summaryrefslogtreecommitdiffstats
path: root/encoder
Commit message (Collapse)AuthorAgeFilesLines
* Encoder: Return error for odd resolutionreplicant-6.0-0004-rc1Akshata Jadhav2018-06-081-0/+18
| | | | | | | | Bug: 73625898 Test: ran POC before/after under ASAN Change-Id: I9765b57f4afc6a2b6ad9cd19c8c7c5000beb9de9 (cherry picked from commit 9fa58d4db3ef176ed54af5f602970b48624be413) CVE-2018-9351
* Remove __DATE__/__TIME__ from Android buildsDan Willemsen2015-11-031-0/+7
| | | | | | | This removes unnecessary changes from build to build. Bug: 24204119 Change-Id: I25258314fe56ea186d38d756fb5fb04aa689a812 (cherry picked from commit f197ccb7562bac6f5bae40c9cb9dbb46a77baeb5)
* Encoder added support for VUI paramsHamsalekha S2015-10-095-6/+145
| | | | | | | SPS will now include vui params Removed a trailing space in ih264e_api.c Change-Id: I67a57741689206e571e9a83f7e3360149f87e06a
* Merge "Encoder: Fixed an issue in handling FPS greater than 60" into mnc-devRonghua Wu2015-08-031-2/+10
|\
| * Encoder: Fixed an issue in handling FPS greater than 60Harish Mahendrakar2015-08-031-2/+10
| | | | | | | | | | | | | | FPS > 60 was resulting in an integer overflow, so using float now. Bug: 22504214 Change-Id: I4626ba6a43167c00c42dd6440df37bd1eb1f151a
* | Don't use side effects in the macro PUT_BITS_SEV parametersMartin Storsjo2015-07-311-4/+8
| | | | | | | | | | | | | | | | | | | | | | | | If ENTROPY_TRACE is defined, the value parameter gets evaluated twice. Since this parameter includes ++, the pointer gets incremented twice. This fixes encoding with trace enabled. Bug: 22860270 Change-Id: I3d71ac33d007301e488d264a33c7b5c4d4ff2b56
* | Made gops closedHarinarayanan K K2015-07-313-10/+13
| | | | | | | | | | | | | | | | | | | | Now no picture in a gop will depend on any picture in the previous gop All calls to force an IDR are respected as soon as possible Bug: 22860270 Change-Id: Id2bcd5687c3fb786336a01894c1b9228d67a87a6
* | Always do recon of 4x4 intra blocksMartin Storsjo2015-07-311-21/+13
| | | | | | | | | | | | | | | | | | | | | | | | Even if this 4x4 block isn't at the bottom or right border, the other 4x4 intra blocks within the same macroblock may use it for intra prediction. This fixes intra coded macroblocks in b-frames. Bug: 22860270 Change-Id: Ifdf48c1b2bbf232e785d6d5b8244aacba1ad3dd6
* | Validate the u4_entropy_coding_mode fieldMartin Storsjo2015-07-312-0/+12
| | | | | | | | | | | | | | | | | | Previously, if this field wasn't set, the invalid coding mode could lead to crashes later. Bug: 22860270 Change-Id: If8dd2f8d5f2fbdd6bb76772344f959df23159167
* | Fixed issue in cost computation for BIPRED MEHarinarayanan K K2015-07-311-4/+4
| | | | | | | | | | | | | | | | | | BIPRED ME was discarding the halfpel motion vector components in cost computation. That is fixed now. Bug: 22860270 Change-Id: If6c77096d701a258920b14134ff120fd1540f254
* | Cabac optimizationsDoney Alex2015-07-313-75/+32
| | | | | | | | | | | | | | | | | | | | Optimized cabac functions for writing bypass bins. Removed unused return from cabac flush function. Removed the macro REV_NBITS. Bug: 22860270 Change-Id: Iece82797f2f45a35281817a2b77a7c7fe4e02bd1
* | Use INT_MAX as i4_max_sad in ih264e_evaluate_bipredMartin Storsjo2015-07-311-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The stored i4_mb_distortion that was used as i4_max_sad previously might come from a different coding mode - even if pf_ime_compute_sad_16x16 returns early when i4_mb_distortion is larger than ps_mb_ctxt_bi->i4_mb_distortion, the calculated i4_mb_cost can still end up smaller than ps_mb_ctxt_bi->i4_mb_cost. Therefore, set i4_max_sad to INT_MAX to avoid the early exit in pf_ime_compute_sad_16x16 here, since it can't be used reliably (and returning early can end up choosing a suboptimal mode). This makes sure that the encoded output is identical when the SIMD optimized routines are disabled, when using B-frames. Bug: 22860270 Change-Id: I9b8d524f88020e557541700133c7a0b48fe6a3b8
* | Make sure that apv_bufs[0] and u4_is_last always are setMartin Storsjo2015-07-312-2/+12
| | | | | | | | | | | | | | | | | | | | | | | | When encoding in header mode, the s_inp_buf struct that was written to ps_video_encode_op->s_ive_op.s_inp_buf was completely uninitialized. In ih264e_input_queue_update, make sure to initialize u4_is_last when skipping frames. Bug: 22860270 Change-Id: I87e677acd00baf4f732ca7d35ee192e7f1f73994
