recon_yoffset, recon_uvoffset; int mb_col; - int recon_y_stride = cm->last_frame.y_stride; - int recon_uv_stride = cm->last_frame.uv_stride; + int ref_fb_idx = cm->lst_fb_idx; + int dst_fb_idx = cm->new_fb_idx; + int recon_y_stride = cm->yv12_fb[ref_fb_idx].y_stride; + int recon_uv_stride = cm->yv12_fb[ref_fb_idx].uv_stride; volatile int *last_row_current_mb_col; if (ithread > 0) @@ -65,11 +68,8 @@ THREAD_FUNCTION thread_encoding_proc(void *p_data) last_row_current_mb_col = &cpi->current_mb_col_main; // reset above block coeffs - xd->above_context[Y1CONTEXT] = cm->above_context[Y1CONTEXT]; - xd->above_context[UCONTEXT ] = cm->above_context[UCONTEXT ]; - xd->above_context[VCONTEXT ] = cm->above_context[VCONTEXT ]; - xd->above_context[Y2CONTEXT] = cm->above_context[Y2CONTEXT]; - xd->left_context = mb_row_left_context; + xd->above_context = cm->above_context; + xd->left_context = &mb_row_left_context; vp8_zero(mb_row_left_context); @@ -106,9 +106,9 @@ THREAD_FUNCTION thread_encoding_proc(void *p_data) x->mv_row_min = -((mb_row * 16) + (VP8BORDERINPIXELS - 16)); x->mv_row_max = ((cm->mb_rows - 1 - mb_row) * 16) + (VP8BORDERINPIXELS - 16); - xd->dst.y_buffer = cm->new_frame.y_buffer + recon_yoffset; - xd->dst.u_buffer = cm->new_frame.u_buffer + recon_uvoffset; - xd->dst.v_buffer = cm->new_frame.v_buffer + recon_uvoffset; + xd->dst.y_buffer = cm->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset; + xd->dst.u_buffer = cm->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset; + xd->dst.v_buffer = cm->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset; xd->left_available = (mb_col != 0); // Is segmentation enabled @@ -117,14 +117,14 @@ THREAD_FUNCTION thread_encoding_proc(void *p_data) { // Code to set segment id in xd->mbmi.segment_id for current MB (with range checking) if (cpi->segmentation_map[seg_map_index+mb_col] <= 3) - xd->mbmi.segment_id = cpi->segmentation_map[seg_map_index+mb_col]; + xd->mode_info_context->mbmi.segment_id = cpi->segmentation_map[seg_map_index+mb_col]; else - xd->mbmi.segment_id = 0; + xd->mode_info_context->mbmi.segment_id = 0; vp8cx_mb_init_quantizer(cpi, x); } else - xd->mbmi.segment_id = 0; // Set to Segment 0 by default + xd->mode_info_context->mbmi.segment_id = 0; // Set to Segment 0 by default if (cm->frame_type == KEY_FRAME) @@ -147,24 +147,21 @@ THREAD_FUNCTION thread_encoding_proc(void *p_data) for (b = 0; b < xd->mbmi.partition_count; b++) { - inter_b_modes[xd->mbmi.partition_bmi[b].mode] ++; + inter_b_modes[x->partition->bmi[b].mode] ++; } } #endif // Count of last ref frame 0,0 useage - if ((xd->mbmi.mode == ZEROMV) && (xd->mbmi.ref_frame == LAST_FRAME)) + if ((xd->mode_info_context->mbmi.mode == ZEROMV) && (xd->mode_info_context->mbmi.ref_frame == LAST_FRAME)) cpi->inter_zz_count ++; } cpi->tplist[mb_row].stop = *tp; - xd->gf_active_ptr++; // Increment pointer into gf useage flags structure for next mb - - // store macroblock mode info into context array - vpx_memcpy(&xd->mode_info_context->mbmi, &xd->mbmi, sizeof(xd->mbmi)); + x->gf_active_ptr++; // Increment pointer into gf useage flags structure for next mb for (i = 0; i < 16; i++) vpx_memcpy(&xd->mode_info_context->bmi[i], &xd->block[i].bmi, sizeof(xd->block[i].bmi)); @@ -178,15 +175,13 @@ THREAD_FUNCTION thread_encoding_proc(void *p_data) recon_uvoffset += 8; // Keep track of segment useage - segment_counts[xd->mbmi.segment_id] ++; + segment_counts[xd->mode_info_context->mbmi.segment_id] ++; // skip to next mb xd->mode_info_context++; + x->partition_info++; - xd->above_context[Y1CONTEXT] += 4; - xd->above_context[UCONTEXT ] += 2; - xd->above_context[VCONTEXT ] += 2; - xd->above_context[Y2CONTEXT] ++; + xd->above_context++; cpi->mb_row_ei[ithread].current_mb_col = mb_col; @@ -194,19 +189,21 @@ THREAD_FUNCTION thread_encoding_proc(void *p_data) //extend the recon for intra prediction vp8_extend_mb_row( - &cm->new_frame, + &cm->yv12_fb[dst_fb_idx], xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8); // this is to account for the border xd->mode_info_context++; + x->partition_info++; x->src.y_buffer += 16 * x->src.y_stride * (cpi->encoding_thread_count + 1) - 16 * cm->mb_cols; x->src.u_buffer += 8 * x->src.uv_stride * (cpi->encoding_thread_count + 1) - 8 * cm->mb_cols; x->src.v_buffer += 8 * x->src.uv_stride * (cpi->encoding_thread_count + 1) - 8 * cm->mb_cols; xd->mode_info_context += xd->mode_info_stride * cpi->encoding_thread_count; + x->partition_info += xd->mode_info_stride * cpi->encoding_thread_count; if (ithread == (cpi->encoding_thread_count - 1) || mb_row == cm->mb_rows - 1) { @@ -256,13 +253,8 @@ static void setup_mbby_copy(MACROBLOCK *mbdst, MACROBLOCK *mbsrc) z->vp8_short_fdct4x4 = x->vp8_short_fdct4x4; z->vp8_short_fdct8x4 = x->vp8_short_fdct8x4; - z->short_fdct4x4rd = x->short_fdct4x4rd; - z->short_fdct8x4rd = x->short_fdct8x4rd; - z->short_fdct8x4rd = x->short_fdct8x4rd; - z->vp8_short_fdct4x4_ptr = x->vp8_short_fdct4x4_ptr; z->short_walsh4x4 = x->short_walsh4x4; z->quantize_b = x->quantize_b; - z->quantize_brd = x->quantize_brd; /* z->mvc = x->mvc; @@ -290,6 +282,7 @@ static void setup_mbby_copy(MACROBLOCK *mbdst, MACROBLOCK *mbsrc) for (i = 0; i < 25; i++) { z->block[i].quant = x->block[i].quant; + z->block[i].quant_shift = x->block[i].quant_shift; z->block[i].zbin = x->block[i].zbin; z->block[i].zrun_zbin_boost = x->block[i].zrun_zbin_boost; z->block[i].round = x->block[i].round; @@ -334,11 +327,6 @@ static void setup_mbby_copy(MACROBLOCK *mbdst, MACROBLOCK *mbsrc) zd->mb_segement_abs_delta = xd->mb_segement_abs_delta; vpx_memcpy(zd->segment_feature_data, xd->segment_feature_data, sizeof(xd->segment_feature_data)); - /* - memcpy(zd->above_context, xd->above_context, sizeof(xd->above_context)); - memcpy(zd->mb_segment_tree_probs, xd->mb_segment_tree_probs, sizeof(xd->mb_segment_tree_probs)); - memcpy(zd->segment_feature_data, xd->segment_feature_data, sizeof(xd->segment_feature_data)); - */ for (i = 0; i < 25; i++) { zd->block[i].dequant = xd->block[i].dequant; @@ -372,14 +360,15 @@ void vp8cx_init_mbrthread_data(VP8_COMP *cpi, #if CONFIG_RUNTIME_CPU_DETECT mbd->rtcd = xd->rtcd; #endif - mbd->gf_active_ptr = xd->gf_active_ptr; + mb->gf_active_ptr = x->gf_active_ptr; mb->vector_range = 32; vpx_memset(mbr_ei[i].segment_counts, 0, sizeof(mbr_ei[i].segment_counts)); mbr_ei[i].totalrate = 0; - mbd->mode_info = cm->mi - 1; + mb->partition_info = x->pi + x->e_mbd.mode_info_stride * (i + 1); + mbd->mode_info_context = cm->mi + x->e_mbd.mode_info_stride * (i + 1); mbd->mode_info_stride = cm->mode_info_stride; @@ -389,8 +378,8 @@ void vp8cx_init_mbrthread_data(VP8_COMP *cpi, mbd->frames_till_alt_ref_frame = cm->frames_till_alt_ref_frame; mb->src = * cpi->Source; - mbd->pre = cm->last_frame; - mbd->dst = cm->new_frame; + mbd->pre = cm->yv12_fb[cm->lst_fb_idx]; + mbd->dst = cm->yv12_fb[cm->new_fb_idx]; mb->src.y_buffer += 16 * x->src.y_stride * (i + 1); mb->src.u_buffer += 8 * x->src.uv_stride * (i + 1); @@ -406,10 +395,7 @@ void vp8cx_init_mbrthread_data(VP8_COMP *cpi, mb->rddiv = cpi->RDDIV; mb->rdmult = cpi->RDMULT; - mbd->mbmi.mode = DC_PRED; - mbd->mbmi.uv_mode = DC_PRED; - - mbd->left_context = cm->left_context; + mbd->left_context = &cm->left_context; mb->mvc = cm->fc.mvc; setup_mbby_copy(&mbr_ei[i].mb, x); diff --git a/vp8/encoder/firstpass.c b/vp8/encoder/firstpass.c index c519080..a65bce6 100644 --- a/vp8/encoder/firstpass.c +++ b/vp8/encoder/firstpass.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -77,9 +78,9 @@ int vp8_encode_intra(VP8_COMP *cpi, MACROBLOCK *x, int use_dc_pred) if (use_dc_pred) { - x->e_mbd.mbmi.mode = DC_PRED; - x->e_mbd.mbmi.uv_mode = DC_PRED; - x->e_mbd.mbmi.ref_frame = INTRA_FRAME; + x->e_mbd.mode_info_context->mbmi.mode = DC_PRED; + x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED; + x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME; vp8_encode_intra16x16mby(IF_RTCD(&cpi->rtcd), x); } @@ -175,7 +176,6 @@ static double calculate_modified_err(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) double vp8_simple_weight(YV12_BUFFER_CONFIG *source) { int i, j; - int Total = 0; unsigned char *src = source->y_buffer; unsigned char value; @@ -217,7 +217,7 @@ int frame_max_bits(VP8_COMP *cpi) // If we are running below the optimal level then we need to gradually tighten up on max_bits. if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER) { - double buffer_fullness_ratio = (double)DOUBLE_DIVIDE_CHECK(cpi->buffer_level) / (double)cpi->oxcf.optimal_buffer_level; + double buffer_fullness_ratio = (double)cpi->buffer_level / DOUBLE_DIVIDE_CHECK((double)cpi->oxcf.optimal_buffer_level); // For CBR base this on the target average bits per frame plus the maximum sedction rate passed in by the user max_bits = (int)(cpi->av_per_frame_bandwidth * ((double)cpi->oxcf.two_pass_vbrmax_section / 100.0)); @@ -536,8 +536,11 @@ void vp8_first_pass(VP8_COMP *cpi) int col_blocks = 4 * cm->mb_cols; int recon_yoffset, recon_uvoffset; - int recon_y_stride = cm->last_frame.y_stride; - int recon_uv_stride = cm->last_frame.uv_stride; + YV12_BUFFER_CONFIG *lst_yv12 = &cm->yv12_fb[cm->lst_fb_idx]; + YV12_BUFFER_CONFIG *new_yv12 = &cm->yv12_fb[cm->new_fb_idx]; + YV12_BUFFER_CONFIG *gld_yv12 = &cm->yv12_fb[cm->gld_fb_idx]; + int recon_y_stride = lst_yv12->y_stride; + int recon_uv_stride = lst_yv12->uv_stride; int intra_error = 0; int coded_error = 0; @@ -559,8 +562,12 @@ void vp8_first_pass(VP8_COMP *cpi) vp8_clear_system_state(); //__asm emms; x->src = * cpi->Source; - xd->pre = cm->last_frame; - xd->dst = cm->new_frame; + xd->pre = *lst_yv12; + xd->dst = *new_yv12; + + x->partition_info = x->pi; + + xd->mode_info_context = cm->mi; vp8_build_block_offsets(x); @@ -569,7 +576,7 @@ void vp8_first_pass(VP8_COMP *cpi) vp8_setup_block_ptrs(x); // set up frame new frame for intra coded blocks - vp8_setup_intra_recon(&cm->new_frame); + vp8_setup_intra_recon(new_yv12); vp8cx_frame_init_quantizer(cpi); // Initialise the MV cost table to the defaults @@ -599,9 +606,9 @@ void vp8_first_pass(VP8_COMP *cpi) int gf_motion_error = INT_MAX; int use_dc_pred = (mb_col || mb_row) && (!mb_col || !mb_row); - xd->dst.y_buffer = cm->new_frame.y_buffer + recon_yoffset; - xd->dst.u_buffer = cm->new_frame.u_buffer + recon_uvoffset; - xd->dst.v_buffer = cm->new_frame.v_buffer + recon_uvoffset; + xd->dst.y_buffer = new_yv12->y_buffer + recon_yoffset; + xd->dst.u_buffer = new_yv12->u_buffer + recon_uvoffset; + xd->dst.v_buffer = new_yv12->v_buffer + recon_uvoffset; xd->left_available = (mb_col != 0); // do intra 16x16 prediction @@ -635,18 +642,22 @@ void vp8_first_pass(VP8_COMP *cpi) int motion_error = INT_MAX; // Simple 0,0 motion with no mv overhead - vp8_zz_motion_search( cpi, x, &cm->last_frame, &motion_error, recon_yoffset ); + vp8_zz_motion_search( cpi, x, lst_yv12, &motion_error, recon_yoffset ); d->bmi.mv.as_mv.row = 0; d->bmi.mv.as_mv.col = 0; - // Test last reference frame using the previous best mv as the starting point (best reference) for the search - vp8_first_pass_motion_search(cpi, x, &best_ref_mv, &d->bmi.mv.as_mv, &cm->last_frame, &motion_error, recon_yoffset); + // Test last reference frame using the previous best mv as the + // starting point (best reference) for the search + vp8_first_pass_motion_search(cpi, x, &best_ref_mv, + &d->bmi.mv.as_mv, lst_yv12, + &motion_error, recon_yoffset); // If the current best reference mv is not centred on 0,0 then do a 0,0 based search as well if ((best_ref_mv.col != 0) || (best_ref_mv.row != 0)) { tmp_err = INT_MAX; - vp8_first_pass_motion_search(cpi, x, &zero_ref_mv, &tmp_mv, &cm->last_frame, &motion_error, recon_yoffset); + vp8_first_pass_motion_search(cpi, x, &zero_ref_mv, &tmp_mv, + lst_yv12, &tmp_err, recon_yoffset); if ( tmp_err < motion_error ) { @@ -660,7 +671,7 @@ void vp8_first_pass(VP8_COMP *cpi) // Experimental search in a second reference frame ((0,0) based only) if (cm->current_video_frame > 1) { - vp8_first_pass_motion_search(cpi, x, &zero_ref_mv, &tmp_mv, &cm->golden_frame, &gf_motion_error, recon_yoffset); + vp8_first_pass_motion_search(cpi, x, &zero_ref_mv, &tmp_mv, gld_yv12, &gf_motion_error, recon_yoffset); if ((gf_motion_error < motion_error) && (gf_motion_error < this_error)) { @@ -678,9 +689,9 @@ void vp8_first_pass(VP8_COMP *cpi) // Reset to last frame as reference buffer - xd->pre.y_buffer = cm->last_frame.y_buffer + recon_yoffset; - xd->pre.u_buffer = cm->last_frame.u_buffer + recon_uvoffset; - xd->pre.v_buffer = cm->last_frame.v_buffer + recon_uvoffset; + xd->pre.y_buffer = lst_yv12->y_buffer + recon_yoffset; + xd->pre.u_buffer = lst_yv12->u_buffer + recon_uvoffset; + xd->pre.v_buffer = lst_yv12->v_buffer + recon_uvoffset; } if (motion_error <= this_error) @@ -772,14 +783,13 @@ void vp8_first_pass(VP8_COMP *cpi) x->src.v_buffer += 8 * x->src.uv_stride - 8 * cm->mb_cols; //extend the recon for intra prediction - vp8_extend_mb_row(&cm->new_frame, xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8); + vp8_extend_mb_row(new_yv12, xd->dst.y_buffer + 16, xd->dst.u_buffer + 8, xd->dst.v_buffer + 8); vp8_clear_system_state(); //__asm emms; } vp8_clear_system_state(); //__asm emms; { double weight = 0.0; - double weigth2 = 0.0; FIRSTPASS_STATS fps; @@ -839,17 +849,17 @@ void vp8_first_pass(VP8_COMP *cpi) (cpi->this_frame_stats.pcnt_inter > 0.20) && ((cpi->this_frame_stats.intra_error / cpi->this_frame_stats.coded_error) > 2.0)) { - vp8_yv12_copy_frame_ptr(&cm->last_frame, &cm->golden_frame); + vp8_yv12_copy_frame_ptr(lst_yv12, gld_yv12); } // swap frame pointers so last frame refers to the frame we just compressed - vp8_swap_yv12_buffer(&cm->last_frame, &cm->new_frame); - vp8_yv12_extend_frame_borders(&cm->last_frame); + vp8_swap_yv12_buffer(lst_yv12, new_yv12); + vp8_yv12_extend_frame_borders(lst_yv12); // Special case for the first frame. Copy into the GF buffer as a second reference. if (cm->current_video_frame == 0) { - vp8_yv12_copy_frame_ptr(&cm->last_frame, &cm->golden_frame); + vp8_yv12_copy_frame_ptr(lst_yv12, gld_yv12); } @@ -865,7 +875,7 @@ void vp8_first_pass(VP8_COMP *cpi) else recon_file = fopen(filename, "ab"); - fwrite(cm->last_frame.buffer_alloc, cm->last_frame.frame_size, 1, recon_file); + fwrite(lst_yv12->buffer_alloc, lst_yv12->frame_size, 1, recon_file); fclose(recon_file); } @@ -1193,8 +1203,9 @@ static void define_gf_group(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) FIRSTPASS_STATS next_frame; FIRSTPASS_STATS *start_pos; int i; - int count = 0; - int image_size = cpi->common.last_frame.y_width * cpi->common.last_frame.y_height; + int y_width = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_width; + int y_height = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_height; + int image_size = y_width * y_height; double boost_score = 0.0; double old_boost_score = 0.0; double gf_group_err = 0.0; @@ -1234,6 +1245,8 @@ static void define_gf_group(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) start_pos = cpi->stats_in; + vpx_memset(&next_frame, 0, sizeof(next_frame)); // assure clean + // Preload the stats for the next frame. mod_frame_err = calculate_modified_err(cpi, this_frame); @@ -1390,6 +1403,7 @@ static void define_gf_group(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) // Should we use the alternate refernce frame if (cpi->oxcf.play_alternate && + cpi->oxcf.lag_in_frames && (i >= MIN_GF_INTERVAL) && (i <= (cpi->frames_to_key - MIN_GF_INTERVAL)) && // dont use ARF very near next kf (((next_frame.pcnt_inter > 0.75) && @@ -1568,26 +1582,36 @@ static void define_gf_group(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) // Calculate the number of bits to be spent on the gf or arf based on the boost number cpi->gf_bits = (int)((double)Boost * (cpi->gf_group_bits / (double)allocation_chunks)); - // If the frame that is to be boosted is simpler than the average for the gf/arf group then use an alternative calculation + // If the frame that is to be boosted is simpler than the average for + // the gf/arf group then use an alternative calculation // based on the error score of the frame itself if (mod_frame_err < gf_group_err / (double)cpi->baseline_gf_interval) { double alt_gf_grp_bits; int alt_gf_bits; - alt_gf_grp_bits = ((double)cpi->kf_group_bits * (mod_frame_err * (double)cpi->baseline_gf_interval) / (double)cpi->kf_group_error_left) ; - alt_gf_bits = (int)((double)Boost * (alt_gf_grp_bits / (double)allocation_chunks)); + alt_gf_grp_bits = + (double)cpi->kf_group_bits * + (mod_frame_err * (double)cpi->baseline_gf_interval) / + DOUBLE_DIVIDE_CHECK((double)cpi->kf_group_error_left); + + alt_gf_bits = (int)((double)Boost * (alt_gf_grp_bits / + (double)allocation_chunks)); if (cpi->gf_bits > alt_gf_bits) { cpi->gf_bits = alt_gf_bits; } } - // Else if it is harder than other frames in the group make sure it at least receives an allocation in keeping with - // its relative error score, otherwise it may be worse off than an "un-boosted" frame + // Else if it is harder than other frames in the group make sure it at + // least receives an allocation in keeping with its relative error + // score, otherwise it may be worse off than an "un-boosted" frame else { - int alt_gf_bits = (int)((double)cpi->kf_group_bits * (mod_frame_err / (double)cpi->kf_group_error_left)); + int alt_gf_bits = + (int)((double)cpi->kf_group_bits * + mod_frame_err / + DOUBLE_DIVIDE_CHECK((double)cpi->kf_group_error_left)); if (alt_gf_bits > cpi->gf_bits) { @@ -1871,6 +1895,18 @@ void vp8_second_pass(VP8_COMP *cpi) } } + // Keep a globally available copy of this and the next frame's iiratio. + cpi->this_iiratio = this_frame_intra_error / + DOUBLE_DIVIDE_CHECK(this_frame_coded_error); + { + FIRSTPASS_STATS next_frame; + if ( lookup_next_frame_stats(cpi, &next_frame) != EOF ) + { + cpi->next_iiratio = next_frame.intra_error / + DOUBLE_DIVIDE_CHECK(next_frame.coded_error); + } + } + // Set nominal per second bandwidth for this frame cpi->target_bandwidth = cpi->per_frame_bandwidth * cpi->output_frame_rate; if (cpi->target_bandwidth < 0) @@ -2028,6 +2064,8 @@ void vp8_find_next_key_frame(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) double kf_group_coded_err = 0.0; double two_pass_min_rate = (double)(cpi->oxcf.target_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100); + vpx_memset(&next_frame, 0, sizeof(next_frame)); // assure clean + vp8_clear_system_state(); //__asm emms; start_position = cpi->stats_in; @@ -2044,7 +2082,7 @@ void vp8_find_next_key_frame(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) // Take a copy of the initial frame details vpx_memcpy(&first_frame, this_frame, sizeof(*this_frame)); - cpi->kf_group_bits = 0; // Estimate of total bits avaialable to kf group + cpi->kf_group_bits = 0; // Total bits avaialable to kf group cpi->kf_group_error_left = 0; // Group modified error score. kf_mod_err = calculate_modified_err(cpi, this_frame); @@ -2060,33 +2098,34 @@ void vp8_find_next_key_frame(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) kf_group_intra_err += this_frame->intra_error; kf_group_coded_err += this_frame->coded_error; + // load a the next frame's stats vpx_memcpy(&last_frame, this_frame, sizeof(*this_frame)); + vp8_input_stats(cpi, this_frame); // Provided that we are not at the end of the file... - if (EOF != vp8_input_stats(cpi, this_frame)) + if (cpi->oxcf.auto_key + && lookup_next_frame_stats(cpi, &next_frame) != EOF) { - if (lookup_next_frame_stats(cpi, &next_frame) != EOF) - { - if (test_candidate_kf(cpi, &last_frame, this_frame, &next_frame)) - break; - } - } - - // Step on to the next frame - cpi->frames_to_key ++; + if (test_candidate_kf(cpi, &last_frame, this_frame, &next_frame)) + break; - // If we don't have a real key frame within the next two - // forcekeyframeevery intervals then break out of the loop. - if (cpi->frames_to_key >= 2 *(int)cpi->key_frame_frequency) - break; + // Step on to the next frame + cpi->frames_to_key ++; + // If we don't have a real key frame within the next two + // forcekeyframeevery intervals then break out of the loop. + if (cpi->frames_to_key >= 2 *(int)cpi->key_frame_frequency) + break; + } else + cpi->frames_to_key ++; } // If there is a max kf interval set by the user we must obey it. // We already breakout of the loop above at 2x max. // This code centers the extra kf if the actual natural // interval is between 1x and 2x - if ( cpi->frames_to_key > (int)cpi->key_frame_frequency ) + if (cpi->oxcf.auto_key + && cpi->frames_to_key > (int)cpi->key_frame_frequency ) { cpi->frames_to_key /= 2; @@ -2111,39 +2150,64 @@ void vp8_find_next_key_frame(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) // Calculate the number of bits that should be assigned to the kf group. if ((cpi->bits_left > 0) && ((int)cpi->modified_total_error_left > 0)) { - int max_bits = frame_max_bits(cpi); // Max for a single normal frame (not key frame) + // Max for a single normal frame (not key frame) + int max_bits = frame_max_bits(cpi); + + // Maximum bits for the kf group + long long max_grp_bits; - // Default allocation based on bits left and relative complexity of the section - cpi->kf_group_bits = (int)(cpi->bits_left * (kf_group_err / cpi->modified_total_error_left)); + // Default allocation based on bits left and relative + // complexity of the section + cpi->kf_group_bits = (long long)( cpi->bits_left * + ( kf_group_err / + cpi->modified_total_error_left )); // Clip based on maximum per frame rate defined by the user. - if (cpi->kf_group_bits > max_bits * cpi->frames_to_key) - cpi->kf_group_bits = max_bits * cpi->frames_to_key; + max_grp_bits = (long long)max_bits * (long long)cpi->frames_to_key; + if (cpi->kf_group_bits > max_grp_bits) + cpi->kf_group_bits = max_grp_bits; // Additional special case for CBR if buffer is getting full. if (cpi->oxcf.end_usage == USAGE_STREAM_FROM_SERVER) { - // If the buffer is near or above the optimal and this kf group is not being allocated much - // then increase the allocation a bit. - if (cpi->buffer_level >= cpi->oxcf.optimal_buffer_level) + int opt_buffer_lvl = cpi->oxcf.optimal_buffer_level; + int buffer_lvl = cpi->buffer_level; + + // If the buffer is near or above the optimal and this kf group is + // not being allocated much then increase the allocation a bit. + if (buffer_lvl >= opt_buffer_lvl) { - int high_water_mark = (cpi->oxcf.optimal_buffer_level + cpi->oxcf.maximum_buffer_size) >> 1; - int min_group_bits; + int high_water_mark = (opt_buffer_lvl + + cpi->oxcf.maximum_buffer_size) >> 1; + + long long av_group_bits; + + // Av bits per frame * number of frames + av_group_bits = (long long)cpi->av_per_frame_bandwidth * + (long long)cpi->frames_to_key; // We are at or above the maximum. if (cpi->buffer_level >= high_water_mark) { - min_group_bits = (cpi->av_per_frame_bandwidth * cpi->frames_to_key) + (cpi->buffer_level - high_water_mark); + long long min_group_bits; + + min_group_bits = av_group_bits + + (long long)(buffer_lvl - + high_water_mark); if (cpi->kf_group_bits < min_group_bits) cpi->kf_group_bits = min_group_bits; } // We are above optimal but below the maximum - else if (cpi->kf_group_bits < (cpi->av_per_frame_bandwidth * cpi->frames_to_key)) + else if (cpi->kf_group_bits < av_group_bits) { - int bits_below_av = (cpi->av_per_frame_bandwidth * cpi->frames_to_key) - cpi->kf_group_bits; - cpi->kf_group_bits += (int)((double)bits_below_av * (double)(cpi->buffer_level - cpi->oxcf.optimal_buffer_level) / - (double)(high_water_mark - cpi->oxcf.optimal_buffer_level)); + long long bits_below_av = av_group_bits - + cpi->kf_group_bits; + + cpi->kf_group_bits += + (long long)((double)bits_below_av * + (double)(buffer_lvl - opt_buffer_lvl) / + (double)(high_water_mark - opt_buffer_lvl)); } } } @@ -2269,7 +2333,7 @@ void vp8_find_next_key_frame(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) int allocation_chunks; int Counter = cpi->frames_to_key; int alt_kf_bits; - + YV12_BUFFER_CONFIG *lst_yv12 = &cpi->common.yv12_fb[cpi->common.lst_fb_idx]; // Min boost based on kf interval #if 0 @@ -2289,10 +2353,10 @@ void vp8_find_next_key_frame(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) } // bigger frame sizes need larger kf boosts, smaller frames smaller boosts... - if ((cpi->common.last_frame.y_width * cpi->common.last_frame.y_height) > (320 * 240)) - kf_boost += 2 * (cpi->common.last_frame.y_width * cpi->common.last_frame.y_height) / (320 * 240); - else if ((cpi->common.last_frame.y_width * cpi->common.last_frame.y_height) < (320 * 240)) - kf_boost -= 4 * (320 * 240) / (cpi->common.last_frame.y_width * cpi->common.last_frame.y_height); + if ((lst_yv12->y_width * lst_yv12->y_height) > (320 * 240)) + kf_boost += 2 * (lst_yv12->y_width * lst_yv12->y_height) / (320 * 240); + else if ((lst_yv12->y_width * lst_yv12->y_height) < (320 * 240)) + kf_boost -= 4 * (320 * 240) / (lst_yv12->y_width * lst_yv12->y_height); kf_boost = (int)((double)kf_boost * 100.0) >> 4; // Scale 16 to 100 @@ -2328,23 +2392,34 @@ void vp8_find_next_key_frame(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame) cpi->kf_bits = (3 * cpi->buffer_level) >> 2; } - // If the key frame is actually easier than the average for the kf group (which does sometimes happen... eg a blank intro frame) - // Then use an alternate calculation based on the kf error score which should give a smaller key frame. + // If the key frame is actually easier than the average for the + // kf group (which does sometimes happen... eg a blank intro frame) + // Then use an alternate calculation based on the kf error score + // which should give a smaller key frame. if (kf_mod_err < kf_group_err / cpi->frames_to_key) { - double alt_kf_grp_bits = ((double)cpi->bits_left * (kf_mod_err * (double)cpi->frames_to_key) / cpi->modified_total_error_left) ; + double alt_kf_grp_bits = + ((double)cpi->bits_left * + (kf_mod_err * (double)cpi->frames_to_key) / + DOUBLE_DIVIDE_CHECK(cpi->modified_total_error_left)); - alt_kf_bits = (int)((double)kf_boost * (alt_kf_grp_bits / (double)allocation_chunks)); + alt_kf_bits = (int)((double)kf_boost * + (alt_kf_grp_bits / (double)allocation_chunks)); if (cpi->kf_bits > alt_kf_bits) { cpi->kf_bits = alt_kf_bits; } } - // Else if it is much harder than other frames in the group make sure it at least receives an allocation in keeping with its relative error score + // Else if it is much harder than other frames in the group make sure + // it at least receives an allocation in keeping with its relative + // error score else { - alt_kf_bits = (int)((double)cpi->bits_left * (kf_mod_err / cpi->modified_total_error_left)); + alt_kf_bits = + (int)((double)cpi->bits_left * + (kf_mod_err / + DOUBLE_DIVIDE_CHECK(cpi->modified_total_error_left))); if (alt_kf_bits > cpi->kf_bits) { diff --git a/vp8/encoder/firstpass.h b/vp8/encoder/firstpass.h index d7b52f3..c7f3e0e 100644 --- a/vp8/encoder/firstpass.h +++ b/vp8/encoder/firstpass.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/generic/csystemdependent.c b/vp8/encoder/generic/csystemdependent.c index 52aab66..1acb73d 100644 --- a/vp8/encoder/generic/csystemdependent.c +++ b/vp8/encoder/generic/csystemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -67,8 +68,8 @@ void vp8_cmachine_specific_config(VP8_COMP *cpi) cpi->rtcd.fdct.short4x4 = vp8_short_fdct4x4_c; cpi->rtcd.fdct.short8x4 = vp8_short_fdct8x4_c; - cpi->rtcd.fdct.fast4x4 = vp8_fast_fdct4x4_c; - cpi->rtcd.fdct.fast8x4 = vp8_fast_fdct8x4_c; + cpi->rtcd.fdct.fast4x4 = vp8_short_fdct4x4_c; + cpi->rtcd.fdct.fast8x4 = vp8_short_fdct8x4_c; cpi->rtcd.fdct.walsh_short4x4 = vp8_short_walsh4x4_c; cpi->rtcd.encodemb.berr = vp8_block_error_c; diff --git a/vp8/encoder/mcomp.c b/vp8/encoder/mcomp.c index d80059d..b89354e 100644 --- a/vp8/encoder/mcomp.c +++ b/vp8/encoder/mcomp.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -996,7 +997,7 @@ int vp8_diamond_search_sadx4 int tot_steps; MV this_mv; - unsigned int bestsad = UINT_MAX; + int bestsad = INT_MAX; int best_site = 0; int last_site = 0; @@ -1137,7 +1138,6 @@ int vp8_diamond_search_sadx4 } -#if !(CONFIG_REALTIME_ONLY) int vp8_full_search_sad(MACROBLOCK *x, BLOCK *b, BLOCKD *d, MV *ref_mv, int error_per_bit, int distance, vp8_variance_fn_ptr_t *fn_ptr, int *mvcost[2], int *mvsadcost[2]) { unsigned char *what = (*(b->base_src) + b->src); @@ -1238,7 +1238,7 @@ int vp8_full_search_sadx3(MACROBLOCK *x, BLOCK *b, BLOCKD *d, MV *ref_mv, int er unsigned char *bestaddress; MV *best_mv = &d->bmi.mv.as_mv; MV this_mv; - unsigned int bestsad = UINT_MAX; + int bestsad = INT_MAX; int r, c; unsigned char *check_here; @@ -1350,7 +1350,6 @@ int vp8_full_search_sadx3(MACROBLOCK *x, BLOCK *b, BLOCKD *d, MV *ref_mv, int er else return INT_MAX; } -#endif #ifdef ENTROPY_STATS void print_mode_context(void) diff --git a/vp8/encoder/mcomp.h b/vp8/encoder/mcomp.h index 921206f..7cc9242 100644 --- a/vp8/encoder/mcomp.h +++ b/vp8/encoder/mcomp.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/modecosts.c b/vp8/encoder/modecosts.c index 73170cf..d23c97e 100644 --- a/vp8/encoder/modecosts.c +++ b/vp8/encoder/modecosts.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/modecosts.h b/vp8/encoder/modecosts.h index 5ade265..99ef119 100644 --- a/vp8/encoder/modecosts.h +++ b/vp8/encoder/modecosts.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/onyx_if.c b/vp8/encoder/onyx_if.c index 7662720..56bac0a 100644 --- a/vp8/encoder/onyx_if.c +++ b/vp8/encoder/onyx_if.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -20,7 +21,7 @@ #include "extend.h" #include "ratectrl.h" #include "quant_common.h" -#include "segmentation_common.h" +#include "segmentation.h" #include "g_common.h" #include "vpx_scale/yv12extend.h" #include "postproc.h" @@ -28,6 +29,8 @@ #include "swapyv12buffer.h" #include "threading.h" #include "vpx_ports/vpx_timer.h" +#include "vpxerrors.h" + #include <math.h> #include <stdio.h> #include <limits.h> @@ -136,8 +139,6 @@ extern unsigned int inter_b_modes[15]; extern void (*vp8_short_fdct4x4)(short *input, short *output, int pitch); extern void (*vp8_short_fdct8x4)(short *input, short *output, int pitch); -extern void (*vp8_fast_fdct4x4)(short *input, short *output, int pitch); -extern void (*vp8_fast_fdct8x4)(short *input, short *output, int pitch); extern const int vp8_bits_per_mb[2][QINDEX_RANGE]; @@ -225,6 +226,17 @@ void vp8_dealloc_compressor_data(VP8_COMP *cpi) vpx_free(cpi->tok); cpi->tok = 0; + // Structure used to minitor GF useage + if (cpi->gf_active_flags != 0) + vpx_free(cpi->gf_active_flags); + + cpi->gf_active_flags = 0; + + if(cpi->mb.pip) + vpx_free(cpi->mb.pip); + + cpi->mb.pip = 0; + } static void enable_segmentation(VP8_PTR ptr) @@ -281,7 +293,6 @@ static void segmentation_test_function(VP8_PTR ptr) unsigned char *seg_map; signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS]; - int i, j; // Create a temporary map for segmentation data. CHECK_MEM_ERROR(seg_map, vpx_calloc(cpi->common.mb_rows * cpi->common.mb_cols, 1)); @@ -545,7 +556,8 @@ void vp8_set_speed_features(VP8_COMP *cpi) sf->thresh_mult[THR_NEWG ] = INT_MAX; sf->thresh_mult[THR_SPLITG ] = INT_MAX; } - else if (!(cpi->ref_frame_flags & VP8_ALT_FLAG)) + + if (!(cpi->ref_frame_flags & VP8_ALT_FLAG)) { sf->thresh_mult[THR_NEARESTA ] = INT_MAX; sf->thresh_mult[THR_ZEROA ] = INT_MAX; @@ -597,7 +609,8 @@ void vp8_set_speed_features(VP8_COMP *cpi) sf->thresh_mult[THR_NEARMV ] = INT_MAX; sf->thresh_mult[THR_SPLITMV ] = INT_MAX; } - else if (!(cpi->ref_frame_flags & VP8_GOLD_FLAG)) + + if (!(cpi->ref_frame_flags & VP8_GOLD_FLAG)) { sf->thresh_mult[THR_NEARESTG ] = INT_MAX; sf->thresh_mult[THR_ZEROG ] = INT_MAX; @@ -605,7 +618,8 @@ void vp8_set_speed_features(VP8_COMP *cpi) sf->thresh_mult[THR_NEWG ] = INT_MAX; sf->thresh_mult[THR_SPLITG ] = INT_MAX; } - else if (!(cpi->ref_frame_flags & VP8_ALT_FLAG)) + + if (!(cpi->ref_frame_flags & VP8_ALT_FLAG)) { sf->thresh_mult[THR_NEARESTA ] = INT_MAX; sf->thresh_mult[THR_ZEROA ] = INT_MAX; @@ -763,7 +777,7 @@ void vp8_set_speed_features(VP8_COMP *cpi) cpi->mode_check_freq[THR_NEWA] = 4; } - if (cpi->ref_frame_flags & VP8_LAST_FLAG & VP8_GOLD_FLAG) + if (cpi->ref_frame_flags & VP8_GOLD_FLAG) { sf->thresh_mult[THR_NEARESTG ] = 2000; sf->thresh_mult[THR_ZEROG ] = 2000; @@ -771,7 +785,7 @@ void vp8_set_speed_features(VP8_COMP *cpi) sf->thresh_mult[THR_NEWG ] = 4000; } - if (cpi->ref_frame_flags & VP8_LAST_FLAG & VP8_ALT_FLAG) + if (cpi->ref_frame_flags & VP8_ALT_FLAG) { sf->thresh_mult[THR_NEARESTA ] = 2000; sf->thresh_mult[THR_ZEROA ] = 2000; @@ -811,7 +825,7 @@ void vp8_set_speed_features(VP8_COMP *cpi) sf->full_freq[1] = 31; sf->search_method = NSTEP; - if (!cpi->ref_frame_flags & VP8_LAST_FLAG) + if (!(cpi->ref_frame_flags & VP8_LAST_FLAG)) { sf->thresh_mult[THR_NEWMV ] = INT_MAX; sf->thresh_mult[THR_NEARESTMV] = INT_MAX; @@ -820,7 +834,7 @@ void vp8_set_speed_features(VP8_COMP *cpi) sf->thresh_mult[THR_SPLITMV ] = INT_MAX; } - if (!cpi->ref_frame_flags & VP8_GOLD_FLAG) + if (!(cpi->ref_frame_flags & VP8_GOLD_FLAG)) { sf->thresh_mult[THR_NEARESTG ] = INT_MAX; sf->thresh_mult[THR_ZEROG ] = INT_MAX; @@ -829,7 +843,7 @@ void vp8_set_speed_features(VP8_COMP *cpi) sf->thresh_mult[THR_SPLITG ] = INT_MAX; } - if (!cpi->ref_frame_flags & VP8_ALT_FLAG) + if (!