Upgrade to the latest .webm project code.

Change-Id: I33907e0c9ded667e54d31e2f9226c77501731c6c

21 files changed, 1225 insertions, 1621 deletions

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;
+    }
+}