(cpi->ref_frame_flags & VP8_ALT_FLAG)) { sf->thresh_mult[THR_NEARESTA ] = INT_MAX; sf->thresh_mult[THR_ZEROA ] = INT_MAX; @@ -1122,40 +1136,33 @@ void vp8_set_speed_features(VP8_COMP *cpi) if (cpi->sf.search_method == NSTEP) { - vp8_init3smotion_compensation(&cpi->mb, cm->last_frame.y_stride); + vp8_init3smotion_compensation(&cpi->mb, cm->yv12_fb[cm->lst_fb_idx].y_stride); } else if (cpi->sf.search_method == DIAMOND) { - vp8_init_dsmotion_compensation(&cpi->mb, cm->last_frame.y_stride); + vp8_init_dsmotion_compensation(&cpi->mb, cm->yv12_fb[cm->lst_fb_idx].y_stride); } if (cpi->sf.improved_dct) { cpi->mb.vp8_short_fdct8x4 = FDCT_INVOKE(&cpi->rtcd.fdct, short8x4); cpi->mb.vp8_short_fdct4x4 = FDCT_INVOKE(&cpi->rtcd.fdct, short4x4); - cpi->mb.short_fdct8x4rd = FDCT_INVOKE(&cpi->rtcd.fdct, short8x4); - cpi->mb.short_fdct4x4rd = FDCT_INVOKE(&cpi->rtcd.fdct, short4x4); } else { cpi->mb.vp8_short_fdct8x4 = FDCT_INVOKE(&cpi->rtcd.fdct, fast8x4); cpi->mb.vp8_short_fdct4x4 = FDCT_INVOKE(&cpi->rtcd.fdct, fast4x4); - cpi->mb.short_fdct8x4rd = FDCT_INVOKE(&cpi->rtcd.fdct, fast8x4); - cpi->mb.short_fdct4x4rd = FDCT_INVOKE(&cpi->rtcd.fdct, fast4x4); } - cpi->mb.vp8_short_fdct4x4_ptr = FDCT_INVOKE(&cpi->rtcd.fdct, short4x4); cpi->mb.short_walsh4x4 = FDCT_INVOKE(&cpi->rtcd.fdct, walsh_short4x4); if (cpi->sf.improved_quant) { cpi->mb.quantize_b = QUANTIZE_INVOKE(&cpi->rtcd.quantize, quantb); - cpi->mb.quantize_brd = QUANTIZE_INVOKE(&cpi->rtcd.quantize, quantb); } else { cpi->mb.quantize_b = QUANTIZE_INVOKE(&cpi->rtcd.quantize, fastquantb); - cpi->mb.quantize_brd = QUANTIZE_INVOKE(&cpi->rtcd.quantize, fastquantb); } #if CONFIG_RUNTIME_CPU_DETECT @@ -1221,6 +1228,20 @@ static void alloc_raw_frame_buffers(VP8_COMP *cpi) cpi->source_buffer_count = 0; } + +static int vp8_alloc_partition_data(VP8_COMP *cpi) +{ + cpi->mb.pip = vpx_calloc((cpi->common.mb_cols + 1) * + (cpi->common.mb_rows + 1), + sizeof(PARTITION_INFO)); + if(!cpi->mb.pip) + return ALLOC_FAILURE; + + cpi->mb.pi = cpi->mb.pip + cpi->common.mode_info_stride + 1; + + return 0; +} + void vp8_alloc_compressor_data(VP8_COMP *cpi) { VP8_COMMON *cm = & cpi->common; @@ -1232,6 +1253,11 @@ void vp8_alloc_compressor_data(VP8_COMP *cpi) vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR, "Failed to allocate frame buffers"); + if (vp8_alloc_partition_data(cpi)) + vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR, + "Failed to allocate partition data"); + + if ((width & 0xf) != 0) width += 16 - (width & 0xf); @@ -1262,6 +1288,15 @@ void vp8_alloc_compressor_data(VP8_COMP *cpi) cpi->inter_zz_count = 0; cpi->gf_bad_count = 0; cpi->gf_update_recommended = 0; + + + // Structures used to minitor GF usage + if (cpi->gf_active_flags != 0) + vpx_free(cpi->gf_active_flags); + + CHECK_MEM_ERROR(cpi->gf_active_flags, vpx_calloc(1, cm->mb_rows * cm->mb_cols)); + + cpi->gf_active_count = cm->mb_rows * cm->mb_cols; } @@ -1309,14 +1344,26 @@ void vp8_new_frame_rate(VP8_COMP *cpi, double framerate) cpi->max_gf_interval = 12; - // Special conditions when altr ref frame enabled - if (cpi->oxcf.play_alternate) + // Special conditions when altr ref frame enabled in lagged compress mode + if (cpi->oxcf.play_alternate && cpi->oxcf.lag_in_frames) { if (cpi->max_gf_interval > cpi->oxcf.lag_in_frames - 1) cpi->max_gf_interval = cpi->oxcf.lag_in_frames - 1; } } + +static int +rescale(int val, int num, int denom) +{ + int64_t llnum = num; + int64_t llden = denom; + int64_t llval = val; + + return llval * llnum / llden; +} + + void vp8_init_config(VP8_PTR ptr, VP8_CONFIG *oxcf) { VP8_COMP *cpi = (VP8_COMP *)(ptr); @@ -1344,9 +1391,9 @@ void vp8_init_config(VP8_PTR ptr, VP8_CONFIG *oxcf) cpi->oxcf.worst_allowed_q = MAXQ; cpi->oxcf.end_usage = USAGE_STREAM_FROM_SERVER; - cpi->oxcf.starting_buffer_level = 4; - cpi->oxcf.optimal_buffer_level = 5; - cpi->oxcf.maximum_buffer_size = 6; + cpi->oxcf.starting_buffer_level = 4000; + cpi->oxcf.optimal_buffer_level = 5000; + cpi->oxcf.maximum_buffer_size = 6000; cpi->oxcf.under_shoot_pct = 90; cpi->oxcf.allow_df = 0; cpi->oxcf.drop_frames_water_mark = 20; @@ -1495,26 +1542,32 @@ void vp8_init_config(VP8_PTR ptr, VP8_CONFIG *oxcf) // local file playback mode == really big buffer if (cpi->oxcf.end_usage == USAGE_LOCAL_FILE_PLAYBACK) { - cpi->oxcf.starting_buffer_level = 60; - cpi->oxcf.optimal_buffer_level = 60; - cpi->oxcf.maximum_buffer_size = 240; + cpi->oxcf.starting_buffer_level = 60000; + cpi->oxcf.optimal_buffer_level = 60000; + cpi->oxcf.maximum_buffer_size = 240000; } // Convert target bandwidth from Kbit/s to Bit/s cpi->oxcf.target_bandwidth *= 1000; - cpi->oxcf.starting_buffer_level *= cpi->oxcf.target_bandwidth; + cpi->oxcf.starting_buffer_level = + rescale(cpi->oxcf.starting_buffer_level, + cpi->oxcf.target_bandwidth, 1000); if (cpi->oxcf.optimal_buffer_level == 0) cpi->oxcf.optimal_buffer_level = cpi->oxcf.target_bandwidth / 8; else - cpi->oxcf.optimal_buffer_level *= cpi->oxcf.target_bandwidth; + cpi->oxcf.optimal_buffer_level = + rescale(cpi->oxcf.optimal_buffer_level, + cpi->oxcf.target_bandwidth, 1000); if (cpi->oxcf.maximum_buffer_size == 0) cpi->oxcf.maximum_buffer_size = cpi->oxcf.target_bandwidth / 8; else - cpi->oxcf.maximum_buffer_size *= cpi->oxcf.target_bandwidth; + cpi->oxcf.maximum_buffer_size = + rescale(cpi->oxcf.maximum_buffer_size, + cpi->oxcf.target_bandwidth, 1000); cpi->buffer_level = cpi->oxcf.starting_buffer_level; cpi->bits_off_target = cpi->oxcf.starting_buffer_level; @@ -1570,9 +1623,9 @@ void vp8_init_config(VP8_PTR ptr, VP8_CONFIG *oxcf) cm->Height = (vs - 1 + cpi->oxcf.Height * vr) / vs; } - if (((cm->Width + 15) & 0xfffffff0) != cm->last_frame.y_width || - ((cm->Height + 15) & 0xfffffff0) != cm->last_frame.y_height || - cm->last_frame.y_width == 0) + if (((cm->Width + 15) & 0xfffffff0) != cm->yv12_fb[cm->lst_fb_idx].y_width || + ((cm->Height + 15) & 0xfffffff0) != cm->yv12_fb[cm->lst_fb_idx].y_height || + cm->yv12_fb[cm->lst_fb_idx].y_width == 0) { alloc_raw_frame_buffers(cpi); vp8_alloc_compressor_data(cpi); @@ -1599,13 +1652,6 @@ void vp8_init_config(VP8_PTR ptr, VP8_CONFIG *oxcf) else if (cpi->oxcf.lag_in_frames > MAX_LAG_BUFFERS) cpi->oxcf.lag_in_frames = MAX_LAG_BUFFERS; - // force play_alternate to 0 if allow_lag is 0, lag_in_frames is too small, Mode is real time or one pass compress enabled. - if (cpi->oxcf.allow_lag == 0 || cpi->oxcf.lag_in_frames <= 5 || (cpi->oxcf.Mode < MODE_SECONDPASS)) - { - cpi->oxcf.play_alternate = 0; - cpi->ref_frame_flags = cpi->ref_frame_flags & ~VP8_ALT_FLAG; - } - // YX Temp cpi->last_alt_ref_sei = -1; cpi->is_src_frame_alt_ref = 0; @@ -1774,26 +1820,32 @@ void vp8_change_config(VP8_PTR ptr, VP8_CONFIG *oxcf) // local file playback mode == really big buffer if (cpi->oxcf.end_usage == USAGE_LOCAL_FILE_PLAYBACK) { - cpi->oxcf.starting_buffer_level = 60; - cpi->oxcf.optimal_buffer_level = 60; - cpi->oxcf.maximum_buffer_size = 240; + cpi->oxcf.starting_buffer_level = 60000; + cpi->oxcf.optimal_buffer_level = 60000; + cpi->oxcf.maximum_buffer_size = 240000; } // Convert target bandwidth from Kbit/s to Bit/s cpi->oxcf.target_bandwidth *= 1000; - cpi->oxcf.starting_buffer_level *= cpi->oxcf.target_bandwidth; + cpi->oxcf.starting_buffer_level = + rescale(cpi->oxcf.starting_buffer_level, + cpi->oxcf.target_bandwidth, 1000); if (cpi->oxcf.optimal_buffer_level == 0) cpi->oxcf.optimal_buffer_level = cpi->oxcf.target_bandwidth / 8; else - cpi->oxcf.optimal_buffer_level *= cpi->oxcf.target_bandwidth; + cpi->oxcf.optimal_buffer_level = + rescale(cpi->oxcf.optimal_buffer_level, + cpi->oxcf.target_bandwidth, 1000); if (cpi->oxcf.maximum_buffer_size == 0) cpi->oxcf.maximum_buffer_size = cpi->oxcf.target_bandwidth / 8; else - cpi->oxcf.maximum_buffer_size *= cpi->oxcf.target_bandwidth; + cpi->oxcf.maximum_buffer_size = + rescale(cpi->oxcf.maximum_buffer_size, + cpi->oxcf.target_bandwidth, 1000); cpi->buffer_level = cpi->oxcf.starting_buffer_level; cpi->bits_off_target = cpi->oxcf.starting_buffer_level; @@ -1849,9 +1901,9 @@ void vp8_change_config(VP8_PTR ptr, VP8_CONFIG *oxcf) cm->Height = (vs - 1 + cpi->oxcf.Height * vr) / vs; } - if (((cm->Width + 15) & 0xfffffff0) != cm->last_frame.y_width || - ((cm->Height + 15) & 0xfffffff0) != cm->last_frame.y_height || - cm->last_frame.y_width == 0) + if (((cm->Width + 15) & 0xfffffff0) != cm->yv12_fb[cm->lst_fb_idx].y_width || + ((cm->Height + 15) & 0xfffffff0) != cm->yv12_fb[cm->lst_fb_idx].y_height || + cm->yv12_fb[cm->lst_fb_idx].y_width == 0) { alloc_raw_frame_buffers(cpi); vp8_alloc_compressor_data(cpi); @@ -1878,13 +1930,6 @@ void vp8_change_config(VP8_PTR ptr, VP8_CONFIG *oxcf) else if (cpi->oxcf.lag_in_frames > MAX_LAG_BUFFERS) cpi->oxcf.lag_in_frames = MAX_LAG_BUFFERS; - // force play_alternate to 0 if allow_lag is 0, lag_in_frames is too small, Mode is real time or one pass compress enabled. - if (cpi->oxcf.allow_lag == 0 || cpi->oxcf.lag_in_frames <= 5 || (cpi->oxcf.Mode < MODE_SECONDPASS)) - { - cpi->oxcf.play_alternate = 0; - cpi->ref_frame_flags = cpi->ref_frame_flags & ~VP8_ALT_FLAG; - } - // YX Temp cpi->last_alt_ref_sei = -1; cpi->is_src_frame_alt_ref = 0; @@ -1925,7 +1970,7 @@ VP8_PTR vp8_create_compressor(VP8_CONFIG *oxcf) VP8_COMP *cpi; VP8_PTR ptr; } ctx; - + VP8_COMP *cpi; VP8_COMMON *cm; @@ -2247,12 +2292,13 @@ void vp8_remove_compressor(VP8_PTR *ptr) if (cpi->b_calculate_psnr) { - double samples = 3.0 / 2 * cpi->count * cpi->common.last_frame.y_width * cpi->common.last_frame.y_height; + YV12_BUFFER_CONFIG *lst_yv12 = &cpi->common.yv12_fb[cpi->common.lst_fb_idx]; + double samples = 3.0 / 2 * cpi->count * lst_yv12->y_width * lst_yv12->y_height; double total_psnr = vp8_mse2psnr(samples, 255.0, cpi->total_sq_error); double total_psnr2 = vp8_mse2psnr(samples, 255.0, cpi->total_sq_error2); double total_ssim = 100 * pow(cpi->summed_quality / cpi->summed_weights, 8.0); - fprintf(f, "Bitrate\AVGPsnr\tGLBPsnr\tAVPsnrP\tGLPsnrP\tVPXSSIM\t Time(us)\n"); + fprintf(f, "Bitrate\tAVGPsnr\tGLBPsnr\tAVPsnrP\tGLPsnrP\tVPXSSIM\t Time(us)\n"); fprintf(f, "%7.3f\t%7.3f\t%7.3f\t%7.3f\t%7.3f\t%7.3f %8.0f\n", dr, cpi->total / cpi->count, total_psnr, cpi->totalp / cpi->count, total_psnr2, total_ssim, total_encode_time); @@ -2586,19 +2632,19 @@ int vp8_get_reference(VP8_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONF { VP8_COMP *cpi = (VP8_COMP *)(ptr); VP8_COMMON *cm = &cpi->common; + int ref_fb_idx; if (ref_frame_flag == VP8_LAST_FLAG) - vp8_yv12_copy_frame_ptr(&cm->last_frame, sd); - + ref_fb_idx = cm->lst_fb_idx; else if (ref_frame_flag == VP8_GOLD_FLAG) - vp8_yv12_copy_frame_ptr(&cm->golden_frame, sd); - + ref_fb_idx = cm->gld_fb_idx; else if (ref_frame_flag == VP8_ALT_FLAG) - vp8_yv12_copy_frame_ptr(&cm->alt_ref_frame, sd); - + ref_fb_idx = cm->alt_fb_idx; else return -1; + vp8_yv12_copy_frame_ptr(&cm->yv12_fb[ref_fb_idx], sd); + return 0; } int vp8_set_reference(VP8_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd) @@ -2606,18 +2652,19 @@ int vp8_set_reference(VP8_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONF VP8_COMP *cpi = (VP8_COMP *)(ptr); VP8_COMMON *cm = &cpi->common; - if (ref_frame_flag == VP8_LAST_FLAG) - vp8_yv12_copy_frame_ptr(sd, &cm->last_frame); + int ref_fb_idx; + if (ref_frame_flag == VP8_LAST_FLAG) + ref_fb_idx = cm->lst_fb_idx; else if (ref_frame_flag == VP8_GOLD_FLAG) - vp8_yv12_copy_frame_ptr(sd, &cm->golden_frame); - + ref_fb_idx = cm->gld_fb_idx; else if (ref_frame_flag == VP8_ALT_FLAG) - vp8_yv12_copy_frame_ptr(sd, &cm->alt_ref_frame); - + ref_fb_idx = cm->alt_fb_idx; else return -1; + vp8_yv12_copy_frame_ptr(sd, &cm->yv12_fb[ref_fb_idx]); + return 0; } int vp8_update_entropy(VP8_PTR comp, int update) @@ -2692,8 +2739,8 @@ static void scale_and_extend_source(YV12_BUFFER_CONFIG *sd, VP8_COMP *cpi) #endif } // we may need to copy to a buffer so we can extend the image... - else if (cm->Width != cm->last_frame.y_width || - cm->Height != cm->last_frame.y_height) + else if (cm->Width != cm->yv12_fb[cm->lst_fb_idx].y_width || + cm->Height != cm->yv12_fb[cm->lst_fb_idx].y_height) { //vp8_yv12_copy_frame_ptr(sd, &cpi->scaled_source); #if HAVE_ARMV7 @@ -2783,23 +2830,17 @@ static int pick_frame_size(VP8_COMP *cpi) cm->frame_type = KEY_FRAME; } - // Auto key frames (Only two pass will enter here) - else if (cm->frame_type == KEY_FRAME) - { - vp8_calc_auto_iframe_target_size(cpi); - } - // Forced key frames (by interval or an external signal) - else if ((cm->frame_flags & FRAMEFLAGS_KEY) || - (cpi->oxcf.auto_key && (cpi->frames_since_key % cpi->key_frame_frequency == 0))) + // Special case for forced key frames + // The frame sizing here is still far from ideal for 2 pass. + else if (cm->frame_flags & FRAMEFLAGS_KEY) { - // Key frame from VFW/auto-keyframe/first frame cm->frame_type = KEY_FRAME; - resize_key_frame(cpi); - - // Compute target frame size - if (cpi->pass != 2) - vp8_calc_iframe_target_size(cpi); + vp8_calc_iframe_target_size(cpi); + } + else if (cm->frame_type == KEY_FRAME) + { + vp8_calc_auto_iframe_target_size(cpi); } else { @@ -2846,7 +2887,7 @@ static void update_alt_ref_frame_and_stats(VP8_COMP *cpi) VP8_COMMON *cm = &cpi->common; // Update the golden frame buffer - vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->alt_ref_frame); + vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->yv12_fb[cm->alt_fb_idx]); // Select an interval before next GF or altref if (!cpi->auto_gold) @@ -2866,8 +2907,8 @@ static void update_alt_ref_frame_and_stats(VP8_COMP *cpi) } // Update data structure that monitors level of reference to last GF - vpx_memset(cm->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols)); - cm->gf_active_count = cm->mb_rows * cm->mb_cols; + vpx_memset(cpi->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols)); + cpi->gf_active_count = cm->mb_rows * cm->mb_cols; // this frame refreshes means next frames don't unless specified by user cpi->common.frames_since_golden = 0; @@ -2888,7 +2929,7 @@ static void update_golden_frame_and_stats(VP8_COMP *cpi) if (cm->refresh_golden_frame) { // Update the golden frame buffer - vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->golden_frame); + vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->yv12_fb[cm->gld_fb_idx]); // Select an interval before next GF if (!cpi->auto_gold) @@ -2914,8 +2955,8 @@ static void update_golden_frame_and_stats(VP8_COMP *cpi) } // Update data structure that monitors level of reference to last GF - vpx_memset(cm->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols)); - cm->gf_active_count = cm->mb_rows * cm->mb_cols; + vpx_memset(cpi->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols)); + cpi->gf_active_count = cm->mb_rows * cm->mb_cols; // this frame refreshes means next frames don't unless specified by user cm->refresh_golden_frame = 0; @@ -3220,10 +3261,31 @@ void write_cx_frame_to_file(YV12_BUFFER_CONFIG *frame, int this_frame) } #endif // return of 0 means drop frame - +#define USE_FILTER_LUT 1 #if VP8_TEMPORAL_ALT_REF + +#if USE_FILTER_LUT +static int modifier_lut[7][19] = +{ + // Strength=0 + {16, 13, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, + // Strength=1 + {16, 15, 10, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, + // Strength=2 + {16, 15, 13, 9, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, + // Strength=3 + {16, 16, 15, 13, 10, 7, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, + // Strength=4 + {16, 16, 15, 14, 13, 11, 9, 7, 4, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0}, + // Strength=5 + {16, 16, 16, 15, 15, 14, 13, 11, 10, 8, 7, 5, 3, 0, 0, 0, 0, 0, 0}, + // Strength=6 + {16, 16, 16, 16, 15, 15, 14, 14, 13, 12, 11, 10, 9, 8, 7, 5, 4, 2, 1} +}; +#endif static void vp8cx_temp_blur1_c ( + VP8_COMP *cpi, unsigned char **frames, int frame_count, unsigned char *src, @@ -3237,22 +3299,24 @@ static void vp8cx_temp_blur1_c unsigned char block_size ) { - int byte = 0; // Buffer offset for the current pixel value being filtered + int byte = 0; // Buffer offset for current pixel being filtered int frame = 0; int modifier = 0; int i, j, k; int block_ofset; - int Cols, Rows; + int cols; unsigned char Shift = (block_size == 16) ? 4 : 3; +#if USE_FILTER_LUT + int *lut = modifier_lut[strength]; +#endif - Cols = width / block_size; - Rows = height / block_size; + cols = cpi->common.mb_cols; for (i = 0; i < height; i++) { - block_ofset = (i >> Shift) * Cols; + block_ofset = (i >> Shift) * cols; - for (j = 0; j < Cols; j ++) + for (j = 0; j < cols; j ++) { if (motion_map_ptr[block_ofset] > 2) { @@ -3270,9 +3334,14 @@ static void vp8cx_temp_blur1_c for (frame = 0; frame < frame_count; frame++) { // get current frame pixel value - int pixel_value = frames[frame][byte]; // int pixel_value = *frameptr; - - modifier = src_byte; // modifier = s[byte]; + int pixel_value = frames[frame][byte]; +#if USE_FILTER_LUT + // LUT implementation -- + // improves precision of filter + modifier = abs(src_byte-pixel_value); + modifier = modifier>18 ? 0 : lut[modifier]; +#else + modifier = src_byte; modifier -= pixel_value; modifier *= modifier; modifier >>= strength; @@ -3282,17 +3351,17 @@ static void vp8cx_temp_blur1_c modifier = 16; modifier = 16 - modifier; - +#endif accumulator += modifier * pixel_value; count += modifier; } accumulator += (count >> 1); - accumulator *= fixed_divide[count]; // accumulator *= ppi->fixed_divide[count]; + accumulator *= fixed_divide[count]; accumulator >>= 16; - dst[byte] = accumulator; // d[byte] = accumulator; + dst[byte] = accumulator; // move to next pixel byte++; @@ -3398,7 +3467,8 @@ static void vp8cx_temp_filter_c { if ((frames_to_blur_backward + frames_to_blur_forward) >= max_frames) { - frames_to_blur_backward = max_frames - frames_to_blur_forward - 1; + frames_to_blur_backward + = max_frames - frames_to_blur_forward - 1; } } else @@ -3441,6 +3511,7 @@ static void vp8cx_temp_filter_c // Blur Y vp8cx_temp_blur1_c( + cpi, cpi->frames, frames_to_blur, temp_source_buffer->y_buffer, // cpi->Source->y_buffer, @@ -3455,7 +3526,7 @@ static void vp8cx_temp_filter_c for (frame = 0; frame < frames_to_blur; frame++) { - int which_buffer = cpi->last_alt_ref_sei - frame; + int which_buffer = start_frame - frame; if (which_buffer < 0) which_buffer += cpi->oxcf.lag_in_frames; @@ -3465,6 +3536,7 @@ static void vp8cx_temp_filter_c // Blur U vp8cx_temp_blur1_c( + cpi, cpi->frames, frames_to_blur, temp_source_buffer->u_buffer, @@ -3479,7 +3551,7 @@ static void vp8cx_temp_filter_c for (frame = 0; frame < frames_to_blur; frame++) { - int which_buffer = cpi->last_alt_ref_sei - frame; + int which_buffer = start_frame - frame; if (which_buffer < 0) which_buffer += cpi->oxcf.lag_in_frames; @@ -3489,6 +3561,7 @@ static void vp8cx_temp_filter_c // Blur V vp8cx_temp_blur1_c( + cpi, cpi->frames, frames_to_blur, temp_source_buffer->v_buffer, @@ -4277,7 +4350,7 @@ static void encode_frame_to_data_rate(VP8_COMP *cpi, unsigned long *size, unsign // Update the GF useage maps. // This is done after completing the compression of a frame when all modes etc. are finalized but before loop filter - vp8_update_gf_useage_maps(cm, &cpi->mb.e_mbd); + vp8_update_gf_useage_maps(cpi, cm, &cpi->mb); if (cm->frame_type == KEY_FRAME) cm->refresh_last_frame = 1; @@ -4285,7 +4358,7 @@ static void encode_frame_to_data_rate(VP8_COMP *cpi, unsigned long *size, unsign if (0) { FILE *f = fopen("gfactive.stt", "a"); - fprintf(f, "%8d %8d %8d %8d %8d\n", cm->current_video_frame, (100 * cpi->common.gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols), cpi->this_iiratio, cpi->next_iiratio, cm->refresh_golden_frame); + fprintf(f, "%8d %8d %8d %8d %8d\n", cm->current_video_frame, (100 * cpi->gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols), cpi->this_iiratio, cpi->next_iiratio, cm->refresh_golden_frame); fclose(f); } @@ -4298,11 +4371,11 @@ static void encode_frame_to_data_rate(VP8_COMP *cpi, unsigned long *size, unsign if (cm->refresh_last_frame) { - vp8_swap_yv12_buffer(&cm->last_frame, &cm->new_frame); - cm->frame_to_show = &cm->last_frame; + vp8_swap_yv12_buffer(&cm->yv12_fb[cm->lst_fb_idx], &cm->yv12_fb[cm->new_fb_idx]); + cm->frame_to_show = &cm->yv12_fb[cm->lst_fb_idx]; } else - cm->frame_to_show = &cm->new_frame; + cm->frame_to_show = &cm->yv12_fb[cm->new_fb_idx]; @@ -4352,43 +4425,48 @@ static void encode_frame_to_data_rate(VP8_COMP *cpi, unsigned long *size, unsign } } - - // At this point the new frame has been encoded coded. - // If any buffer copy / swaping is signalled it should be done here. - if (cm->frame_type == KEY_FRAME) { - vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->golden_frame); - vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->alt_ref_frame); - } - else // For non key frames - { - // Code to copy between reference buffers - if (cm->copy_buffer_to_arf) + YV12_BUFFER_CONFIG *lst_yv12 = &cm->yv12_fb[cm->lst_fb_idx]; + YV12_BUFFER_CONFIG *new_yv12 = &cm->yv12_fb[cm->new_fb_idx]; + YV12_BUFFER_CONFIG *gld_yv12 = &cm->yv12_fb[cm->gld_fb_idx]; + YV12_BUFFER_CONFIG *alt_yv12 = &cm->yv12_fb[cm->alt_fb_idx]; + // At this point the new frame has been encoded coded. + // If any buffer copy / swaping is signalled it should be done here. + if (cm->frame_type == KEY_FRAME) { - if (cm->copy_buffer_to_arf == 1) + vp8_yv12_copy_frame_ptr(cm->frame_to_show, gld_yv12); + vp8_yv12_copy_frame_ptr(cm->frame_to_show, alt_yv12); + } + else // For non key frames + { + // Code to copy between reference buffers + if (cm->copy_buffer_to_arf) { - if (cm->refresh_last_frame) - // We copy new_frame here because last and new buffers will already have been swapped if cm->refresh_last_frame is set. - vp8_yv12_copy_frame_ptr(&cm->new_frame, &cm->alt_ref_frame); - else - vp8_yv12_copy_frame_ptr(&cm->last_frame, &cm->alt_ref_frame); + if (cm->copy_buffer_to_arf == 1) + { + if (cm->refresh_last_frame) + // We copy new_frame here because last and new buffers will already have been swapped if cm->refresh_last_frame is set. + vp8_yv12_copy_frame_ptr(new_yv12, alt_yv12); + else + vp8_yv12_copy_frame_ptr(lst_yv12, alt_yv12); + } + else if (cm->copy_buffer_to_arf == 2) + vp8_yv12_copy_frame_ptr(gld_yv12, alt_yv12); } - else if (cm->copy_buffer_to_arf == 2) - vp8_yv12_copy_frame_ptr(&cm->golden_frame, &cm->alt_ref_frame); - } - if (cm->copy_buffer_to_gf) - { - if (cm->copy_buffer_to_gf == 1) + if (cm->copy_buffer_to_gf) { - if (cm->refresh_last_frame) - // We copy new_frame here because last and new buffers will already have been swapped if cm->refresh_last_frame is set. - vp8_yv12_copy_frame_ptr(&cm->new_frame, &cm->golden_frame); - else - vp8_yv12_copy_frame_ptr(&cm->last_frame, &cm->golden_frame); + if (cm->copy_buffer_to_gf == 1) + { + if (cm->refresh_last_frame) + // We copy new_frame here because last and new buffers will already have been swapped if cm->refresh_last_frame is set. + vp8_yv12_copy_frame_ptr(new_yv12, gld_yv12); + else + vp8_yv12_copy_frame_ptr(lst_yv12, gld_yv12); + } + else if (cm->copy_buffer_to_gf == 2) + vp8_yv12_copy_frame_ptr(alt_yv12, gld_yv12); } - else if (cm->copy_buffer_to_gf == 2) - vp8_yv12_copy_frame_ptr(&cm->alt_ref_frame, &cm->golden_frame); } } @@ -4591,23 +4669,23 @@ static void encode_frame_to_data_rate(VP8_COMP *cpi, unsigned long *size, unsign cpi->ref_frame_flags = VP8_ALT_FLAG | VP8_GOLD_FLAG | VP8_LAST_FLAG; if (cpi->gold_is_last) - cpi->ref_frame_flags &= !VP8_GOLD_FLAG; + cpi->ref_frame_flags &= ~VP8_GOLD_FLAG; if (cpi->alt_is_last) - cpi->ref_frame_flags &= !VP8_ALT_FLAG; + cpi->ref_frame_flags &= ~VP8_ALT_FLAG; if (cpi->gold_is_alt) - cpi->ref_frame_flags &= !VP8_ALT_FLAG; + cpi->ref_frame_flags &= ~VP8_ALT_FLAG; if (cpi->oxcf.error_resilient_mode) { // Is this an alternate reference update if (cpi->common.refresh_alt_ref_frame) - vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->alt_ref_frame); + vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->yv12_fb[cm->alt_fb_idx]); if (cpi->common.refresh_golden_frame) - vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->golden_frame); + vp8_yv12_copy_frame_ptr(cm->frame_to_show, &cm->yv12_fb[cm->gld_fb_idx]); } else { @@ -4659,7 +4737,8 @@ static void encode_frame_to_data_rate(VP8_COMP *cpi, unsigned long *size, unsign FILE *recon_file; sprintf(filename, "enc%04d.yuv", (int) cm->current_video_frame); recon_file = fopen(filename, "wb"); - fwrite(cm->last_frame.buffer_alloc, cm->last_frame.frame_size, 1, recon_file); + fwrite(cm->yv12_fb[cm->lst_fb_idx].buffer_alloc, + cm->yv12_fb[cm->lst_fb_idx].frame_size, 1, recon_file); fclose(recon_file); } @@ -4683,7 +4762,7 @@ int vp8_is_gf_update_needed(VP8_PTR ptr) void vp8_check_gf_quality(VP8_COMP *cpi) { VP8_COMMON *cm = &cpi->common; - int gf_active_pct = (100 * cm->gf_active_count) / (cm->mb_rows * cm->mb_cols); + int gf_active_pct = (100 * cpi->gf_active_count) / (cm->mb_rows * cm->mb_cols); int gf_ref_usage_pct = (cpi->count_mb_ref_frame_usage[GOLDEN_FRAME] * 100) / (cm->mb_rows * cm->mb_cols); int last_ref_zz_useage = (cpi->inter_zz_count * 100) / (cm->mb_rows * cm->mb_cols); @@ -4739,7 +4818,6 @@ void vp8_check_gf_quality(VP8_COMP *cpi) #if !(CONFIG_REALTIME_ONLY) static void Pass2Encode(VP8_COMP *cpi, unsigned long *size, unsigned char *dest, unsigned int *frame_flags) { - double two_pass_min_rate = (double)(cpi->oxcf.target_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100); if (!cpi->common.refresh_alt_ref_frame) vp8_second_pass(cpi); @@ -4748,7 +4826,11 @@ static void Pass2Encode(VP8_COMP *cpi, unsigned long *size, unsigned char *dest, cpi->bits_left -= 8 * *size; if (!cpi->common.refresh_alt_ref_frame) + { + double two_pass_min_rate = (double)(cpi->oxcf.target_bandwidth + *cpi->oxcf.two_pass_vbrmin_section / 100); cpi->bits_left += (long long)(two_pass_min_rate / cpi->oxcf.frame_rate); + } } #endif @@ -4959,20 +5041,7 @@ int vp8_get_compressed_data(VP8_PTR ptr, unsigned int *frame_flags, unsigned lon { if (cpi->source_encode_index == cpi->last_alt_ref_sei) { -#if VP8_TEMPORAL_ALT_REF - - if (cpi->oxcf.arnr_max_frames == 0) - { - cpi->is_src_frame_alt_ref = 1; // copy alt ref - } - else - { - cpi->is_src_frame_alt_ref = 0; - } - -#else cpi->is_src_frame_alt_ref = 1; -#endif cpi->last_alt_ref_sei = -1; } else @@ -5140,8 +5209,6 @@ int vp8_get_compressed_data(VP8_PTR ptr, unsigned int *frame_flags, unsigned lon { // return to normal state - cpi->ref_frame_flags = VP8_ALT_FLAG | VP8_GOLD_FLAG | VP8_LAST_FLAG; - cm->refresh_entropy_probs = 1; cm->refresh_alt_ref_frame = 0; cm->refresh_golden_frame = 0; @@ -5402,7 +5469,7 @@ int vp8_calc_low_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest, co { for (j = 0; j < source->y_width; j += 16) { - unsigned int sse, sse2, sum2; + unsigned int sse; VARIANCE_INVOKE(rtcd, mse16x16)(src + j, source->y_stride, dst + j, dest->y_stride, &sse); if (sse < 8096) diff --git a/vp8/encoder/onyx_int.h b/vp8/encoder/onyx_int.h index 29b120e..211f659 100644 --- a/vp8/encoder/onyx_int.h +++ b/vp8/encoder/onyx_int.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -25,7 +26,7 @@ #include "entropy.h" #include "threading.h" #include "vpx_ports/mem.h" -#include "vpx_codec/internal/vpx_codec_internal.h" +#include "vpx/internal/vpx_codec_internal.h" #include "mcomp.h" #define INTRARDOPT @@ -233,14 +234,17 @@ typedef struct { DECLARE_ALIGNED(16, short, Y1quant[QINDEX_RANGE][4][4]); + DECLARE_ALIGNED(16, short, Y1quant_shift[QINDEX_RANGE][4][4]); DECLARE_ALIGNED(16, short, Y1zbin[QINDEX_RANGE][4][4]); DECLARE_ALIGNED(16, short, Y1round[QINDEX_RANGE][4][4]); DECLARE_ALIGNED(16, short, Y2quant[QINDEX_RANGE][4][4]); + DECLARE_ALIGNED(16, short, Y2quant_shift[QINDEX_RANGE][4][4]); DECLARE_ALIGNED(16, short, Y2zbin[QINDEX_RANGE][4][4]); DECLARE_ALIGNED(16, short, Y2round[QINDEX_RANGE][4][4]); DECLARE_ALIGNED(16, short, UVquant[QINDEX_RANGE][4][4]); + DECLARE_ALIGNED(16, short, UVquant_shift[QINDEX_RANGE][4][4]); DECLARE_ALIGNED(16, short, UVzbin[QINDEX_RANGE][4][4]); DECLARE_ALIGNED(16, short, UVround[QINDEX_RANGE][4][4]); @@ -355,9 +359,14 @@ typedef struct int gf_bits; // Bits for the golden frame or ARF - 2 pass only int mid_gf_extra_bits; // A few extra bits for the frame half way between two gfs. - int kf_group_bits; // Projected total bits available for a key frame group of frames - int kf_group_error_left; // Error score of frames still to be coded in kf group - int kf_bits; // Bits for the key frame in a key frame group - 2 pass only + // Projected total bits available for a key frame group of frames + long long kf_group_bits; + + // Error score of frames still to be coded in kf group + long long kf_group_error_left; + + // Bits for the key frame in a key frame group - 2 pass only + int kf_bits; int non_gf_bitrate_adjustment; // Used in the few frames following a GF to recover the extra bits spent in that GF int initial_gf_use; // percentage use of gf 2 frames after gf @@ -522,8 +531,8 @@ typedef struct int motion_lvl; int motion_speed; int motion_var; - int next_iiratio; - int this_iiratio; + unsigned int next_iiratio; + unsigned int this_iiratio; int this_frame_modified_error; double norm_intra_err_per_mb; @@ -637,6 +646,12 @@ typedef struct int b_calculate_ssimg; #endif int b_calculate_psnr; + + + unsigned char *gf_active_flags; // Record of which MBs still refer to last golden frame either directly or through 0,0 + int gf_active_count; + + } VP8_COMP; void control_data_rate(VP8_COMP *cpi); diff --git a/vp8/encoder/parms.cpp b/vp8/encoder/parms.cpp index 66fdafb..6cc4501 100644 --- a/vp8/encoder/parms.cpp +++ b/vp8/encoder/parms.cpp @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/pickinter.c b/vp8/encoder/pickinter.c index d61e2ce..eeeddcc 100644 --- a/vp8/encoder/pickinter.c +++ b/vp8/encoder/pickinter.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -219,13 +220,20 @@ int vp8_pick_intra4x4mby_modes(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *mb, int { MACROBLOCKD *const xd = &mb->e_mbd; int i; - TEMP_CONTEXT t; int cost = mb->mbmode_cost [xd->frame_type] [B_PRED]; int error = RD_ESTIMATE(mb->rdmult, mb->rddiv, cost, 0); // Rd estimate for the cost of the block prediction mode int distortion = 0; + ENTROPY_CONTEXT_PLANES t_above, t_left; + ENTROPY_CONTEXT *ta; + ENTROPY_CONTEXT *tl; + + vpx_memcpy(&t_above, mb->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memcpy(&t_left, mb->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES)); + + ta = (ENTROPY_CONTEXT *)&t_above; + tl = (ENTROPY_CONTEXT *)&t_left; vp8_intra_prediction_down_copy(xd); - vp8_setup_temp_context(&t, xd->above_context[Y1CONTEXT], xd->left_context[Y1CONTEXT], 4); for (i = 0; i < 16; i++) { @@ -238,8 +246,8 @@ int vp8_pick_intra4x4mby_modes(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *mb, int error += pick_intra4x4block(rtcd, mb, mb->block + i, xd->block + i, &best_mode, A, L, - t.a + vp8_block2above[i], - t.l + vp8_block2left[i], &r, &d); + ta + vp8_block2above[i], + tl + vp8_block2left[i], &r, &d); cost += r; distortion += d; @@ -409,7 +417,7 @@ int vp8_pick_intra_mbuv_mode(MACROBLOCK *mb) } - mb->e_mbd.mbmi.uv_mode = best_mode; + mb->e_mbd.mode_info_context->mbmi.uv_mode = best_mode; return best_error; } @@ -422,6 +430,7 @@ int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int rec MACROBLOCKD *xd = &x->e_mbd; B_MODE_INFO best_bmodes[16]; MB_MODE_INFO best_mbmode; + PARTITION_INFO best_partition; MV best_ref_mv1; MV mode_mv[MB_MODE_COUNT]; MB_PREDICTION_MODE this_mode; @@ -453,41 +462,48 @@ int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int rec vpx_memset(mode_mv, 0, sizeof(mode_mv)); vpx_memset(nearest_mv, 0, sizeof(nearest_mv)); vpx_memset(near_mv, 0, sizeof(near_mv)); + vpx_memset(&best_mbmode, 0, sizeof(best_mbmode)); // set up all the refframe dependent pointers. if (cpi->ref_frame_flags & VP8_LAST_FLAG) { + YV12_BUFFER_CONFIG *lst_yv12 = &cpi->common.yv12_fb[cpi->common.lst_fb_idx]; + vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &nearest_mv[LAST_FRAME], &near_mv[LAST_FRAME], &best_ref_mv[LAST_FRAME], MDCounts[LAST_FRAME], LAST_FRAME, cpi->common.ref_frame_sign_bias); - y_buffer[LAST_FRAME] = cpi->common.last_frame.y_buffer + recon_yoffset; - u_buffer[LAST_FRAME] = cpi->common.last_frame.u_buffer + recon_uvoffset; - v_buffer[LAST_FRAME] = cpi->common.last_frame.v_buffer + recon_uvoffset; + y_buffer[LAST_FRAME] = lst_yv12->y_buffer + recon_yoffset; + u_buffer[LAST_FRAME] = lst_yv12->u_buffer + recon_uvoffset; + v_buffer[LAST_FRAME] = lst_yv12->v_buffer + recon_uvoffset; } else skip_mode[LAST_FRAME] = 1; if (cpi->ref_frame_flags & VP8_GOLD_FLAG) { + YV12_BUFFER_CONFIG *gld_yv12 = &cpi->common.yv12_fb[cpi->common.gld_fb_idx]; + vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &nearest_mv[GOLDEN_FRAME], &near_mv[GOLDEN_FRAME], &best_ref_mv[GOLDEN_FRAME], MDCounts[GOLDEN_FRAME], GOLDEN_FRAME, cpi->common.ref_frame_sign_bias); - y_buffer[GOLDEN_FRAME] = cpi->common.golden_frame.y_buffer + recon_yoffset; - u_buffer[GOLDEN_FRAME] = cpi->common.golden_frame.u_buffer + recon_uvoffset; - v_buffer[GOLDEN_FRAME] = cpi->common.golden_frame.v_buffer + recon_uvoffset; + y_buffer[GOLDEN_FRAME] = gld_yv12->y_buffer + recon_yoffset; + u_buffer[GOLDEN_FRAME] = gld_yv12->u_buffer + recon_uvoffset; + v_buffer[GOLDEN_FRAME] = gld_yv12->v_buffer + recon_uvoffset; } else skip_mode[GOLDEN_FRAME] = 1; if (cpi->ref_frame_flags & VP8_ALT_FLAG && cpi->source_alt_ref_active) { + YV12_BUFFER_CONFIG *alt_yv12 = &cpi->common.yv12_fb[cpi->common.alt_fb_idx]; + vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &nearest_mv[ALTREF_FRAME], &near_mv[ALTREF_FRAME], &best_ref_mv[ALTREF_FRAME], MDCounts[ALTREF_FRAME], ALTREF_FRAME, cpi->common.ref_frame_sign_bias); - y_buffer[ALTREF_FRAME] = cpi->common.alt_ref_frame.y_buffer + recon_yoffset; - u_buffer[ALTREF_FRAME] = cpi->common.alt_ref_frame.u_buffer + recon_uvoffset; - v_buffer[ALTREF_FRAME] = cpi->common.alt_ref_frame.v_buffer + recon_uvoffset; + y_buffer[ALTREF_FRAME] = alt_yv12->y_buffer + recon_yoffset; + u_buffer[ALTREF_FRAME] = alt_yv12->u_buffer + recon_uvoffset; + v_buffer[ALTREF_FRAME] = alt_yv12->v_buffer + recon_uvoffset; } else skip_mode[ALTREF_FRAME] = 1; @@ -527,7 +543,7 @@ int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int rec best_rd = INT_MAX; - x->e_mbd.mbmi.ref_frame = INTRA_FRAME; + x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME; // if we encode a new mv this is important // find the best new motion vector @@ -539,9 +555,9 @@ int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int rec if (best_rd <= cpi->rd_threshes[mode_index]) continue; - x->e_mbd.mbmi.ref_frame = vp8_ref_frame_order[mode_index]; + x->e_mbd.mode_info_context->mbmi.ref_frame = vp8_ref_frame_order[mode_index]; - if (skip_mode[x->e_mbd.mbmi.ref_frame]) + if (skip_mode[x->e_mbd.mode_info_context->mbmi.ref_frame]) continue; // Check to see if the testing frequency for this mode is at its max @@ -570,33 +586,33 @@ int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int rec distortion2 = 0; this_mode = vp8_mode_order[mode_index]; - + // Experimental debug code. //all_rds[mode_index] = -1; - x->e_mbd.mbmi.mode = this_mode; - x->e_mbd.mbmi.uv_mode = DC_PRED; + x->e_mbd.mode_info_context->mbmi.mode = this_mode; + x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED; // Work out the cost assosciated with selecting the reference frame - frame_cost = ref_frame_cost[x->e_mbd.mbmi.ref_frame]; + frame_cost = ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame]; rate2 += frame_cost; // everything but intra - if (x->e_mbd.mbmi.ref_frame) + if (x->e_mbd.mode_info_context->mbmi.ref_frame) { - x->e_mbd.pre.y_buffer = y_buffer[x->e_mbd.mbmi.ref_frame]; - x->e_mbd.pre.u_buffer = u_buffer[x->e_mbd.mbmi.ref_frame]; - x->e_mbd.pre.v_buffer = v_buffer[x->e_mbd.mbmi.ref_frame]; - mode_mv[NEARESTMV] = nearest_mv[x->e_mbd.mbmi.ref_frame]; - mode_mv[NEARMV] = near_mv[x->e_mbd.mbmi.ref_frame]; - best_ref_mv1 = best_ref_mv[x->e_mbd.mbmi.ref_frame]; - memcpy(mdcounts, MDCounts[x->e_mbd.mbmi.ref_frame], sizeof(mdcounts)); + x->e_mbd.pre.y_buffer = y_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame]; + x->e_mbd.pre.u_buffer = u_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame]; + x->e_mbd.pre.v_buffer = v_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame]; + mode_mv[NEARESTMV] = nearest_mv[x->e_mbd.mode_info_context->mbmi.ref_frame]; + mode_mv[NEARMV] = near_mv[x->e_mbd.mode_info_context->mbmi.ref_frame]; + best_ref_mv1 = best_ref_mv[x->e_mbd.mode_info_context->mbmi.ref_frame]; + memcpy(mdcounts, MDCounts[x->e_mbd.mode_info_context->mbmi.ref_frame], sizeof(mdcounts)); } //Only consider ZEROMV/ALTREF_FRAME for alt ref frame. if (cpi->is_src_frame_alt_ref) { - if (this_mode != ZEROMV || x->e_mbd.mbmi.ref_frame != ALTREF_FRAME) + if (this_mode != ZEROMV || x->e_mbd.mode_info_context->mbmi.ref_frame != ALTREF_FRAME) continue; } @@ -636,7 +652,7 @@ int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int rec case TM_PRED: vp8_build_intra_predictors_mby_ptr(&x->e_mbd); distortion2 = VARIANCE_INVOKE(&cpi->rtcd.variance, get16x16prederror)(x->src.y_buffer, x->src.y_stride, x->e_mbd.predictor, 16, 0x7fffffff); - rate2 += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mbmi.mode]; + rate2 += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.mode]; this_rd = RD_ESTIMATE(x->rdmult, x->rddiv, rate2, distortion2); if (this_rd < best_intra_rd) @@ -774,10 +790,10 @@ int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int rec continue; rate2 += vp8_cost_mv_ref(this_mode, mdcounts); - x->e_mbd.mbmi.mode = this_mode; - x->e_mbd.mbmi.mv.as_mv = mode_mv[this_mode]; + x->e_mbd.mode_info_context->mbmi.mode = this_mode; + x->e_mbd.mode_info_context->mbmi.mv.as_mv = mode_mv[this_mode]; x->e_mbd.block[0].bmi.mode = this_mode; - x->e_mbd.block[0].bmi.mv.as_int = x->e_mbd.mbmi.mv.as_int; + x->e_mbd.block[0].bmi.mv.as_int = x->e_mbd.mode_info_context->mbmi.mv.as_int; distortion2 = get_inter_mbpred_error(x, cpi->fn_ptr.svf, cpi->fn_ptr.vf, (unsigned int *)(&sse)); @@ -816,7 +832,8 @@ int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int rec *returnrate = rate2; *returndistortion = distortion2; best_rd = this_rd; - vpx_memcpy(&best_mbmode, &x->e_mbd.mbmi, sizeof(MB_MODE_INFO)); + vpx_memcpy(&best_mbmode, &x->e_mbd.mode_info_context->mbmi, sizeof(MB_MODE_INFO)); + vpx_memcpy(&best_partition, x->partition_info, sizeof(PARTITION_INFO)); if (this_mode == B_PRED || this_mode == SPLITMV) for (i = 0; i < 16; i++) @@ -862,9 +879,9 @@ int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int rec if (best_mbmode.mode <= B_PRED) { - x->e_mbd.mbmi.ref_frame = INTRA_FRAME; + x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME; vp8_pick_intra_mbuv_mode(x); - best_mbmode.uv_mode = x->e_mbd.mbmi.uv_mode; + best_mbmode.uv_mode = x->e_mbd.mode_info_context->mbmi.uv_mode; } @@ -890,23 +907,25 @@ int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int rec best_mbmode.partitioning = 0; best_mbmode.dc_diff = 0; - vpx_memcpy(&x->e_mbd.mbmi, &best_mbmode, sizeof(MB_MODE_INFO)); + vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO)); + vpx_memcpy(x->partition_info, &best_partition, sizeof(PARTITION_INFO)); for (i = 0; i < 16; i++) { vpx_memset(&x->e_mbd.block[i].bmi, 0, sizeof(B_MODE_INFO)); } - x->e_mbd.mbmi.mv.as_int = 0; + x->e_mbd.mode_info_context->mbmi.mv.as_int = 0; return best_rd; } // macroblock modes - vpx_memcpy(&x->e_mbd.mbmi, &best_mbmode, sizeof(MB_MODE_INFO)); + vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO)); + vpx_memcpy(x->partition_info, &best_partition, sizeof(PARTITION_INFO)); - if (x->e_mbd.mbmi.mode == B_PRED || x->e_mbd.mbmi.mode == SPLITMV) + if (x->e_mbd.mode_info_context->mbmi.mode == B_PRED || x->e_mbd.mode_info_context->mbmi.mode == SPLITMV) for (i = 0; i < 16; i++) { vpx_memcpy(&x->e_mbd.block[i].bmi, &best_bmodes[i], sizeof(B_MODE_INFO)); @@ -914,10 +933,10 @@ int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int rec } else { - vp8_set_mbmode_and_mvs(x, x->e_mbd.mbmi.mode, &best_bmodes[0].mv.as_mv); + vp8_set_mbmode_and_mvs(x, x->e_mbd.mode_info_context->mbmi.mode, &best_bmodes[0].mv.as_mv); } - x->e_mbd.mbmi.mv.as_mv = x->e_mbd.block[15].bmi.mv.as_mv; + x->e_mbd.mode_info_context->mbmi.mv.as_mv = x->e_mbd.block[15].bmi.mv.as_mv; return best_rd; } diff --git a/vp8/encoder/pickinter.h b/vp8/encoder/pickinter.h index fb28837..b80e4c8 100644 --- a/vp8/encoder/pickinter.h +++ b/vp8/encoder/pickinter.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/picklpf.c b/vp8/encoder/picklpf.c index bbd7840..79e07db 100644 --- a/vp8/encoder/picklpf.c +++ b/vp8/encoder/picklpf.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/ppc/csystemdependent.c b/vp8/encoder/ppc/csystemdependent.c index f99277f..588656b 100644 --- a/vp8/encoder/ppc/csystemdependent.c +++ b/vp8/encoder/ppc/csystemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/ppc/encodemb_altivec.asm b/vp8/encoder/ppc/encodemb_altivec.asm index e0e976d..6e0099d 100644 --- a/vp8/encoder/ppc/encodemb_altivec.asm +++ b/vp8/encoder/ppc/encodemb_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/ppc/fdct_altivec.asm b/vp8/encoder/ppc/fdct_altivec.asm index eaab14c..935d0cb 100644 --- a/vp8/encoder/ppc/fdct_altivec.asm +++ b/vp8/encoder/ppc/fdct_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/ppc/rdopt_altivec.asm b/vp8/encoder/ppc/rdopt_altivec.asm index 917bfe0..ba48230 100644 --- a/vp8/encoder/ppc/rdopt_altivec.asm +++ b/vp8/encoder/ppc/rdopt_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/ppc/sad_altivec.asm b/vp8/encoder/ppc/sad_altivec.asm index 1102ccf..e5f2638 100644 --- a/vp8/encoder/ppc/sad_altivec.asm +++ b/vp8/encoder/ppc/sad_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/ppc/variance_altivec.asm b/vp8/encoder/ppc/variance_altivec.asm index 952bf72..a1ebf66 100644 --- a/vp8/encoder/ppc/variance_altivec.asm +++ b/vp8/encoder/ppc/variance_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/ppc/variance_subpixel_altivec.asm b/vp8/encoder/ppc/variance_subpixel_altivec.asm index 148a8d2..301360b 100644 --- a/vp8/encoder/ppc/variance_subpixel_altivec.asm +++ b/vp8/encoder/ppc/variance_subpixel_altivec.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/preproc.c b/vp8/encoder/preproc.c index d2a13dc..bd918fa 100644 --- a/vp8/encoder/preproc.c +++ b/vp8/encoder/preproc.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/psnr.c b/vp8/encoder/psnr.c index 0e34cec..dc2a03b 100644 --- a/vp8/encoder/psnr.c +++ b/vp8/encoder/psnr.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/psnr.h b/vp8/encoder/psnr.h index 9f6ca0b..8ae4448 100644 --- a/vp8/encoder/psnr.h +++ b/vp8/encoder/psnr.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/quantize.c b/vp8/encoder/quantize.c index 6028ebf..20ec9d1 100644 --- a/vp8/encoder/quantize.c +++ b/vp8/encoder/quantize.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -15,6 +16,8 @@ #include "entropy.h" #include "predictdc.h" +//#define EXACT_QUANT +#ifdef EXACT_QUANT void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d) { int i, rc, eob; @@ -24,6 +27,7 @@ void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d) short *zbin_ptr = &b->zbin[0][0]; short *round_ptr = &b->round[0][0]; short *quant_ptr = &b->quant[0][0]; + short *quant_shift_ptr = &b->quant_shift[0][0]; short *qcoeff_ptr = d->qcoeff; short *dqcoeff_ptr = d->dqcoeff; short *dequant_ptr = &d->dequant[0][0]; @@ -44,7 +48,9 @@ void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d) if (x >= zbin) { - y = ((x + round_ptr[rc]) * quant_ptr[rc]) >> 16; // quantize (x) + x += round_ptr[rc]; + y = (((x * quant_ptr[rc]) >> 16) + x) + >> quant_shift_ptr[rc]; // quantize (x) x = (y ^ sz) - sz; // get the sign back qcoeff_ptr[rc] = x; // write to destination dqcoeff_ptr[rc] = x * dequant_ptr[rc]; // dequantized value @@ -55,9 +61,7 @@ void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d) } } } - d->eob = eob + 1; - } void vp8_regular_quantize_b(BLOCK *b, BLOCKD *d) @@ -70,6 +74,7 @@ void vp8_regular_quantize_b(BLOCK *b, BLOCKD *d) short *zbin_ptr = &b->zbin[0][0]; short *round_ptr = &b->round[0][0]; short *quant_ptr = &b->quant[0][0]; + short *quant_shift_ptr = &b->quant_shift[0][0]; short *qcoeff_ptr = d->qcoeff; short *dqcoeff_ptr = d->dqcoeff; short *dequant_ptr = &d->dequant[0][0]; @@ -96,7 +101,9 @@ void vp8_regular_quantize_b(BLOCK *b, BLOCKD *d) if (x >= zbin) { - y = ((x + round_ptr[rc]) * quant_ptr[rc]) >> 16; // quantize (x) + x += round_ptr[rc]; + y = (((x * quant_ptr[rc]) >> 16) + x) + >> quant_shift_ptr[rc]; // quantize (x) x = (y ^ sz) - sz; // get the sign back qcoeff_ptr[rc] = x; // write to destination dqcoeff_ptr[rc] = x * dequant_ptr[rc]; // dequantized value @@ -111,139 +118,205 @@ void vp8_regular_quantize_b(BLOCK *b, BLOCKD *d) d->eob = eob + 1; } -void vp8_quantize_mby(MACROBLOCK *x) +#else +void vp8_fast_quantize_b_c(BLOCK *b, BLOCKD *d) { - int i; + int i, rc, eob; + int zbin; + int x, y, z, sz; + short *coeff_ptr = &b->coeff[0]; + short *zbin_ptr = &b->zbin[0][0]; + short *round_ptr = &b->round[0][0]; + short *quant_ptr = &b->quant[0][0]; + short *qcoeff_ptr = d->qcoeff; + short *dqcoeff_ptr = d->dqcoeff; + short *dequant_ptr = &d->dequant[0][0]; - if (x->e_mbd.mbmi.mode != B_PRED && x->e_mbd.mbmi.mode != SPLITMV) - { - for (i = 0; i < 16; i++) - { - x->quantize_b(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (x->e_mbd.block[i].eob < 2); - } + vpx_memset(qcoeff_ptr, 0, 32); + vpx_memset(dqcoeff_ptr, 0, 32); - x->quantize_b(&x->block[24], &x->e_mbd.block[24]); - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[24].eob); + eob = -1; - } - else + for (i = 0; i < 16; i++) { - for (i = 0; i < 16; i++) + rc = vp8_default_zig_zag1d[i]; + z = coeff_ptr[rc]; + zbin = zbin_ptr[rc] ; + + sz = (z >> 31); // sign of z + x = (z ^ sz) - sz; // x = abs(z) + + if (x >= zbin) { - x->quantize_b(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); + y = ((x + round_ptr[rc]) * quant_ptr[rc]) >> 16; // quantize (x) + x = (y ^ sz) - sz; // get the sign back + qcoeff_ptr[rc] = x; // write to destination + dqcoeff_ptr[rc] = x * dequant_ptr[rc]; // dequantized value + + if (y) + { + eob = i; // last nonzero coeffs + } } } + d->eob = eob + 1; } -void vp8_quantize_mb(MACROBLOCK *x) +void vp8_regular_quantize_b(BLOCK *b, BLOCKD *d) { - int i; + int i, rc, eob; + int zbin; + int x, y, z, sz; + short *zbin_boost_ptr = &b->zrun_zbin_boost[0]; + short *coeff_ptr = &b->coeff[0]; + short *zbin_ptr = &b->zbin[0][0]; + short *round_ptr = &b->round[0][0]; + short *quant_ptr = &b->quant[0][0]; + short *qcoeff_ptr = d->qcoeff; + short *dqcoeff_ptr = d->dqcoeff; + short *dequant_ptr = &d->dequant[0][0]; + short zbin_oq_value = b->zbin_extra; - x->e_mbd.mbmi.mb_skip_coeff = 1; + vpx_memset(qcoeff_ptr, 0, 32); + vpx_memset(dqcoeff_ptr, 0, 32); - if (x->e_mbd.mbmi.mode != B_PRED && x->e_mbd.mbmi.mode != SPLITMV) - { - for (i = 0; i < 16; i++) - { - x->quantize_b(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (x->e_mbd.block[i].eob < 2); - } + eob = -1; - for (i = 16; i < 25; i++) - { - x->quantize_b(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); - } - } - else + for (i = 0; i < 16; i++) { - for (i = 0; i < 24; i++) + rc = vp8_default_zig_zag1d[i]; + z = coeff_ptr[rc]; + + //if ( i == 0 ) + // zbin = zbin_ptr[rc] + *zbin_boost_ptr + zbin_oq_value/2; + //else + zbin = zbin_ptr[rc] + *zbin_boost_ptr + zbin_oq_value; + + zbin_boost_ptr ++; + sz = (z >> 31); // sign of z + x = (z ^ sz) - sz; // x = abs(z) + + if (x >= zbin) { - x->quantize_b(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); + y = ((x + round_ptr[rc]) * quant_ptr[rc]) >> 16; // quantize (x) + x = (y ^ sz) - sz; // get the sign back + qcoeff_ptr[rc] = x; // write to destination + dqcoeff_ptr[rc] = x * dequant_ptr[rc]; // dequantized value + + if (y) + { + eob = i; // last nonzero coeffs + zbin_boost_ptr = &b->zrun_zbin_boost[0]; // reset zero runlength + } } } + d->eob = eob + 1; } +#endif -void vp8_quantize_mbuv(MACROBLOCK *x) +/* Perform regular quantization, with unbiased rounding and no zero bin. */ +void vp8_strict_quantize_b(BLOCK *b, BLOCKD *d) { int i; - - for (i = 16; i < 24; i++) + int rc; + int eob; + int x; + int y; + int z; + int sz; + short *coeff_ptr; + short *quant_ptr; + short *quant_shift_ptr; + short *qcoeff_ptr; + short *dqcoeff_ptr; + short *dequant_ptr; + + coeff_ptr = &b->coeff[0]; + quant_ptr = &b->quant[0][0]; + quant_shift_ptr = &b->quant_shift[0][0]; + qcoeff_ptr = d->qcoeff; + dqcoeff_ptr = d->dqcoeff; + dequant_ptr = &d->dequant[0][0]; + eob = - 1; + vpx_memset(qcoeff_ptr, 0, 32); + vpx_memset(dqcoeff_ptr, 0, 32); + for (i = 0; i < 16; i++) { - x->quantize_b(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); + int dq; + int round; + + /*TODO: These arrays should be stored in zig-zag order.*/ + rc = vp8_default_zig_zag1d[i]; + z = coeff_ptr[rc]; + dq = dequant_ptr[rc]; + round = dq >> 1; + /* Sign of z. */ + sz = -(z < 0); + x = (z + sz) ^ sz; + x += round; + if (x >= dq) + { + /* Quantize x. */ + y = (((x * quant_ptr[rc]) >> 16) + x) >> quant_shift_ptr[rc]; + /* Put the sign back. */ + x = (y + sz) ^ sz; + /* Save the coefficient and its dequantized value. */ + qcoeff_ptr[rc] = x; + dqcoeff_ptr[rc] = x * dq; + /* Remember the last non-zero coefficient. */ + if (y) + eob = i; + } } + + d->eob = eob + 1; } -// This function is not currently called -void vp8_quantize_mbrd(MACROBLOCK *x) +void vp8_quantize_mby(MACROBLOCK *x) { int i; + int has_2nd_order = (x->e_mbd.mode_info_context->mbmi.mode != B_PRED + && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV); - x->e_mbd.mbmi.mb_skip_coeff = 1; - - if (x->e_mbd.mbmi.mode != B_PRED && x->e_mbd.mbmi.mode != SPLITMV) + for (i = 0; i < 16; i++) { - for (i = 0; i < 16; i++) - { - x->quantize_brd(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (x->e_mbd.block[i].eob < 2); - } - - for (i = 16; i < 25; i++) - { - x->quantize_brd(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); - } + x->quantize_b(&x->block[i], &x->e_mbd.block[i]); + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff &= + (x->e_mbd.block[i].eob <= has_2nd_order); } - else + + if(has_2nd_order) { - for (i = 0; i < 24; i++) - { - x->quantize_brd(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); - } + x->quantize_b(&x->block[24], &x->e_mbd.block[24]); + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff &= (!x->e_mbd.block[24].eob); } } -void vp8_quantize_mbuvrd(MACROBLOCK *x) +void vp8_quantize_mb(MACROBLOCK *x) { int i; + int has_2nd_order=(x->e_mbd.mode_info_context->mbmi.mode != B_PRED + && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV); - for (i = 16; i < 24; i++) + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff = 1; + for (i = 0; i < 24+has_2nd_order; i++) { - x->quantize_brd(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); + x->quantize_b(&x->block[i], &x->e_mbd.block[i]); + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff &= + (x->e_mbd.block[i].eob <= (has_2nd_order && i<16)); } } -void vp8_quantize_mbyrd(MACROBLOCK *x) + +void vp8_quantize_mbuv(MACROBLOCK *x) { int i; - if (x->e_mbd.mbmi.mode != B_PRED && x->e_mbd.mbmi.mode != SPLITMV) - { - for (i = 0; i < 16; i++) - { - x->quantize_brd(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (x->e_mbd.block[i].eob < 2); - } - - x->quantize_brd(&x->block[24], &x->e_mbd.block[24]); - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[24].eob); - - } - else + for (i = 16; i < 24; i++) { - for (i = 0; i < 16; i++) - { - x->quantize_brd(&x->block[i], &x->e_mbd.block[i]); - x->e_mbd.mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); - } + x->quantize_b(&x->block[i], &x->e_mbd.block[i]); + x->e_mbd.mode_info_context->mbmi.mb_skip_coeff &= (!x->e_mbd.block[i].eob); } } diff --git a/vp8/encoder/quantize.h b/vp8/encoder/quantize.h index 868e8e3..b74718b 100644 --- a/vp8/encoder/quantize.h +++ b/vp8/encoder/quantize.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -16,6 +17,10 @@ #define prototype_quantize_block(sym) \ void (sym)(BLOCK *b,BLOCKD *d) +#if ARCH_X86 || ARCH_X86_64 +#include "x86/quantize_x86.h" +#endif + #if ARCH_ARM #include "arm/quantize_arm.h" #endif @@ -42,11 +47,10 @@ typedef struct #define QUANTIZE_INVOKE(ctx,fn) vp8_quantize_##fn #endif +extern void vp8_strict_quantize_b(BLOCK *b,BLOCKD *d); + extern void vp8_quantize_mb(MACROBLOCK *x); extern void vp8_quantize_mbuv(MACROBLOCK *x); extern void vp8_quantize_mby(MACROBLOCK *x); -extern void vp8_quantize_mbyrd(MACROBLOCK *x); -extern void vp8_quantize_mbuvrd(MACROBLOCK *x); -extern void vp8_quantize_mbrd(MACROBLOCK *x); #endif diff --git a/vp8/encoder/ratectrl.c b/vp8/encoder/ratectrl.c index 05040d3..50f4db0 100644 --- a/vp8/encoder/ratectrl.c +++ b/vp8/encoder/ratectrl.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -407,7 +408,7 @@ static void calc_gf_params(VP8_COMP *cpi) cpi->recent_ref_frame_usage[GOLDEN_FRAME] + cpi->recent_ref_frame_usage[ALTREF_FRAME]; - int pct_gf_active = (100 * cpi->common.gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols); + int pct_gf_active = (100 * cpi->gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols); // Reset the last boost indicator //cpi->last_boost = 100; @@ -1021,7 +1022,7 @@ void vp8_calc_pframe_target_size(VP8_COMP *cpi) cpi->recent_ref_frame_usage[GOLDEN_FRAME] + cpi->recent_ref_frame_usage[ALTREF_FRAME]; - int pct_gf_active = (100 * cpi->common.gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols); + int pct_gf_active = (100 * cpi->gf_active_count) / (cpi->common.mb_rows * cpi->common.mb_cols); // Reset the last boost indicator //cpi->last_boost = 100; @@ -1063,7 +1064,6 @@ void vp8_calc_pframe_target_size(VP8_COMP *cpi) if (cpi->common.refresh_golden_frame == TRUE) { - int isize_adjustment = 0; #if 0 if (0) // p_gw @@ -1119,10 +1119,13 @@ void vp8_calc_pframe_target_size(VP8_COMP *cpi) cpi->this_frame_target = (baseline_bits_at_q(1, Q, cpi->common.MBs) * cpi->last_boost) / 100; } - // If there is an active ARF at this location use the minimum bits on this frame + // If there is an active ARF at this location use the minimum + // bits on this frame even if it is a contructed arf. + // The active maximum quantizer insures that an appropriate + // number of bits will be spent if needed for contstructed ARFs. else { - cpi->this_frame_target = 0; // Minimial spend on gf that is replacing an arf + cpi->this_frame_target = 0; } cpi->current_gf_interval = cpi->frames_till_gf_update_due; @@ -1171,7 +1174,8 @@ void vp8_update_rate_correction_factors(VP8_COMP *cpi, int damp_var) while (Z > 0) { Z --; - projected_size_based_on_q *= (int)Factor; + projected_size_based_on_q = + (int)(Factor * projected_size_based_on_q); Factor += factor_adjustment; if (Factor >= 0.999) @@ -1362,7 +1366,8 @@ int vp8_regulate_q(VP8_COMP *cpi, int target_bits_per_frame) if (cpi->zbin_over_quant > zbin_oqmax) cpi->zbin_over_quant = zbin_oqmax; - bits_per_mb_at_this_q *= (int)Factor; // Each over-ruin step is assumed to equate to approximately 3% reduction in bitrate + // Adjust bits_per_mb_at_this_q estimate + bits_per_mb_at_this_q = (int)(Factor * bits_per_mb_at_this_q); Factor += factor_adjustment; if (Factor >= 0.999) @@ -1439,6 +1444,9 @@ void vp8_adjust_key_frame_context(VP8_COMP *cpi) } else { + int last_kf_interval = + (cpi->frames_since_key > 0) ? cpi->frames_since_key : 1; + // reset keyframe context and calculate weighted average of last KEY_FRAME_CONTEXT keyframes for (i = 0; i < KEY_FRAME_CONTEXT; i++) { @@ -1449,8 +1457,8 @@ void vp8_adjust_key_frame_context(VP8_COMP *cpi) } else { - cpi->prior_key_frame_size[KEY_FRAME_CONTEXT - 1] = cpi->projected_frame_size; - cpi->prior_key_frame_distance[KEY_FRAME_CONTEXT - 1] = cpi->frames_since_key; + cpi->prior_key_frame_size[i] = cpi->projected_frame_size; + cpi->prior_key_frame_distance[i] = last_kf_interval; } av_key_frame_bits += prior_key_frame_weight[i] * cpi->prior_key_frame_size[i]; diff --git a/vp8/encoder/ratectrl.h b/vp8/encoder/ratectrl.h index 588c7a8..766dfdf 100644 --- a/vp8/encoder/ratectrl.h +++ b/vp8/encoder/ratectrl.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/rdopt.c b/vp8/encoder/rdopt.c index 0846996..dbef85b 100644 --- a/vp8/encoder/rdopt.c +++ b/vp8/encoder/rdopt.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -63,11 +64,6 @@ void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x); #define MAXF(a,b) (((a) > (b)) ? (a) : (b)) -extern const TOKENEXTRA vp8_dct_value_tokens[DCT_MAX_VALUE*2]; -extern const TOKENEXTRA *vp8_dct_value_tokens_ptr; -extern int vp8_dct_value_cost[DCT_MAX_VALUE*2]; -extern int *vp8_dct_value_cost_ptr; - const int vp8_auto_speed_thresh[17] = { @@ -170,15 +166,13 @@ static void fill_token_costs( } -static int rd_iifactor [ 32 ] = { 16, 16, 16, 12, 8, 4, 2, 0, +static int rd_iifactor [ 32 ] = { 4, 4, 3, 2, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, }; - - // The values in this table should be reviewed static int sad_per_bit16lut[128] = { @@ -232,43 +226,41 @@ void vp8_initialize_rd_consts(VP8_COMP *cpi, int Qvalue) int i; int *thresh; int threshmult; - - int capped_q = (Qvalue < 160) ? Qvalue : 160; + double capped_q = (Qvalue < 160) ? (double)Qvalue : 160.0; + double rdconst = 3.00; vp8_clear_system_state(); //__asm emms; - cpi->RDMULT = (int)((0.00007 * (capped_q * capped_q * capped_q * capped_q)) - (0.0125 * (capped_q * capped_q * capped_q)) + - (2.25 * (capped_q * capped_q)) - (12.5 * capped_q) + 25.0); - - if (cpi->RDMULT < 25) - cpi->RDMULT = 25; - - if (cpi->pass == 2) - { - if (cpi->common.frame_type == KEY_FRAME) - cpi->RDMULT += (cpi->RDMULT * rd_iifactor[0]) / 16; - else if (cpi->next_iiratio > 31) - cpi->RDMULT += (cpi->RDMULT * rd_iifactor[31]) / 16; - else - cpi->RDMULT += (cpi->RDMULT * rd_iifactor[cpi->next_iiratio]) / 16; - } - + // Further tests required to see if optimum is different + // for key frames, golden frames and arf frames. + // if (cpi->common.refresh_golden_frame || + // cpi->common.refresh_alt_ref_frame) + cpi->RDMULT = (int)(rdconst * (capped_q * capped_q)); // Extend rate multiplier along side quantizer zbin increases if (cpi->zbin_over_quant > 0) { - // Extend rate multiplier along side quantizer zbin increases - if (cpi->zbin_over_quant > 0) - { - double oq_factor = pow(1.006, cpi->zbin_over_quant); - - if (oq_factor > (1.0 + ((double)cpi->zbin_over_quant / 64.0))) - oq_factor = (1.0 + (double)cpi->zbin_over_quant / 64.0); + double oq_factor; + double modq; + + // Experimental code using the same basic equation as used for Q above + // The units of cpi->zbin_over_quant are 1/128 of Q bin size + oq_factor = 1.0 + ((double)0.0015625 * cpi->zbin_over_quant); + modq = (int)((double)capped_q * oq_factor); + cpi->RDMULT = (int)(rdconst * (modq * modq)); + } - cpi->RDMULT *= (int)oq_factor; - } + if (cpi->pass == 2 && (cpi->common.frame_type != KEY_FRAME)) + { + if (cpi->next_iiratio > 31) + cpi->RDMULT += (cpi->RDMULT * rd_iifactor[31]) >> 4; + else + cpi->RDMULT += (cpi->RDMULT * rd_iifactor[cpi->next_iiratio]) >> 4; } + if (cpi->RDMULT < 125) + cpi->RDMULT = 125; + cpi->mb.errorperbit = (cpi->RDMULT / 100); if (cpi->mb.errorperbit < 1) @@ -494,7 +486,7 @@ static int macro_block_max_error(MACROBLOCK *mb) int i, j; int berror; - dc = !(mb->e_mbd.mbmi.mode == B_PRED || mb->e_mbd.mbmi.mode == SPLITMV); + dc = !(mb->e_mbd.mode_info_context->mbmi.mode == B_PRED || mb->e_mbd.mode_info_context->mbmi.mode == SPLITMV); for (i = 0; i < 16; i++) { @@ -622,24 +614,28 @@ int vp8_rdcost_mby(MACROBLOCK *mb) { int cost = 0; int b; - TEMP_CONTEXT t, t2; int type = 0; - MACROBLOCKD *x = &mb->e_mbd; + ENTROPY_CONTEXT_PLANES t_above, t_left; + ENTROPY_CONTEXT *ta; + ENTROPY_CONTEXT *tl; - vp8_setup_temp_context(&t, x->above_context[Y1CONTEXT], x->left_context[Y1CONTEXT], 4); - vp8_setup_temp_context(&t2, x->above_context[Y2CONTEXT], x->left_context[Y2CONTEXT], 1); + vpx_memcpy(&t_above, mb->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memcpy(&t_left, mb->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES)); - if (x->mbmi.mode == SPLITMV) + ta = (ENTROPY_CONTEXT *)&t_above; + tl = (ENTROPY_CONTEXT *)&t_left; + + if (x->mode_info_context->mbmi.mode == SPLITMV) type = 3; for (b = 0; b < 16; b++) cost += cost_coeffs(mb, x->block + b, type, - t.a + vp8_block2above[b], t.l + vp8_block2left[b]); + ta + vp8_block2above[b], tl + vp8_block2left[b]); - if (x->mbmi.mode != SPLITMV) + if (x->mode_info_context->mbmi.mode != SPLITMV) cost += cost_coeffs(mb, x->block + 24, 1, - t2.a + vp8_block2above[24], t2.l + vp8_block2left[24]); + ta + vp8_block2above[24], tl + vp8_block2left[24]); return cost; } @@ -718,13 +714,20 @@ int vp8_rd_pick_intra4x4mby_modes(VP8_COMP *cpi, MACROBLOCK *mb, int *Rate, int { MACROBLOCKD *const xd = &mb->e_mbd; int i; - TEMP_CONTEXT t; int cost = mb->mbmode_cost [xd->frame_type] [B_PRED]; int distortion = 0; int tot_rate_y = 0; + ENTROPY_CONTEXT_PLANES t_above, t_left; + ENTROPY_CONTEXT *ta; + ENTROPY_CONTEXT *tl; + + vpx_memcpy(&t_above, mb->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memcpy(&t_left, mb->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES)); + + ta = (ENTROPY_CONTEXT *)&t_above; + tl = (ENTROPY_CONTEXT *)&t_left; vp8_intra_prediction_down_copy(xd); - vp8_setup_temp_context(&t, xd->above_context[Y1CONTEXT], xd->left_context[Y1CONTEXT], 4); for (i = 0; i < 16; i++) { @@ -737,8 +740,8 @@ int vp8_rd_pick_intra4x4mby_modes(VP8_COMP *cpi, MACROBLOCK *mb, int *Rate, int rd_pick_intra4x4block( cpi, mb, mb->block + i, xd->block + i, &best_mode, A, L, - t.a + vp8_block2above[i], - t.l + vp8_block2left[i], &r, &ry, &d); + ta + vp8_block2above[i], + tl + vp8_block2left[i], &r, &ry, &d); cost += r; distortion += d; @@ -769,9 +772,9 @@ int vp8_rd_pick_intra16x16mby_mode(VP8_COMP *cpi, MACROBLOCK *x, int *Rate, int int dummy; rate = 0; - x->e_mbd.mbmi.mode = mode; + x->e_mbd.mode_info_context->mbmi.mode = mode; - rate += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mbmi.mode]; + rate += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.mode]; vp8_encode_intra16x16mbyrd(IF_RTCD(&cpi->rtcd), x); @@ -793,28 +796,33 @@ int vp8_rd_pick_intra16x16mby_mode(VP8_COMP *cpi, MACROBLOCK *x, int *Rate, int } } - x->e_mbd.mbmi.mode = mode_selected; + x->e_mbd.mode_info_context->mbmi.mode = mode_selected; return best_rd; } static int rd_cost_mbuv(MACROBLOCK *mb) { - TEMP_CONTEXT t, t2; int b; int cost = 0; MACROBLOCKD *x = &mb->e_mbd; + ENTROPY_CONTEXT_PLANES t_above, t_left; + ENTROPY_CONTEXT *ta; + ENTROPY_CONTEXT *tl; + + vpx_memcpy(&t_above, mb->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memcpy(&t_left, mb->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES)); - vp8_setup_temp_context(&t, x->above_context[UCONTEXT], x->left_context[UCONTEXT], 2); - vp8_setup_temp_context(&t2, x->above_context[VCONTEXT], x->left_context[VCONTEXT], 2); + ta = (ENTROPY_CONTEXT *)&t_above; + tl = (ENTROPY_CONTEXT *)&t_left; for (b = 16; b < 20; b++) cost += cost_coeffs(mb, x->block + b, vp8_block2type[b], - t.a + vp8_block2above[b], t.l + vp8_block2left[b]); + ta + vp8_block2above[b], tl + vp8_block2left[b]); for (b = 20; b < 24; b++) cost += cost_coeffs(mb, x->block + b, vp8_block2type[b], - t2.a + vp8_block2above[b], t2.l + vp8_block2left[b]); + ta + vp8_block2above[b], tl + vp8_block2left[b]); return cost; } @@ -855,11 +863,11 @@ int vp8_rd_pick_intra_mbuv_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate, int *ra int distortion; int this_rd; - x->e_mbd.mbmi.uv_mode = mode; + x->e_mbd.mode_info_context->mbmi.uv_mode = mode; vp8_encode_intra16x16mbuvrd(IF_RTCD(&cpi->rtcd), x); rate_to = rd_cost_mbuv(x); - rate = rate_to + x->intra_uv_mode_cost[x->e_mbd.frame_type][x->e_mbd.mbmi.uv_mode]; + rate = rate_to + x->intra_uv_mode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.uv_mode]; distortion = vp8_get_mbuvrecon_error(IF_RTCD(&cpi->rtcd.variance), x); @@ -878,7 +886,7 @@ int vp8_rd_pick_intra_mbuv_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate, int *ra *rate = r; *distortion = d; - x->e_mbd.mbmi.uv_mode = mode_selected; + x->e_mbd.mode_info_context->mbmi.uv_mode = mode_selected; return best_rd; } #endif @@ -888,16 +896,17 @@ int vp8_cost_mv_ref(MB_PREDICTION_MODE m, const int near_mv_ref_ct[4]) vp8_prob p [VP8_MVREFS-1]; assert(NEARESTMV <= m && m <= SPLITMV); vp8_mv_ref_probs(p, near_mv_ref_ct); - return vp8_cost_token(vp8_mv_ref_tree, p, VP8_MVREFENCODINGS + m); + return vp8_cost_token(vp8_mv_ref_tree, p, + vp8_mv_ref_encoding_array - NEARESTMV + m); } void vp8_set_mbmode_and_mvs(MACROBLOCK *x, MB_PREDICTION_MODE mb, MV *mv) { int i; - x->e_mbd.mbmi.mode = mb; - x->e_mbd.mbmi.mv.as_mv.row = mv->row; - x->e_mbd.mbmi.mv.as_mv.col = mv->col; + x->e_mbd.mode_info_context->mbmi.mode = mb; + x->e_mbd.mode_info_context->mbmi.mv.as_mv.row = mv->row; + x->e_mbd.mode_info_context->mbmi.mv.as_mv.col = mv->col; for (i = 0; i < 16; i++) { @@ -1002,18 +1011,19 @@ static int labels2mode( return cost; } -static int rdcost_mbsegment_y(MACROBLOCK *mb, const int *labels, int which_label, TEMP_CONTEXT *t) +static int rdcost_mbsegment_y(MACROBLOCK *mb, const int *labels, + int which_label, ENTROPY_CONTEXT *ta, + ENTROPY_CONTEXT *tl) { int cost = 0; int b; MACROBLOCKD *x = &mb->e_mbd; - for (b = 0; b < 16; b++) if (labels[ b] == which_label) cost += cost_coeffs(mb, x->block + b, 3, - t->a + vp8_block2above[b], - t->l + vp8_block2left[b]); + ta + vp8_block2above[b], + tl + vp8_block2left[b]); return cost; @@ -1033,11 +1043,11 @@ static unsigned int vp8_encode_inter_mb_segment(MACROBLOCK *x, int const *labels vp8_build_inter_predictors_b(bd, 16, x->e_mbd.subpixel_predict); ENCODEMB_INVOKE(rtcd, subb)(be, bd, 16); - x->short_fdct4x4rd(be->src_diff, be->coeff, 32); + x->vp8_short_fdct4x4(be->src_diff, be->coeff, 32); // set to 0 no way to account for 2nd order DC so discount //be->coeff[0] = 0; - x->quantize_brd(be, bd); + x->quantize_b(be, bd); distortion += ENCODEMB_INVOKE(rtcd, berr)(be->coeff, bd->dqcoeff); } @@ -1061,13 +1071,13 @@ static void macro_block_yrd(MACROBLOCK *mb, int *Rate, int *Distortion, const vp // Fdct and building the 2nd order block for (beptr = mb->block; beptr < mb->block + 16; beptr += 2) { - mb->short_fdct8x4rd(beptr->src_diff, beptr->coeff, 32); + mb->vp8_short_fdct8x4(beptr->src_diff, beptr->coeff, 32); *Y2DCPtr++ = beptr->coeff[0]; *Y2DCPtr++ = beptr->coeff[16]; } // 2nd order fdct - if (x->mbmi.mode != SPLITMV) + if (x->mode_info_context->mbmi.mode != SPLITMV) { mb->short_walsh4x4(mb_y2->src_diff, mb_y2->coeff, 8); } @@ -1075,20 +1085,20 @@ static void macro_block_yrd(MACROBLOCK *mb, int *Rate, int *Distortion, const vp // Quantization for (b = 0; b < 16; b++) { - mb->quantize_brd(&mb->block[b], &mb->e_mbd.block[b]); + mb->quantize_b(&mb->block[b], &mb->e_mbd.block[b]); } // DC predication and Quantization of 2nd Order block - if (x->mbmi.mode != SPLITMV) + if (x->mode_info_context->mbmi.mode != SPLITMV) { { - mb->quantize_brd(mb_y2, x_y2); + mb->quantize_b(mb_y2, x_y2); } } // Distortion - if (x->mbmi.mode == SPLITMV) + if (x->mode_info_context->mbmi.mode == SPLITMV) d = ENCODEMB_INVOKE(rtcd, mberr)(mb, 0) << 2; else { @@ -1130,6 +1140,9 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, MV *bes MV bmvs[16]; int beobs[16]; + vpx_memset(beobs, 0, sizeof(beobs)); + + for (segmentation = 0; segmentation < VP8_NUMMBSPLITS; segmentation++) { int label_count; @@ -1143,9 +1156,20 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, MV *bes vp8_variance_fn_ptr_t v_fn_ptr; - TEMP_CONTEXT t; - TEMP_CONTEXT tb; - vp8_setup_temp_context(&t, xc->above_context[Y1CONTEXT], xc->left_context[Y1CONTEXT], 4); + ENTROPY_CONTEXT_PLANES t_above, t_left; + ENTROPY_CONTEXT *ta; + ENTROPY_CONTEXT *tl; + ENTROPY_CONTEXT_PLANES t_above_b, t_left_b; + ENTROPY_CONTEXT *ta_b; + ENTROPY_CONTEXT *tl_b; + + vpx_memcpy(&t_above, x->e_mbd.above_context, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memcpy(&t_left, x->e_mbd.left_context, sizeof(ENTROPY_CONTEXT_PLANES)); + + ta = (ENTROPY_CONTEXT *)&t_above; + tl = (ENTROPY_CONTEXT *)&t_left; + ta_b = (ENTROPY_CONTEXT *)&t_above_b; + tl_b = (ENTROPY_CONTEXT *)&t_left_b; br = 0; bd = 0; @@ -1230,9 +1254,15 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, MV *bes int this_rd; int num00; int labelyrate; + ENTROPY_CONTEXT_PLANES t_above_s, t_left_s; + ENTROPY_CONTEXT *ta_s; + ENTROPY_CONTEXT *tl_s; + + vpx_memcpy(&t_above_s, &t_above, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memcpy(&t_left_s, &t_left, sizeof(ENTROPY_CONTEXT_PLANES)); - TEMP_CONTEXT ts; - vp8_setup_temp_context(&ts, &t.a[0], &t.l[0], 4); + ta_s = (ENTROPY_CONTEXT *)&t_above_s; + tl_s = (ENTROPY_CONTEXT *)&t_left_s; if (this_mode == NEW4X4) { @@ -1317,7 +1347,7 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, MV *bes distortion = vp8_encode_inter_mb_segment(x, labels, i, IF_RTCD(&cpi->rtcd.encodemb)) / 4; - labelyrate = rdcost_mbsegment_y(x, labels, i, &ts); + labelyrate = rdcost_mbsegment_y(x, labels, i, ta_s, tl_s); rate += labelyrate; this_rd = RDFUNC(x->rdmult, x->rddiv, rate, distortion, cpi->target_bits_per_mb); @@ -1329,12 +1359,15 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, MV *bes bestlabelyrate = labelyrate; mode_selected = this_mode; best_label_rd = this_rd; - vp8_setup_temp_context(&tb, &ts.a[0], &ts.l[0], 4); + + vpx_memcpy(ta_b, ta_s, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memcpy(tl_b, tl_s, sizeof(ENTROPY_CONTEXT_PLANES)); } } - vp8_setup_temp_context(&t, &tb.a[0], &tb.l[0], 4); + vpx_memcpy(ta, ta_b, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memcpy(tl, tl_b, sizeof(ENTROPY_CONTEXT_PLANES)); labels2mode(x, labels, i, mode_selected, &mode_mv[mode_selected], best_ref_mv, mvcost); @@ -1405,10 +1438,10 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, MV *bes // save partitions labels = vp8_mbsplits[best_seg]; - x->e_mbd.mbmi.partitioning = best_seg; - x->e_mbd.mbmi.partition_count = vp8_count_labels(labels); + x->e_mbd.mode_info_context->mbmi.partitioning = best_seg; + x->partition_info->count = vp8_count_labels(labels); - for (i = 0; i < x->e_mbd.mbmi.partition_count; i++) + for (i = 0; i < x->partition_info->count; i++) { int j; @@ -1418,8 +1451,8 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, MV *bes break; } - x->e_mbd.mbmi.partition_bmi[i].mode = x->e_mbd.block[j].bmi.mode; - x->e_mbd.mbmi.partition_bmi[i].mv.as_mv = x->e_mbd.block[j].bmi.mv.as_mv; + x->partition_info->bmi[i].mode = x->e_mbd.block[j].bmi.mode; + x->partition_info->bmi[i].mv.as_mv = x->e_mbd.block[j].bmi.mv.as_mv; } return best_segment_rd; @@ -1433,6 +1466,7 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int MACROBLOCKD *xd = &x->e_mbd; B_MODE_INFO best_bmodes[16]; MB_MODE_INFO best_mbmode; + PARTITION_INFO best_partition; MV best_ref_mv; MV mode_mv[MB_MODE_COUNT]; MB_PREDICTION_MODE this_mode; @@ -1464,6 +1498,8 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int *returnintra = INT_MAX; + vpx_memset(&best_mbmode, 0, sizeof(best_mbmode)); // clean + cpi->mbs_tested_so_far++; // Count of the number of MBs tested so far this frame x->skip = 0; @@ -1517,9 +1553,9 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int vpx_memset(mode_mv, 0, sizeof(mode_mv)); - x->e_mbd.mbmi.ref_frame = INTRA_FRAME; + x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME; vp8_rd_pick_intra_mbuv_mode(cpi, x, &uv_intra_rate, &uv_intra_rate_tokenonly, &uv_intra_distortion); - uv_intra_mode = x->e_mbd.mbmi.uv_mode; + uv_intra_mode = x->e_mbd.mode_info_context->mbmi.uv_mode; { uvintra_eob = 0; @@ -1541,7 +1577,7 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int //all_rds[mode_index] = -1; //all_rates[mode_index] = -1; //all_dist[mode_index] = -1; - //intermodecost[mode_index] = -1; + //intermodecost[mode_index] = -1; // Test best rd so far against threshold for trying this mode. if (best_rd <= cpi->rd_threshes[mode_index]) @@ -1563,31 +1599,34 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int this_mode = vp8_mode_order[mode_index]; - x->e_mbd.mbmi.mode = this_mode; - x->e_mbd.mbmi.uv_mode = DC_PRED; - x->e_mbd.mbmi.ref_frame = vp8_ref_frame_order[mode_index]; + x->e_mbd.mode_info_context->mbmi.mode = this_mode; + x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED; + x->e_mbd.mode_info_context->mbmi.ref_frame = vp8_ref_frame_order[mode_index]; //Only consider ZEROMV/ALTREF_FRAME for alt ref frame. if (cpi->is_src_frame_alt_ref) { - if (this_mode != ZEROMV || x->e_mbd.mbmi.ref_frame != ALTREF_FRAME) + if (this_mode != ZEROMV || x->e_mbd.mode_info_context->mbmi.ref_frame != ALTREF_FRAME) continue; } - if (x->e_mbd.mbmi.ref_frame == LAST_FRAME) + if (x->e_mbd.mode_info_context->mbmi.ref_frame == LAST_FRAME) { + YV12_BUFFER_CONFIG *lst_yv12 = &cpi->common.yv12_fb[cpi->common.lst_fb_idx]; + if (!(cpi->ref_frame_flags & VP8_LAST_FLAG)) continue; lf_or_gf = 0; // Local last frame vs Golden frame flag // Set up pointers for this macro block into the previous frame recon buffer - x->e_mbd.pre.y_buffer = cpi->common.last_frame.y_buffer + recon_yoffset; - x->e_mbd.pre.u_buffer = cpi->common.last_frame.u_buffer + recon_uvoffset; - x->e_mbd.pre.v_buffer = cpi->common.last_frame.v_buffer + recon_uvoffset; + x->e_mbd.pre.y_buffer = lst_yv12->y_buffer + recon_yoffset; + x->e_mbd.pre.u_buffer = lst_yv12->u_buffer + recon_uvoffset; + x->e_mbd.pre.v_buffer = lst_yv12->v_buffer + recon_uvoffset; } - else if (x->e_mbd.mbmi.ref_frame == GOLDEN_FRAME) + else if (x->e_mbd.mode_info_context->mbmi.ref_frame == GOLDEN_FRAME) { + YV12_BUFFER_CONFIG *gld_yv12 = &cpi->common.yv12_fb[cpi->common.gld_fb_idx]; // not supposed to reference gold frame if (!(cpi->ref_frame_flags & VP8_GOLD_FLAG)) @@ -1596,12 +1635,14 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int lf_or_gf = 1; // Local last frame vs Golden frame flag // Set up pointers for this macro block into the previous frame recon buffer - x->e_mbd.pre.y_buffer = cpi->common.golden_frame.y_buffer + recon_yoffset; - x->e_mbd.pre.u_buffer = cpi->common.golden_frame.u_buffer + recon_uvoffset; - x->e_mbd.pre.v_buffer = cpi->common.golden_frame.v_buffer + recon_uvoffset; + x->e_mbd.pre.y_buffer = gld_yv12->y_buffer + recon_yoffset; + x->e_mbd.pre.u_buffer = gld_yv12->u_buffer + recon_uvoffset; + x->e_mbd.pre.v_buffer = gld_yv12->v_buffer + recon_uvoffset; } - else if (x->e_mbd.mbmi.ref_frame == ALTREF_FRAME) + else if (x->e_mbd.mode_info_context->mbmi.ref_frame == ALTREF_FRAME) { + YV12_BUFFER_CONFIG *alt_yv12 = &cpi->common.yv12_fb[cpi->common.alt_fb_idx]; + // not supposed to reference alt ref frame if (!(cpi->ref_frame_flags & VP8_ALT_FLAG)) continue; @@ -1612,19 +1653,19 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int lf_or_gf = 1; // Local last frame vs Golden frame flag // Set up pointers for this macro block into the previous frame recon buffer - x->e_mbd.pre.y_buffer = cpi->common.alt_ref_frame.y_buffer + recon_yoffset; - x->e_mbd.pre.u_buffer = cpi->common.alt_ref_frame.u_buffer + recon_uvoffset; - x->e_mbd.pre.v_buffer = cpi->common.alt_ref_frame.v_buffer + recon_uvoffset; + x->e_mbd.pre.y_buffer = alt_yv12->y_buffer + recon_yoffset; + x->e_mbd.pre.u_buffer = alt_yv12->u_buffer + recon_uvoffset; + x->e_mbd.pre.v_buffer = alt_yv12->v_buffer + recon_uvoffset; } vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &mode_mv[NEARESTMV], &mode_mv[NEARMV], &best_ref_mv, - mdcounts, x->e_mbd.mbmi.ref_frame, cpi->common.ref_frame_sign_bias); + mdcounts, x->e_mbd.mode_info_context->mbmi.ref_frame, cpi->common.ref_frame_sign_bias); // Estimate the reference frame signaling cost and add it to the rolling cost variable. - frame_cost = ref_frame_cost[x->e_mbd.mbmi.ref_frame]; + frame_cost = ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame]; rate2 += frame_cost; if (this_mode <= B_PRED) @@ -1691,9 +1732,9 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int int breakout_rd = best_rd - frame_cost_rd; int tmp_rd; - if (x->e_mbd.mbmi.ref_frame == LAST_FRAME) + if (x->e_mbd.mode_info_context->mbmi.ref_frame == LAST_FRAME) tmp_rd = vp8_rd_pick_best_mbsegmentation(cpi, x, &best_ref_mv, breakout_rd, mdcounts, &rate, &rate_y, &distortion, cpi->compressor_speed, x->mvcost, cpi->rd_threshes[THR_NEWMV], cpi->common.full_pixel) ; - else if (x->e_mbd.mbmi.ref_frame == GOLDEN_FRAME) + else if (x->e_mbd.mode_info_context->mbmi.ref_frame == GOLDEN_FRAME) tmp_rd = vp8_rd_pick_best_mbsegmentation(cpi, x, &best_ref_mv, breakout_rd, mdcounts, &rate, &rate_y, &distortion, cpi->compressor_speed, x->mvcost, cpi->rd_threshes[THR_NEWG], cpi->common.full_pixel) ; else tmp_rd = vp8_rd_pick_best_mbsegmentation(cpi, x, &best_ref_mv, breakout_rd, mdcounts, &rate, &rate_y, &distortion, cpi->compressor_speed, x->mvcost, cpi->rd_threshes[THR_NEWA], cpi->common.full_pixel) ; @@ -1747,19 +1788,19 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int } // trap cases where the 8x8s can be promoted to 8x16s or 16x8s - if (0)//x->e_mbd.mbmi.partition_count == 4) + if (0)//x->partition_info->count == 4) { - if (x->e_mbd.mbmi.partition_bmi[0].mv.as_int == x->e_mbd.mbmi.partition_bmi[1].mv.as_int - && x->e_mbd.mbmi.partition_bmi[2].mv.as_int == x->e_mbd.mbmi.partition_bmi[3].mv.as_int) + if (x->partition_info->bmi[0].mv.as_int == x->partition_info->bmi[1].mv.as_int + && x->partition_info->bmi[2].mv.as_int == x->partition_info->bmi[3].mv.as_int) { const int *labels = vp8_mbsplits[2]; - x->e_mbd.mbmi.partitioning = 0; + x->e_mbd.mode_info_context->mbmi.partitioning = 0; rate -= vp8_cost_token(vp8_mbsplit_tree, vp8_mbsplit_probs, vp8_mbsplit_encodings + 2); rate += vp8_cost_token(vp8_mbsplit_tree, vp8_mbsplit_probs, vp8_mbsplit_encodings); - //rate -= x->inter_bmode_costs[ x->e_mbd.mbmi.partition_bmi[1]]; - //rate -= x->inter_bmode_costs[ x->e_mbd.mbmi.partition_bmi[3]]; - x->e_mbd.mbmi.partition_bmi[1] = x->e_mbd.mbmi.partition_bmi[2]; + //rate -= x->inter_bmode_costs[ x->partition_info->bmi[1]]; + //rate -= x->inter_bmode_costs[ x->partition_info->bmi[3]]; + x->partition_info->bmi[1] = x->partition_info->bmi[2]; } } @@ -1769,14 +1810,14 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int case V_PRED: case H_PRED: case TM_PRED: - x->e_mbd.mbmi.ref_frame = INTRA_FRAME; + x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME; vp8_build_intra_predictors_mby_ptr(&x->e_mbd); { macro_block_yrd(x, &rate, &distortion, IF_RTCD(&cpi->rtcd.encodemb)) ; rate2 += rate; rate_y = rate; distortion2 += distortion; - rate2 += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mbmi.mode]; + rate2 += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.mode]; rate2 += uv_intra_rate; rate_uv = uv_intra_rate_tokenonly; distortion2 += uv_intra_distortion; @@ -2082,7 +2123,7 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int //all_rates[mode_index] = rate2; //all_dist[mode_index] = distortion2; - if ((x->e_mbd.mbmi.ref_frame == INTRA_FRAME) && (this_rd < *returnintra)) + if ((x->e_mbd.mode_info_context->mbmi.ref_frame == INTRA_FRAME) && (this_rd < *returnintra)) { *returnintra = this_rd ; } @@ -2092,17 +2133,18 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int { // Note index of best mode so far best_mode_index = mode_index; - x->e_mbd.mbmi.force_no_skip = force_no_skip; + x->e_mbd.mode_info_context->mbmi.force_no_skip = force_no_skip; if (this_mode <= B_PRED) { - x->e_mbd.mbmi.uv_mode = uv_intra_mode; + x->e_mbd.mode_info_context->mbmi.uv_mode = uv_intra_mode; } *returnrate = rate2; *returndistortion = distortion2; best_rd = this_rd; - vpx_memcpy(&best_mbmode, &x->e_mbd.mbmi, sizeof(MB_MODE_INFO)); + vpx_memcpy(&best_mbmode, &x->e_mbd.mode_info_context->mbmi, sizeof(MB_MODE_INFO)); + vpx_memcpy(&best_partition, x->partition_info, sizeof(PARTITION_INFO)); for (i = 0; i < 16; i++) { @@ -2183,28 +2225,30 @@ int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int best_mbmode.partitioning = 0; best_mbmode.dc_diff = 0; - vpx_memcpy(&x->e_mbd.mbmi, &best_mbmode, sizeof(MB_MODE_INFO)); + vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO)); + vpx_memcpy(x->partition_info, &best_partition, sizeof(PARTITION_INFO)); for (i = 0; i < 16; i++) { vpx_memset(&x->e_mbd.block[i].bmi, 0, sizeof(B_MODE_INFO)); } - x->e_mbd.mbmi.mv.as_int = 0; + x->e_mbd.mode_info_context->mbmi.mv.as_int = 0; return best_rd; } // macroblock modes - vpx_memcpy(&x->e_mbd.mbmi, &best_mbmode, sizeof(MB_MODE_INFO)); + vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO)); + vpx_memcpy(x->partition_info, &best_partition, sizeof(PARTITION_INFO)); for (i = 0; i < 16; i++) { vpx_memcpy(&x->e_mbd.block[i].bmi, &best_bmodes[i], sizeof(B_MODE_INFO)); } - x->e_mbd.mbmi.mv.as_mv = x->e_mbd.block[15].bmi.mv.as_mv; + x->e_mbd.mode_info_context->mbmi.mv.as_mv = x->e_mbd.block[15].bmi.mv.as_mv; return best_rd; } diff --git a/vp8/encoder/rdopt.h b/vp8/encoder/rdopt.h index c6eae4b..fb74dd4 100644 --- a/vp8/encoder/rdopt.h +++ b/vp8/encoder/rdopt.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/sad_c.c b/vp8/encoder/sad_c.c index 74c6bd7..e63be2b 100644 --- a/vp8/encoder/sad_c.c +++ b/vp8/encoder/sad_c.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/common/segmentation_common.c b/vp8/encoder/segmentation.c index 72b8c87..fc0967d 100644 --- a/vp8/common/segmentation_common.c +++ b/vp8/encoder/segmentation.c @@ -1,29 +1,30 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ -#include "segmentation_common.h" +#include "segmentation.h" #include "vpx_mem/vpx_mem.h" -void vp8_update_gf_useage_maps(VP8_COMMON *cm, MACROBLOCKD *xd) +void vp8_update_gf_useage_maps(VP8_COMP *cpi, VP8_COMMON *cm, MACROBLOCK *x) { int mb_row, mb_col; MODE_INFO *this_mb_mode_info = cm->mi; - xd->gf_active_ptr = (signed char *)cm->gf_active_flags; + x->gf_active_ptr = (signed char *)cpi->gf_active_flags; if ((cm->frame_type == KEY_FRAME) || (cm->refresh_golden_frame)) { // Reset Gf useage monitors - vpx_memset(cm->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols)); - cm->gf_active_count = cm->mb_rows * cm->mb_cols; + vpx_memset(cpi->gf_active_flags, 1, (cm->mb_rows * cm->mb_cols)); + cpi->gf_active_count = cm->mb_rows * cm->mb_cols; } else { @@ -39,19 +40,19 @@ void vp8_update_gf_useage_maps(VP8_COMMON *cm, MACROBLOCKD *xd) // else if using non 0,0 motion or intra modes then clear flag if it is currently set if ((this_mb_mode_info->mbmi.ref_frame == GOLDEN_FRAME) || (this_mb_mode_info->mbmi.ref_frame == ALTREF_FRAME)) { - if (*(xd->gf_active_ptr) == 0) + if (*(x->gf_active_ptr) == 0) { - *(xd->gf_active_ptr) = 1; - cm->gf_active_count ++; + *(x->gf_active_ptr) = 1; + cpi->gf_active_count ++; } } - else if ((this_mb_mode_info->mbmi.mode != ZEROMV) && *(xd->gf_active_ptr)) + else if ((this_mb_mode_info->mbmi.mode != ZEROMV) && *(x->gf_active_ptr)) { - *(xd->gf_active_ptr) = 0; - cm->gf_active_count--; + *(x->gf_active_ptr) = 0; + cpi->gf_active_count--; } - xd->gf_active_ptr++; // Step onto next entry + x->gf_active_ptr++; // Step onto next entry this_mb_mode_info++; // skip to next mb } diff --git a/vp8/encoder/segmentation.h b/vp8/encoder/segmentation.h new file mode 100644 index 0000000..216e194 --- /dev/null +++ b/vp8/encoder/segmentation.h @@ -0,0 +1,16 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + + +#include "string.h" +#include "blockd.h" +#include "onyx_int.h" + +extern void vp8_update_gf_useage_maps(VP8_COMP *cpi, VP8_COMMON *cm, MACROBLOCK *x); diff --git a/vp8/encoder/ssim.c b/vp8/encoder/ssim.c index df214a8..4ebcba1 100644 --- a/vp8/encoder/ssim.c +++ b/vp8/encoder/ssim.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/tokenize.c b/vp8/encoder/tokenize.c index 33ddd64..d9b8d36 100644 --- a/vp8/encoder/tokenize.c +++ b/vp8/encoder/tokenize.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -23,12 +24,12 @@ _int64 context_counters[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens]; #endif void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) ; -void vp8_fix_contexts(VP8_COMP *cpi, MACROBLOCKD *x); +void vp8_fix_contexts(MACROBLOCKD *x); TOKENEXTRA vp8_dct_value_tokens[DCT_MAX_VALUE*2]; -TOKENEXTRA *vp8_dct_value_tokens_ptr; +const TOKENEXTRA *vp8_dct_value_tokens_ptr; int vp8_dct_value_cost[DCT_MAX_VALUE*2]; -int *vp8_dct_value_cost_ptr; +const int *vp8_dct_value_cost_ptr; #if 0 int skip_true_count = 0; int skip_false_count = 0; @@ -196,86 +197,18 @@ static void tokenize1st_order_b *a = *l = pt; } -#if 0 -void vp8_tokenize_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) -{ - //int i; - ENTROPY_CONTEXT **const A = x->above_context; - ENTROPY_CONTEXT(* const L)[4] = x->left_context; - int plane_type; - int b; - - TOKENEXTRA *start = *t; - TOKENEXTRA *tp = *t; - - x->mbmi.dc_diff = 1; - - vpx_memcpy(cpi->coef_counts_backup, cpi->coef_counts, sizeof(cpi->coef_counts)); - - if (x->mbmi.mode == B_PRED || x->mbmi.mode == SPLITMV) - { - plane_type = 3; - } - else - { - tokenize2nd_order_b(x->block + 24, t, 1, x->frame_type, - A[Y2CONTEXT] + vp8_block2above[24], L[Y2CONTEXT] + vp8_block2left[24], cpi); - plane_type = 0; - - } - - for (b = 0; b < 16; b++) - tokenize1st_order_b(x->block + b, t, plane_type, x->frame_type, - A[vp8_block2context[b]] + vp8_block2above[b], - L[vp8_block2context[b]] + vp8_block2left[b], cpi); - - for (b = 16; b < 24; b++) - tokenize1st_order_b(x->block + b, t, 2, x->frame_type, - A[vp8_block2context[b]] + vp8_block2above[b], - L[vp8_block2context[b]] + vp8_block2left[b], cpi); - - if (cpi->common.mb_no_coeff_skip) - { - x->mbmi.mb_skip_coeff = 1; - while ((tp != *t) && x->mbmi.mb_skip_coeff) - { - x->mbmi.mb_skip_coeff = (x->mbmi.mb_skip_coeff && (tp->Token == DCT_EOB_TOKEN)); - tp ++; - } - - if (x->mbmi.mb_skip_coeff == 1) - { - x->mbmi.dc_diff = 0; - //redo the coutnts - vpx_memcpy(cpi->coef_counts, cpi->coef_counts_backup, sizeof(cpi->coef_counts)); - - *t = start; - cpi->skip_true_count++; - - //skip_true_count++; - } - else - { - - cpi->skip_false_count++; - //skip_false_count++; - } - } -} -#else void vp8_tokenize_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) { - //int i; - ENTROPY_CONTEXT **const A = x->above_context; - ENTROPY_CONTEXT(* const L)[4] = x->left_context; + ENTROPY_CONTEXT * A = (ENTROPY_CONTEXT *)x->above_context; + ENTROPY_CONTEXT * L = (ENTROPY_CONTEXT *)x->left_context; int plane_type; int b; TOKENEXTRA *start = *t; TOKENEXTRA *tp = *t; - x->mbmi.dc_diff = 1; + x->mode_info_context->mbmi.dc_diff = 1; #if 0 @@ -290,7 +223,7 @@ void vp8_tokenize_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) #if 1 - if (x->mbmi.mb_skip_coeff) + if (x->mode_info_context->mbmi.mb_skip_coeff) { cpi->skip_true_count++; @@ -299,13 +232,13 @@ void vp8_tokenize_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) vp8_stuff_mb(cpi, x, t) ; else { - vp8_fix_contexts(cpi, x); + vp8_fix_contexts(x); } - if (x->mbmi.mode != B_PRED && x->mbmi.mode != SPLITMV) - x->mbmi.dc_diff = 0; + if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV) + x->mode_info_context->mbmi.dc_diff = 0; else - x->mbmi.dc_diff = 1; + x->mode_info_context->mbmi.dc_diff = 1; return; @@ -346,27 +279,27 @@ void vp8_tokenize_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) vpx_memcpy(cpi->coef_counts_backup, cpi->coef_counts, sizeof(cpi->coef_counts)); #endif - if (x->mbmi.mode == B_PRED || x->mbmi.mode == SPLITMV) + if (x->mode_info_context->mbmi.mode == B_PRED || x->mode_info_context->mbmi.mode == SPLITMV) { plane_type = 3; } else { tokenize2nd_order_b(x->block + 24, t, 1, x->frame_type, - A[Y2CONTEXT] + vp8_block2above[24], L[Y2CONTEXT] + vp8_block2left[24], cpi); + A + vp8_block2above[24], L + vp8_block2left[24], cpi); plane_type = 0; } for (b = 0; b < 16; b++) tokenize1st_order_b(x->block + b, t, plane_type, x->frame_type, - A[vp8_block2context[b]] + vp8_block2above[b], - L[vp8_block2context[b]] + vp8_block2left[b], cpi); + A + vp8_block2above[b], + L + vp8_block2left[b], cpi); for (b = 16; b < 24; b++) tokenize1st_order_b(x->block + b, t, 2, x->frame_type, - A[vp8_block2context[b]] + vp8_block2above[b], - L[vp8_block2context[b]] + vp8_block2left[b], cpi); + A + vp8_block2above[b], + L + vp8_block2left[b], cpi); #if 0 @@ -405,7 +338,7 @@ void vp8_tokenize_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) #endif } -#endif + #ifdef ENTROPY_STATS @@ -580,57 +513,45 @@ void stuff1st_order_buv void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) { - //int i; - ENTROPY_CONTEXT **const A = x->above_context; - ENTROPY_CONTEXT(* const L)[4] = x->left_context; + ENTROPY_CONTEXT * A = (ENTROPY_CONTEXT *)x->above_context; + ENTROPY_CONTEXT * L = (ENTROPY_CONTEXT *)x->left_context; int plane_type; int b; stuff2nd_order_b(x->block + 24, t, 1, x->frame_type, - A[Y2CONTEXT] + vp8_block2above[24], L[Y2CONTEXT] + vp8_block2left[24], cpi); + A + vp8_block2above[24], L + vp8_block2left[24], cpi); plane_type = 0; - if (x->mbmi.mode != B_PRED && x->mbmi.mode != SPLITMV) - x->mbmi.dc_diff = 0; + if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV) + x->mode_info_context->mbmi.dc_diff = 0; else - x->mbmi.dc_diff = 1; + x->mode_info_context->mbmi.dc_diff = 1; for (b = 0; b < 16; b++) stuff1st_order_b(x->block + b, t, plane_type, x->frame_type, - A[vp8_block2context[b]] + vp8_block2above[b], - L[vp8_block2context[b]] + vp8_block2left[b], cpi); + A + vp8_block2above[b], + L + vp8_block2left[b], cpi); for (b = 16; b < 24; b++) stuff1st_order_buv(x->block + b, t, 2, x->frame_type, - A[vp8_block2context[b]] + vp8_block2above[b], - L[vp8_block2context[b]] + vp8_block2left[b], cpi); + A + vp8_block2above[b], + L + vp8_block2left[b], cpi); } -void vp8_fix_contexts(VP8_COMP *cpi, MACROBLOCKD *x) +void vp8_fix_contexts(MACROBLOCKD *x) { - x->left_context[Y1CONTEXT][0] = 0; - x->left_context[Y1CONTEXT][1] = 0; - x->left_context[Y1CONTEXT][2] = 0; - x->left_context[Y1CONTEXT][3] = 0; - x->left_context[UCONTEXT][0] = 0; - x->left_context[VCONTEXT][0] = 0; - x->left_context[UCONTEXT][1] = 0; - x->left_context[VCONTEXT][1] = 0; - - x->above_context[Y1CONTEXT][0] = 0; - x->above_context[Y1CONTEXT][1] = 0; - x->above_context[Y1CONTEXT][2] = 0; - x->above_context[Y1CONTEXT][3] = 0; - x->above_context[UCONTEXT][0] = 0; - x->above_context[VCONTEXT][0] = 0; - x->above_context[UCONTEXT][1] = 0; - x->above_context[VCONTEXT][1] = 0; - - if (x->mbmi.mode != B_PRED && x->mbmi.mode != SPLITMV) + /* Clear entropy contexts for Y2 blocks */ + if (x->mode_info_context->mbmi.mode != B_PRED && x->mode_info_context->mbmi.mode != SPLITMV) { - x->left_context[Y2CONTEXT][0] = 0; - x->above_context[Y2CONTEXT][0] = 0; + vpx_memset(x->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)); + vpx_memset(x->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)); } + else + { + vpx_memset(x->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1); + vpx_memset(x->left_context, 0, sizeof(ENTROPY_CONTEXT_PLANES)-1); + } + } diff --git a/vp8/encoder/tokenize.h b/vp8/encoder/tokenize.h index 02aacc2..7b9fc9e 100644 --- a/vp8/encoder/tokenize.h +++ b/vp8/encoder/tokenize.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -34,5 +35,11 @@ void print_context_counters(); extern _int64 context_counters[BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens]; #endif +extern const int *vp8_dct_value_cost_ptr; +/* TODO: The Token field should be broken out into a separate char array to + * improve cache locality, since it's needed for costing when the rest of the + * fields are not. + */ +extern const TOKENEXTRA *vp8_dct_value_tokens_ptr; #endif /* tokenize_h */ diff --git a/vp8/encoder/treewriter.c b/vp8/encoder/treewriter.c index e398044..03967c8 100644 --- a/vp8/encoder/treewriter.c +++ b/vp8/encoder/treewriter.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/treewriter.h b/vp8/encoder/treewriter.h index 05ac74c..88096d8 100644 --- a/vp8/encoder/treewriter.h +++ b/vp8/encoder/treewriter.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/variance.h b/vp8/encoder/variance.h index b1aede2..0341fbd 100644 --- a/vp8/encoder/variance.h +++ b/vp8/encoder/variance.h @@ -1,13 +1,13 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ -#include "config.h" #ifndef VARIANCE_H #define VARIANCE_H diff --git a/vp8/encoder/variance_c.c b/vp8/encoder/variance_c.c index 85269b9..179cd0d 100644 --- a/vp8/encoder/variance_c.c +++ b/vp8/encoder/variance_c.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/x86/csystemdependent.c b/vp8/encoder/x86/csystemdependent.c index 186ee68..9fb6761 100644 --- a/vp8/encoder/x86/csystemdependent.c +++ b/vp8/encoder/x86/csystemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -179,11 +180,21 @@ void vp8_cmachine_specific_config(void) { // Willamette instruction set available: vp8_mbuverror = vp8_mbuverror_xmm; - vp8_fast_quantize_b = vp8_fast_quantize_b_sse; + /* The sse quantizer has not been updated to match the new exact + * quantizer introduced in commit e04e2935 + */ + vp8_fast_quantize_b = vp8_fast_quantize_b_c; +#if 0 //new fdct vp8_short_fdct4x4 = vp8_short_fdct4x4_mmx; vp8_short_fdct8x4 = vp8_short_fdct8x4_mmx; - vp8_fast_fdct4x4 = vp8_fast_fdct4x4_mmx; - vp8_fast_fdct8x4 = vp8_fast_fdct8x4_wmt; + vp8_fast_fdct4x4 = vp8_short_fdct4x4_mmx; + vp8_fast_fdct8x4 = vp8_short_fdct8x4_wmt; +#else + vp8_short_fdct4x4 = vp8_short_fdct4x4_c; + vp8_short_fdct8x4 = vp8_short_fdct8x4_c; + vp8_fast_fdct4x4 = vp8_short_fdct4x4_c; + vp8_fast_fdct8x4 = vp8_fast_fdct8x4_c; +#endif vp8_subtract_b = vp8_subtract_b_mmx; vp8_subtract_mbuv = vp8_subtract_mbuv_mmx; vp8_variance4x4 = vp8_variance4x4_mmx; @@ -216,11 +227,21 @@ void vp8_cmachine_specific_config(void) { // MMX instruction set available: vp8_mbuverror = vp8_mbuverror_mmx; - vp8_fast_quantize_b = vp8_fast_quantize_b_mmx; + /* The mmx quantizer has not been updated to match the new exact + * quantizer introduced in commit e04e2935 + */ + vp8_fast_quantize_b = vp8_fast_quantize_b_c; +#if 0 // new fdct vp8_short_fdct4x4 = vp8_short_fdct4x4_mmx; vp8_short_fdct8x4 = vp8_short_fdct8x4_mmx; - vp8_fast_fdct4x4 = vp8_fast_fdct4x4_mmx; - vp8_fast_fdct8x4 = vp8_fast_fdct8x4_mmx; + vp8_fast_fdct4x4 = vp8_short_fdct4x4_mmx; + vp8_fast_fdct8x4 = vp8_short_fdct8x4_mmx; +#else + vp8_short_fdct4x4 = vp8_short_fdct4x4_c; + vp8_short_fdct8x4 = vp8_short_fdct8x4_c; + vp8_fast_fdct4x4 = vp8_short_fdct4x4_c; + vp8_fast_fdct8x4 = vp8_fast_fdct8x4_c; +#endif vp8_subtract_b = vp8_subtract_b_mmx; vp8_subtract_mbuv = vp8_subtract_mbuv_mmx; vp8_variance4x4 = vp8_variance4x4_mmx; @@ -253,10 +274,10 @@ void vp8_cmachine_specific_config(void) { // Pure C: vp8_mbuverror = vp8_mbuverror_c; - vp8_fast_quantize_b = vp8_fast_quantize_b_c; + vp8_fast_quantize_b = vp8_fast_quantize_b_c; vp8_short_fdct4x4 = vp8_short_fdct4x4_c; vp8_short_fdct8x4 = vp8_short_fdct8x4_c; - vp8_fast_fdct4x4 = vp8_fast_fdct4x4_c; + vp8_fast_fdct4x4 = vp8_short_fdct4x4_c; vp8_fast_fdct8x4 = vp8_fast_fdct8x4_c; vp8_subtract_b = vp8_subtract_b_c; vp8_subtract_mbuv = vp8_subtract_mbuv_c; diff --git a/vp8/encoder/x86/dct_mmx.asm b/vp8/encoder/x86/dct_mmx.asm index e134237..b6cfc5c 100644 --- a/vp8/encoder/x86/dct_mmx.asm +++ b/vp8/encoder/x86/dct_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -12,8 +13,7 @@ section .text global sym(vp8_short_fdct4x4_mmx) - global sym(vp8_fast_fdct4x4_mmx) - global sym(vp8_fast_fdct8x4_wmt) + global sym(vp8_short_fdct8x4_wmt) %define DCTCONSTANTSBITS (16) @@ -23,10 +23,6 @@ section .text %define x_c3 (25080) ; cos(pi*3/8) * (1<<15) -%define _1STSTAGESHIFT 14 -%define _2NDSTAGESHIFT 16 - -; using matrix multiply with source and destbuffer has a pitch ;void vp8_short_fdct4x4_mmx(short *input, short *output, int pitch) sym(vp8_short_fdct4x4_mmx): push rbp @@ -36,333 +32,6 @@ sym(vp8_short_fdct4x4_mmx): push rsi push rdi ; end prolog - - mov rsi, arg(0) ;input - mov rdi, arg(1) ;output - - movsxd rax, dword ptr arg(2) ;pitch - lea rdx, [dct_matrix GLOBAL] - - movq mm0, [rsi ] - movq mm1, [rsi + rax] - - movq mm2, [rsi + rax*2] - lea rsi, [rsi + rax*2] - - movq mm3, [rsi + rax] - - ; first column - movq mm4, mm0 - movq mm7, [rdx] - - pmaddwd mm4, mm7 - movq mm5, mm1 - - pmaddwd mm5, mm7 - movq mm6, mm4 - - punpckldq mm4, mm5 - punpckhdq mm6, mm5 - - paddd mm4, mm6 - movq mm5, mm2 - - - pmaddwd mm5, mm7 - movq mm6, mm3 - - pmaddwd mm6, mm7 - movq mm7, mm5 - - punpckldq mm5, mm6 - punpckhdq mm7, mm6 - - paddd mm5, mm7 - movq mm6, [dct1st_stage_rounding_mmx GLOBAL] - - paddd mm4, mm6 - paddd mm5, mm6 - - psrad mm4, _1STSTAGESHIFT - psrad mm5, _1STSTAGESHIFT - - packssdw mm4, mm5 - movq [rdi], mm4 - - ;second column - movq mm4, mm0 - - pmaddwd mm4, [rdx+8] - movq mm5, mm1 - - pmaddwd mm5, [rdx+8] - movq mm6, mm4 - - punpckldq mm4, mm5 - punpckhdq mm6, mm5 - - paddd mm4, mm6 - movq mm5, mm2 - - pmaddwd mm5, [rdx+8] - movq mm6, mm3 - - pmaddwd mm6, [rdx+8] - movq mm7, mm5 - - punpckldq mm5, mm6 - punpckhdq mm7, mm6 - - paddd mm5, mm7 - movq mm6, [dct1st_stage_rounding_mmx GLOBAL] - - paddd mm4, mm6 - paddd mm5, mm6 - - psrad mm4, _1STSTAGESHIFT - psrad mm5, _1STSTAGESHIFT - - packssdw mm4, mm5 - movq [rdi+8], mm4 - - - ;third column - movq mm4, mm0 - - pmaddwd mm4, [rdx+16] - movq mm5, mm1 - - pmaddwd mm5, [rdx+16] - movq mm6, mm4 - - punpckldq mm4, mm5 - punpckhdq mm6, mm5 - - paddd mm4, mm6 - movq mm5, mm2 - - pmaddwd mm5, [rdx+16] - movq mm6, mm3 - - pmaddwd mm6, [rdx+16] - movq mm7, mm5 - - punpckldq mm5, mm6 - punpckhdq mm7, mm6 - - paddd mm5, mm7 - movq mm6, [dct1st_stage_rounding_mmx GLOBAL] - - paddd mm4, mm6 - paddd mm5, mm6 - - psrad mm4, _1STSTAGESHIFT - psrad mm5, _1STSTAGESHIFT - - packssdw mm4, mm5 - movq [rdi+16], mm4 - - ;fourth column (this is the last column, so we do not have save the source any more) - - pmaddwd mm0, [rdx+24] - - pmaddwd mm1, [rdx+24] - movq mm6, mm0 - - punpckldq mm0, mm1 - punpckhdq mm6, mm1 - - paddd mm0, mm6 - - pmaddwd mm2, [rdx+24] - - pmaddwd mm3, [rdx+24] - movq mm7, mm2 - - punpckldq mm2, mm3 - punpckhdq mm7, mm3 - - paddd mm2, mm7 - movq mm6, [dct1st_stage_rounding_mmx GLOBAL] - - paddd mm0, mm6 - paddd mm2, mm6 - - psrad mm0, _1STSTAGESHIFT - psrad mm2, _1STSTAGESHIFT - - packssdw mm0, mm2 - - movq mm3, mm0 - - ; done with one pass - ; now start second pass - movq mm0, [rdi ] - movq mm1, [rdi+ 8] - movq mm2, [rdi+ 16] - - movq mm4, mm0 - - pmaddwd mm4, [rdx] - movq mm5, mm1 - - pmaddwd mm5, [rdx] - movq mm6, mm4 - - punpckldq mm4, mm5 - punpckhdq mm6, mm5 - - paddd mm4, mm6 - movq mm5, mm2 - - pmaddwd mm5, [rdx] - movq mm6, mm3 - - pmaddwd mm6, [rdx] - movq mm7, mm5 - - punpckldq mm5, mm6 - punpckhdq mm7, mm6 - - paddd mm5, mm7 - movq mm6, [dct2nd_stage_rounding_mmx GLOBAL] - - paddd mm4, mm6 - paddd mm5, mm6 - - psrad mm4, _2NDSTAGESHIFT - psrad mm5, _2NDSTAGESHIFT - - packssdw mm4, mm5 - movq [rdi], mm4 - - ;second column - movq mm4, mm0 - - pmaddwd mm4, [rdx+8] - movq mm5, mm1 - - pmaddwd mm5, [rdx+8] - movq mm6, mm4 - - punpckldq mm4, mm5 - punpckhdq mm6, mm5 - - paddd mm4, mm6 - movq mm5, mm2 - - pmaddwd mm5, [rdx+8] - movq mm6, mm3 - - pmaddwd mm6, [rdx+8] - movq mm7, mm5 - - punpckldq mm5, mm6 - punpckhdq mm7, mm6 - - paddd mm5, mm7 - movq mm6, [dct2nd_stage_rounding_mmx GLOBAL] - - paddd mm4, mm6 - paddd mm5, mm6 - - psrad mm4, _2NDSTAGESHIFT - psrad mm5, _2NDSTAGESHIFT - - packssdw mm4, mm5 - movq [rdi+8], mm4 - - - ;third column - movq mm4, mm0 - - pmaddwd mm4, [rdx+16] - movq mm5, mm1 - - pmaddwd mm5, [rdx+16] - movq mm6, mm4 - - punpckldq mm4, mm5 - punpckhdq mm6, mm5 - - paddd mm4, mm6 - movq mm5, mm2 - - pmaddwd mm5, [rdx+16] - movq mm6, mm3 - - pmaddwd mm6, [rdx+16] - movq mm7, mm5 - - punpckldq mm5, mm6 - punpckhdq mm7, mm6 - - paddd mm5, mm7 - movq mm6, [dct2nd_stage_rounding_mmx GLOBAL] - - paddd mm4, mm6 - paddd mm5, mm6 - - psrad mm4, _2NDSTAGESHIFT - psrad mm5, _2NDSTAGESHIFT - - packssdw mm4, mm5 - movq [rdi+16], mm4 - - ;fourth column - movq mm4, mm0 - - pmaddwd mm4, [rdx+24] - movq mm5, mm1 - - pmaddwd mm5, [rdx+24] - movq mm6, mm4 - - punpckldq mm4, mm5 - punpckhdq mm6, mm5 - - paddd mm4, mm6 - movq mm5, mm2 - - pmaddwd mm5, [rdx+24] - movq mm6, mm3 - - pmaddwd mm6, [rdx+24] - movq mm7, mm5 - - punpckldq mm5, mm6 - punpckhdq mm7, mm6 - - paddd mm5, mm7 - movq mm6, [dct2nd_stage_rounding_mmx GLOBAL] - - paddd mm4, mm6 - paddd mm5, mm6 - - psrad mm4, _2NDSTAGESHIFT - psrad mm5, _2NDSTAGESHIFT - - packssdw mm4, mm5 - movq [rdi+24], mm4 - - ; begin epilog - pop rdi - pop rsi - RESTORE_GOT - UNSHADOW_ARGS - pop rbp - ret - - -;void vp8_fast_fdct4x4_mmx(short *input, short *output, int pitch) -sym(vp8_fast_fdct4x4_mmx): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 3 - GET_GOT rbx - push rsi - push rdi - ; end prolog mov rsi, arg(0) ;input mov rdi, arg(1) ;output @@ -378,11 +47,11 @@ sym(vp8_fast_fdct4x4_mmx): movq mm3, [rcx + rax] ; get the constants ;shift to left by 1 for prescision - paddw mm0, mm0 - paddw mm1, mm1 + psllw mm0, 3 + psllw mm1, 3 - psllw mm2, 1 - psllw mm3, 1 + psllw mm2, 3 + psllw mm3, 3 ; transpose for the second stage movq mm4, mm0 ; 00 01 02 03 @@ -530,20 +199,23 @@ sym(vp8_fast_fdct4x4_mmx): movq mm3, mm5 ; done with vertical - pcmpeqw mm4, mm4 - pcmpeqw mm5, mm5 - psrlw mm4, 15 - psrlw mm5, 15 + pcmpeqw mm4, mm4 + pcmpeqw mm5, mm5 + psrlw mm4, 15 + psrlw mm5, 15 + + psllw mm4, 2 + psllw mm5, 2 paddw mm0, mm4 paddw mm1, mm5 paddw mm2, mm4 paddw mm3, mm5 - psraw mm0, 1 - psraw mm1, 1 - psraw mm2, 1 - psraw mm3, 1 + psraw mm0, 3 + psraw mm1, 3 + psraw mm2, 3 + psraw mm3, 3 movq [rdi ], mm0 movq [rdi+ 8], mm1 @@ -559,8 +231,8 @@ sym(vp8_fast_fdct4x4_mmx): ret -;void vp8_fast_fdct8x4_wmt(short *input, short *output, int pitch) -sym(vp8_fast_fdct8x4_wmt): +;void vp8_short_fdct8x4_wmt(short *input, short *output, int pitch) +sym(vp8_short_fdct8x4_wmt): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 3 @@ -583,11 +255,11 @@ sym(vp8_fast_fdct8x4_wmt): movdqa xmm3, [rcx + rax] ; get the constants ;shift to left by 1 for prescision - psllw xmm0, 1 - psllw xmm2, 1 + psllw xmm0, 3 + psllw xmm2, 3 - psllw xmm4, 1 - psllw xmm3, 1 + psllw xmm4, 3 + psllw xmm3, 3 ; transpose for the second stage movdqa xmm1, xmm0 ; 00 01 02 03 04 05 06 07 @@ -757,20 +429,23 @@ sym(vp8_fast_fdct8x4_wmt): ; done with vertical - pcmpeqw xmm4, xmm4 - pcmpeqw xmm5, xmm5; - psrlw xmm4, 15 - psrlw xmm5, 15 + pcmpeqw xmm4, xmm4 + pcmpeqw xmm5, xmm5; + psrlw xmm4, 15 + psrlw xmm5, 15 + + psllw xmm4, 2 + psllw xmm5, 2 paddw xmm0, xmm4 paddw xmm1, xmm5 paddw xmm2, xmm4 paddw xmm3, xmm5 - psraw xmm0, 1 - psraw xmm1, 1 - psraw xmm2, 1 - psraw xmm3, 1 + psraw xmm0, 3 + psraw xmm1, 3 + psraw xmm2, 3 + psraw xmm3, 3 movq QWORD PTR[rdi ], xmm0 movq QWORD PTR[rdi+ 8], xmm1 diff --git a/vp8/encoder/x86/dct_sse2.asm b/vp8/encoder/x86/dct_sse2.asm index 3e5e9a7..f7a1843 100644 --- a/vp8/encoder/x86/dct_sse2.asm +++ b/vp8/encoder/x86/dct_sse2.asm @@ -1,260 +1,189 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; %include "vpx_ports/x86_abi_support.asm" -global sym(vp8_short_fdct4x4_wmt) - -%define DCTCONSTANTSBITS (16) -%define DCTROUNDINGVALUE (1<< (DCTCONSTANTSBITS-1)) -%define x_c1 (60547) ; cos(pi /8) * (1<<15) -%define x_c2 (46341) ; cos(pi*2/8) * (1<<15) -%define x_c3 (25080) ; cos(pi*3/8) * (1<<15) - -%define _1STSTAGESHIFT 14 -%define _2NDSTAGESHIFT 16 - - -;; using matrix multiply -;void vp8_short_fdct4x4_wmt(short *input, short *output) -sym(vp8_short_fdct4x4_wmt): +;void vp8_short_fdct4x4_sse2(short *input, short *output, int pitch) +global sym(vp8_short_fdct4x4_sse2) +sym(vp8_short_fdct4x4_sse2): push rbp mov rbp, rsp - SHADOW_ARGS_TO_STACK 2 + SHADOW_ARGS_TO_STACK 3 +;; SAVE_XMM GET_GOT rbx + push rsi + push rdi ; end prolog - mov rax, arg(0) ;input - mov rcx, arg(1) ;output - - lea rdx, [dct_matrix_sse2 GLOBAL] - - movdqu xmm0, [rax ] - movdqu xmm1, [rax+16] - - ; first column - movdqa xmm2, xmm0 - movdqa xmm7, [rdx] - - pmaddwd xmm2, xmm7 - movdqa xmm3, xmm1 - - pmaddwd xmm3, xmm7 - movdqa xmm4, xmm2 - - punpckldq xmm2, xmm3 - punpckhdq xmm4, xmm3 - - movdqa xmm3, xmm2 - punpckldq xmm2, xmm4 - - punpckhdq xmm3, xmm4 - paddd xmm2, xmm3 - - - paddd xmm2, XMMWORD PTR [dct1st_stage_rounding_sse2 GLOBAL] - psrad xmm2, _1STSTAGESHIFT - ;second column - movdqa xmm3, xmm0 - pmaddwd xmm3, [rdx+16] - - movdqa xmm4, xmm1 - pmaddwd xmm4, [rdx+16] - - movdqa xmm5, xmm3 - punpckldq xmm3, xmm4 - - punpckhdq xmm5, xmm4 - movdqa xmm4, xmm3 - - punpckldq xmm3, xmm5 - punpckhdq xmm4, xmm5 - - paddd xmm3, xmm4 - paddd xmm3, XMMWORD PTR [dct1st_stage_rounding_sse2 GLOBAL] - - - psrad xmm3, _1STSTAGESHIFT - packssdw xmm2, xmm3 - - ;third column - movdqa xmm3, xmm0 - pmaddwd xmm3, [rdx+32] - - movdqa xmm4, xmm1 - pmaddwd xmm4, [rdx+32] - - movdqa xmm5, xmm3 - punpckldq xmm3, xmm4 - - punpckhdq xmm5, xmm4 - movdqa xmm4, xmm3 - - punpckldq xmm3, xmm5 - punpckhdq xmm4, xmm5 - - paddd xmm3, xmm4 - paddd xmm3, XMMWORD PTR [dct1st_stage_rounding_sse2 GLOBAL] - - psrad xmm3, _1STSTAGESHIFT - - ;fourth column (this is the last column, so we do not have save the source any more) - pmaddwd xmm0, [rdx+48] - pmaddwd xmm1, [rdx+48] - - movdqa xmm4, xmm0 - punpckldq xmm0, xmm1 - - punpckhdq xmm4, xmm1 - movdqa xmm1, xmm0 - - punpckldq xmm0, xmm4 - punpckhdq xmm1, xmm4 - - paddd xmm0, xmm1 - paddd xmm0, XMMWORD PTR [dct1st_stage_rounding_sse2 GLOBAL] - - - psrad xmm0, _1STSTAGESHIFT - packssdw xmm3, xmm0 - ; done with one pass - ; now start second pass - movdqa xmm0, xmm2 - movdqa xmm1, xmm3 - - pmaddwd xmm2, xmm7 - pmaddwd xmm3, xmm7 - - movdqa xmm4, xmm2 - punpckldq xmm2, xmm3 + mov rsi, arg(0) + movsxd rax, DWORD PTR arg(2) + lea rdi, [rsi + rax*2] + + movq xmm0, MMWORD PTR[rsi ] ;03 02 01 00 + movq xmm2, MMWORD PTR[rsi + rax] ;13 12 11 10 + movq xmm1, MMWORD PTR[rsi + rax*2] ;23 22 21 20 + movq xmm3, MMWORD PTR[rdi + rax] ;33 32 31 30 + + punpcklqdq xmm0, xmm2 ;13 12 11 10 03 02 01 00 + punpcklqdq xmm1, xmm3 ;33 32 31 30 23 22 21 20 + + mov rdi, arg(1) + + movdqa xmm2, xmm0 + punpckldq xmm0, xmm1 ;23 22 03 02 21 20 01 00 + punpckhdq xmm2, xmm1 ;33 32 13 12 31 30 11 10 + movdqa xmm1, xmm0 + punpckldq xmm0, xmm2 ;31 21 30 20 11 10 01 00 + pshufhw xmm1, xmm1, 0b1h ;22 23 02 03 xx xx xx xx + pshufhw xmm2, xmm2, 0b1h ;32 33 12 13 xx xx xx xx + + punpckhdq xmm1, xmm2 ;32 33 22 23 12 13 02 03 + movdqa xmm3, xmm0 + paddw xmm0, xmm1 ;b1 a1 b1 a1 b1 a1 b1 a1 + psubw xmm3, xmm1 ;c1 d1 c1 d1 c1 d1 c1 d1 + psllw xmm0, 3 ;b1 <<= 3 a1 <<= 3 + psllw xmm3, 3 ;c1 <<= 3 d1 <<= 3 + movdqa xmm1, xmm0 + pmaddwd xmm0, XMMWORD PTR[_mult_add GLOBAL] ;a1 + b1 + pmaddwd xmm1, XMMWORD PTR[_mult_sub GLOBAL] ;a1 - b1 + movdqa xmm4, xmm3 + pmaddwd xmm3, XMMWORD PTR[_5352_2217 GLOBAL] ;c1*2217 + d1*5352 + pmaddwd xmm4, XMMWORD PTR[_2217_neg5352 GLOBAL] ;d1*2217 - c1*5352 + + paddd xmm3, XMMWORD PTR[_14500 GLOBAL] + paddd xmm4, XMMWORD PTR[_7500 GLOBAL] + psrad xmm3, 12 ;(c1 * 2217 + d1 * 5352 + 14500)>>12 + psrad xmm4, 12 ;(d1 * 2217 - c1 * 5352 + 7500)>>12 + + packssdw xmm0, xmm1 ;op[2] op[0] + packssdw xmm3, xmm4 ;op[3] op[1] + ; 23 22 21 20 03 02 01 00 + ; + ; 33 32 31 30 13 12 11 10 + ; + movdqa xmm2, xmm0 + punpcklqdq xmm0, xmm3 ;13 12 11 10 03 02 01 00 + punpckhqdq xmm2, xmm3 ;23 22 21 20 33 32 31 30 + + movdqa xmm3, xmm0 + punpcklwd xmm0, xmm2 ;32 30 22 20 12 10 02 00 + punpckhwd xmm3, xmm2 ;33 31 23 21 13 11 03 01 + movdqa xmm2, xmm0 + punpcklwd xmm0, xmm3 ;13 12 11 10 03 02 01 00 + punpckhwd xmm2, xmm3 ;33 32 31 30 23 22 21 20 + + movdqa xmm5, XMMWORD PTR[_7 GLOBAL] + pshufd xmm2, xmm2, 04eh + movdqa xmm3, xmm0 + paddw xmm0, xmm2 ;b1 b1 b1 b1 a1 a1 a1 a1 + psubw xmm3, xmm2 ;c1 c1 c1 c1 d1 d1 d1 d1 + + pshufd xmm0, xmm0, 0d8h ;b1 b1 a1 a1 b1 b1 a1 a1 + movdqa xmm2, xmm3 ;save d1 for compare + pshufd xmm3, xmm3, 0d8h ;c1 c1 d1 d1 c1 c1 d1 d1 + pshuflw xmm0, xmm0, 0d8h ;b1 b1 a1 a1 b1 a1 b1 a1 + pshuflw xmm3, xmm3, 0d8h ;c1 c1 d1 d1 c1 d1 c1 d1 + pshufhw xmm0, xmm0, 0d8h ;b1 a1 b1 a1 b1 a1 b1 a1 + pshufhw xmm3, xmm3, 0d8h ;c1 d1 c1 d1 c1 d1 c1 d1 + movdqa xmm1, xmm0 + pmaddwd xmm0, XMMWORD PTR[_mult_add GLOBAL] ;a1 + b1 + pmaddwd xmm1, XMMWORD PTR[_mult_sub GLOBAL] ;a1 - b1 + + pxor xmm4, xmm4 ;zero out for compare + paddd xmm0, xmm5 + paddd xmm1, xmm5 + pcmpeqw xmm2, xmm4 + psrad xmm0, 4 ;(a1 + b1 + 7)>>4 + psrad xmm1, 4 ;(a1 - b1 + 7)>>4 + pandn xmm2, XMMWORD PTR[_cmp_mask GLOBAL] ;clear upper, + ;and keep bit 0 of lower + + movdqa xmm4, xmm3 + pmaddwd xmm3, XMMWORD PTR[_5352_2217 GLOBAL] ;c1*2217 + d1*5352 + pmaddwd xmm4, XMMWORD PTR[_2217_neg5352 GLOBAL] ;d1*2217 - c1*5352 + paddd xmm3, XMMWORD PTR[_12000 GLOBAL] + paddd xmm4, XMMWORD PTR[_51000 GLOBAL] + packssdw xmm0, xmm1 ;op[8] op[0] + psrad xmm3, 16 ;(c1 * 2217 + d1 * 5352 + 12000)>>16 + psrad xmm4, 16 ;(d1 * 2217 - c1 * 5352 + 51000)>>16 + + packssdw xmm3, xmm4 ;op[12] op[4] + movdqa xmm1, xmm0 + paddw xmm3, xmm2 ;op[4] += (d1!=0) + punpcklqdq xmm0, xmm3 ;op[4] op[0] + punpckhqdq xmm1, xmm3 ;op[12] op[8] + + movdqa XMMWORD PTR[rdi + 0], xmm0 + movdqa XMMWORD PTR[rdi + 16], xmm1 - punpckhdq xmm4, xmm3 - movdqa xmm3, xmm2 - - punpckldq xmm2, xmm4 - punpckhdq xmm3, xmm4 - - paddd xmm2, xmm3 - paddd xmm2, XMMWORD PTR [dct2nd_stage_rounding_sse2 GLOBAL] - - psrad xmm2, _2NDSTAGESHIFT - - ;second column - movdqa xmm3, xmm0 - pmaddwd xmm3, [rdx+16] - - movdqa xmm4, xmm1 - pmaddwd xmm4, [rdx+16] - - movdqa xmm5, xmm3 - punpckldq xmm3, xmm4 - - punpckhdq xmm5, xmm4 - movdqa xmm4, xmm3 - - punpckldq xmm3, xmm5 - punpckhdq xmm4, xmm5 - - paddd xmm3, xmm4 - paddd xmm3, XMMWORD PTR [dct2nd_stage_rounding_sse2 GLOBAL] - - psrad xmm3, _2NDSTAGESHIFT - packssdw xmm2, xmm3 - - movdqu [rcx], xmm2 - ;third column - movdqa xmm3, xmm0 - pmaddwd xmm3, [rdx+32] - - movdqa xmm4, xmm1 - pmaddwd xmm4, [rdx+32] - - movdqa xmm5, xmm3 - punpckldq xmm3, xmm4 - - punpckhdq xmm5, xmm4 - movdqa xmm4, xmm3 - - punpckldq xmm3, xmm5 - punpckhdq xmm4, xmm5 - - paddd xmm3, xmm4 - paddd xmm3, XMMWORD PTR [dct2nd_stage_rounding_sse2 GLOBAL] - - psrad xmm3, _2NDSTAGESHIFT - ;fourth column - pmaddwd xmm0, [rdx+48] - pmaddwd xmm1, [rdx+48] - - movdqa xmm4, xmm0 - punpckldq xmm0, xmm1 - - punpckhdq xmm4, xmm1 - movdqa xmm1, xmm0 - - punpckldq xmm0, xmm4 - punpckhdq xmm1, xmm4 - - paddd xmm0, xmm1 - paddd xmm0, XMMWORD PTR [dct2nd_stage_rounding_sse2 GLOBAL] - - psrad xmm0, _2NDSTAGESHIFT - packssdw xmm3, xmm0 - - movdqu [rcx+16], xmm3 - - mov rsp, rbp ; begin epilog + pop rdi + pop rsi RESTORE_GOT +;; RESTORE_XMM UNSHADOW_ARGS pop rbp ret - SECTION_RODATA -;static unsigned int dct1st_stage_rounding_sse2[4] = align 16 -dct1st_stage_rounding_sse2: - times 4 dd 8192 - - -;static unsigned int dct2nd_stage_rounding_sse2[4] = +_5352_2217: + dw 5352 + dw 2217 + dw 5352 + dw 2217 + dw 5352 + dw 2217 + dw 5352 + dw 2217 align 16 -dct2nd_stage_rounding_sse2: - times 4 dd 32768 - -;static short dct_matrix_sse2[4][8]= +_2217_neg5352: + dw 2217 + dw -5352 + dw 2217 + dw -5352 + dw 2217 + dw -5352 + dw 2217 + dw -5352 align 16 -dct_matrix_sse2: - times 8 dw 23170 - - dw 30274 - dw 12540 - dw -12540 - dw -30274 - dw 30274 - dw 12540 - dw -12540 - dw -30274 - - dw 23170 - times 2 dw -23170 - times 2 dw 23170 - times 2 dw -23170 - dw 23170 +_mult_add: + times 8 dw 1 +align 16 +_cmp_mask: + times 4 dw 1 + times 4 dw 0 - dw 12540 - dw -30274 - dw 30274 - dw -12540 - dw 12540 - dw -30274 - dw 30274 - dw -12540 +align 16 +_mult_sub: + dw 1 + dw -1 + dw 1 + dw -1 + dw 1 + dw -1 + dw 1 + dw -1 +align 16 +_7: + times 4 dd 7 +align 16 +_14500: + times 4 dd 14500 +align 16 +_7500: + times 4 dd 7500 +align 16 +_12000: + times 4 dd 12000 +align 16 +_51000: + times 4 dd 51000 diff --git a/vp8/encoder/x86/dct_x86.h b/vp8/encoder/x86/dct_x86.h index bc80e64..05824c6 100644 --- a/vp8/encoder/x86/dct_x86.h +++ b/vp8/encoder/x86/dct_x86.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -21,46 +22,41 @@ #if HAVE_MMX extern prototype_fdct(vp8_short_fdct4x4_mmx); extern prototype_fdct(vp8_short_fdct8x4_mmx); -extern prototype_fdct(vp8_fast_fdct4x4_mmx); -extern prototype_fdct(vp8_fast_fdct8x4_mmx); #if !CONFIG_RUNTIME_CPU_DETECT +#if 0 #undef vp8_fdct_short4x4 #define vp8_fdct_short4x4 vp8_short_fdct4x4_mmx #undef vp8_fdct_short8x4 #define vp8_fdct_short8x4 vp8_short_fdct8x4_mmx - -#undef vp8_fdct_fast4x4 -#define vp8_fdct_fast4x4 vp8_fast_fdct4x4_mmx - -#undef vp8_fdct_fast8x4 -#define vp8_fdct_fast8x4 vp8_fast_fdct8x4_mmx +#endif #endif #endif #if HAVE_SSE2 -extern prototype_fdct(vp8_short_fdct4x4_wmt); extern prototype_fdct(vp8_short_fdct8x4_wmt); -extern prototype_fdct(vp8_fast_fdct8x4_wmt); - extern prototype_fdct(vp8_short_walsh4x4_sse2); -#if !CONFIG_RUNTIME_CPU_DETECT +extern prototype_fdct(vp8_short_fdct4x4_sse2); -#if 0 +#if !CONFIG_RUNTIME_CPU_DETECT +#if 1 /* short SSE2 DCT currently disabled, does not match the MMX version */ #undef vp8_fdct_short4x4 -#define vp8_fdct_short4x4 vp8_short_fdct4x4_wmt +#define vp8_fdct_short4x4 vp8_short_fdct4x4_sse2 #undef vp8_fdct_short8x4 -#define vp8_fdct_short8x4 vp8_short_fdct8x4_wmt +#define vp8_fdct_short8x4 vp8_short_fdct8x4_sse2 #endif +#undef vp8_fdct_fast4x4 +#define vp8_fdct_fast4x4 vp8_short_fdct4x4_sse2 + #undef vp8_fdct_fast8x4 -#define vp8_fdct_fast8x4 vp8_fast_fdct8x4_wmt +#define vp8_fdct_fast8x4 vp8_short_fdct8x4_sse2 #undef vp8_fdct_walsh_short4x4 #define vp8_fdct_walsh_short4x4 vp8_short_walsh4x4_sse2 diff --git a/vp8/encoder/x86/encodemb_x86.h b/vp8/encoder/x86/encodemb_x86.h index 9397a6c..d090b2d 100644 --- a/vp8/encoder/x86/encodemb_x86.h +++ b/vp8/encoder/x86/encodemb_x86.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/x86/encodeopt.asm b/vp8/encoder/x86/encodeopt.asm index 1940471..413d74d 100644 --- a/vp8/encoder/x86/encodeopt.asm +++ b/vp8/encoder/x86/encodeopt.asm @@ -1,16 +1,16 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; %include "vpx_ports/x86_abi_support.asm" - ;int vp8_block_error_xmm(short *coeff_ptr, short *dcoef_ptr) global sym(vp8_block_error_xmm) sym(vp8_block_error_xmm): @@ -19,11 +19,9 @@ sym(vp8_block_error_xmm): SHADOW_ARGS_TO_STACK 2 push rsi push rdi - ; end prolog - + ; end prologue mov rsi, arg(0) ;coeff_ptr - pxor xmm7, xmm7 mov rdi, arg(1) ;dcoef_ptr movdqa xmm3, [rsi] @@ -32,31 +30,25 @@ sym(vp8_block_error_xmm): movdqa xmm5, [rsi+16] movdqa xmm6, [rdi+16] - pxor xmm1, xmm1 ; from movd xmm1, dc; dc=0 + psubw xmm3, xmm4 - movdqa xmm2, xmm7 psubw xmm5, xmm6 - - por xmm1, xmm2 + pmaddwd xmm3, xmm3 pmaddwd xmm5, xmm5 - pcmpeqw xmm1, xmm7 - psubw xmm3, xmm4 + paddd xmm3, xmm5 - pand xmm1, xmm3 - pmaddwd xmm1, xmm1 - - paddd xmm1, xmm5 - movdqa xmm0, xmm1 + pxor xmm7, xmm7 + movdqa xmm0, xmm3 punpckldq xmm0, xmm7 - punpckhdq xmm1, xmm7 + punpckhdq xmm3, xmm7 - paddd xmm0, xmm1 - movdqa xmm1, xmm0 + paddd xmm0, xmm3 + movdqa xmm3, xmm0 psrldq xmm0, 8 - paddd xmm0, xmm1 + paddd xmm0, xmm3 movd rax, xmm0 @@ -67,7 +59,6 @@ sym(vp8_block_error_xmm): pop rbp ret - ;int vp8_block_error_mmx(short *coeff_ptr, short *dcoef_ptr) global sym(vp8_block_error_mmx) sym(vp8_block_error_mmx): diff --git a/vp8/encoder/x86/fwalsh_sse2.asm b/vp8/encoder/x86/fwalsh_sse2.asm index 7d86201..38812c8 100644 --- a/vp8/encoder/x86/fwalsh_sse2.asm +++ b/vp8/encoder/x86/fwalsh_sse2.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/x86/mcomp_x86.h b/vp8/encoder/x86/mcomp_x86.h index 5661491..e8d658b 100644 --- a/vp8/encoder/x86/mcomp_x86.h +++ b/vp8/encoder/x86/mcomp_x86.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/x86/preproc_mmx.c b/vp8/encoder/x86/preproc_mmx.c index 69617ca..a182c88 100644 --- a/vp8/encoder/x86/preproc_mmx.c +++ b/vp8/encoder/x86/preproc_mmx.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/x86/quantize_mmx.asm b/vp8/encoder/x86/quantize_mmx.asm index 847fc6e..a867409 100644 --- a/vp8/encoder/x86/quantize_mmx.asm +++ b/vp8/encoder/x86/quantize_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/x86/quantize_sse2.asm b/vp8/encoder/x86/quantize_sse2.asm new file mode 100644 index 0000000..a1b1c40 --- /dev/null +++ b/vp8/encoder/x86/quantize_sse2.asm @@ -0,0 +1,254 @@ +; +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. +; +; Use of this source code is governed by a BSD-style license and patent +; grant that can be found in the LICENSE file in the root of the source +; tree. All contributing project authors may be found in the AUTHORS +; file in the root of the source tree. +; + + +%include "vpx_ports/x86_abi_support.asm" + + +;int vp8_regular_quantize_b_impl_sse2(short *coeff_ptr, short *zbin_ptr, +; short *qcoeff_ptr,short *dequant_ptr, +; const int *default_zig_zag, short *round_ptr, +; short *quant_ptr, short *dqcoeff_ptr, +; unsigned short zbin_oq_value, +; short *zbin_boost_ptr); +; +global sym(vp8_regular_quantize_b_impl_sse2) +sym(vp8_regular_quantize_b_impl_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 10 + push rsi + push rdi + push rbx + ; end prolog + + ALIGN_STACK 16, rax + + %define abs_minus_zbin_lo 0 + %define abs_minus_zbin_hi 16 + %define temp_qcoeff_lo 32 + %define temp_qcoeff_hi 48 + %define save_xmm6 64 + %define save_xmm7 80 + %define eob 96 + + %define vp8_regularquantizeb_stack_size eob + 16 + + sub rsp, vp8_regularquantizeb_stack_size + + movdqa OWORD PTR[rsp + save_xmm6], xmm6 + movdqa OWORD PTR[rsp + save_xmm7], xmm7 + + mov rdx, arg(0) ;coeff_ptr + mov eax, arg(8) ;zbin_oq_value + + mov rcx, arg(1) ;zbin_ptr + movd xmm7, eax + + movdqa xmm0, OWORD PTR[rdx] + movdqa xmm4, OWORD PTR[rdx + 16] + + movdqa xmm1, xmm0 + movdqa xmm5, xmm4 + + psraw xmm0, 15 ;sign of z (aka sz) + psraw xmm4, 15 ;sign of z (aka sz) + + pxor xmm1, xmm0 + pxor xmm5, xmm4 + + movdqa xmm2, OWORD PTR[rcx] ;load zbin_ptr + movdqa xmm3, OWORD PTR[rcx + 16] ;load zbin_ptr + + pshuflw xmm7, xmm7, 0 + psubw xmm1, xmm0 ;x = abs(z) + + punpcklwd xmm7, xmm7 ;duplicated zbin_oq_value + psubw xmm5, xmm4 ;x = abs(z) + + paddw xmm2, xmm7 + paddw xmm3, xmm7 + + psubw xmm1, xmm2 ;sub (zbin_ptr + zbin_oq_value) + psubw xmm5, xmm3 ;sub (zbin_ptr + zbin_oq_value) + + mov rdi, arg(5) ;round_ptr + mov rsi, arg(6) ;quant_ptr + + movdqa OWORD PTR[rsp + abs_minus_zbin_lo], xmm1 + movdqa OWORD PTR[rsp + abs_minus_zbin_hi], xmm5 + + paddw xmm1, xmm2 ;add (zbin_ptr + zbin_oq_value) back + paddw xmm5, xmm3 ;add (zbin_ptr + zbin_oq_value) back + + movdqa xmm2, OWORD PTR[rdi] + movdqa xmm3, OWORD PTR[rsi] + + movdqa xmm6, OWORD PTR[rdi + 16] + movdqa xmm7, OWORD PTR[rsi + 16] + + paddw xmm1, xmm2 + paddw xmm5, xmm6 + + pmulhw xmm1, xmm3 + pmulhw xmm5, xmm7 + + mov rsi, arg(2) ;qcoeff_ptr + pxor xmm6, xmm6 + + pxor xmm1, xmm0 + pxor xmm5, xmm4 + + psubw xmm1, xmm0 + psubw xmm5, xmm4 + + movdqa OWORD PTR[rsp + temp_qcoeff_lo], xmm1 + movdqa OWORD PTR[rsp + temp_qcoeff_hi], xmm5 + + movdqa OWORD PTR[rsi], xmm6 ;zero qcoeff + movdqa OWORD PTR[rsi + 16], xmm6 ;zero qcoeff + + xor rax, rax + mov rcx, -1 + + mov [rsp + eob], rcx + mov rsi, arg(9) ;zbin_boost_ptr + + mov rbx, arg(4) ;default_zig_zag + +rq_zigzag_loop: + movsxd rcx, DWORD PTR[rbx + rax*4] ;now we have rc + movsx edi, WORD PTR [rsi] ;*zbin_boost_ptr aka zbin + lea rsi, [rsi + 2] ;zbin_boost_ptr++ + + movsx edx, WORD PTR[rsp + abs_minus_zbin_lo + rcx *2] + + sub edx, edi ;x - zbin + jl rq_zigzag_1 + + mov rdi, arg(2) ;qcoeff_ptr + + movsx edx, WORD PTR[rsp + temp_qcoeff_lo + rcx *2] + + cmp edx, 0 + je rq_zigzag_1 + + mov WORD PTR[rdi + rcx * 2], dx ;qcoeff_ptr[rc] = temp_qcoeff[rc] + + mov rsi, arg(9) ;zbin_boost_ptr + mov [rsp + eob], rax ;eob = i + +rq_zigzag_1: + movsxd rcx, DWORD PTR[rbx + rax*4 + 4] + movsx edi, WORD PTR [rsi] ;*zbin_boost_ptr aka zbin + lea rsi, [rsi + 2] ;zbin_boost_ptr++ + + movsx edx, WORD PTR[rsp + abs_minus_zbin_lo + rcx *2] + lea rax, [rax + 1] + + sub edx, edi ;x - zbin + jl rq_zigzag_1a + + mov rdi, arg(2) ;qcoeff_ptr + + movsx edx, WORD PTR[rsp + temp_qcoeff_lo + rcx *2] + + cmp edx, 0 + je rq_zigzag_1a + + mov WORD PTR[rdi + rcx * 2], dx ;qcoeff_ptr[rc] = temp_qcoeff[rc] + + mov rsi, arg(9) ;zbin_boost_ptr + mov [rsp + eob], rax ;eob = i + +rq_zigzag_1a: + movsxd rcx, DWORD PTR[rbx + rax*4 + 4] + movsx edi, WORD PTR [rsi] ;*zbin_boost_ptr aka zbin + lea rsi, [rsi + 2] ;zbin_boost_ptr++ + + movsx edx, WORD PTR[rsp + abs_minus_zbin_lo + rcx *2] + lea rax, [rax + 1] + + sub edx, edi ;x - zbin + jl rq_zigzag_1b + + mov rdi, arg(2) ;qcoeff_ptr + + movsx edx, WORD PTR[rsp + temp_qcoeff_lo + rcx *2] + + cmp edx, 0 + je rq_zigzag_1b + + mov WORD PTR[rdi + rcx * 2], dx ;qcoeff_ptr[rc] = temp_qcoeff[rc] + + mov rsi, arg(9) ;zbin_boost_ptr + mov [rsp + eob], rax ;eob = i + +rq_zigzag_1b: + movsxd rcx, DWORD PTR[rbx + rax*4 + 4] + movsx edi, WORD PTR [rsi] ;*zbin_boost_ptr aka zbin + lea rsi, [rsi + 2] ;zbin_boost_ptr++ + + movsx edx, WORD PTR[rsp + abs_minus_zbin_lo + rcx *2] + lea rax, [rax + 1] + + sub edx, edi ;x - zbin + jl rq_zigzag_1c + + mov rdi, arg(2) ;qcoeff_ptr + + movsx edx, WORD PTR[rsp + temp_qcoeff_lo + rcx *2] + + cmp edx, 0 + je rq_zigzag_1c + + mov WORD PTR[rdi + rcx * 2], dx ;qcoeff_ptr[rc] = temp_qcoeff[rc] + + mov rsi, arg(9) ;zbin_boost_ptr + mov [rsp + eob], rax ;eob = i + +rq_zigzag_1c: + lea rax, [rax + 1] + + cmp rax, 16 + jl rq_zigzag_loop + + mov rdi, arg(2) ;qcoeff_ptr + mov rcx, arg(3) ;dequant_ptr + mov rsi, arg(7) ;dqcoeff_ptr + + movdqa xmm2, OWORD PTR[rdi] + movdqa xmm3, OWORD PTR[rdi + 16] + + movdqa xmm0, OWORD PTR[rcx] + movdqa xmm1, OWORD PTR[rcx + 16] + + pmullw xmm0, xmm2 + pmullw xmm1, xmm3 + + movdqa OWORD PTR[rsi], xmm0 ;store dqcoeff + movdqa OWORD PTR[rsi + 16], xmm1 ;store dqcoeff + + mov rax, [rsp + eob] + + movdqa xmm6, OWORD PTR[rsp + save_xmm6] + movdqa xmm7, OWORD PTR[rsp + save_xmm7] + + add rax, 1 + + add rsp, vp8_regularquantizeb_stack_size + pop rsp + + ; begin epilog + pop rbx + pop rdi + pop rsi + UNSHADOW_ARGS + pop rbp + ret diff --git a/vp8/encoder/x86/quantize_x86.h b/vp8/encoder/x86/quantize_x86.h new file mode 100644 index 0000000..b5b22c0 --- /dev/null +++ b/vp8/encoder/x86/quantize_x86.h @@ -0,0 +1,41 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license and patent + * grant that can be found in the LICENSE file in the root of the source + * tree. All contributing project authors may be found in the AUTHORS + * file in the root of the source tree. + */ + +#ifndef QUANTIZE_X86_H +#define QUANTIZE_X86_H + + +/* Note: + * + * This platform is commonly built for runtime CPU detection. If you modify + * any of the function mappings present in this file, be sure to also update + * them in the function pointer initialization code + */ +#if HAVE_MMX + +#endif + + +#if HAVE_SSE2 +extern prototype_quantize_block(vp8_regular_quantize_b_sse2); + +#if !CONFIG_RUNTIME_CPU_DETECT + +/* The sse2 quantizer has not been updated to match the new exact + * quantizer introduced in commit e04e2935 + *#undef vp8_quantize_quantb + *#define vp8_quantize_quantb vp8_regular_quantize_b_sse2 + */ + +#endif + +#endif + + +#endif diff --git a/vp8/encoder/x86/sad_mmx.asm b/vp8/encoder/x86/sad_mmx.asm index a825698..ad9658b 100644 --- a/vp8/encoder/x86/sad_mmx.asm +++ b/vp8/encoder/x86/sad_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/x86/sad_sse2.asm b/vp8/encoder/x86/sad_sse2.asm index 53240bb..9f34a7a 100644 --- a/vp8/encoder/x86/sad_sse2.asm +++ b/vp8/encoder/x86/sad_sse2.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/x86/sad_sse3.asm b/vp8/encoder/x86/sad_sse3.asm index 38cc029..c2a1ae7 100644 --- a/vp8/encoder/x86/sad_sse3.asm +++ b/vp8/encoder/x86/sad_sse3.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/x86/sad_ssse3.asm b/vp8/encoder/x86/sad_ssse3.asm index 1bb9561..94bbfff 100644 --- a/vp8/encoder/x86/sad_ssse3.asm +++ b/vp8/encoder/x86/sad_ssse3.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/x86/subtract_mmx.asm b/vp8/encoder/x86/subtract_mmx.asm index ce3e610..8fe3ee1 100644 --- a/vp8/encoder/x86/subtract_mmx.asm +++ b/vp8/encoder/x86/subtract_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -14,7 +15,7 @@ ; unsigned short *diff, unsigned char *Predictor, ; int pitch); global sym(vp8_subtract_b_mmx_impl) -sym(vp8_subtract_b_mmx_impl) +sym(vp8_subtract_b_mmx_impl): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 5 @@ -150,7 +151,7 @@ submby_loop: ;void vp8_subtract_mbuv_mmx(short *diff, unsigned char *usrc, unsigned char *vsrc, unsigned char *pred, int stride) global sym(vp8_subtract_mbuv_mmx) -sym(vp8_subtract_mbuv_mmx) +sym(vp8_subtract_mbuv_mmx): push rbp mov rbp, rsp SHADOW_ARGS_TO_STACK 5 diff --git a/vp8/encoder/x86/variance_impl_mmx.asm b/vp8/encoder/x86/variance_impl_mmx.asm index d0da82a..173238e 100644 --- a/vp8/encoder/x86/variance_impl_mmx.asm +++ b/vp8/encoder/x86/variance_impl_mmx.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/x86/variance_impl_sse2.asm b/vp8/encoder/x86/variance_impl_sse2.asm index 7e5ee28..f47d9cc 100644 --- a/vp8/encoder/x86/variance_impl_sse2.asm +++ b/vp8/encoder/x86/variance_impl_sse2.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vp8/encoder/x86/variance_mmx.c b/vp8/encoder/x86/variance_mmx.c index 4a5b25b..2600ce9 100644 --- a/vp8/encoder/x86/variance_mmx.c +++ b/vp8/encoder/x86/variance_mmx.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/x86/variance_sse2.c b/vp8/encoder/x86/variance_sse2.c index ea80753..5e750ba 100644 --- a/vp8/encoder/x86/variance_sse2.c +++ b/vp8/encoder/x86/variance_sse2.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vp8/encoder/x86/variance_x86.h b/vp8/encoder/x86/variance_x86.h index 35fc90c..3c9f9c7 100644 --- a/vp8/encoder/x86/variance_x86.h +++ b/vp8/encoder/x86/variance_x86.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -240,7 +241,7 @@ extern prototype_sad_multi_dif_address(vp8_sad4x4x4d_sse3); #define vp8_variance_sad4x4x3 vp8_sad4x4x3_sse3 #undef vp8_variance_sad16x16x4d -#define vp8_variance_sad16x16x4 vp8_sad16x16x4d_sse3 +#define vp8_variance_sad16x16x4d vp8_sad16x16x4d_sse3 #undef vp8_variance_sad16x8x4d #define vp8_variance_sad16x8x4d vp8_sad16x8x4d_sse3 diff --git a/vp8/encoder/x86/x86_csystemdependent.c b/vp8/encoder/x86/x86_csystemdependent.c index f1391ba..18dc49c 100644 --- a/vp8/encoder/x86/x86_csystemdependent.c +++ b/vp8/encoder/x86/x86_csystemdependent.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -17,15 +18,10 @@ #if HAVE_MMX void vp8_short_fdct8x4_mmx(short *input, short *output, int pitch) { - vp8_short_fdct4x4_mmx(input, output, pitch); - vp8_short_fdct4x4_mmx(input + 4, output + 16, pitch); + vp8_short_fdct4x4_c(input, output, pitch); + vp8_short_fdct4x4_c(input + 4, output + 16, pitch); } -void vp8_fast_fdct8x4_mmx(short *input, short *output, int pitch) -{ - vp8_fast_fdct4x4_mmx(input, output , pitch); - vp8_fast_fdct4x4_mmx(input + 4, output + 16, pitch); -} int vp8_fast_quantize_b_impl_mmx(short *coeff_ptr, short *zbin_ptr, short *qcoeff_ptr, short *dequant_ptr, @@ -86,10 +82,10 @@ void vp8_subtract_b_mmx(BLOCK *be, BLOCKD *bd, int pitch) #endif #if HAVE_SSE2 -void vp8_short_fdct8x4_wmt(short *input, short *output, int pitch) +void vp8_short_fdct8x4_sse2(short *input, short *output, int pitch) { - vp8_short_fdct4x4_wmt(input, output, pitch); - vp8_short_fdct4x4_wmt(input + 4, output + 16, pitch); + vp8_short_fdct4x4_sse2(input, output, pitch); + vp8_short_fdct4x4_sse2(input + 4, output + 16, pitch); } int vp8_fast_quantize_b_impl_sse(short *coeff_ptr, short *zbin_ptr, @@ -120,6 +116,40 @@ void vp8_fast_quantize_b_sse(BLOCK *b, BLOCKD *d) ); } +int vp8_regular_quantize_b_impl_sse2(short *coeff_ptr, short *zbin_ptr, + short *qcoeff_ptr,short *dequant_ptr, + const int *default_zig_zag, short *round_ptr, + short *quant_ptr, short *dqcoeff_ptr, + unsigned short zbin_oq_value, + short *zbin_boost_ptr); + +void vp8_regular_quantize_b_sse2(BLOCK *b,BLOCKD *d) +{ + short *zbin_boost_ptr = &b->zrun_zbin_boost[0]; + short *coeff_ptr = &b->coeff[0]; + short *zbin_ptr = &b->zbin[0][0]; + short *round_ptr = &b->round[0][0]; + short *quant_ptr = &b->quant[0][0]; + short *qcoeff_ptr = d->qcoeff; + short *dqcoeff_ptr = d->dqcoeff; + short *dequant_ptr = &d->dequant[0][0]; + short zbin_oq_value = b->zbin_extra; + + d->eob = vp8_regular_quantize_b_impl_sse2( + coeff_ptr, + zbin_ptr, + qcoeff_ptr, + dequant_ptr, + vp8_default_zig_zag1d, + + round_ptr, + quant_ptr, + dqcoeff_ptr, + zbin_oq_value, + zbin_boost_ptr + ); +} + int vp8_mbblock_error_xmm_impl(short *coeff_ptr, short *dcoef_ptr, int dc); int vp8_mbblock_error_xmm(MACROBLOCK *mb, int dc) { @@ -186,11 +216,19 @@ void vp8_arch_x86_encoder_init(VP8_COMP *cpi) cpi->rtcd.variance.get8x8var = vp8_get8x8var_mmx; cpi->rtcd.variance.get16x16var = vp8_get16x16var_mmx; cpi->rtcd.variance.get4x4sse_cs = vp8_get4x4sse_cs_mmx; - +#if 0 // new fdct cpi->rtcd.fdct.short4x4 = vp8_short_fdct4x4_mmx; cpi->rtcd.fdct.short8x4 = vp8_short_fdct8x4_mmx; - cpi->rtcd.fdct.fast4x4 = vp8_fast_fdct4x4_mmx; - cpi->rtcd.fdct.fast8x4 = vp8_fast_fdct8x4_mmx; + cpi->rtcd.fdct.fast4x4 = vp8_short_fdct4x4_mmx; + cpi->rtcd.fdct.fast8x4 = vp8_short_fdct8x4_mmx; +#else + cpi->rtcd.fdct.short4x4 = vp8_short_fdct4x4_c; + cpi->rtcd.fdct.short8x4 = vp8_short_fdct8x4_c; + cpi->rtcd.fdct.fast4x4 = vp8_short_fdct4x4_c; + cpi->rtcd.fdct.fast8x4 = vp8_short_fdct8x4_c; + +#endif + cpi->rtcd.fdct.walsh_short4x4 = vp8_short_walsh4x4_c; cpi->rtcd.encodemb.berr = vp8_block_error_mmx; @@ -200,7 +238,7 @@ void vp8_arch_x86_encoder_init(VP8_COMP *cpi) cpi->rtcd.encodemb.submby = vp8_subtract_mby_mmx; cpi->rtcd.encodemb.submbuv = vp8_subtract_mbuv_mmx; - cpi->rtcd.quantize.fastquantb = vp8_fast_quantize_b_mmx; + /*cpi->rtcd.quantize.fastquantb = vp8_fast_quantize_b_mmx;*/ } #endif @@ -235,21 +273,20 @@ void vp8_arch_x86_encoder_init(VP8_COMP *cpi) cpi->rtcd.variance.get16x16var = vp8_get16x16var_sse2; /* cpi->rtcd.variance.get4x4sse_cs not implemented for wmt */; -#if 0 - /* short SSE2 DCT currently disabled, does not match the MMX version */ - cpi->rtcd.fdct.short4x4 = vp8_short_fdct4x4_wmt; - cpi->rtcd.fdct.short8x4 = vp8_short_fdct8x4_wmt; -#endif - /* cpi->rtcd.fdct.fast4x4 not implemented for wmt */; - cpi->rtcd.fdct.fast8x4 = vp8_fast_fdct8x4_wmt; - cpi->rtcd.fdct.walsh_short4x4 = vp8_short_walsh4x4_sse2; + cpi->rtcd.fdct.short4x4 = vp8_short_fdct4x4_sse2; + cpi->rtcd.fdct.short8x4 = vp8_short_fdct8x4_sse2; + cpi->rtcd.fdct.fast4x4 = vp8_short_fdct4x4_sse2; + cpi->rtcd.fdct.fast8x4 = vp8_short_fdct8x4_sse2; + + cpi->rtcd.fdct.walsh_short4x4 = vp8_short_walsh4x4_c ; cpi->rtcd.encodemb.berr = vp8_block_error_xmm; cpi->rtcd.encodemb.mberr = vp8_mbblock_error_xmm; cpi->rtcd.encodemb.mbuverr = vp8_mbuverror_xmm; /* cpi->rtcd.encodemb.sub* not implemented for wmt */ - cpi->rtcd.quantize.fastquantb = vp8_fast_quantize_b_sse; + /*cpi->rtcd.quantize.fastquantb = vp8_fast_quantize_b_sse; + cpi->rtcd.quantize.quantb = vp8_regular_quantize_b_sse2;*/ } #endif diff --git a/vp8/exports_dec b/vp8/exports_dec new file mode 100644 index 0000000..f9b985c --- /dev/null +++ b/vp8/exports_dec @@ -0,0 +1 @@ +data vpx_codec_vp8_dx_algo diff --git a/vp8/exports_enc b/vp8/exports_enc new file mode 100644 index 0000000..9967011 --- /dev/null +++ b/vp8/exports_enc @@ -0,0 +1 @@ +data vpx_codec_vp8_cx_algo diff --git a/vp8/vp8_common.mk b/vp8/vp8_common.mk index ec467c5..ecca18a 100644 --- a/vp8/vp8_common.mk +++ b/vp8/vp8_common.mk @@ -1,19 +1,17 @@ ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## #add this file to the installed sources list VP8_COMMON_SRCS-yes += vp8_common.mk -#common interface -VP8_COMMON_SRCS-yes += vp8.h - CFLAGS+=-I$(SRC_PATH_BARE)/$(VP8_PREFIX)common VP8_COMMON_SRCS-yes += common/type_aliases.h VP8_COMMON_SRCS-yes += common/pragmas.h @@ -29,7 +27,6 @@ VP8_COMMON_SRCS-yes += common/onyxd.h CFLAGS+=-I$(SRC_PATH_BARE)/$(VP8_PREFIX)common -VP8_COMMON_SRCS-yes += common/segmentation_common.c VP8_COMMON_SRCS-yes += common/alloccommon.c VP8_COMMON_SRCS-yes += common/blockd.c VP8_COMMON_SRCS-yes += common/coefupdateprobs.h @@ -66,7 +63,6 @@ VP8_COMMON_SRCS-yes += common/recon.h VP8_COMMON_SRCS-yes += common/reconinter.h VP8_COMMON_SRCS-yes += common/reconintra.h VP8_COMMON_SRCS-yes += common/reconintra4x4.h -VP8_COMMON_SRCS-yes += common/segmentation_common.h VP8_COMMON_SRCS-yes += common/setupintrarecon.h VP8_COMMON_SRCS-yes += common/subpixel.h VP8_COMMON_SRCS-yes += common/swapyv12buffer.h @@ -100,17 +96,17 @@ VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/vp8_asm_stubs.c VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/loopfilter_x86.c VP8_COMMON_SRCS-$(CONFIG_POSTPROC) += common/postproc.h VP8_COMMON_SRCS-$(CONFIG_POSTPROC) += common/postproc.c -VP8_COMMON_SRCS-$(CONFIG_VP8_ENCODER) += common/postproc.h -VP8_COMMON_SRCS-$(CONFIG_VP8_ENCODER) += common/postproc.c VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/idctllm_mmx.asm VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/iwalsh_mmx.asm VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/recon_mmx.asm VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/subpixel_mmx.asm VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/loopfilter_mmx.asm +VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/idctllm_sse2.asm VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/recon_sse2.asm VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/subpixel_sse2.asm VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/loopfilter_sse2.asm VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/iwalsh_sse2.asm +VP8_COMMON_SRCS-$(HAVE_SSSE3) += common/x86/subpixel_ssse3.asm ifeq ($(CONFIG_POSTPROC),yes) VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/postproc_mmx.asm VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/postproc_sse2.asm @@ -127,6 +123,7 @@ VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/systemdependent.c VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/vpx_asm_offsets.c VP8_COMMON_SRCS_REMOVE-$(HAVE_ARMV6) += common/filter_c.c +VP8_COMMON_SRCS_REMOVE-$(HAVE_ARMV6) += common/idctllm.c VP8_COMMON_SRCS_REMOVE-$(HAVE_ARMV6) += common/recon.c VP8_COMMON_SRCS_REMOVE-$(HAVE_ARMV6) += common/reconintra4x4.c VP8_COMMON_SRCS_REMOVE-$(HAVE_ARMV6) += common/generic/systemdependent.c @@ -136,6 +133,7 @@ VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/bilinearfilter_v6$(ASM) VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/copymem8x4_v6$(ASM) VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/copymem8x8_v6$(ASM) VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/copymem16x16_v6$(ASM) +VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/dc_only_idct_add_v6$(ASM) VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/iwalsh_v6$(ASM) VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/filter_v6$(ASM) VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/idct_v6$(ASM) @@ -152,6 +150,7 @@ VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/bilinearpredict16x16_neon$(ASM VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/copymem8x4_neon$(ASM) VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/copymem8x8_neon$(ASM) VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/copymem16x16_neon$(ASM) +VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/dc_only_idct_add_neon$(ASM) VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/iwalsh_neon$(ASM) VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/loopfiltersimplehorizontaledge_neon$(ASM) VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/loopfiltersimpleverticaledge_neon$(ASM) diff --git a/vp8/vp8_cx_iface.c b/vp8/vp8_cx_iface.c index e129ec9..8845368 100644 --- a/vp8/vp8_cx_iface.c +++ b/vp8/vp8_cx_iface.c @@ -1,18 +1,19 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ -#include "vpx_codec/vpx_codec.h" -#include "vpx_codec/internal/vpx_codec_internal.h" +#include "vpx/vpx_codec.h" +#include "vpx/internal/vpx_codec_internal.h" #include "vpx_version.h" #include "onyx_int.h" -#include "vp8e.h" +#include "vpx/vp8e.h" #include "onyx.h" #include <stdlib.h> #include <string.h> @@ -52,15 +53,15 @@ static const struct extraconfig_map extracfg_map[] = NULL, #if !(CONFIG_REALTIME_ONLY) VP8_BEST_QUALITY_ENCODING, /* Encoding Mode */ - -4, /* cpu_used */ + 0, /* cpu_used */ #else VP8_REAL_TIME_ENCODING, /* Encoding Mode */ - -8, /* cpu_used */ + 4, /* cpu_used */ #endif 0, /* enable_auto_alt_ref */ 0, /* noise_sensitivity */ 0, /* Sharpness */ - 800, /* static_thresh */ + 0, /* static_thresh */ VP8_ONE_TOKENPARTITION, /* token_partitions */ 0, /* arnr_max_frames */ 0, /* arnr_strength */ @@ -81,7 +82,7 @@ struct vpx_codec_alg_priv vpx_image_t preview_img; unsigned int next_frame_flag; vp8_postproc_cfg_t preview_ppcfg; - vpx_codec_pkt_list_decl(26) pkt_list; // changed to accomendate the maximum number of lagged frames allowed + vpx_codec_pkt_list_decl(64) pkt_list; // changed to accomendate the maximum number of lagged frames allowed int deprecated_mode; unsigned int fixed_kf_cntr; }; @@ -108,7 +109,7 @@ update_error_state(vpx_codec_alg_priv_t *ctx, } while(0) #define RANGE_CHECK(p,memb,lo,hi) do {\ - if(!((p)->memb >= (lo) && (p)->memb <= hi)) \ + if(!(((p)->memb == lo || (p)->memb > (lo)) && (p)->memb <= hi)) \ ERROR(#memb " out of range ["#lo".."#hi"]");\ } while(0) @@ -177,7 +178,7 @@ static vpx_codec_err_t validate_config(vpx_codec_alg_priv_t *ctx, RANGE_CHECK(vp8_cfg, token_partitions, VP8_ONE_TOKENPARTITION, VP8_EIGHT_TOKENPARTITION); RANGE_CHECK(vp8_cfg, Sharpness, 0, 7); - RANGE_CHECK(vp8_cfg, arnr_max_frames, 0, 25); + RANGE_CHECK(vp8_cfg, arnr_max_frames, 0, 15); RANGE_CHECK(vp8_cfg, arnr_strength, 0, 6); RANGE_CHECK(vp8_cfg, arnr_type, 0, 0xffffffff); @@ -211,10 +212,10 @@ static vpx_codec_err_t validate_img(vpx_codec_alg_priv_t *ctx, { switch (img->fmt) { - case IMG_FMT_YV12: - case IMG_FMT_I420: - case IMG_FMT_VPXI420: - case IMG_FMT_VPXYV12: + case VPX_IMG_FMT_YV12: + case VPX_IMG_FMT_I420: + case VPX_IMG_FMT_VPXI420: + case VPX_IMG_FMT_VPXYV12: break; default: ERROR("Invalid image format. Only YV12 and I420 images are supported"); @@ -295,9 +296,9 @@ static vpx_codec_err_t set_vp8e_config(VP8_CONFIG *oxcf, oxcf->under_shoot_pct = cfg.rc_undershoot_pct; //oxcf->over_shoot_pct = cfg.rc_overshoot_pct; - oxcf->maximum_buffer_size = cfg.rc_buf_sz / 1000; - oxcf->starting_buffer_level = cfg.rc_buf_initial_sz / 1000; - oxcf->optimal_buffer_level = cfg.rc_buf_optimal_sz / 1000; + oxcf->maximum_buffer_size = cfg.rc_buf_sz; + oxcf->starting_buffer_level = cfg.rc_buf_initial_sz; + oxcf->optimal_buffer_level = cfg.rc_buf_optimal_sz; oxcf->two_pass_vbrbias = cfg.rc_2pass_vbr_bias_pct; oxcf->two_pass_vbrmin_section = cfg.rc_2pass_vbr_minsection_pct; @@ -536,20 +537,20 @@ static vpx_codec_err_t image2yuvconfig(const vpx_image_t *img, YV12_BUFFER_CONFIG *yv12) { vpx_codec_err_t res = VPX_CODEC_OK; - yv12->y_buffer = img->planes[PLANE_Y]; - yv12->u_buffer = img->planes[PLANE_U]; - yv12->v_buffer = img->planes[PLANE_V]; + yv12->y_buffer = img->planes[VPX_PLANE_Y]; + yv12->u_buffer = img->planes[VPX_PLANE_U]; + yv12->v_buffer = img->planes[VPX_PLANE_V]; yv12->y_width = img->d_w; yv12->y_height = img->d_h; yv12->uv_width = (1 + yv12->y_width) / 2; yv12->uv_height = (1 + yv12->y_height) / 2; - yv12->y_stride = img->stride[PLANE_Y]; - yv12->uv_stride = img->stride[PLANE_U]; + yv12->y_stride = img->stride[VPX_PLANE_Y]; + yv12->uv_stride = img->stride[VPX_PLANE_U]; - yv12->border = (img->stride[PLANE_Y] - img->w) / 2; - yv12->clrtype = (img->fmt == IMG_FMT_VPXI420 || img->fmt == IMG_FMT_VPXYV12); //REG_YUV = 0 + yv12->border = (img->stride[VPX_PLANE_Y] - img->w) / 2; + yv12->clrtype = (img->fmt == VPX_IMG_FMT_VPXI420 || img->fmt == VPX_IMG_FMT_VPXYV12); //REG_YUV = 0 return res; } @@ -825,7 +826,9 @@ static vpx_codec_err_t vp8e_set_previewpp(vpx_codec_alg_priv_t *ctx, int ctr_id, va_list args) { +#if CONFIG_POSTPROC vp8_postproc_cfg_t *data = va_arg(args, vp8_postproc_cfg_t *); + (void)ctr_id; if (data) { @@ -834,6 +837,12 @@ static vpx_codec_err_t vp8e_set_previewpp(vpx_codec_alg_priv_t *ctx, } else return VPX_CODEC_INVALID_PARAM; +#else + (void)ctx; + (void)ctr_id; + (void)args; + return VPX_CODEC_INCAPABLE; +#endif } @@ -846,7 +855,7 @@ static vpx_image_t *vp8e_get_preview(vpx_codec_alg_priv_t *ctx) { /* - vpx_img_wrap(&ctx->preview_img, IMG_FMT_YV12, + vpx_img_wrap(&ctx->preview_img, VPX_IMG_FMT_YV12, sd.y_width + 2*VP8BORDERINPIXELS, sd.y_height + 2*VP8BORDERINPIXELS, 1, @@ -857,23 +866,23 @@ static vpx_image_t *vp8e_get_preview(vpx_codec_alg_priv_t *ctx) */ ctx->preview_img.bps = 12; - ctx->preview_img.planes[PLANE_Y] = sd.y_buffer; - ctx->preview_img.planes[PLANE_U] = sd.u_buffer; - ctx->preview_img.planes[PLANE_V] = sd.v_buffer; + ctx->preview_img.planes[VPX_PLANE_Y] = sd.y_buffer; + ctx->preview_img.planes[VPX_PLANE_U] = sd.u_buffer; + ctx->preview_img.planes[VPX_PLANE_V] = sd.v_buffer; if (sd.clrtype == REG_YUV) - ctx->preview_img.fmt = IMG_FMT_I420; + ctx->preview_img.fmt = VPX_IMG_FMT_I420; else - ctx->preview_img.fmt = IMG_FMT_VPXI420; + ctx->preview_img.fmt = VPX_IMG_FMT_VPXI420; ctx->preview_img.x_chroma_shift = 1; ctx->preview_img.y_chroma_shift = 1; ctx->preview_img.d_w = ctx->cfg.g_w; ctx->preview_img.d_h = ctx->cfg.g_h; - ctx->preview_img.stride[PLANE_Y] = sd.y_stride; - ctx->preview_img.stride[PLANE_U] = sd.uv_stride; - ctx->preview_img.stride[PLANE_V] = sd.uv_stride; + ctx->preview_img.stride[VPX_PLANE_Y] = sd.y_stride; + ctx->preview_img.stride[VPX_PLANE_U] = sd.uv_stride; + ctx->preview_img.stride[VPX_PLANE_V] = sd.uv_stride; ctx->preview_img.w = sd.y_width; ctx->preview_img.h = sd.y_height; @@ -1023,7 +1032,7 @@ static vpx_codec_enc_cfg_map_t vp8e_usage_cfg_map[] = 0, /* g_lag_in_frames */ - 70, /* rc_dropframe_thresh */ + 0, /* rc_dropframe_thresh */ 0, /* rc_resize_allowed */ 60, /* rc_resize_down_thresold */ 30, /* rc_resize_up_thresold */ @@ -1067,7 +1076,7 @@ static vpx_codec_enc_cfg_map_t vp8e_usage_cfg_map[] = #endif vpx_codec_iface_t vpx_codec_vp8_cx_algo = { - "vpx Technologies VP8 Encoder" VERSION_STRING, + "WebM Project VP8 Encoder" VERSION_STRING, VPX_CODEC_INTERNAL_ABI_VERSION, VPX_CODEC_CAP_ENCODER | VPX_CODEC_CAP_PSNR, /* vpx_codec_caps_t caps; */ @@ -1156,7 +1165,7 @@ static vpx_codec_err_t api1_encode(vpx_codec_alg_priv_t *ctx, vpx_codec_iface_t vpx_enc_vp8_algo = { - "vpx Technologies VP8 Encoder (Deprecated API)" VERSION_STRING, + "WebM Project VP8 Encoder (Deprecated API)" VERSION_STRING, VPX_CODEC_INTERNAL_ABI_VERSION, VPX_CODEC_CAP_ENCODER, /* vpx_codec_caps_t caps; */ diff --git a/vp8/vp8_dx_iface.c b/vp8/vp8_dx_iface.c index 3e6cdf4..e7e5356 100644 --- a/vp8/vp8_dx_iface.c +++ b/vp8/vp8_dx_iface.c @@ -1,18 +1,19 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ #include <stdlib.h> #include <string.h> -#include "vpx_codec/vpx_decoder.h" -#include "vp8dx.h" -#include "vpx_codec/internal/vpx_codec_internal.h" +#include "vpx/vpx_decoder.h" +#include "vpx/vp8dx.h" +#include "vpx/internal/vpx_codec_internal.h" #include "vpx_version.h" #include "onyxd.h" #include "onyxd_int.h" @@ -195,9 +196,6 @@ static void vp8_finalize_mmaps(vpx_codec_alg_priv_t *ctx) ctx->pbi->fb_storage_ptr[0] = mmap_lkup(ctx, VP6_SEG_IMG0_STRG); ctx->pbi->fb_storage_ptr[1] = mmap_lkup(ctx, VP6_SEG_IMG1_STRG); ctx->pbi->fb_storage_ptr[2] = mmap_lkup(ctx, VP6_SEG_IMG2_STRG); - #if CONFIG_NEW_TOKENS - ctx->pbi->token_graph = mmap_lkup(ctx, VP6_SEG_TOKEN_GRAPH); - #endif #if CONFIG_POSTPROC ctx->pbi->postproc.deblock.fragment_variances = mmap_lkup(ctx, VP6_SEG_DEBLOCKER); ctx->pbi->fb_storage_ptr[3] = mmap_lkup(ctx, VP6_SEG_PP_IMG_STRG); @@ -225,11 +223,12 @@ static vpx_codec_err_t vp8_init(vpx_codec_ctx_t *ctx) res = vp8_mmap_alloc(&mmap); if (!res) + { vp8_init_ctx(ctx, &mmap); - ctx->priv->alg_priv->defer_alloc = 1; - /*post processing level initialized to do nothing */ - + ctx->priv->alg_priv->defer_alloc = 1; + /*post processing level initialized to do nothing */ + } } return res; @@ -257,12 +256,12 @@ static vpx_codec_err_t vp8_peek_si(const uint8_t *data, vpx_codec_err_t res = VPX_CODEC_OK; { - /*Parse from VP8 compressed data, the implies knowledge of the - *VP8 bitsteam. - * First 3 byte header including version, frame type and an offset - * Next 3 bytes are image sizewith 12 bit each for width and height + /* Parse uncompresssed part of key frame header. + * 3 bytes:- including version, frame type and an offset + * 3 bytes:- sync code (0x9d, 0x01, 0x2a) + * 4 bytes:- including image width and height in the lowest 14 bits + * of each 2-byte value. */ - si->is_kf = 0; if (data_sz >= 10 && !(data[0] & 0x01)) /* I-Frame */ @@ -434,7 +433,7 @@ static vpx_codec_err_t vp8_decode(vpx_codec_alg_priv_t *ctx, unsigned int a_w = (sd.y_width + 15) & ~15; unsigned int a_h = (sd.y_height + 15) & ~15; - vpx_img_wrap(&ctx->img, IMG_FMT_I420, + vpx_img_wrap(&ctx->img, VPX_IMG_FMT_I420, a_w + 2 * VP8BORDERINPIXELS, a_h + 2 * VP8BORDERINPIXELS, 1, @@ -529,7 +528,7 @@ static vpx_codec_err_t vp8_xma_set_mmap(vpx_codec_ctx_t *ctx, done = 1; - if (ctx->priv->alg_priv) + if (!res && ctx->priv->alg_priv) { for (i = 0; i < NELEMENTS(vp8_mem_req_segs); i++) { @@ -557,20 +556,20 @@ static vpx_codec_err_t image2yuvconfig(const vpx_image_t *img, YV12_BUFFER_CONFIG *yv12) { vpx_codec_err_t res = VPX_CODEC_OK; - yv12->y_buffer = img->planes[PLANE_Y]; - yv12->u_buffer = img->planes[PLANE_U]; - yv12->v_buffer = img->planes[PLANE_V]; + yv12->y_buffer = img->planes[VPX_PLANE_Y]; + yv12->u_buffer = img->planes[VPX_PLANE_U]; + yv12->v_buffer = img->planes[VPX_PLANE_V]; yv12->y_width = img->d_w; yv12->y_height = img->d_h; yv12->uv_width = yv12->y_width / 2; yv12->uv_height = yv12->y_height / 2; - yv12->y_stride = img->stride[PLANE_Y]; - yv12->uv_stride = img->stride[PLANE_U]; + yv12->y_stride = img->stride[VPX_PLANE_Y]; + yv12->uv_stride = img->stride[VPX_PLANE_U]; - yv12->border = (img->stride[PLANE_Y] - img->d_w) / 2; - yv12->clrtype = (img->fmt == IMG_FMT_VPXI420 || img->fmt == IMG_FMT_VPXYV12); + yv12->border = (img->stride[VPX_PLANE_Y] - img->d_w) / 2; + yv12->clrtype = (img->fmt == VPX_IMG_FMT_VPXI420 || img->fmt == VPX_IMG_FMT_VPXYV12); return res; } @@ -656,7 +655,7 @@ vpx_codec_ctrl_fn_map_t vp8_ctf_maps[] = #endif vpx_codec_iface_t vpx_codec_vp8_dx_algo = { - "vpx Technologies VP8 Decoder" VERSION_STRING, + "WebM Project VP8 Decoder" VERSION_STRING, VPX_CODEC_INTERNAL_ABI_VERSION, VPX_CODEC_CAP_DECODER | VP8_CAP_POSTPROC, /* vpx_codec_caps_t caps; */ @@ -679,7 +678,7 @@ vpx_codec_iface_t vpx_codec_vp8_dx_algo = */ vpx_codec_iface_t vpx_codec_vp8_algo = { - "vpx Technologies VP8 Decoder (Deprecated API)" VERSION_STRING, + "WebM Project VP8 Decoder (Deprecated API)" VERSION_STRING, VPX_CODEC_INTERNAL_ABI_VERSION, VPX_CODEC_CAP_DECODER | VP8_CAP_POSTPROC, /* vpx_codec_caps_t caps; */ diff --git a/vp8/vp8cx.mk b/vp8/vp8cx.mk index e7e7663..4ce18b6 100644 --- a/vp8/vp8cx.mk +++ b/vp8/vp8cx.mk @@ -1,14 +1,18 @@ ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8_common.mk + +VP8_CX_EXPORTS += exports_enc + VP8_CX_SRCS-yes += $(VP8_COMMON_SRCS-yes) VP8_CX_SRCS-no += $(VP8_COMMON_SRCS-no) VP8_CX_SRCS_REMOVE-yes += $(VP8_COMMON_SRCS_REMOVE-yes) @@ -18,7 +22,7 @@ ifeq ($(ARCH_ARM),yes) include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8cx_arm.mk endif -VP8_CX_SRCS-yes += vp8cx.h vp8e.h vp8_cx_iface.c +VP8_CX_SRCS-yes += vp8_cx_iface.c # encoder #INCLUDES += algo/vpx_common/vpx_mem/include @@ -70,10 +74,14 @@ VP8_CX_SRCS-yes += encoder/quantize.c VP8_CX_SRCS-yes += encoder/ratectrl.c VP8_CX_SRCS-yes += encoder/rdopt.c VP8_CX_SRCS-yes += encoder/sad_c.c -VP8_CX_SRCS-yes += encoder/ssim.c +VP8_CX_SRCS-yes += encoder/segmentation.c +VP8_CX_SRCS-yes += encoder/segmentation.h +VP8_CX_SRCS-$(CONFIG_PSNR) += encoder/ssim.c VP8_CX_SRCS-yes += encoder/tokenize.c VP8_CX_SRCS-yes += encoder/treewriter.c VP8_CX_SRCS-yes += encoder/variance_c.c +VP8_CX_SRCS-$(CONFIG_PSNR) += common/postproc.h +VP8_CX_SRCS-$(CONFIG_PSNR) += common/postproc.c ifeq ($(CONFIG_REALTIME_ONLY),yes) VP8_CX_SRCS_REMOVE-yes += encoder/firstpass.c @@ -83,22 +91,22 @@ VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/encodemb_x86.h VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/dct_x86.h VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/mcomp_x86.h VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/variance_x86.h +VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/quantize_x86.h VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/x86_csystemdependent.c VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/variance_mmx.c VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/variance_impl_mmx.asm VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/sad_mmx.asm VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/dct_mmx.asm VP8_CX_SRCS-$(HAVE_MMX) += encoder/x86/subtract_mmx.asm +VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/dct_sse2.asm VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/variance_sse2.c VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/variance_impl_sse2.asm VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/sad_sse2.asm -VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/dct_sse2.asm VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/fwalsh_sse2.asm +VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/quantize_sse2.asm VP8_CX_SRCS-$(HAVE_SSE3) += encoder/x86/sad_sse3.asm VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/sad_ssse3.asm VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/quantize_mmx.asm VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/encodeopt.asm VP8_CX_SRCS-yes := $(filter-out $(VP8_CX_SRCS_REMOVE-yes),$(VP8_CX_SRCS-yes)) - -INSTALL-LIBS-yes += include/vp8.h include/vp8e.h include/vp8cx.h diff --git a/vp8/vp8cx_arm.mk b/vp8/vp8cx_arm.mk index f0753d9..1424bd1 100644 --- a/vp8/vp8cx_arm.mk +++ b/vp8/vp8cx_arm.mk @@ -1,10 +1,11 @@ ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## diff --git a/vp8/vp8dx.mk b/vp8/vp8dx.mk index e6af543..9419617 100644 --- a/vp8/vp8dx.mk +++ b/vp8/vp8dx.mk @@ -1,14 +1,18 @@ ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8_common.mk + +VP8_DX_EXPORTS += exports_dec + VP8_DX_SRCS-yes += $(VP8_COMMON_SRCS-yes) VP8_DX_SRCS-no += $(VP8_COMMON_SRCS-no) VP8_DX_SRCS_REMOVE-yes += $(VP8_COMMON_SRCS_REMOVE-yes) @@ -18,7 +22,7 @@ ifeq ($(ARCH_ARM),yes) include $(SRC_PATH_BARE)/$(VP8_PREFIX)vp8dx_arm.mk endif -VP8_DX_SRCS-yes += vp8dx.h vp8_dx_iface.c +VP8_DX_SRCS-yes += vp8_dx_iface.c CFLAGS+=-I$(SRC_PATH_BARE)/$(VP8_PREFIX)decoder @@ -26,7 +30,6 @@ CFLAGS+=-I$(SRC_PATH_BARE)/$(VP8_PREFIX)decoder # common #define ARM #define DISABLE_THREAD -#define INLINE=__forceinline #INCLUDES += algo/vpx_common/vpx_mem/include #INCLUDES += common @@ -40,7 +43,6 @@ CFLAGS+=-I$(SRC_PATH_BARE)/$(VP8_PREFIX)decoder # decoder #define ARM #define DISABLE_THREAD -#define INLINE=__forceinline #INCLUDES += algo/vpx_common/vpx_mem/include #INCLUDES += common @@ -52,25 +54,24 @@ CFLAGS+=-I$(SRC_PATH_BARE)/$(VP8_PREFIX)decoder VP8_DX_SRCS-yes += decoder/dboolhuff.c VP8_DX_SRCS-yes += decoder/decodemv.c VP8_DX_SRCS-yes += decoder/decodframe.c -VP8_DX_SRCS-yes += decoder/demode.c VP8_DX_SRCS-yes += decoder/dequantize.c VP8_DX_SRCS-yes += decoder/detokenize.c VP8_DX_SRCS-yes += decoder/generic/dsystemdependent.c VP8_DX_SRCS-yes += decoder/dboolhuff.h VP8_DX_SRCS-yes += decoder/decodemv.h VP8_DX_SRCS-yes += decoder/decoderthreading.h -VP8_DX_SRCS-yes += decoder/demode.h VP8_DX_SRCS-yes += decoder/dequantize.h VP8_DX_SRCS-yes += decoder/detokenize.h VP8_DX_SRCS-yes += decoder/onyxd_int.h VP8_DX_SRCS-yes += decoder/treereader.h VP8_DX_SRCS-yes += decoder/onyxd_if.c VP8_DX_SRCS-yes += decoder/threading.c +VP8_DX_SRCS-yes += decoder/idct_blk.c VP8_DX_SRCS-yes := $(filter-out $(VP8_DX_SRCS_REMOVE-yes),$(VP8_DX_SRCS-yes)) -INSTALL-LIBS-yes += include/vp8.h include/vp8dx.h - VP8_DX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += decoder/x86/dequantize_x86.h VP8_DX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += decoder/x86/x86_dsystemdependent.c VP8_DX_SRCS-$(HAVE_MMX) += decoder/x86/dequantize_mmx.asm +VP8_DX_SRCS-$(HAVE_MMX) += decoder/x86/idct_blk_mmx.c +VP8_DX_SRCS-$(HAVE_SSE2) += decoder/x86/idct_blk_sse2.c diff --git a/vp8/vp8dx_arm.mk b/vp8/vp8dx_arm.mk index 1b4a7ec..989232c 100644 --- a/vp8/vp8dx_arm.mk +++ b/vp8/vp8dx_arm.mk @@ -1,44 +1,31 @@ ## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. ## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. ## #VP8_DX_SRCS list is modified according to different platforms. -#File list for arm -# decoder -#VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/decodframe_arm.c VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/dequantize_arm.c VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/dsystemdependent.c - -#VP8_DX_SRCS_REMOVE-$(HAVE_ARMV6) += decoder/decodframe.c -VP8_DX_SRCS_REMOVE-$(HAVE_ARMV6) += decoder/dequantize.c VP8_DX_SRCS_REMOVE-$(HAVE_ARMV6) += decoder/generic/dsystemdependent.c +VP8_DX_SRCS_REMOVE-$(HAVE_ARMV6) += decoder/dequantize.c +VP8_DX_SRCS_REMOVE-$(HAVE_ARMV6) += decoder/idct_blk.c +VP8_DX_SRCS-$(CONFIG_ARM_ASM_DETOK) += decoder/arm/detokenize$(ASM) #File list for armv6 -# decoder -VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/dequantdcidct_v6$(ASM) -VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/dequantidct_v6$(ASM) +VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/dequant_dc_idct_v6$(ASM) +VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/dequant_idct_v6$(ASM) VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/dequantize_v6$(ASM) +VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/idct_blk_v6.c #File list for neon -# decoder -VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/dequantdcidct_neon$(ASM) -VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/dequantidct_neon$(ASM) +VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/dequant_dc_idct_neon$(ASM) +VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/dequant_idct_neon$(ASM) VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/dequantizeb_neon$(ASM) - - -#for new token test -ifeq ($(ARCH_ARM),yes) -VP8_DX_SRCS-$(CONFIG_NEW_TOKENS) += decoder/arm/detokenize_arm_sjl.c -VP8_DX_SRCS-$(CONFIG_NEW_TOKENS) += decoder/arm/detokenize_arm_v6$(ASM) -VP8_DX_SRCS-$(CONFIG_NEW_TOKENS) += decoder/onyxd_if_sjl.c - -VP8_DX_SRCS_REMOVE-$(CONFIG_NEW_TOKENS) += decoder/arm/detokenize_arm.c -VP8_DX_SRCS_REMOVE-$(CONFIG_NEW_TOKENS) += decoder/onyxd_if.c -endif +VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/idct_blk_neon.c diff --git a/vpx/exports_com b/vpx/exports_com new file mode 100644 index 0000000..2ab0509 --- /dev/null +++ b/vpx/exports_com @@ -0,0 +1,16 @@ +text vpx_codec_build_config +text vpx_codec_control_ +text vpx_codec_destroy +text vpx_codec_err_to_string +text vpx_codec_error +text vpx_codec_error_detail +text vpx_codec_get_caps +text vpx_codec_iface_name +text vpx_codec_version +text vpx_codec_version_extra_str +text vpx_codec_version_str +text vpx_img_alloc +text vpx_img_flip +text vpx_img_free +text vpx_img_set_rect +text vpx_img_wrap diff --git a/vpx/exports_dec b/vpx/exports_dec new file mode 100644 index 0000000..ed121f7 --- /dev/null +++ b/vpx/exports_dec @@ -0,0 +1,9 @@ +text vpx_codec_dec_init_ver +text vpx_codec_decode +text vpx_codec_get_frame +text vpx_codec_get_mem_map +text vpx_codec_get_stream_info +text vpx_codec_peek_stream_info +text vpx_codec_register_put_frame_cb +text vpx_codec_register_put_slice_cb +text vpx_codec_set_mem_map diff --git a/vpx/exports_enc b/vpx/exports_enc new file mode 100644 index 0000000..3d56749 --- /dev/null +++ b/vpx/exports_enc @@ -0,0 +1,8 @@ +text vpx_codec_enc_config_default +text vpx_codec_enc_config_set +text vpx_codec_enc_init_ver +text vpx_codec_encode +text vpx_codec_get_cx_data +text vpx_codec_get_global_headers +text vpx_codec_get_preview_frame +text vpx_codec_set_cx_data_buf diff --git a/vpx_codec/internal/vpx_codec_internal.h b/vpx/internal/vpx_codec_internal.h index 0867552..ab4cad1 100644 --- a/vpx_codec/internal/vpx_codec_internal.h +++ b/vpx/internal/vpx_codec_internal.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -55,7 +56,7 @@ * types, removing or reassigning enums, adding/removing/rearranging * fields to structures */ -#define VPX_CODEC_INTERNAL_ABI_VERSION (2) /**<\hideinitializer*/ +#define VPX_CODEC_INTERNAL_ABI_VERSION (3) /**<\hideinitializer*/ typedef struct vpx_codec_alg_priv vpx_codec_alg_priv_t; @@ -137,7 +138,7 @@ typedef vpx_codec_err_t (*vpx_codec_get_si_fn_t)(vpx_codec_alg_priv_t *ctx, * provide type safety for the exchanged data or assign meanings to the * control codes. Those details should be specified in the algorithm's * header file. In particular, the ctrl_id parameter is guaranteed to exist - * in the algorithm's control mapping table, and the data paramter may be NULL. + * in the algorithm's control mapping table, and the data parameter may be NULL. * * * \param[in] ctx Pointer to this instance's context @@ -314,7 +315,7 @@ struct vpx_codec_iface }; /*!\brief Callback function pointer / user data pair storage */ -typedef struct +typedef struct vpx_codec_priv_cb_pair { union { @@ -339,7 +340,6 @@ struct vpx_codec_priv vpx_codec_iface_t *iface; struct vpx_codec_alg_priv *alg_priv; const char *err_detail; - unsigned int eval_counter; vpx_codec_flags_t init_flags; struct { diff --git a/vpx_codec/src/vpx_codec.c b/vpx/src/vpx_codec.c index 6366416..9c1558c 100644 --- a/vpx_codec/src/vpx_codec.c +++ b/vpx/src/vpx_codec.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -12,9 +13,9 @@ * \brief Provides the high level interface to wrap decoder algorithms. * */ -#include <stdlib.h> -#include <string.h> -#include "vpx_codec/internal/vpx_codec_internal.h" +#include <stdarg.h> +#include "vpx/vpx_integer.h" +#include "vpx/internal/vpx_codec_internal.h" #include "vpx_version.h" #define SAVE_STATUS(ctx,var) (ctx?(ctx->err = var):var) @@ -86,53 +87,6 @@ const char *vpx_codec_error_detail(vpx_codec_ctx_t *ctx) } -vpx_codec_err_t vpx_codec_dec_init_ver(vpx_codec_ctx_t *ctx, - vpx_codec_iface_t *iface, - vpx_codec_dec_cfg_t *cfg, - vpx_codec_flags_t flags, - int ver) -{ - vpx_codec_err_t res; - - if (ver != VPX_DECODER_ABI_VERSION) - res = VPX_CODEC_ABI_MISMATCH; - else if (!ctx || !iface) - res = VPX_CODEC_INVALID_PARAM; - else if (iface->abi_version != VPX_CODEC_INTERNAL_ABI_VERSION) - res = VPX_CODEC_ABI_MISMATCH; - else if ((flags & VPX_CODEC_USE_XMA) && !(iface->caps & VPX_CODEC_CAP_XMA)) - res = VPX_CODEC_INCAPABLE; - else if ((flags & VPX_CODEC_USE_POSTPROC) && !(iface->caps & VPX_CODEC_CAP_POSTPROC)) - res = VPX_CODEC_INCAPABLE; - else - { - memset(ctx, 0, sizeof(*ctx)); - ctx->iface = iface; - ctx->name = iface->name; - ctx->priv = NULL; - ctx->init_flags = flags; - ctx->config.dec = cfg; - res = VPX_CODEC_OK; - - if (!(flags & VPX_CODEC_USE_XMA)) - { - res = ctx->iface->init(ctx); - - if (res) - { - ctx->err_detail = ctx->priv ? ctx->priv->err_detail : NULL; - vpx_codec_destroy(ctx); - } - - if (ctx->priv) - ctx->priv->iface = ctx->iface; - } - } - - return SAVE_STATUS(ctx, res); -} - - vpx_codec_err_t vpx_codec_destroy(vpx_codec_ctx_t *ctx) { vpx_codec_err_t res; diff --git a/vpx_codec/src/vpx_decoder.c b/vpx/src/vpx_decoder.c index 7e8575f..b52470b 100644 --- a/vpx_codec/src/vpx_decoder.c +++ b/vpx/src/vpx_decoder.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -12,11 +13,58 @@ * \brief Provides the high level interface to wrap decoder algorithms. * */ -#include <stdlib.h> -#include "vpx_codec/internal/vpx_codec_internal.h" +#include <string.h> +#include "vpx/internal/vpx_codec_internal.h" #define SAVE_STATUS(ctx,var) (ctx?(ctx->err = var):var) +vpx_codec_err_t vpx_codec_dec_init_ver(vpx_codec_ctx_t *ctx, + vpx_codec_iface_t *iface, + vpx_codec_dec_cfg_t *cfg, + vpx_codec_flags_t flags, + int ver) +{ + vpx_codec_err_t res; + + if (ver != VPX_DECODER_ABI_VERSION) + res = VPX_CODEC_ABI_MISMATCH; + else if (!ctx || !iface) + res = VPX_CODEC_INVALID_PARAM; + else if (iface->abi_version != VPX_CODEC_INTERNAL_ABI_VERSION) + res = VPX_CODEC_ABI_MISMATCH; + else if ((flags & VPX_CODEC_USE_XMA) && !(iface->caps & VPX_CODEC_CAP_XMA)) + res = VPX_CODEC_INCAPABLE; + else if ((flags & VPX_CODEC_USE_POSTPROC) && !(iface->caps & VPX_CODEC_CAP_POSTPROC)) + res = VPX_CODEC_INCAPABLE; + else + { + memset(ctx, 0, sizeof(*ctx)); + ctx->iface = iface; + ctx->name = iface->name; + ctx->priv = NULL; + ctx->init_flags = flags; + ctx->config.dec = cfg; + res = VPX_CODEC_OK; + + if (!(flags & VPX_CODEC_USE_XMA)) + { + res = ctx->iface->init(ctx); + + if (res) + { + ctx->err_detail = ctx->priv ? ctx->priv->err_detail : NULL; + vpx_codec_destroy(ctx); + } + + if (ctx->priv) + ctx->priv->iface = ctx->iface; + } + } + + return SAVE_STATUS(ctx, res); +} + + vpx_codec_err_t vpx_codec_peek_stream_info(vpx_codec_iface_t *iface, const uint8_t *data, unsigned int data_sz, @@ -74,22 +122,10 @@ vpx_codec_err_t vpx_codec_decode(vpx_codec_ctx_t *ctx, res = VPX_CODEC_INVALID_PARAM; else if (!ctx->iface || !ctx->priv) res = VPX_CODEC_ERROR; - -#if CONFIG_EVAL_LIMIT - else if (ctx->priv->eval_counter >= 500) - { - ctx->priv->err_detail = "Evaluation limit exceeded."; - res = VPX_CODEC_ERROR; - } - -#endif else { res = ctx->iface->dec.decode(ctx->priv->alg_priv, data, data_sz, user_priv, deadline); -#if CONFIG_EVAL_LIMIT - ctx->priv->eval_counter++; -#endif } return SAVE_STATUS(ctx, res); diff --git a/vpx_codec/src/vpx_decoder_compat.c b/vpx/src/vpx_decoder_compat.c index d5b04ae..e264734 100644 --- a/vpx_codec/src/vpx_decoder_compat.c +++ b/vpx/src/vpx_decoder_compat.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -14,8 +15,8 @@ */ #include <stdlib.h> #include <string.h> -#include "vpx_codec/vpx_decoder.h" -#include "vpx_codec/internal/vpx_codec_internal.h" +#include "vpx/vpx_decoder.h" +#include "vpx/internal/vpx_codec_internal.h" #define SAVE_STATUS(ctx,var) (ctx?(ctx->err = var):var) diff --git a/vpx_codec/src/vpx_encoder.c b/vpx/src/vpx_encoder.c index 98ad8ba..ddbd654 100644 --- a/vpx_codec/src/vpx_encoder.c +++ b/vpx/src/vpx_encoder.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -14,7 +15,7 @@ */ #include <limits.h> #include <string.h> -#include "vpx_codec/internal/vpx_codec_internal.h" +#include "vpx/internal/vpx_codec_internal.h" #define SAVE_STATUS(ctx,var) (ctx?(ctx->err = var):var) @@ -126,15 +127,6 @@ vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx, res = VPX_CODEC_ERROR; else if (!(ctx->iface->caps & VPX_CODEC_CAP_ENCODER)) res = VPX_CODEC_INCAPABLE; - -#if CONFIG_EVAL_LIMIT - else if (ctx->priv->eval_counter >= 500) - { - ctx->priv->err_detail = "Evaluation limit exceeded."; - res = VPX_CODEC_ERROR; - } - -#endif else { /* Execute in a normalized floating point environment, if the platform @@ -144,10 +136,6 @@ vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx, res = ctx->iface->enc.encode(ctx->priv->alg_priv, img, pts, duration, flags, deadline); FLOATING_POINT_RESTORE(); - -#if CONFIG_EVAL_LIMIT - ctx->priv->eval_counter++; -#endif } return SAVE_STATUS(ctx, res); diff --git a/vpx_codec/src/vpx_image.c b/vpx/src/vpx_image.c index 8a16e58..7a4e270 100644 --- a/vpx_codec/src/vpx_image.c +++ b/vpx/src/vpx_image.c @@ -1,19 +1,20 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ #include <stdlib.h> #include <string.h> -#include "vpx_codec/vpx_image.h" +#include "vpx/vpx_image.h" static vpx_image_t *img_alloc_helper(vpx_image_t *img, - img_fmt_t fmt, + vpx_img_fmt_t fmt, unsigned int d_w, unsigned int d_h, unsigned int stride_align, @@ -34,29 +35,29 @@ static vpx_image_t *img_alloc_helper(vpx_image_t *img, /* Get sample size for this format */ switch (fmt) { - case IMG_FMT_RGB32: - case IMG_FMT_RGB32_LE: - case IMG_FMT_ARGB: - case IMG_FMT_ARGB_LE: + case VPX_IMG_FMT_RGB32: + case VPX_IMG_FMT_RGB32_LE: + case VPX_IMG_FMT_ARGB: + case VPX_IMG_FMT_ARGB_LE: bps = 32; break; - case IMG_FMT_RGB24: - case IMG_FMT_BGR24: + case VPX_IMG_FMT_RGB24: + case VPX_IMG_FMT_BGR24: bps = 24; break; - case IMG_FMT_RGB565: - case IMG_FMT_RGB565_LE: - case IMG_FMT_RGB555: - case IMG_FMT_RGB555_LE: - case IMG_FMT_UYVY: - case IMG_FMT_YUY2: - case IMG_FMT_YVYU: + case VPX_IMG_FMT_RGB565: + case VPX_IMG_FMT_RGB565_LE: + case VPX_IMG_FMT_RGB555: + case VPX_IMG_FMT_RGB555_LE: + case VPX_IMG_FMT_UYVY: + case VPX_IMG_FMT_YUY2: + case VPX_IMG_FMT_YVYU: bps = 16; break; - case IMG_FMT_I420: - case IMG_FMT_YV12: - case IMG_FMT_VPXI420: - case IMG_FMT_VPXYV12: + case VPX_IMG_FMT_I420: + case VPX_IMG_FMT_YV12: + case VPX_IMG_FMT_VPXI420: + case VPX_IMG_FMT_VPXYV12: bps = 12; break; default: @@ -67,10 +68,10 @@ static vpx_image_t *img_alloc_helper(vpx_image_t *img, /* Get chroma shift values for this format */ switch (fmt) { - case IMG_FMT_I420: - case IMG_FMT_YV12: - case IMG_FMT_VPXI420: - case IMG_FMT_VPXYV12: + case VPX_IMG_FMT_I420: + case VPX_IMG_FMT_YV12: + case VPX_IMG_FMT_VPXI420: + case VPX_IMG_FMT_VPXYV12: xcs = 1; break; default: @@ -80,10 +81,10 @@ static vpx_image_t *img_alloc_helper(vpx_image_t *img, switch (fmt) { - case IMG_FMT_I420: - case IMG_FMT_YV12: - case IMG_FMT_VPXI420: - case IMG_FMT_VPXYV12: + case VPX_IMG_FMT_I420: + case VPX_IMG_FMT_YV12: + case VPX_IMG_FMT_VPXI420: + case VPX_IMG_FMT_VPXYV12: ycs = 1; break; default: @@ -96,7 +97,7 @@ static vpx_image_t *img_alloc_helper(vpx_image_t *img, w = (d_w + align) & ~align; align = (1 << ycs) - 1; h = (d_h + align) & ~align; - s = (fmt & IMG_FMT_PLANAR) ? w : bps * w / 8; + s = (fmt & VPX_IMG_FMT_PLANAR) ? w : bps * w / 8; s = (s + stride_align - 1) & ~(stride_align - 1); /* Allocate the new image */ @@ -118,7 +119,7 @@ static vpx_image_t *img_alloc_helper(vpx_image_t *img, if (!img_data) { - img->img_data = malloc((fmt & IMG_FMT_PLANAR) ? h * w * bps / 8 : h * s); + img->img_data = malloc((fmt & VPX_IMG_FMT_PLANAR) ? h * w * bps / 8 : h * s); img->img_data_owner = 1; } @@ -133,8 +134,8 @@ static vpx_image_t *img_alloc_helper(vpx_image_t *img, img->bps = bps; /* Calculate strides */ - img->stride[PLANE_Y] = img->stride[PLANE_ALPHA] = s; - img->stride[PLANE_U] = img->stride[PLANE_V] = s >> xcs; + img->stride[VPX_PLANE_Y] = img->stride[VPX_PLANE_ALPHA] = s; + img->stride[VPX_PLANE_U] = img->stride[VPX_PLANE_V] = s >> xcs; /* Default viewport to entire image */ if (!vpx_img_set_rect(img, 0, 0, d_w, d_h)) @@ -146,7 +147,7 @@ fail: } vpx_image_t *vpx_img_alloc(vpx_image_t *img, - img_fmt_t fmt, + vpx_img_fmt_t fmt, unsigned int d_w, unsigned int d_h, unsigned int stride_align) @@ -155,7 +156,7 @@ vpx_image_t *vpx_img_alloc(vpx_image_t *img, } vpx_image_t *vpx_img_wrap(vpx_image_t *img, - img_fmt_t fmt, + vpx_img_fmt_t fmt, unsigned int d_w, unsigned int d_h, unsigned int stride_align, @@ -178,44 +179,44 @@ int vpx_img_set_rect(vpx_image_t *img, img->d_h = h; /* Calculate plane pointers */ - if (!(img->fmt & IMG_FMT_PLANAR)) + if (!(img->fmt & VPX_IMG_FMT_PLANAR)) { - img->planes[PLANE_PACKED] = - img->img_data + x * img->bps / 8 + y * img->stride[PLANE_PACKED]; + img->planes[VPX_PLANE_PACKED] = + img->img_data + x * img->bps / 8 + y * img->stride[VPX_PLANE_PACKED]; } else { data = img->img_data; - if (img->fmt & IMG_FMT_HAS_ALPHA) + if (img->fmt & VPX_IMG_FMT_HAS_ALPHA) { - img->planes[PLANE_ALPHA] = - data + x + y * img->stride[PLANE_ALPHA]; - data += img->h * img->stride[PLANE_ALPHA]; + img->planes[VPX_PLANE_ALPHA] = + data + x + y * img->stride[VPX_PLANE_ALPHA]; + data += img->h * img->stride[VPX_PLANE_ALPHA]; } - img->planes[PLANE_Y] = data + x + y * img->stride[PLANE_Y]; - data += img->h * img->stride[PLANE_Y]; + img->planes[VPX_PLANE_Y] = data + x + y * img->stride[VPX_PLANE_Y]; + data += img->h * img->stride[VPX_PLANE_Y]; - if (!(img->fmt & IMG_FMT_UV_FLIP)) + if (!(img->fmt & VPX_IMG_FMT_UV_FLIP)) { - img->planes[PLANE_U] = data + img->planes[VPX_PLANE_U] = data + (x >> img->x_chroma_shift) - + (y >> img->y_chroma_shift) * img->stride[PLANE_U]; - data += (img->h >> img->y_chroma_shift) * img->stride[PLANE_U]; - img->planes[PLANE_V] = data + + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_U]; + data += (img->h >> img->y_chroma_shift) * img->stride[VPX_PLANE_U]; + img->planes[VPX_PLANE_V] = data + (x >> img->x_chroma_shift) - + (y >> img->y_chroma_shift) * img->stride[PLANE_V]; + + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_V]; } else { - img->planes[PLANE_V] = data + img->planes[VPX_PLANE_V] = data + (x >> img->x_chroma_shift) - + (y >> img->y_chroma_shift) * img->stride[PLANE_V]; - data += (img->h >> img->y_chroma_shift) * img->stride[PLANE_V]; - img->planes[PLANE_U] = data + + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_V]; + data += (img->h >> img->y_chroma_shift) * img->stride[VPX_PLANE_V]; + img->planes[VPX_PLANE_U] = data + (x >> img->x_chroma_shift) - + (y >> img->y_chroma_shift) * img->stride[PLANE_U]; + + (y >> img->y_chroma_shift) * img->stride[VPX_PLANE_U]; } } @@ -233,19 +234,19 @@ void vpx_img_flip(vpx_image_t *img) * stride parameter will be promoted to unsigned, causing errors when * the lhs is a larger type than the rhs. */ - img->planes[PLANE_Y] += (signed)(img->d_h - 1) * img->stride[PLANE_Y]; - img->stride[PLANE_Y] = -img->stride[PLANE_Y]; + img->planes[VPX_PLANE_Y] += (signed)(img->d_h - 1) * img->stride[VPX_PLANE_Y]; + img->stride[VPX_PLANE_Y] = -img->stride[VPX_PLANE_Y]; - img->planes[PLANE_U] += (signed)((img->d_h >> img->y_chroma_shift) - 1) - * img->stride[PLANE_U]; - img->stride[PLANE_U] = -img->stride[PLANE_U]; + img->planes[VPX_PLANE_U] += (signed)((img->d_h >> img->y_chroma_shift) - 1) + * img->stride[VPX_PLANE_U]; + img->stride[VPX_PLANE_U] = -img->stride[VPX_PLANE_U]; - img->planes[PLANE_V] += (signed)((img->d_h >> img->y_chroma_shift) - 1) - * img->stride[PLANE_V]; - img->stride[PLANE_V] = -img->stride[PLANE_V]; + img->planes[VPX_PLANE_V] += (signed)((img->d_h >> img->y_chroma_shift) - 1) + * img->stride[VPX_PLANE_V]; + img->stride[VPX_PLANE_V] = -img->stride[VPX_PLANE_V]; - img->planes[PLANE_ALPHA] += (signed)(img->d_h - 1) * img->stride[PLANE_ALPHA]; - img->stride[PLANE_ALPHA] = -img->stride[PLANE_ALPHA]; + img->planes[VPX_PLANE_ALPHA] += (signed)(img->d_h - 1) * img->stride[VPX_PLANE_ALPHA]; + img->stride[VPX_PLANE_ALPHA] = -img->stride[VPX_PLANE_ALPHA]; } void vpx_img_free(vpx_image_t *img) @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -29,7 +30,7 @@ */ #ifndef VP8_H #define VP8_H -#include "vpx_codec_impl_top.h" +#include "vpx/vpx_codec_impl_top.h" /*!\brief Control functions * @@ -112,5 +113,5 @@ VPX_CTRL_USE_TYPE(VP8_SET_POSTPROC, vp8_postproc_cfg_t *) DECLSPEC_DEPRECATED extern vpx_codec_iface_t vpx_codec_vp8_algo DEPRECATED; #endif -#include "vpx_codec_impl_bottom.h" +#include "vpx/vpx_codec_impl_bottom.h" #endif diff --git a/vp8/vp8cx.h b/vpx/vp8cx.h index dd48c07..e1c8211 100644 --- a/vp8/vp8cx.h +++ b/vpx/vp8cx.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -21,7 +22,7 @@ */ #ifndef VP8CX_H #define VP8CX_H -#include "vpx_codec_impl_top.h" +#include "vpx/vpx_codec_impl_top.h" /*!\brief Algorithm interface for VP8 * @@ -257,5 +258,5 @@ VPX_CTRL_USE_TYPE(VP8E_GET_LAST_QUANTIZER, int *) VPX_CTRL_USE_TYPE(VP8E_GET_LAST_QUANTIZER_64, int *) /*! @} - end defgroup vp8_encoder */ -#include "vpx_codec_impl_bottom.h" +#include "vpx/vpx_codec_impl_bottom.h" #endif diff --git a/vp8/vp8dx.h b/vpx/vp8dx.h index 7310b3b..4cad838 100644 --- a/vp8/vp8dx.h +++ b/vpx/vp8dx.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -21,7 +22,7 @@ */ #ifndef VP8DX_H #define VP8DX_H -#include "vpx_codec_impl_top.h" +#include "vpx/vpx_codec_impl_top.h" /*!\brief Algorithm interface for VP8 * @@ -37,5 +38,5 @@ extern vpx_codec_iface_t vpx_codec_vp8_dx_algo; /*! @} - end defgroup vp8_decoder */ -#include "vpx_codec_impl_bottom.h" +#include "vpx/vpx_codec_impl_bottom.h" #endif @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -13,7 +14,7 @@ */ #ifndef VP8E_H #define VP8E_H -#include "vpx_codec_impl_top.h" +#include "vpx/vpx_codec_impl_top.h" #if defined(VPX_CODEC_DISABLE_COMPAT) && VPX_CODEC_DISABLE_COMPAT #error "Backwards compatibility disabled: don't include vp8e.h" @@ -58,5 +59,5 @@ VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_SET_FRAMETYPE, int) * #VPX_DL_BEST_QUALITY constants to that parameter instead. */ VPX_CTRL_USE_TYPE_DEPRECATED(VP8E_SET_ENCODING_MODE, vp8e_encoding_mode) -#include "vpx_codec_impl_bottom.h" +#include "vpx/vpx_codec_impl_bottom.h" #endif diff --git a/vpx_codec/vpx_codec.h b/vpx/vpx_codec.h index e2a79f9..371df00 100644 --- a/vpx_codec/vpx_codec.h +++ b/vpx/vpx_codec.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -41,14 +42,7 @@ extern "C" { #ifndef VPX_CODEC_H #define VPX_CODEC_H -#ifdef HAVE_CONFIG_H -# include "vpx_config.h" -#endif -#if defined(HAVE_VPX_PORTS) && HAVE_VPX_PORTS -# include "vpx_ports/vpx_integer.h" -#else -# include "vpx_integer.h" -#endif +#include "vpx_integer.h" #include "vpx_image.h" /*!\brief Decorator indicating a function is deprecated */ @@ -196,7 +190,7 @@ extern "C" { * may reference the 'name' member to get a printable description of the * algorithm. */ - typedef struct + typedef struct vpx_codec_ctx { const char *name; /**< Printable interface name */ vpx_codec_iface_t *iface; /**< Interface pointers */ diff --git a/vpx/vpx_codec.mk b/vpx/vpx_codec.mk new file mode 100644 index 0000000..4f1d74b --- /dev/null +++ b/vpx/vpx_codec.mk @@ -0,0 +1,27 @@ +## +## Copyright (c) 2010 The WebM project authors. All Rights Reserved. +## +## Use of this source code is governed by a BSD-style license +## that can be found in the LICENSE file in the root of the source +## tree. An additional intellectual property rights grant can be found +## in the file PATENTS. All contributing project authors may +## be found in the AUTHORS file in the root of the source tree. +## + + +API_EXPORTS += exports + +API_SRCS-$(CONFIG_DECODERS) += src/vpx_decoder.c +API_SRCS-$(CONFIG_DECODERS) += src/vpx_decoder_compat.c +API_SRCS-$(CONFIG_DECODERS) += vpx_decoder.h +API_SRCS-$(CONFIG_DECODERS) += vpx_decoder_compat.h +API_SRCS-$(CONFIG_ENCODERS) += src/vpx_encoder.c +API_SRCS-$(CONFIG_ENCODERS) += vpx_encoder.h +API_SRCS-yes += internal/vpx_codec_internal.h +API_SRCS-yes += src/vpx_codec.c +API_SRCS-yes += src/vpx_image.c +API_SRCS-yes += vpx_codec.h +API_SRCS-yes += vpx_codec.mk +API_SRCS-yes += vpx_codec_impl_bottom.h +API_SRCS-yes += vpx_codec_impl_top.h +API_SRCS-yes += vpx_image.h diff --git a/vpx/vpx_codec_impl_bottom.h b/vpx/vpx_codec_impl_bottom.h new file mode 100644 index 0000000..6eb79a8 --- /dev/null +++ b/vpx/vpx_codec_impl_bottom.h @@ -0,0 +1,19 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + + +/* + * This file is to be included at the bottom of the header files defining the + * interface to individual codecs and contains matching blocks to those defined + * in vpx_codec_impl_top.h + */ +#ifdef __cplusplus +} +#endif diff --git a/vpx/vpx_codec_impl_top.h b/vpx/vpx_codec_impl_top.h new file mode 100644 index 0000000..c9b8cfa --- /dev/null +++ b/vpx/vpx_codec_impl_top.h @@ -0,0 +1,19 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + */ + + +/* + * This file is to be included at the top of the header files defining the + * interface to individual codecs and contains various workarounds common + * to all codec implementations. + */ +#ifdef __cplusplus +extern "C" { +#endif diff --git a/vpx_codec/vpx_decoder.h b/vpx/vpx_decoder.h index 5e4968d..6ffc2d4 100644 --- a/vpx_codec/vpx_decoder.h +++ b/vpx/vpx_decoder.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -68,7 +69,7 @@ extern "C" { * stream. Algorithms may extend this structure with data specific * to their bitstream by setting the sz member appropriately. */ - typedef struct + typedef struct vpx_codec_stream_info { unsigned int sz; /**< Size of this structure */ unsigned int w; /**< Width (or 0 for unknown/default) */ diff --git a/vpx_codec/vpx_decoder_compat.h b/vpx/vpx_decoder_compat.h index 25bb5eb..9e1e492 100644 --- a/vpx_codec/vpx_decoder_compat.h +++ b/vpx/vpx_decoder_compat.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_codec/vpx_encoder.h b/vpx/vpx_encoder.h index 67393be..f894cd8 100644 --- a/vpx_codec/vpx_encoder.h +++ b/vpx/vpx_encoder.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_codec/vpx_image.h b/vpx/vpx_image.h index a8a9416..4506dd3 100644 --- a/vpx_codec/vpx_image.h +++ b/vpx/vpx_image.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -30,39 +31,72 @@ extern "C" { #define VPX_IMAGE_ABI_VERSION (1) /**<\hideinitializer*/ -#define IMG_FMT_PLANAR 0x100 /**< Image is a planar format */ -#define IMG_FMT_UV_FLIP 0x200 /**< V plane precedes U plane in memory */ -#define IMG_FMT_HAS_ALPHA 0x400 /**< Image has an alpha channel componnent */ +#define VPX_IMG_FMT_PLANAR 0x100 /**< Image is a planar format */ +#define VPX_IMG_FMT_UV_FLIP 0x200 /**< V plane precedes U plane in memory */ +#define VPX_IMG_FMT_HAS_ALPHA 0x400 /**< Image has an alpha channel componnent */ /*!\brief List of supported image formats */ - typedef enum img_fmt { - IMG_FMT_NONE, - IMG_FMT_RGB24, /**< 24 bit per pixel packed RGB */ - IMG_FMT_RGB32, /**< 32 bit per pixel packed 0RGB */ - IMG_FMT_RGB565, /**< 16 bit per pixel, 565 */ - IMG_FMT_RGB555, /**< 16 bit per pixel, 555 */ - IMG_FMT_UYVY, /**< UYVY packed YUV */ - IMG_FMT_YUY2, /**< YUYV packed YUV */ - IMG_FMT_YVYU, /**< YVYU packed YUV */ - IMG_FMT_BGR24, /**< 24 bit per pixel packed BGR */ - IMG_FMT_RGB32_LE, /**< 32 bit packed BGR0 */ - IMG_FMT_ARGB, /**< 32 bit packed ARGB, alpha=255 */ - IMG_FMT_ARGB_LE, /**< 32 bit packed BGRA, alpha=255 */ - IMG_FMT_RGB565_LE, /**< 16 bit per pixel, gggbbbbb rrrrrggg */ - IMG_FMT_RGB555_LE, /**< 16 bit per pixel, gggbbbbb 0rrrrrgg */ - IMG_FMT_YV12 = IMG_FMT_PLANAR | IMG_FMT_UV_FLIP | 1, /**< planar YVU */ - IMG_FMT_I420 = IMG_FMT_PLANAR | 2, - IMG_FMT_VPXYV12 = IMG_FMT_PLANAR | IMG_FMT_UV_FLIP | 3, /** < planar 4:2:0 format with vpx color space */ - IMG_FMT_VPXI420 = IMG_FMT_PLANAR | 4, /** < planar 4:2:0 format with vpx color space */ + typedef enum vpx_img_fmt { + VPX_IMG_FMT_NONE, + VPX_IMG_FMT_RGB24, /**< 24 bit per pixel packed RGB */ + VPX_IMG_FMT_RGB32, /**< 32 bit per pixel packed 0RGB */ + VPX_IMG_FMT_RGB565, /**< 16 bit per pixel, 565 */ + VPX_IMG_FMT_RGB555, /**< 16 bit per pixel, 555 */ + VPX_IMG_FMT_UYVY, /**< UYVY packed YUV */ + VPX_IMG_FMT_YUY2, /**< YUYV packed YUV */ + VPX_IMG_FMT_YVYU, /**< YVYU packed YUV */ + VPX_IMG_FMT_BGR24, /**< 24 bit per pixel packed BGR */ + VPX_IMG_FMT_RGB32_LE, /**< 32 bit packed BGR0 */ + VPX_IMG_FMT_ARGB, /**< 32 bit packed ARGB, alpha=255 */ + VPX_IMG_FMT_ARGB_LE, /**< 32 bit packed BGRA, alpha=255 */ + VPX_IMG_FMT_RGB565_LE, /**< 16 bit per pixel, gggbbbbb rrrrrggg */ + VPX_IMG_FMT_RGB555_LE, /**< 16 bit per pixel, gggbbbbb 0rrrrrgg */ + VPX_IMG_FMT_YV12 = VPX_IMG_FMT_PLANAR | VPX_IMG_FMT_UV_FLIP | 1, /**< planar YVU */ + VPX_IMG_FMT_I420 = VPX_IMG_FMT_PLANAR | 2, + VPX_IMG_FMT_VPXYV12 = VPX_IMG_FMT_PLANAR | VPX_IMG_FMT_UV_FLIP | 3, /** < planar 4:2:0 format with vpx color space */ + VPX_IMG_FMT_VPXI420 = VPX_IMG_FMT_PLANAR | 4, /** < planar 4:2:0 format with vpx color space */ } - img_fmt_t; /**< alias for enum img_fmt */ + vpx_img_fmt_t; /**< alias for enum vpx_img_fmt */ +#if !defined(VPX_CODEC_DISABLE_COMPAT) || !VPX_CODEC_DISABLE_COMPAT +#define IMG_FMT_PLANAR VPX_IMG_FMT_PLANAR /**< \deprecated Use #VPX_IMG_FMT_PLANAR */ +#define IMG_FMT_UV_FLIP VPX_IMG_FMT_UV_FLIP /**< \deprecated Use #VPX_IMG_FMT_UV_FLIP */ +#define IMG_FMT_HAS_ALPHA VPX_IMG_FMT_HAS_ALPHA /**< \deprecated Use #VPX_IMG_FMT_HAS_ALPHA */ + + /*!\brief Deprecated list of supported image formats + * \deprecated New code should use #vpx_img_fmt + */ +#define img_fmt vpx_img_fmt + /*!\brief alias for enum img_fmt. + * \deprecated New code should use #vpx_img_fmt_t + */ +#define img_fmt_t vpx_img_fmt_t + +#define IMG_FMT_NONE VPX_IMG_FMT_NONE /**< \deprecated Use #VPX_IMG_FMT_NONE */ +#define IMG_FMT_RGB24 VPX_IMG_FMT_RGB24 /**< \deprecated Use #VPX_IMG_FMT_RGB24 */ +#define IMG_FMT_RGB32 VPX_IMG_FMT_RGB32 /**< \deprecated Use #VPX_IMG_FMT_RGB32 */ +#define IMG_FMT_RGB565 VPX_IMG_FMT_RGB565 /**< \deprecated Use #VPX_IMG_FMT_RGB565 */ +#define IMG_FMT_RGB555 VPX_IMG_FMT_RGB555 /**< \deprecated Use #VPX_IMG_FMT_RGB555 */ +#define IMG_FMT_UYVY VPX_IMG_FMT_UYVY /**< \deprecated Use #VPX_IMG_FMT_UYVY */ +#define IMG_FMT_YUY2 VPX_IMG_FMT_YUY2 /**< \deprecated Use #VPX_IMG_FMT_YUY2 */ +#define IMG_FMT_YVYU VPX_IMG_FMT_YVYU /**< \deprecated Use #VPX_IMG_FMT_YVYU */ +#define IMG_FMT_BGR24 VPX_IMG_FMT_BGR24 /**< \deprecated Use #VPX_IMG_FMT_BGR24 */ +#define IMG_FMT_RGB32_LE VPX_IMG_FMT_RGB32_LE /**< \deprecated Use #VPX_IMG_FMT_RGB32_LE */ +#define IMG_FMT_ARGB VPX_IMG_FMT_ARGB /**< \deprecated Use #VPX_IMG_FMT_ARGB */ +#define IMG_FMT_ARGB_LE VPX_IMG_FMT_ARGB_LE /**< \deprecated Use #VPX_IMG_FMT_ARGB_LE */ +#define IMG_FMT_RGB565_LE VPX_IMG_FMT_RGB565_LE /**< \deprecated Use #VPX_IMG_FMT_RGB565_LE */ +#define IMG_FMT_RGB555_LE VPX_IMG_FMT_RGB555_LE /**< \deprecated Use #VPX_IMG_FMT_RGB555_LE */ +#define IMG_FMT_YV12 VPX_IMG_FMT_YV12 /**< \deprecated Use #VPX_IMG_FMT_YV12 */ +#define IMG_FMT_I420 VPX_IMG_FMT_I420 /**< \deprecated Use #VPX_IMG_FMT_I420 */ +#define IMG_FMT_VPXYV12 VPX_IMG_FMT_VPXYV12 /**< \deprecated Use #VPX_IMG_FMT_VPXYV12 */ +#define IMG_FMT_VPXI420 VPX_IMG_FMT_VPXI420 /**< \deprecated Use #VPX_IMG_FMT_VPXI420 */ +#endif /* VPX_CODEC_DISABLE_COMPAT */ /**\brief Image Descriptor */ - typedef struct + typedef struct vpx_image { - img_fmt_t fmt; /**< Image Format */ + vpx_img_fmt_t fmt; /**< Image Format */ /* Image storage dimensions */ unsigned int w; /**< Stored image width */ @@ -77,11 +111,18 @@ extern "C" { unsigned int y_chroma_shift; /**< subsampling order, Y */ /* Image data pointers. */ -#define PLANE_PACKED 0 /**< To be used for all packed formats */ -#define PLANE_Y 0 /**< Y (Luminance) plane */ -#define PLANE_U 1 /**< U (Chroma) plane */ -#define PLANE_V 2 /**< V (Chroma) plane */ -#define PLANE_ALPHA 3 /**< A (Transparancy) plane */ +#define VPX_PLANE_PACKED 0 /**< To be used for all packed formats */ +#define VPX_PLANE_Y 0 /**< Y (Luminance) plane */ +#define VPX_PLANE_U 1 /**< U (Chroma) plane */ +#define VPX_PLANE_V 2 /**< V (Chroma) plane */ +#define VPX_PLANE_ALPHA 3 /**< A (Transparancy) plane */ +#if !defined(VPX_CODEC_DISABLE_COMPAT) || !VPX_CODEC_DISABLE_COMPAT +#define PLANE_PACKED VPX_PLANE_PACKED +#define PLANE_Y VPX_PLANE_Y +#define PLANE_U VPX_PLANE_U +#define PLANE_V VPX_PLANE_V +#define PLANE_ALPHA VPX_PLANE_ALPHA +#endif unsigned char *planes[4]; /**< pointer to the top left pixel for each plane */ int stride[4]; /**< stride between rows for each plane */ @@ -126,7 +167,7 @@ extern "C" { * returned. */ vpx_image_t *vpx_img_alloc(vpx_image_t *img, - img_fmt_t fmt, + vpx_img_fmt_t fmt, unsigned int d_w, unsigned int d_h, unsigned int align); @@ -151,7 +192,7 @@ extern "C" { * returned. */ vpx_image_t *vpx_img_wrap(vpx_image_t *img, - img_fmt_t fmt, + vpx_img_fmt_t fmt, unsigned int d_w, unsigned int d_h, unsigned int align, diff --git a/vpx_ports/vpx_integer.h b/vpx/vpx_integer.h index d3f7ddd..9a06c1a 100644 --- a/vpx_ports/vpx_integer.h +++ b/vpx/vpx_integer.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -14,13 +15,7 @@ /* get ptrdiff_t, size_t, wchar_t, NULL */ #include <stddef.h> -#if defined(HAVE_STDINT_H) && HAVE_STDINT_H -#if defined(__cplusplus) && !defined(__STDC_FORMAT_MACROS) -#define __STDC_FORMAT_MACROS -#endif -#include <stdint.h> -#include <inttypes.h> -#else +#if defined(_MSC_VER) || defined(VPX_EMULATE_INTTYPES) typedef signed char int8_t; typedef signed short int16_t; typedef signed int int32_t; @@ -47,6 +42,16 @@ typedef unsigned char uint_fast8_t; typedef unsigned int uintptr_t; #endif +#else + +/* Most platforms have the C99 standard integer types. */ + +#if defined(__cplusplus) && !defined(__STDC_FORMAT_MACROS) +#define __STDC_FORMAT_MACROS +#endif +#include <stdint.h> +#include <inttypes.h> + #endif #endif diff --git a/vpx_codec/exports b/vpx_codec/exports deleted file mode 100644 index f5e7473..0000000 --- a/vpx_codec/exports +++ /dev/null @@ -1,17 +0,0 @@ -text vpx_dec_control -text vpx_dec_decode -text vpx_dec_destroy -text vpx_dec_err_to_string -text vpx_dec_error -text vpx_dec_error_detail -text vpx_dec_get_caps -text vpx_dec_get_frame -text vpx_dec_get_mem_map -text vpx_dec_get_stream_info -text vpx_dec_iface_name -text vpx_dec_init_ver -text vpx_dec_peek_stream_info -text vpx_dec_register_put_frame_cb -text vpx_dec_register_put_slice_cb -text vpx_dec_set_mem_map -text vpx_dec_xma_init_ver diff --git a/vpx_codec/vpx_codec.mk b/vpx_codec/vpx_codec.mk deleted file mode 100644 index 75fbeea..0000000 --- a/vpx_codec/vpx_codec.mk +++ /dev/null @@ -1,26 +0,0 @@ -## -## Copyright (c) 2010 The VP8 project authors. All Rights Reserved. -## -## Use of this source code is governed by a BSD-style license and patent -## grant that can be found in the LICENSE file in the root of the source -## tree. All contributing project authors may be found in the AUTHORS -## file in the root of the source tree. -## - - -API_EXPORTS += exports - -API_SRCS-yes += internal/vpx_codec_internal.h -API_SRCS-yes += vpx_codec.h -API_SRCS-yes += vpx_codec.mk -API_SRCS-yes += vpx_codec_impl_top.h -API_SRCS-yes += vpx_codec_impl_bottom.h -API_SRCS-yes += vpx_decoder.h -API_SRCS-yes += vpx_decoder_compat.h -API_SRCS-yes += vpx_encoder.h -API_SRCS-yes += vpx_image.h -API_SRCS-yes += src/vpx_codec.c -API_SRCS-yes += src/vpx_decoder.c -API_SRCS-yes += src/vpx_decoder_compat.c -API_SRCS-yes += src/vpx_image.c -API_SRCS-yes += src/vpx_encoder.c diff --git a/vpx_codec/vpx_codec_impl_bottom.h b/vpx_codec/vpx_codec_impl_bottom.h deleted file mode 100644 index c52654c..0000000 --- a/vpx_codec/vpx_codec_impl_bottom.h +++ /dev/null @@ -1,18 +0,0 @@ -/* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. - * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. - */ - - -/* - * This file is to be included at the bottom of the header files defining the - * interface to individual codecs and contains matching blocks to those defined - * in vpx_codec_impl_top.h - */ -#ifdef __cplusplus -} -#endif diff --git a/vpx_codec/vpx_codec_impl_top.h b/vpx_codec/vpx_codec_impl_top.h deleted file mode 100644 index f73809a..0000000 --- a/vpx_codec/vpx_codec_impl_top.h +++ /dev/null @@ -1,18 +0,0 @@ -/* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. - * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. - */ - - -/* - * This file is to be included at the top of the header files defining the - * interface to individual codecs and contains various workarounds common - * to all codec implementations. - */ -#ifdef __cplusplus -extern "C" { -#endif diff --git a/vpx_mem/include/nds/vpx_mem_nds.h b/vpx_mem/include/nds/vpx_mem_nds.h index c332403..e54f54d 100644 --- a/vpx_mem/include/nds/vpx_mem_nds.h +++ b/vpx_mem/include/nds/vpx_mem_nds.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/include/vpx_mem_intrnl.h b/vpx_mem/include/vpx_mem_intrnl.h index 3b68d86..00f9c90 100644 --- a/vpx_mem/include/vpx_mem_intrnl.h +++ b/vpx_mem/include/vpx_mem_intrnl.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/include/vpx_mem_tracker.h b/vpx_mem/include/vpx_mem_tracker.h index ab85d19..ef2b29b 100644 --- a/vpx_mem/include/vpx_mem_tracker.h +++ b/vpx_mem/include/vpx_mem_tracker.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/intel_linux/vpx_mem.c b/vpx_mem/intel_linux/vpx_mem.c index 002e407..00150ac 100644 --- a/vpx_mem/intel_linux/vpx_mem.c +++ b/vpx_mem/intel_linux/vpx_mem.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/intel_linux/vpx_mem_tracker.c b/vpx_mem/intel_linux/vpx_mem_tracker.c index fa023e3..5bed4b5 100644 --- a/vpx_mem/intel_linux/vpx_mem_tracker.c +++ b/vpx_mem/intel_linux/vpx_mem_tracker.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/hmm_alloc.c b/vpx_mem/memory_manager/hmm_alloc.c index 9abd81e..22c4a54 100644 --- a/vpx_mem/memory_manager/hmm_alloc.c +++ b/vpx_mem/memory_manager/hmm_alloc.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/hmm_base.c b/vpx_mem/memory_manager/hmm_base.c index 0cacc3f..ad1da03 100644 --- a/vpx_mem/memory_manager/hmm_base.c +++ b/vpx_mem/memory_manager/hmm_base.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/hmm_dflt_abort.c b/vpx_mem/memory_manager/hmm_dflt_abort.c index dc59f55..d92435c 100644 --- a/vpx_mem/memory_manager/hmm_dflt_abort.c +++ b/vpx_mem/memory_manager/hmm_dflt_abort.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/hmm_grow.c b/vpx_mem/memory_manager/hmm_grow.c index 79d75a7..9a4b6e4 100644 --- a/vpx_mem/memory_manager/hmm_grow.c +++ b/vpx_mem/memory_manager/hmm_grow.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/hmm_largest.c b/vpx_mem/memory_manager/hmm_largest.c index 5ebe398..c3c6f2c 100644 --- a/vpx_mem/memory_manager/hmm_largest.c +++ b/vpx_mem/memory_manager/hmm_largest.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/hmm_resize.c b/vpx_mem/memory_manager/hmm_resize.c index 6e3f2f0..f90da96 100644 --- a/vpx_mem/memory_manager/hmm_resize.c +++ b/vpx_mem/memory_manager/hmm_resize.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/hmm_shrink.c b/vpx_mem/memory_manager/hmm_shrink.c index 5ef9b23..78fe268 100644 --- a/vpx_mem/memory_manager/hmm_shrink.c +++ b/vpx_mem/memory_manager/hmm_shrink.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/hmm_true.c b/vpx_mem/memory_manager/hmm_true.c index 41103c8..3f7be8f 100644 --- a/vpx_mem/memory_manager/hmm_true.c +++ b/vpx_mem/memory_manager/hmm_true.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/include/cavl_if.h b/vpx_mem/memory_manager/include/cavl_if.h index e2733ef..1b2c9b7 100644 --- a/vpx_mem/memory_manager/include/cavl_if.h +++ b/vpx_mem/memory_manager/include/cavl_if.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/include/cavl_impl.h b/vpx_mem/memory_manager/include/cavl_impl.h index 267bc73..5e165dd 100644 --- a/vpx_mem/memory_manager/include/cavl_impl.h +++ b/vpx_mem/memory_manager/include/cavl_impl.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/include/heapmm.h b/vpx_mem/memory_manager/include/heapmm.h index 933e30d..33004ca 100644 --- a/vpx_mem/memory_manager/include/heapmm.h +++ b/vpx_mem/memory_manager/include/heapmm.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/include/hmm_cnfg.h b/vpx_mem/memory_manager/include/hmm_cnfg.h index 86e4e9f..30b9f50 100644 --- a/vpx_mem/memory_manager/include/hmm_cnfg.h +++ b/vpx_mem/memory_manager/include/hmm_cnfg.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/memory_manager/include/hmm_intrnl.h b/vpx_mem/memory_manager/include/hmm_intrnl.h index 6e2be08..5d62abc 100644 --- a/vpx_mem/memory_manager/include/hmm_intrnl.h +++ b/vpx_mem/memory_manager/include/hmm_intrnl.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/nds/vpx_mem_nds.c b/vpx_mem/nds/vpx_mem_nds.c index f2a3043..11ac95c 100644 --- a/vpx_mem/nds/vpx_mem_nds.c +++ b/vpx_mem/nds/vpx_mem_nds.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/ti_c6x/vpx_mem_ti_6cx.c b/vpx_mem/ti_c6x/vpx_mem_ti_6cx.c index 6501855..d55b7d9 100644 --- a/vpx_mem/ti_c6x/vpx_mem_ti_6cx.c +++ b/vpx_mem/ti_c6x/vpx_mem_ti_6cx.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/vpx_mem.c b/vpx_mem/vpx_mem.c index f6b1a35..85b05ab 100644 --- a/vpx_mem/vpx_mem.c +++ b/vpx_mem/vpx_mem.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/vpx_mem.h b/vpx_mem/vpx_mem.h index 6ccb9be..31f8f9c 100644 --- a/vpx_mem/vpx_mem.h +++ b/vpx_mem/vpx_mem.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_mem/vpx_mem_tracker.c b/vpx_mem/vpx_mem_tracker.c index 4427e27..938ad07 100644 --- a/vpx_mem/vpx_mem_tracker.c +++ b/vpx_mem/vpx_mem_tracker.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_ports/config.h b/vpx_ports/config.h index da38137..1abe70d 100644 --- a/vpx_ports/config.h +++ b/vpx_ports/config.h @@ -1,19 +1,10 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ - - -/* This file uses some preprocessor magic to expand the value of HAVE_CONFIG_H, - * as defined by the build system, so that different projects can use the file - * name for config.h that suits them. - */ -#define QUOTE_(x) #x -#define QUOTE(x) QUOTE_(x) -#include QUOTE(HAVE_CONFIG_H) -#undef QUOTE -#undef QUOTE_ +#include "vpx_config.h" diff --git a/vpx_ports/emms.asm b/vpx_ports/emms.asm index 03e3499..87eece8 100644 --- a/vpx_ports/emms.asm +++ b/vpx_ports/emms.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vpx_ports/mem.h b/vpx_ports/mem.h index 10942f1..9ec34fe 100644 --- a/vpx_ports/mem.h +++ b/vpx_ports/mem.h @@ -1,17 +1,18 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ #ifndef VPX_PORTS_MEM_H #define VPX_PORTS_MEM_H #include "vpx_config.h" -#include "vpx_integer.h" +#include "vpx/vpx_integer.h" #if defined(__GNUC__) && __GNUC__ #define DECLARE_ALIGNED(n,typ,val) typ val __attribute__ ((aligned (n))) diff --git a/vpx_ports/mem_ops.h b/vpx_ports/mem_ops.h index 869d583..c178b8b 100644 --- a/vpx_ports/mem_ops.h +++ b/vpx_ports/mem_ops.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -59,7 +60,7 @@ #undef mem_get_be16 #define mem_get_be16 mem_ops_wrap_symbol(mem_get_be16) -static INLINE unsigned MEM_VALUE_T mem_get_be16(const void *vmem) +static unsigned MEM_VALUE_T mem_get_be16(const void *vmem) { unsigned MEM_VALUE_T val; const MAU_T *mem = (const MAU_T *)vmem; @@ -71,7 +72,7 @@ static INLINE unsigned MEM_VALUE_T mem_get_be16(const void *vmem) #undef mem_get_be24 #define mem_get_be24 mem_ops_wrap_symbol(mem_get_be24) -static INLINE unsigned MEM_VALUE_T mem_get_be24(const void *vmem) +static unsigned MEM_VALUE_T mem_get_be24(const void *vmem) { unsigned MEM_VALUE_T val; const MAU_T *mem = (const MAU_T *)vmem; @@ -84,7 +85,7 @@ static INLINE unsigned MEM_VALUE_T mem_get_be24(const void *vmem) #undef mem_get_be32 #define mem_get_be32 mem_ops_wrap_symbol(mem_get_be32) -static INLINE unsigned MEM_VALUE_T mem_get_be32(const void *vmem) +static unsigned MEM_VALUE_T mem_get_be32(const void *vmem) { unsigned MEM_VALUE_T val; const MAU_T *mem = (const MAU_T *)vmem; @@ -98,7 +99,7 @@ static INLINE unsigned MEM_VALUE_T mem_get_be32(const void *vmem) #undef mem_get_le16 #define mem_get_le16 mem_ops_wrap_symbol(mem_get_le16) -static INLINE unsigned MEM_VALUE_T mem_get_le16(const void *vmem) +static unsigned MEM_VALUE_T mem_get_le16(const void *vmem) { unsigned MEM_VALUE_T val; const MAU_T *mem = (const MAU_T *)vmem; @@ -110,7 +111,7 @@ static INLINE unsigned MEM_VALUE_T mem_get_le16(const void *vmem) #undef mem_get_le24 #define mem_get_le24 mem_ops_wrap_symbol(mem_get_le24) -static INLINE unsigned MEM_VALUE_T mem_get_le24(const void *vmem) +static unsigned MEM_VALUE_T mem_get_le24(const void *vmem) { unsigned MEM_VALUE_T val; const MAU_T *mem = (const MAU_T *)vmem; @@ -123,7 +124,7 @@ static INLINE unsigned MEM_VALUE_T mem_get_le24(const void *vmem) #undef mem_get_le32 #define mem_get_le32 mem_ops_wrap_symbol(mem_get_le32) -static INLINE unsigned MEM_VALUE_T mem_get_le32(const void *vmem) +static unsigned MEM_VALUE_T mem_get_le32(const void *vmem) { unsigned MEM_VALUE_T val; const MAU_T *mem = (const MAU_T *)vmem; @@ -136,7 +137,7 @@ static INLINE unsigned MEM_VALUE_T mem_get_le32(const void *vmem) } #define mem_get_s_generic(end,sz) \ - static INLINE signed MEM_VALUE_T mem_get_s##end##sz(const void *vmem) {\ + static signed MEM_VALUE_T mem_get_s##end##sz(const void *vmem) {\ const MAU_T *mem = (const MAU_T*)vmem;\ signed MEM_VALUE_T val = mem_get_##end##sz(mem);\ return (val << (MEM_VALUE_T_SZ_BITS - sz)) >> (MEM_VALUE_T_SZ_BITS - sz);\ @@ -168,7 +169,7 @@ mem_get_s_generic(le, 32); #undef mem_put_be16 #define mem_put_be16 mem_ops_wrap_symbol(mem_put_be16) -static INLINE void mem_put_be16(void *vmem, MEM_VALUE_T val) +static void mem_put_be16(void *vmem, MEM_VALUE_T val) { MAU_T *mem = (MAU_T *)vmem; @@ -178,7 +179,7 @@ static INLINE void mem_put_be16(void *vmem, MEM_VALUE_T val) #undef mem_put_be24 #define mem_put_be24 mem_ops_wrap_symbol(mem_put_be24) -static INLINE void mem_put_be24(void *vmem, MEM_VALUE_T val) +static void mem_put_be24(void *vmem, MEM_VALUE_T val) { MAU_T *mem = (MAU_T *)vmem; @@ -189,7 +190,7 @@ static INLINE void mem_put_be24(void *vmem, MEM_VALUE_T val) #undef mem_put_be32 #define mem_put_be32 mem_ops_wrap_symbol(mem_put_be32) -static INLINE void mem_put_be32(void *vmem, MEM_VALUE_T val) +static void mem_put_be32(void *vmem, MEM_VALUE_T val) { MAU_T *mem = (MAU_T *)vmem; @@ -201,7 +202,7 @@ static INLINE void mem_put_be32(void *vmem, MEM_VALUE_T val) #undef mem_put_le16 #define mem_put_le16 mem_ops_wrap_symbol(mem_put_le16) -static INLINE void mem_put_le16(void *vmem, MEM_VALUE_T val) +static void mem_put_le16(void *vmem, MEM_VALUE_T val) { MAU_T *mem = (MAU_T *)vmem; @@ -211,7 +212,7 @@ static INLINE void mem_put_le16(void *vmem, MEM_VALUE_T val) #undef mem_put_le24 #define mem_put_le24 mem_ops_wrap_symbol(mem_put_le24) -static INLINE void mem_put_le24(void *vmem, MEM_VALUE_T val) +static void mem_put_le24(void *vmem, MEM_VALUE_T val) { MAU_T *mem = (MAU_T *)vmem; @@ -222,7 +223,7 @@ static INLINE void mem_put_le24(void *vmem, MEM_VALUE_T val) #undef mem_put_le32 #define mem_put_le32 mem_ops_wrap_symbol(mem_put_le32) -static INLINE void mem_put_le32(void *vmem, MEM_VALUE_T val) +static void mem_put_le32(void *vmem, MEM_VALUE_T val) { MAU_T *mem = (MAU_T *)vmem; diff --git a/vpx_ports/mem_ops_aligned.h b/vpx_ports/mem_ops_aligned.h index 1d0db2c..4c44aa2 100644 --- a/vpx_ports/mem_ops_aligned.h +++ b/vpx_ports/mem_ops_aligned.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -39,19 +40,19 @@ #define swap_endian_32_se(val,raw) swap_endian_32(val,raw) #define mem_get_ne_aligned_generic(end,sz) \ - static INLINE unsigned MEM_VALUE_T mem_get_##end##sz##_aligned(const void *vmem) {\ + static unsigned MEM_VALUE_T mem_get_##end##sz##_aligned(const void *vmem) {\ const uint##sz##_t *mem = (const uint##sz##_t *)vmem;\ return *mem;\ } #define mem_get_sne_aligned_generic(end,sz) \ - static INLINE signed MEM_VALUE_T mem_get_s##end##sz##_aligned(const void *vmem) {\ + static signed MEM_VALUE_T mem_get_s##end##sz##_aligned(const void *vmem) {\ const int##sz##_t *mem = (const int##sz##_t *)vmem;\ return *mem;\ } #define mem_get_se_aligned_generic(end,sz) \ - static INLINE unsigned MEM_VALUE_T mem_get_##end##sz##_aligned(const void *vmem) {\ + static unsigned MEM_VALUE_T mem_get_##end##sz##_aligned(const void *vmem) {\ const uint##sz##_t *mem = (const uint##sz##_t *)vmem;\ unsigned MEM_VALUE_T val, raw = *mem;\ swap_endian_##sz(val,raw);\ @@ -59,7 +60,7 @@ } #define mem_get_sse_aligned_generic(end,sz) \ - static INLINE signed MEM_VALUE_T mem_get_s##end##sz##_aligned(const void *vmem) {\ + static signed MEM_VALUE_T mem_get_s##end##sz##_aligned(const void *vmem) {\ const int##sz##_t *mem = (const int##sz##_t *)vmem;\ unsigned MEM_VALUE_T val, raw = *mem;\ swap_endian_##sz##_se(val,raw);\ @@ -67,13 +68,13 @@ } #define mem_put_ne_aligned_generic(end,sz) \ - static INLINE void mem_put_##end##sz##_aligned(void *vmem, MEM_VALUE_T val) {\ + static void mem_put_##end##sz##_aligned(void *vmem, MEM_VALUE_T val) {\ uint##sz##_t *mem = (uint##sz##_t *)vmem;\ *mem = (uint##sz##_t)val;\ } #define mem_put_se_aligned_generic(end,sz) \ - static INLINE void mem_put_##end##sz##_aligned(void *vmem, MEM_VALUE_T val) {\ + static void mem_put_##end##sz##_aligned(void *vmem, MEM_VALUE_T val) {\ uint##sz##_t *mem = (uint##sz##_t *)vmem, raw;\ swap_endian_##sz(raw,val);\ *mem = (uint##sz##_t)raw;\ diff --git a/vpx_ports/vpx_timer.h b/vpx_ports/vpx_timer.h index 5c04538..37a0c7c 100644 --- a/vpx_ports/vpx_timer.h +++ b/vpx_ports/vpx_timer.h @@ -1,17 +1,18 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ #ifndef VPX_TIMER_H #define VPX_TIMER_H -#if defined(_MSC_VER) +#if defined(_WIN32) /* * Win32 specific includes */ @@ -42,7 +43,7 @@ struct vpx_usec_timer { -#if defined(_MSC_VER) +#if defined(_WIN32) LARGE_INTEGER begin, end; #else struct timeval begin, end; @@ -50,10 +51,10 @@ struct vpx_usec_timer }; -static INLINE void +static void vpx_usec_timer_start(struct vpx_usec_timer *t) { -#if defined(_MSC_VER) +#if defined(_WIN32) QueryPerformanceCounter(&t->begin); #else gettimeofday(&t->begin, NULL); @@ -61,10 +62,10 @@ vpx_usec_timer_start(struct vpx_usec_timer *t) } -static INLINE void +static void vpx_usec_timer_mark(struct vpx_usec_timer *t) { -#if defined(_MSC_VER) +#if defined(_WIN32) QueryPerformanceCounter(&t->end); #else gettimeofday(&t->end, NULL); @@ -72,10 +73,10 @@ vpx_usec_timer_mark(struct vpx_usec_timer *t) } -static INLINE long +static long vpx_usec_timer_elapsed(struct vpx_usec_timer *t) { -#if defined(_MSC_VER) +#if defined(_WIN32) LARGE_INTEGER freq, diff; diff.QuadPart = t->end.QuadPart - t->begin.QuadPart; diff --git a/vpx_ports/vpxtypes.h b/vpx_ports/vpxtypes.h index 86525b7..2ab66b1 100644 --- a/vpx_ports/vpxtypes.h +++ b/vpx_ports/vpxtypes.h @@ -1,19 +1,18 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ #ifndef __VPXTYPES_H__ #define __VPXTYPES_H__ -#ifdef HAVE_CONFIG_H -#include HAVE_CONFIG_H -#endif +#include "vpx_ports/config.h" //#include <sys/types.h> #ifdef _MSC_VER diff --git a/vpx_ports/x86.h b/vpx_ports/x86.h index 935d037..a8e4607 100644 --- a/vpx_ports/x86.h +++ b/vpx_ports/x86.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_ports/x86_abi_support.asm b/vpx_ports/x86_abi_support.asm index db8208f..dc9e2d9 100644 --- a/vpx_ports/x86_abi_support.asm +++ b/vpx_ports/x86_abi_support.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; @@ -89,7 +90,7 @@ %macro ALIGN_STACK 2 mov %2, rsp and rsp, -%1 - sub rsp, %1 - REG_SZ_BYTES + lea rsp, [rsp - (%1 - REG_SZ_BYTES)] push %2 %endmacro @@ -104,7 +105,6 @@ %idefine XMMWORD %idefine MMWORD - ; PIC macros ; %if ABI_IS_32BIT @@ -115,9 +115,13 @@ extern _GLOBAL_OFFSET_TABLE_ push %1 call %%get_got + %%sub_offset: + jmp %%exitGG %%get_got: - pop %1 - add %1, _GLOBAL_OFFSET_TABLE_ + $$ - %%get_got wrt ..gotpc + mov %1, [esp] + add %1, _GLOBAL_OFFSET_TABLE_ + $$ - %%sub_offset wrt ..gotpc + ret + %%exitGG: %undef GLOBAL %define GLOBAL + %1 wrt ..gotoff %undef RESTORE_GOT @@ -127,9 +131,13 @@ %macro GET_GOT 1 push %1 call %%get_got + %%sub_offset: + jmp %%exitGG %%get_got: - pop %1 - add %1, fake_got - %%get_got + mov %1, [esp] + add %1, fake_got - %%sub_offset + ret + %%exitGG: %undef GLOBAL %define GLOBAL + %1 - fake_got %undef RESTORE_GOT @@ -137,12 +145,16 @@ %endmacro %endif %endif + %define HIDDEN_DATA(x) x %else %macro GET_GOT 1 %endmacro %define GLOBAL wrt rip %ifidn __OUTPUT_FORMAT__,elf64 %define WRT_PLT wrt ..plt + %define HIDDEN_DATA(x) x:data hidden + %else + %define HIDDEN_DATA(x) x %endif %endif %ifnmacro GET_GOT @@ -198,22 +210,38 @@ push r9 %endif %if %1 > 6 - mov rax,[rbp+16] - push rax - %endif - %if %1 > 7 - mov rax,[rbp+24] - push rax - %endif - %if %1 > 8 - mov rax,[rbp+32] + %assign i %1-6 + %assign off 16 + %rep i + mov rax,[rbp+off] push rax + %assign off off+8 + %endrep %endif %endm %endif %define UNSHADOW_ARGS mov rsp, rbp %endif +; must keep XMM6:XMM15 (libvpx uses XMM6 and XMM7) on Win64 ABI +; rsp register has to be aligned +%ifidn __OUTPUT_FORMAT__,x64 +%macro SAVE_XMM 0 + sub rsp, 32 + movdqa XMMWORD PTR [rsp], xmm6 + movdqa XMMWORD PTR [rsp+16], xmm7 +%endmacro +%macro RESTORE_XMM 0 + movdqa xmm6, XMMWORD PTR [rsp] + movdqa xmm7, XMMWORD PTR [rsp+16] + add rsp, 32 +%endmacro +%else +%macro SAVE_XMM 0 +%endmacro +%macro RESTORE_XMM 0 +%endmacro +%endif ; Name of the rodata section ; @@ -229,3 +257,14 @@ fake_got: %else %define SECTION_RODATA section .rodata %endif + + +; Tell GNU ld that we don't require an executable stack. +%ifidn __OUTPUT_FORMAT__,elf32 +section .note.GNU-stack noalloc noexec nowrite progbits +section .text +%elifidn __OUTPUT_FORMAT__,elf64 +section .note.GNU-stack noalloc noexec nowrite progbits +section .text +%endif + diff --git a/vpx_scale/arm/armv4/gen_scalers_armv4.asm b/vpx_scale/arm/armv4/gen_scalers_armv4.asm index 1c904ed..e495184 100644 --- a/vpx_scale/arm/armv4/gen_scalers_armv4.asm +++ b/vpx_scale/arm/armv4/gen_scalers_armv4.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vpx_scale/arm/nds/yv12extend.c b/vpx_scale/arm/nds/yv12extend.c index 56959cb..48c0dfb 100644 --- a/vpx_scale/arm/nds/yv12extend.c +++ b/vpx_scale/arm/nds/yv12extend.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/arm/neon/vp8_vpxyv12_copyframe_func_neon.asm b/vpx_scale/arm/neon/vp8_vpxyv12_copyframe_func_neon.asm index 26384c4..24d46cb 100644 --- a/vpx_scale/arm/neon/vp8_vpxyv12_copyframe_func_neon.asm +++ b/vpx_scale/arm/neon/vp8_vpxyv12_copyframe_func_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vpx_scale/arm/neon/vp8_vpxyv12_copyframeyonly_neon.asm b/vpx_scale/arm/neon/vp8_vpxyv12_copyframeyonly_neon.asm index a50ae60..6534827 100644 --- a/vpx_scale/arm/neon/vp8_vpxyv12_copyframeyonly_neon.asm +++ b/vpx_scale/arm/neon/vp8_vpxyv12_copyframeyonly_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vpx_scale/arm/neon/vp8_vpxyv12_copysrcframe_func_neon.asm b/vpx_scale/arm/neon/vp8_vpxyv12_copysrcframe_func_neon.asm index c8923d5..dfc8db5 100644 --- a/vpx_scale/arm/neon/vp8_vpxyv12_copysrcframe_func_neon.asm +++ b/vpx_scale/arm/neon/vp8_vpxyv12_copysrcframe_func_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vpx_scale/arm/neon/vp8_vpxyv12_extendframeborders_neon.asm b/vpx_scale/arm/neon/vp8_vpxyv12_extendframeborders_neon.asm index 8c9ce19..e475b92 100644 --- a/vpx_scale/arm/neon/vp8_vpxyv12_extendframeborders_neon.asm +++ b/vpx_scale/arm/neon/vp8_vpxyv12_extendframeborders_neon.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vpx_scale/arm/scalesystemdependant.c b/vpx_scale/arm/scalesystemdependant.c index 3c355be..1e8bcb8 100644 --- a/vpx_scale/arm/scalesystemdependant.c +++ b/vpx_scale/arm/scalesystemdependant.c @@ -1,18 +1,17 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ +#include "vpx_ports/config.h" #include "vpx_scale/vpxscale.h" -#ifdef HAVE_CONFIG_H -#include "vpx_config.h" -#endif void (*vp8_yv12_extend_frame_borders_ptr)(YV12_BUFFER_CONFIG *ybf); extern void vp8_yv12_extend_frame_borders(YV12_BUFFER_CONFIG *ybf); diff --git a/vpx_scale/arm/yv12extend_arm.c b/vpx_scale/arm/yv12extend_arm.c index 7c3f7cd..d7a8289 100644 --- a/vpx_scale/arm/yv12extend_arm.c +++ b/vpx_scale/arm/yv12extend_arm.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/blackfin/yv12config.c b/vpx_scale/blackfin/yv12config.c index 7cb083f..42538af 100644 --- a/vpx_scale/blackfin/yv12config.c +++ b/vpx_scale/blackfin/yv12config.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/blackfin/yv12extend.c b/vpx_scale/blackfin/yv12extend.c index d5be495..cfd3de0 100644 --- a/vpx_scale/blackfin/yv12extend.c +++ b/vpx_scale/blackfin/yv12extend.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/dm642/bicubic_scaler_c64.c b/vpx_scale/dm642/bicubic_scaler_c64.c index 9bd3797..5166ace 100644 --- a/vpx_scale/dm642/bicubic_scaler_c64.c +++ b/vpx_scale/dm642/bicubic_scaler_c64.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/dm642/gen_scalers_c64.c b/vpx_scale/dm642/gen_scalers_c64.c index 2126a75..87ff998 100644 --- a/vpx_scale/dm642/gen_scalers_c64.c +++ b/vpx_scale/dm642/gen_scalers_c64.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/dm642/yv12extend.c b/vpx_scale/dm642/yv12extend.c index ca25a5f..646e69a 100644 --- a/vpx_scale/dm642/yv12extend.c +++ b/vpx_scale/dm642/yv12extend.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/generic/bicubic_scaler.c b/vpx_scale/generic/bicubic_scaler.c index e3c2b4a..420f719 100644 --- a/vpx_scale/generic/bicubic_scaler.c +++ b/vpx_scale/generic/bicubic_scaler.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -45,7 +46,7 @@ static float a = -0.6; // 3 2 // C0 = a*t - a*t // -static INLINE short c0_fixed(unsigned int t) +static short c0_fixed(unsigned int t) { // put t in Q16 notation unsigned short v1, v2; @@ -66,7 +67,7 @@ static INLINE short c0_fixed(unsigned int t) // 2 3 // C1 = a*t + (3-2*a)*t - (2-a)*t // -static INLINE short c1_fixed(unsigned int t) +static short c1_fixed(unsigned int t) { unsigned short v1, v2, v3; unsigned short two, three; @@ -95,7 +96,7 @@ static INLINE short c1_fixed(unsigned int t) // 2 3 // C2 = 1 - (3-a)*t + (2-a)*t // -static INLINE short c2_fixed(unsigned int t) +static short c2_fixed(unsigned int t) { unsigned short v1, v2, v3; unsigned short two, three; @@ -123,7 +124,7 @@ static INLINE short c2_fixed(unsigned int t) // 2 3 // C3 = a*t - 2*a*t + a*t // -static INLINE short c3_fixed(unsigned int t) +static short c3_fixed(unsigned int t) { int v1, v2, v3; diff --git a/vpx_scale/generic/gen_scalers.c b/vpx_scale/generic/gen_scalers.c index a5e545f..b084e81 100644 --- a/vpx_scale/generic/gen_scalers.c +++ b/vpx_scale/generic/gen_scalers.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ @@ -936,12 +937,13 @@ void vp8cx_vertical_band_2_1_scale_c(unsigned char *source, unsigned int src_pit void vp8cx_vertical_band_2_1_scale_i_c(unsigned char *source, unsigned int src_pitch, unsigned char *dest, unsigned int dest_pitch, unsigned int dest_width) { - unsigned int i; + int i; int temp; + int width = dest_width; (void) dest_pitch; - for (i = 0; i < dest_width; i++) + for (i = 0; i < width; i++) { temp = 8; temp += source[i-(int)src_pitch] * 3; diff --git a/vpx_scale/generic/scalesystemdependant.c b/vpx_scale/generic/scalesystemdependant.c index 28f5c72..926feb7 100644 --- a/vpx_scale/generic/scalesystemdependant.c +++ b/vpx_scale/generic/scalesystemdependant.c @@ -1,18 +1,17 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ +#include "vpx_ports/config.h" #include "vpx_scale/vpxscale.h" -#ifdef HAVE_CONFIG_H -#include "vpx_config.h" -#endif void (*vp8_yv12_extend_frame_borders_ptr)(YV12_BUFFER_CONFIG *ybf); extern void vp8_yv12_extend_frame_borders(YV12_BUFFER_CONFIG *ybf); diff --git a/vpx_scale/generic/vpxscale.c b/vpx_scale/generic/vpxscale.c index 206cd55..e1c8028 100644 --- a/vpx_scale/generic/vpxscale.c +++ b/vpx_scale/generic/vpxscale.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/generic/yv12config.c b/vpx_scale/generic/yv12config.c index 04617be..3034cc3 100644 --- a/vpx_scale/generic/yv12config.c +++ b/vpx_scale/generic/yv12config.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/generic/yv12extend.c b/vpx_scale/generic/yv12extend.c index 4906625..39d2fa8 100644 --- a/vpx_scale/generic/yv12extend.c +++ b/vpx_scale/generic/yv12extend.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/include/arm/vpxscale_nofp.h b/vpx_scale/include/arm/vpxscale_nofp.h index d6181d2..3e1a9fa 100644 --- a/vpx_scale/include/arm/vpxscale_nofp.h +++ b/vpx_scale/include/arm/vpxscale_nofp.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/include/generic/vpxscale_arbitrary.h b/vpx_scale/include/generic/vpxscale_arbitrary.h index 2b50f24..39de181 100644 --- a/vpx_scale/include/generic/vpxscale_arbitrary.h +++ b/vpx_scale/include/generic/vpxscale_arbitrary.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/include/generic/vpxscale_depricated.h b/vpx_scale/include/generic/vpxscale_depricated.h index 015eed0..3f7fe0f 100644 --- a/vpx_scale/include/generic/vpxscale_depricated.h +++ b/vpx_scale/include/generic/vpxscale_depricated.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/include/generic/vpxscale_nofp.h b/vpx_scale/include/generic/vpxscale_nofp.h index c4d5f4c..7b8205a 100644 --- a/vpx_scale/include/generic/vpxscale_nofp.h +++ b/vpx_scale/include/generic/vpxscale_nofp.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/include/leapster/vpxscale.h b/vpx_scale/include/leapster/vpxscale.h index f70029c..5021849 100644 --- a/vpx_scale/include/leapster/vpxscale.h +++ b/vpx_scale/include/leapster/vpxscale.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/include/symbian/vpxscale_nofp.h b/vpx_scale/include/symbian/vpxscale_nofp.h index d6181d2..3e1a9fa 100644 --- a/vpx_scale/include/symbian/vpxscale_nofp.h +++ b/vpx_scale/include/symbian/vpxscale_nofp.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/include/vpxscale_nofp.h b/vpx_scale/include/vpxscale_nofp.h index f6482f9..a704bd9 100644 --- a/vpx_scale/include/vpxscale_nofp.h +++ b/vpx_scale/include/vpxscale_nofp.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/intel_linux/scaleopt.c b/vpx_scale/intel_linux/scaleopt.c index 6555600..499f0ed 100644 --- a/vpx_scale/intel_linux/scaleopt.c +++ b/vpx_scale/intel_linux/scaleopt.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/intel_linux/scalesystemdependant.c b/vpx_scale/intel_linux/scalesystemdependant.c index 9ed48bf..eab741f 100644 --- a/vpx_scale/intel_linux/scalesystemdependant.c +++ b/vpx_scale/intel_linux/scalesystemdependant.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/leapster/doptsystemdependant_lf.c b/vpx_scale/leapster/doptsystemdependant_lf.c index ca13167..7cbb1c5 100644 --- a/vpx_scale/leapster/doptsystemdependant_lf.c +++ b/vpx_scale/leapster/doptsystemdependant_lf.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/leapster/gen_scalers_lf.c b/vpx_scale/leapster/gen_scalers_lf.c index 1b9c7c7..f1ee056 100644 --- a/vpx_scale/leapster/gen_scalers_lf.c +++ b/vpx_scale/leapster/gen_scalers_lf.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/leapster/vpxscale_lf.c b/vpx_scale/leapster/vpxscale_lf.c index 5f05e5d..616ddf5 100644 --- a/vpx_scale/leapster/vpxscale_lf.c +++ b/vpx_scale/leapster/vpxscale_lf.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/leapster/yv12extend.c b/vpx_scale/leapster/yv12extend.c index 480d971..1cddd95 100644 --- a/vpx_scale/leapster/yv12extend.c +++ b/vpx_scale/leapster/yv12extend.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/scale_mode.h b/vpx_scale/scale_mode.h index 2a9ab76..1476e64 100644 --- a/vpx_scale/scale_mode.h +++ b/vpx_scale/scale_mode.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/symbian/gen_scalers_armv4.asm b/vpx_scale/symbian/gen_scalers_armv4.asm index 1c904ed..e495184 100644 --- a/vpx_scale/symbian/gen_scalers_armv4.asm +++ b/vpx_scale/symbian/gen_scalers_armv4.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vpx_scale/symbian/scalesystemdependant.c b/vpx_scale/symbian/scalesystemdependant.c index a2acc3e..b0bffdd 100644 --- a/vpx_scale/symbian/scalesystemdependant.c +++ b/vpx_scale/symbian/scalesystemdependant.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/vpxscale.h b/vpx_scale/vpxscale.h index 9a86b75..86fc128 100644 --- a/vpx_scale/vpxscale.h +++ b/vpx_scale/vpxscale.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/wce/gen_scalers_armv4.asm b/vpx_scale/wce/gen_scalers_armv4.asm index 1c904ed..e495184 100644 --- a/vpx_scale/wce/gen_scalers_armv4.asm +++ b/vpx_scale/wce/gen_scalers_armv4.asm @@ -1,10 +1,11 @@ ; -; Copyright (c) 2010 The VP8 project authors. All Rights Reserved. +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. ; -; Use of this source code is governed by a BSD-style license and patent -; grant that can be found in the LICENSE file in the root of the source -; tree. All contributing project authors may be found in the AUTHORS -; file in the root of the source tree. +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. ; diff --git a/vpx_scale/wce/scalesystemdependant.c b/vpx_scale/wce/scalesystemdependant.c index a5a6a52..e9f2c26 100644 --- a/vpx_scale/wce/scalesystemdependant.c +++ b/vpx_scale/wce/scalesystemdependant.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/win32/scaleopt.c b/vpx_scale/win32/scaleopt.c index da0533e..3711fe5 100644 --- a/vpx_scale/win32/scaleopt.c +++ b/vpx_scale/win32/scaleopt.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/win32/scalesystemdependant.c b/vpx_scale/win32/scalesystemdependant.c index 9ed48bf..eab741f 100644 --- a/vpx_scale/win32/scalesystemdependant.c +++ b/vpx_scale/win32/scalesystemdependant.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/x86_64/scaleopt.c b/vpx_scale/x86_64/scaleopt.c index 3d2d5f2..101f5ff 100644 --- a/vpx_scale/x86_64/scaleopt.c +++ b/vpx_scale/x86_64/scaleopt.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/x86_64/scalesystemdependant.c b/vpx_scale/x86_64/scalesystemdependant.c index 43f05a6..89ae86e 100644 --- a/vpx_scale/x86_64/scalesystemdependant.c +++ b/vpx_scale/x86_64/scalesystemdependant.c @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/yv12config.h b/vpx_scale/yv12config.h index a8d0ce4..50d6e3b 100644 --- a/vpx_scale/yv12config.h +++ b/vpx_scale/yv12config.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/vpx_scale/yv12extend.h b/vpx_scale/yv12extend.h index 9968fea..0c239b9 100644 --- a/vpx_scale/yv12extend.h +++ b/vpx_scale/yv12extend.h @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/wince_wmain_adapter.cpp b/wince_wmain_adapter.cpp index db2119c..57f880a 100644 --- a/wince_wmain_adapter.cpp +++ b/wince_wmain_adapter.cpp @@ -1,10 +1,11 @@ /* - * Copyright (c) 2010 The VP8 project authors. All Rights Reserved. + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. * - * Use of this source code is governed by a BSD-style license and patent - * grant that can be found in the LICENSE file in the root of the source - * tree. All contributing project authors may be found in the AUTHORS - * file in the root of the source tree. + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. */ diff --git a/y4minput.c b/y4minput.c new file mode 100644 index 0000000..3eaec4e --- /dev/null +++ b/y4minput.c @@ -0,0 +1,881 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + * + * Based on code from the OggTheora software codec source code, + * Copyright (C) 2002-2010 The Xiph.Org Foundation and contributors. + */ +#include <stdlib.h> +#include <string.h> +#include "y4minput.h" + +static int y4m_parse_tags(y4m_input *_y4m,char *_tags){ + int got_w; + int got_h; + int got_fps; + int got_interlace; + int got_par; + int got_chroma; + char *p; + char *q; + got_w=got_h=got_fps=got_interlace=got_par=got_chroma=0; + for(p=_tags;;p=q){ + /*Skip any leading spaces.*/ + while(*p==' ')p++; + /*If that's all we have, stop.*/ + if(p[0]=='\0')break; + /*Find the end of this tag.*/ + for(q=p+1;*q!='\0'&&*q!=' ';q++); + /*Process the tag.*/ + switch(p[0]){ + case 'W':{ + if(sscanf(p+1,"%d",&_y4m->pic_w)!=1)return -1; + got_w=1; + }break; + case 'H':{ + if(sscanf(p+1,"%d",&_y4m->pic_h)!=1)return -1; + got_h=1; + }break; + case 'F':{ + if(sscanf(p+1,"%d:%d",&_y4m->fps_n,&_y4m->fps_d)!=2){ + return -1; + } + got_fps=1; + }break; + case 'I':{ + _y4m->interlace=p[1]; + got_interlace=1; + }break; + case 'A':{ + if(sscanf(p+1,"%d:%d",&_y4m->par_n,&_y4m->par_d)!=2){ + return -1; + } + got_par=1; + }break; + case 'C':{ + if(q-p>16)return -1; + memcpy(_y4m->chroma_type,p+1,q-p-1); + _y4m->chroma_type[q-p-1]='\0'; + got_chroma=1; + }break; + /*Ignore unknown tags.*/ + } + } + if(!got_w||!got_h||!got_fps)return -1; + if(!got_interlace)_y4m->interlace='?'; + if(!got_par)_y4m->par_n=_y4m->par_d=0; + /*Chroma-type is not specified in older files, e.g., those generated by + mplayer.*/ + if(!got_chroma)strcpy(_y4m->chroma_type,"420"); + return 0; +} + + + +/*All anti-aliasing filters in the following conversion functions are based on + one of two window functions: + The 6-tap Lanczos window (for down-sampling and shifts): + sinc(\pi*t)*sinc(\pi*t/3), |t|<3 (sinc(t)==sin(t)/t) + 0, |t|>=3 + The 4-tap Mitchell window (for up-sampling): + 7|t|^3-12|t|^2+16/3, |t|<1 + -(7/3)|x|^3+12|x|^2-20|x|+32/3, |t|<2 + 0, |t|>=2 + The number of taps is intentionally kept small to reduce computational + overhead and limit ringing. + + The taps from these filters are scaled so that their sum is 1, and the result + is scaled by 128 and rounded to integers to create a filter whose + intermediate values fit inside 16 bits. + Coefficients are rounded in such a way as to ensure their sum is still 128, + which is usually equivalent to normal rounding. + + Conversions which require both horizontal and vertical filtering could + have these steps pipelined, for less memory consumption and better cache + performance, but we do them separately for simplicity.*/ + +#define OC_MINI(_a,_b) ((_a)>(_b)?(_b):(_a)) +#define OC_MAXI(_a,_b) ((_a)<(_b)?(_b):(_a)) +#define OC_CLAMPI(_a,_b,_c) (OC_MAXI(_a,OC_MINI(_b,_c))) + +/*420jpeg chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + 420mpeg2 chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + BR | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + BR | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + We use a resampling filter to shift the site locations one quarter pixel (at + the chroma plane's resolution) to the right. + The 4:2:2 modes look exactly the same, except there are twice as many chroma + lines, and they are vertically co-sited with the luma samples in both the + mpeg2 and jpeg cases (thus requiring no vertical resampling).*/ +static void y4m_42xmpeg2_42xjpeg_helper(unsigned char *_dst, + const unsigned char *_src,int _c_w,int _c_h){ + int pli; + int y; + int x; + for(y=0;y<_c_h;y++){ + /*Filter: [4 -17 114 35 -9 1]/128, derived from a 6-tap Lanczos + window.*/ + for(x=0;x<OC_MINI(_c_w,2);x++){ + _dst[x]=(unsigned char)OC_CLAMPI(0,(4*_src[0]-17*_src[OC_MAXI(x-1,0)]+ + 114*_src[x]+35*_src[OC_MINI(x+1,_c_w-1)]-9*_src[OC_MINI(x+2,_c_w-1)]+ + _src[OC_MINI(x+3,_c_w-1)]+64)>>7,255); + } + for(;x<_c_w-3;x++){ + _dst[x]=(unsigned char)OC_CLAMPI(0,(4*_src[x-2]-17*_src[x-1]+ + 114*_src[x]+35*_src[x+1]-9*_src[x+2]+_src[x+3]+64)>>7,255); + } + for(;x<_c_w;x++){ + _dst[x]=(unsigned char)OC_CLAMPI(0,(4*_src[x-2]-17*_src[x-1]+ + 114*_src[x]+35*_src[OC_MINI(x+1,_c_w-1)]-9*_src[OC_MINI(x+2,_c_w-1)]+ + _src[_c_w-1]+64)>>7,255); + } + _dst+=_c_w; + _src+=_c_w; + } +} + +/*Handles both 422 and 420mpeg2 to 422jpeg and 420jpeg, respectively.*/ +static void y4m_convert_42xmpeg2_42xjpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + int c_w; + int c_h; + int c_sz; + int pli; + int y; + int x; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + for(pli=1;pli<3;pli++){ + y4m_42xmpeg2_42xjpeg_helper(_dst,_aux,c_w,c_h); + _dst+=c_sz; + _aux+=c_sz; + } +} + +/*This format is only used for interlaced content, but is included for + completeness. + + 420jpeg chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + 420paldv chroma samples are sited like: + YR------Y-------YR------Y------- + | | | | + | | | | + | | | | + YB------Y-------YB------Y------- + | | | | + | | | | + | | | | + YR------Y-------YR------Y------- + | | | | + | | | | + | | | | + YB------Y-------YB------Y------- + | | | | + | | | | + | | | | + + We use a resampling filter to shift the site locations one quarter pixel (at + the chroma plane's resolution) to the right. + Then we use another filter to move the C_r location down one quarter pixel, + and the C_b location up one quarter pixel.*/ +static void y4m_convert_42xpaldv_42xjpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + unsigned char *tmp; + int c_w; + int c_h; + int c_sz; + int pli; + int y; + int x; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+1)/2; + c_h=(_y4m->pic_h+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + c_sz=c_w*c_h; + tmp=_aux+2*c_sz; + for(pli=1;pli<3;pli++){ + /*First do the horizontal re-sampling. + This is the same as the mpeg2 case, except that after the horizontal + case, we need to apply a second vertical filter.*/ + y4m_42xmpeg2_42xjpeg_helper(tmp,_aux,c_w,c_h); + _aux+=c_sz; + switch(pli){ + case 1:{ + /*Slide C_b up a quarter-pel. + This is the same filter used above, but in the other order.*/ + for(x=0;x<c_w;x++){ + for(y=0;y<OC_MINI(c_h,3);y++){ + _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(tmp[0] + -9*tmp[OC_MAXI(y-2,0)*c_w]+35*tmp[OC_MAXI(y-1,0)*c_w] + +114*tmp[y*c_w]-17*tmp[OC_MINI(y+1,c_h-1)*c_w] + +4*tmp[OC_MINI(y+2,c_h-1)*c_w]+64)>>7,255); + } + for(;y<c_h-2;y++){ + _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(tmp[(y-3)*c_w] + -9*tmp[(y-2)*c_w]+35*tmp[(y-1)*c_w]+114*tmp[y*c_w] + -17*tmp[(y+1)*c_w]+4*tmp[(y+2)*c_w]+64)>>7,255); + } + for(;y<c_h;y++){ + _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(tmp[(y-3)*c_w] + -9*tmp[(y-2)*c_w]+35*tmp[(y-1)*c_w]+114*tmp[y*c_w] + -17*tmp[OC_MINI(y+1,c_h-1)*c_w]+4*tmp[(c_h-1)*c_w]+64)>>7,255); + } + _dst++; + tmp++; + } + _dst+=c_sz-c_w; + tmp-=c_w; + }break; + case 2:{ + /*Slide C_r down a quarter-pel. + This is the same as the horizontal filter.*/ + for(x=0;x<c_w;x++){ + for(y=0;y<OC_MINI(c_h,2);y++){ + _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(4*tmp[0] + -17*tmp[OC_MAXI(y-1,0)*c_w]+114*tmp[y*c_w] + +35*tmp[OC_MINI(y+1,c_h-1)*c_w]-9*tmp[OC_MINI(y+2,c_h-1)*c_w] + +tmp[OC_MINI(y+3,c_h-1)*c_w]+64)>>7,255); + } + for(;y<c_h-3;y++){ + _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(4*tmp[(y-2)*c_w] + -17*tmp[(y-1)*c_w]+114*tmp[y*c_w]+35*tmp[(y+1)*c_w] + -9*tmp[(y+2)*c_w]+tmp[(y+3)*c_w]+64)>>7,255); + } + for(;y<c_h;y++){ + _dst[y*c_w]=(unsigned char)OC_CLAMPI(0,(4*tmp[(y-2)*c_w] + -17*tmp[(y-1)*c_w]+114*tmp[y*c_w]+35*tmp[OC_MINI(y+1,c_h-1)*c_w] + -9*tmp[OC_MINI(y+2,c_h-1)*c_w]+tmp[(c_h-1)*c_w]+64)>>7,255); + } + _dst++; + tmp++; + } + }break; + } + /*For actual interlaced material, this would have to be done separately on + each field, and the shift amounts would be different. + C_r moves down 1/8, C_b up 3/8 in the top field, and C_r moves down 3/8, + C_b up 1/8 in the bottom field. + The corresponding filters would be: + Down 1/8 (reverse order for up): [3 -11 125 15 -4 0]/128 + Down 3/8 (reverse order for up): [4 -19 98 56 -13 2]/128*/ + } +} + +/*Perform vertical filtering to reduce a single plane from 4:2:2 to 4:2:0. + This is used as a helper by several converation routines.*/ +static void y4m_422jpeg_420jpeg_helper(unsigned char *_dst, + const unsigned char *_src,int _c_w,int _c_h){ + int y; + int x; + /*Filter: [3 -17 78 78 -17 3]/128, derived from a 6-tap Lanczos window.*/ + for(x=0;x<_c_w;x++){ + for(y=0;y<OC_MINI(_c_h,2);y+=2){ + _dst[(y>>1)*_c_w]=OC_CLAMPI(0,(64*_src[0] + +78*_src[OC_MINI(1,_c_h-1)*_c_w] + -17*_src[OC_MINI(2,_c_h-1)*_c_w] + +3*_src[OC_MINI(3,_c_h-1)*_c_w]+64)>>7,255); + } + for(;y<_c_h-3;y+=2){ + _dst[(y>>1)*_c_w]=OC_CLAMPI(0,(3*(_src[(y-2)*_c_w]+_src[(y+3)*_c_w]) + -17*(_src[(y-1)*_c_w]+_src[(y+2)*_c_w]) + +78*(_src[y*_c_w]+_src[(y+1)*_c_w])+64)>>7,255); + } + for(;y<_c_h;y+=2){ + _dst[(y>>1)*_c_w]=OC_CLAMPI(0,(3*(_src[(y-2)*_c_w] + +_src[(_c_h-1)*_c_w])-17*(_src[(y-1)*_c_w] + +_src[OC_MINI(y+2,_c_h-1)*_c_w]) + +78*(_src[y*_c_w]+_src[OC_MINI(y+1,_c_h-1)*_c_w])+64)>>7,255); + } + _src++; + _dst++; + } +} + +/*420jpeg chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + 422jpeg chroma samples are sited like: + Y---BR--Y-------Y---BR--Y------- + | | | | + | | | | + | | | | + Y---BR--Y-------Y---BR--Y------- + | | | | + | | | | + | | | | + Y---BR--Y-------Y---BR--Y------- + | | | | + | | | | + | | | | + Y---BR--Y-------Y---BR--Y------- + | | | | + | | | | + | | | | + + We use a resampling filter to decimate the chroma planes by two in the + vertical direction.*/ +static void y4m_convert_422jpeg_420jpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + int c_w; + int c_h; + int c_sz; + int dst_c_w; + int dst_c_h; + int dst_c_sz; + int tmp_sz; + int pic_sz; + int pli; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h; + c_h=_y4m->pic_h; + dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + dst_c_sz=dst_c_w*dst_c_h; + for(pli=1;pli<3;pli++){ + y4m_422jpeg_420jpeg_helper(_dst,_aux,c_w,c_h); + _aux+=c_sz; + _dst+=dst_c_sz; + } +} + +/*420jpeg chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + 422 chroma samples are sited like: + YBR-----Y-------YBR-----Y------- + | | | | + | | | | + | | | | + YBR-----Y-------YBR-----Y------- + | | | | + | | | | + | | | | + YBR-----Y-------YBR-----Y------- + | | | | + | | | | + | | | | + YBR-----Y-------YBR-----Y------- + | | | | + | | | | + | | | | + + We use a resampling filter to shift the original site locations one quarter + pixel (at the original chroma resolution) to the right. + Then we use a second resampling filter to decimate the chroma planes by two + in the vertical direction.*/ +static void y4m_convert_422_420jpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + unsigned char *tmp; + int c_w; + int c_h; + int c_sz; + int dst_c_w; + int dst_c_h; + int dst_c_sz; + int pli; + int y; + int x; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h; + c_h=_y4m->pic_h; + dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + dst_c_sz=c_w*dst_c_h; + tmp=_aux+2*c_sz; + for(pli=1;pli<3;pli++){ + /*In reality, the horizontal and vertical steps could be pipelined, for + less memory consumption and better cache performance, but we do them + separately for simplicity.*/ + /*First do horizontal filtering (convert to 422jpeg)*/ + y4m_42xmpeg2_42xjpeg_helper(tmp,_aux,c_w,c_h); + /*Now do the vertical filtering.*/ + y4m_422jpeg_420jpeg_helper(_dst,tmp,c_w,c_h); + _aux+=c_sz; + _dst+=dst_c_sz; + } +} + +/*420jpeg chroma samples are sited like: + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | BR | | BR | + | | | | + Y-------Y-------Y-------Y------- + | | | | + | | | | + | | | | + + 411 chroma samples are sited like: + YBR-----Y-------Y-------Y------- + | | | | + | | | | + | | | | + YBR-----Y-------Y-------Y------- + | | | | + | | | | + | | | | + YBR-----Y-------Y-------Y------- + | | | | + | | | | + | | | | + YBR-----Y-------Y-------Y------- + | | | | + | | | | + | | | | + + We use a filter to resample at site locations one eighth pixel (at the source + chroma plane's horizontal resolution) and five eighths of a pixel to the + right. + Then we use another filter to decimate the planes by 2 in the vertical + direction.*/ +static void y4m_convert_411_420jpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + unsigned char *tmp; + int c_w; + int c_h; + int c_sz; + int dst_c_w; + int dst_c_h; + int dst_c_sz; + int tmp_sz; + int pli; + int y; + int x; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h; + c_h=_y4m->pic_h; + dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + dst_c_sz=dst_c_w*dst_c_h; + tmp_sz=dst_c_w*c_h; + tmp=_aux+2*c_sz; + for(pli=1;pli<3;pli++){ + /*In reality, the horizontal and vertical steps could be pipelined, for + less memory consumption and better cache performance, but we do them + separately for simplicity.*/ + /*First do horizontal filtering (convert to 422jpeg)*/ + for(y=0;y<c_h;y++){ + /*Filters: [1 110 18 -1]/128 and [-3 50 86 -5]/128, both derived from a + 4-tap Mitchell window.*/ + for(x=0;x<OC_MINI(c_w,1);x++){ + tmp[x<<1]=(unsigned char)OC_CLAMPI(0,(111*_aux[0] + +18*_aux[OC_MINI(1,c_w-1)]-_aux[OC_MINI(2,c_w-1)]+64)>>7,255); + tmp[x<<1|1]=(unsigned char)OC_CLAMPI(0,(47*_aux[0] + +86*_aux[OC_MINI(1,c_w-1)]-5*_aux[OC_MINI(2,c_w-1)]+64)>>7,255); + } + for(;x<c_w-2;x++){ + tmp[x<<1]=(unsigned char)OC_CLAMPI(0,(_aux[x-1]+110*_aux[x] + +18*_aux[x+1]-_aux[x+2]+64)>>7,255); + tmp[x<<1|1]=(unsigned char)OC_CLAMPI(0,(-3*_aux[x-1]+50*_aux[x] + +86*_aux[x+1]-5*_aux[x+2]+64)>>7,255); + } + for(;x<c_w;x++){ + tmp[x<<1]=(unsigned char)OC_CLAMPI(0,(_aux[x-1]+110*_aux[x] + +18*_aux[OC_MINI(x+1,c_w-1)]-_aux[c_w-1]+64)>>7,255); + if((x<<1|1)<dst_c_w){ + tmp[x<<1|1]=(unsigned char)OC_CLAMPI(0,(-3*_aux[x-1]+50*_aux[x] + +86*_aux[OC_MINI(x+1,c_w-1)]-5*_aux[c_w-1]+64)>>7,255); + } + } + tmp+=dst_c_w; + _aux+=c_w; + } + tmp-=tmp_sz; + /*Now do the vertical filtering.*/ + y4m_422jpeg_420jpeg_helper(_dst,tmp,dst_c_w,c_h); + _dst+=dst_c_sz; + } +} + +/*Convert 444 to 420jpeg.*/ +static void y4m_convert_444_420jpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + unsigned char *tmp; + int c_w; + int c_h; + int c_sz; + int dst_c_w; + int dst_c_h; + int dst_c_sz; + int tmp_sz; + int pli; + int y; + int x; + /*Skip past the luma data.*/ + _dst+=_y4m->pic_w*_y4m->pic_h; + /*Compute the size of each chroma plane.*/ + c_w=(_y4m->pic_w+_y4m->src_c_dec_h-1)/_y4m->src_c_dec_h; + c_h=_y4m->pic_h; + dst_c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + dst_c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + dst_c_sz=dst_c_w*dst_c_h; + tmp_sz=dst_c_w*c_h; + tmp=_aux+2*c_sz; + for(pli=1;pli<3;pli++){ + /*Filter: [3 -17 78 78 -17 3]/128, derived from a 6-tap Lanczos window.*/ + for(y=0;y<c_h;y++){ + for(x=0;x<OC_MINI(c_w,2);x+=2){ + tmp[x>>1]=OC_CLAMPI(0,(64*_aux[0]+78*_aux[OC_MINI(1,c_w-1)] + -17*_aux[OC_MINI(2,c_w-1)] + +3*_aux[OC_MINI(3,c_w-1)]+64)>>7,255); + } + for(;x<c_w-3;x+=2){ + tmp[x>>1]=OC_CLAMPI(0,(3*(_aux[x-2]+_aux[x+3]) + -17*(_aux[x-1]+_aux[x+2])+78*(_aux[x]+_aux[x+1])+64)>>7,255); + } + for(;x<c_w;x+=2){ + tmp[x>>1]=OC_CLAMPI(0,(3*(_aux[x-2]+_aux[c_w-1])- + 17*(_aux[x-1]+_aux[OC_MINI(x+2,c_w-1)])+ + 78*(_aux[x]+_aux[OC_MINI(x+1,c_w-1)])+64)>>7,255); + } + tmp+=dst_c_w; + _aux+=c_w; + } + tmp-=tmp_sz; + /*Now do the vertical filtering.*/ + y4m_422jpeg_420jpeg_helper(_dst,tmp,dst_c_w,c_h); + _dst+=dst_c_sz; + } +} + +/*The image is padded with empty chroma components at 4:2:0.*/ +static void y4m_convert_mono_420jpeg(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ + int c_sz; + _dst+=_y4m->pic_w*_y4m->pic_h; + c_sz=((_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h)* + ((_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v); + memset(_dst,128,c_sz*2); +} + +/*No conversion function needed.*/ +static void y4m_convert_null(y4m_input *_y4m,unsigned char *_dst, + unsigned char *_aux){ +} + +int y4m_input_open(y4m_input *_y4m,FILE *_fin,char *_skip,int _nskip){ + char buffer[80]; + int ret; + int i; + /*Read until newline, or 80 cols, whichever happens first.*/ + for(i=0;i<79;i++){ + if(_nskip>0){ + buffer[i]=*_skip++; + _nskip--; + } + else{ + ret=fread(buffer+i,1,1,_fin); + if(ret<1)return -1; + } + if(buffer[i]=='\n')break; + } + /*We skipped too much header data.*/ + if(_nskip>0)return -1; + if(i==79){ + fprintf(stderr,"Error parsing header; not a YUV2MPEG2 file?\n"); + return -1; + } + buffer[i]='\0'; + if(memcmp(buffer,"YUV4MPEG",8)){ + fprintf(stderr,"Incomplete magic for YUV4MPEG file.\n"); + return -1; + } + if(buffer[8]!='2'){ + fprintf(stderr,"Incorrect YUV input file version; YUV4MPEG2 required.\n"); + } + ret=y4m_parse_tags(_y4m,buffer+5); + if(ret<0){ + fprintf(stderr,"Error parsing YUV4MPEG2 header.\n"); + return ret; + } + if(_y4m->interlace=='?'){ + fprintf(stderr,"Warning: Input video interlacing format unknown; " + "assuming progressive scan.\n"); + } + else if(_y4m->interlace!='p'){ + fprintf(stderr,"Input video is interlaced; " + "Only progressive scan handled.\n"); + return -1; + } + if(strcmp(_y4m->chroma_type,"420")==0|| + strcmp(_y4m->chroma_type,"420jpeg")==0){ + _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h + +2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2); + /*Natively supported: no conversion required.*/ + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=0; + _y4m->convert=y4m_convert_null; + } + else if(strcmp(_y4m->chroma_type,"420mpeg2")==0){ + _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first.*/ + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz= + 2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2); + _y4m->convert=y4m_convert_42xmpeg2_42xjpeg; + } + else if(strcmp(_y4m->chroma_type,"420paldv")==0){ + _y4m->src_c_dec_h=_y4m->dst_c_dec_h=_y4m->src_c_dec_v=_y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first. + We need to make two filter passes, so we need some extra space in the + aux buffer.*/ + _y4m->aux_buf_sz=3*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2); + _y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*((_y4m->pic_h+1)/2); + _y4m->convert=y4m_convert_42xpaldv_42xjpeg; + } + else if(strcmp(_y4m->chroma_type,"422jpeg")==0){ + _y4m->src_c_dec_h=_y4m->dst_c_dec_h=2; + _y4m->src_c_dec_v=1; + _y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first.*/ + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*_y4m->pic_h; + _y4m->convert=y4m_convert_422jpeg_420jpeg; + } + else if(strcmp(_y4m->chroma_type,"422")==0){ + _y4m->src_c_dec_h=_y4m->dst_c_dec_h=2; + _y4m->src_c_dec_v=1; + _y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first. + We need to make two filter passes, so we need some extra space in the + aux buffer.*/ + _y4m->aux_buf_read_sz=2*((_y4m->pic_w+1)/2)*_y4m->pic_h; + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h; + _y4m->convert=y4m_convert_422_420jpeg; + } + else if(strcmp(_y4m->chroma_type,"411")==0){ + _y4m->src_c_dec_h=4; + _y4m->dst_c_dec_h=2; + _y4m->src_c_dec_v=1; + _y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first. + We need to make two filter passes, so we need some extra space in the + aux buffer.*/ + _y4m->aux_buf_read_sz=2*((_y4m->pic_w+3)/4)*_y4m->pic_h; + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h; + _y4m->convert=y4m_convert_411_420jpeg; + } + else if(strcmp(_y4m->chroma_type,"444")==0){ + _y4m->src_c_dec_h=1; + _y4m->dst_c_dec_h=2; + _y4m->src_c_dec_v=1; + _y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first. + We need to make two filter passes, so we need some extra space in the + aux buffer.*/ + _y4m->aux_buf_read_sz=2*_y4m->pic_w*_y4m->pic_h; + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz+((_y4m->pic_w+1)/2)*_y4m->pic_h; + _y4m->convert=y4m_convert_444_420jpeg; + } + else if(strcmp(_y4m->chroma_type,"444alpha")==0){ + _y4m->src_c_dec_h=1; + _y4m->dst_c_dec_h=2; + _y4m->src_c_dec_v=1; + _y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*Chroma filter required: read into the aux buf first. + We need to make two filter passes, so we need some extra space in the + aux buffer. + The extra plane also gets read into the aux buf. + It will be discarded.*/ + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=3*_y4m->pic_w*_y4m->pic_h; + _y4m->convert=y4m_convert_444_420jpeg; + } + else if(strcmp(_y4m->chroma_type,"mono")==0){ + _y4m->src_c_dec_h=_y4m->src_c_dec_v=0; + _y4m->dst_c_dec_h=_y4m->dst_c_dec_v=2; + _y4m->dst_buf_read_sz=_y4m->pic_w*_y4m->pic_h; + /*No extra space required, but we need to clear the chroma planes.*/ + _y4m->aux_buf_sz=_y4m->aux_buf_read_sz=0; + _y4m->convert=y4m_convert_mono_420jpeg; + } + else{ + fprintf(stderr,"Unknown chroma sampling type: %s\n",_y4m->chroma_type); + return -1; + } + /*The size of the final frame buffers is always computed from the + destination chroma decimation type.*/ + _y4m->dst_buf_sz=_y4m->pic_w*_y4m->pic_h + +2*((_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h)* + ((_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v); + _y4m->dst_buf=(unsigned char *)malloc(_y4m->dst_buf_sz); + _y4m->aux_buf=(unsigned char *)malloc(_y4m->aux_buf_sz); + return 0; +} + +void y4m_input_close(y4m_input *_y4m){ + free(_y4m->dst_buf); + free(_y4m->aux_buf); +} + +int y4m_input_fetch_frame(y4m_input *_y4m,FILE *_fin,vpx_image_t *_img){ + char frame[6]; + int pic_sz; + int frame_c_w; + int frame_c_h; + int c_w; + int c_h; + int c_sz; + int ret; + /*Read and skip the frame header.*/ + ret=fread(frame,1,6,_fin); + if(ret<6)return 0; + if(memcmp(frame,"FRAME",5)){ + fprintf(stderr,"Loss of framing in Y4M input data\n"); + return -1; + } + if(frame[5]!='\n'){ + char c; + int j; + for(j=0;j<79&&fread(&c,1,1,_fin)&&c!='\n';j++); + if(j==79){ + fprintf(stderr,"Error parsing Y4M frame header\n"); + return -1; + } + } + /*Read the frame data that needs no conversion.*/ + if(fread(_y4m->dst_buf,1,_y4m->dst_buf_read_sz,_fin)!=_y4m->dst_buf_read_sz){ + fprintf(stderr,"Error reading Y4M frame data.\n"); + return -1; + } + /*Read the frame data that does need conversion.*/ + if(fread(_y4m->aux_buf,1,_y4m->aux_buf_read_sz,_fin)!=_y4m->aux_buf_read_sz){ + fprintf(stderr,"Error reading Y4M frame data.\n"); + return -1; + } + /*Now convert the just read frame.*/ + (*_y4m->convert)(_y4m,_y4m->dst_buf,_y4m->aux_buf); + /*Fill in the frame buffer pointers. + We don't use vpx_img_wrap() because it forces padding for odd picture + sizes, which would require a separate fread call for every row.*/ + memset(_img,0,sizeof(*_img)); + /*Y4M has the planes in Y'CbCr order, which libvpx calls Y, U, and V.*/ + _img->fmt=IMG_FMT_I420; + _img->w=_img->d_w=_y4m->pic_w; + _img->h=_img->d_h=_y4m->pic_h; + /*This is hard-coded to 4:2:0 for now, as that's all VP8 supports.*/ + _img->x_chroma_shift=1; + _img->y_chroma_shift=1; + _img->bps=12; + /*Set up the buffer pointers.*/ + pic_sz=_y4m->pic_w*_y4m->pic_h; + c_w=(_y4m->pic_w+_y4m->dst_c_dec_h-1)/_y4m->dst_c_dec_h; + c_h=(_y4m->pic_h+_y4m->dst_c_dec_v-1)/_y4m->dst_c_dec_v; + c_sz=c_w*c_h; + _img->stride[PLANE_Y]=_y4m->pic_w; + _img->stride[PLANE_U]=_img->stride[PLANE_V]=c_w; + _img->planes[PLANE_Y]=_y4m->dst_buf; + _img->planes[PLANE_U]=_y4m->dst_buf+pic_sz; + _img->planes[PLANE_V]=_y4m->dst_buf+pic_sz+c_sz; + return 0; +} diff --git a/y4minput.h b/y4minput.h new file mode 100644 index 0000000..1a01bcd --- /dev/null +++ b/y4minput.h @@ -0,0 +1,60 @@ +/* + * Copyright (c) 2010 The WebM project authors. All Rights Reserved. + * + * Use of this source code is governed by a BSD-style license + * that can be found in the LICENSE file in the root of the source + * tree. An additional intellectual property rights grant can be found + * in the file PATENTS. All contributing project authors may + * be found in the AUTHORS file in the root of the source tree. + * + * Based on code from the OggTheora software codec source code, + * Copyright (C) 2002-2010 The Xiph.Org Foundation and contributors. + */ +#if !defined(_y4minput_H) +# define _y4minput_H (1) +# include <stdio.h> +# include "vpx/vpx_image.h" + + + +typedef struct y4m_input y4m_input; + + + +/*The function used to perform chroma conversion.*/ +typedef void (*y4m_convert_func)(y4m_input *_y4m, + unsigned char *_dst,unsigned char *_src); + + + +struct y4m_input{ + int pic_w; + int pic_h; + int fps_n; + int fps_d; + int par_n; + int par_d; + char interlace; + int src_c_dec_h; + int src_c_dec_v; + int dst_c_dec_h; + int dst_c_dec_v; + char chroma_type[16]; + /*The size of each converted frame buffer.*/ + size_t dst_buf_sz; + /*The amount to read directly into the converted frame buffer.*/ + size_t dst_buf_read_sz; + /*The size of the auxilliary buffer.*/ + size_t aux_buf_sz; + /*The amount to read into the auxilliary buffer.*/ + size_t aux_buf_read_sz; + y4m_convert_func convert; + unsigned char *dst_buf; + unsigned char *aux_buf; +}; + +int y4m_input_open(y4m_input *_y4m,FILE *_fin,char *_skip,int _nskip); +void y4m_input_close(y4m_input *_y4m); +int y4m_input_fetch_frame(y4m_input *_y4m,FILE *_fin,vpx_image_t *img); + +#endif |