* | Check the width/height and adjust level used for RCMartin Storsjo2015-07-311-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Since 6cb6772805, the actual level used is MAX(u4_max_level, ih264e_get_min_level()), which means that it isn't a fatal error to set u4_max_level to a too low level (if the caller doesn't know or care). Despite this, the actual value set by the caller was still used for initializing the RC. This allows the caller to not have to set u4_max_level (currently every caller has to duplicate a table or code snippet for determining the right level for each resolution). The caller still can set it to a higher value if he wants to explicitly use a higher level than what is necessary. This makes sure that the output is identical even if u4_max_level is set too low. Bug: 22860270 Change-Id: Icdc6c0652a97568d1eafc050226357a9d643d809
* | Fixed encode of trailing B framesHarinarayanan K K2015-07-311-0/+80
| | | | | | | | | | | | | | | | | | In case a P frame was marked as the last frame to encode, the B frames previous to it in the same gop were not getting encoded. This is fixed now. Bug: 22860270 Change-Id: I871f2a1d52b8db9ee75426b00287b58203586f23
* | Remove the now unused stride field from the set dimensions structMartin Storsjo2015-07-312-12/+0
| | | | | | | | | | | | | | | | | | Since 90a3904fd3 and d020be5ac7, this field doesn't have any actual effect - the strides are taken from the input pictures. Bug: 22860270 Change-Id: I9b08cbf8ee558e038d0017a6176d7b3bd3c428bd
* | Fixed encode of single frameHarinarayanan K K2015-07-311-1/+1
| | | | | | | | | | | | | | | | | | Encoding a single frame used to make encoder to go into infinite loop. That is fixed now Bug: 22860270 Change-Id: I51752bdc5fd6071868884c2ff6074c14d05c8564
* | Don't declare variables after statementsMartin Storsjo2015-07-313-20/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | This fixes building with compilers that are strict about the pre-C99 rule about having all variable declarations before statements. (The previous version of this commit was originally made before the commit adding support for main profile, which added a bunch more similar issues.) Bug: 22860270 Change-Id: I0a927deb2ef4125f79429d108439e249fa531607
* | Remove double semicolons (empty statements)Martin Storsjo2015-07-311-1/+1
|/ | | | | | | | Some compilers can't handle this, and these occurrances are typos. Bug: 22860270 Change-Id: I82349f103cef1f196d1a85dae307aff83c7d6090
* Fix MIPS buildMarco Nelissen2015-06-251-1/+1
| | | | Change-Id: I8190ce01d313f987f142a0c7ef5ae31f01562510
* Don't declare variables after statementsMartin Storsjo2015-06-253-11/+16
| | | | | | | This fixes building with compilers that are strict about the pre-C99 rule about having all variable declarations before statements. Change-Id: I06dd0cc42a6dbe90025c3c913a0c8d9ffb0acbcb
* Reorder code to avoid reading uninitialized dataMartin Storsjo2015-06-251-2/+2
| | | | | | | | | If the current slice isn't BSLICE, u1_weighted_bipred_idc is uninitialized, which valgrind can complain about (mostly in uninitialized builds). By reordering the conditions, we avoid doing conditional jumps based on uninitialized data. Change-Id: I4d601fd1a4e34d85fe5992ecb934ac3747953960
* Force regenerating the header via i4_gen_header instead of i4_header_modeMartin Storsjo2015-06-252-1/+2
| | | | | | | | | | | | If i4_header_mode is set (without the caller explicitly requesting it), the next frame to be encoded will be missed altogether (when only a header is output). Instead force a header to be generated prepended to the next frame. If the caller explicitly requests header mode, it will be output in a separate call. Change-Id: If475717edf980bccb2076f880c60c9dd7dede904
* Don't use static structs for rc initMartin Storsjo2015-06-2512-13/+13
| | | | | | | | | | | | | Most of them was benign and didn't risk any data race; they were used for write-only memory during init. The array as_itt_memtab in ih264e_get_rate_control_mem_tab could however cause a real data race, if multiple threads try to initialize an encoder at the same time, since it used as temporary storage when mapping between different structs. Change-Id: I3ca29f0e3be2fad154becc3ee29867e591975a74
* Initialize i4_non_ref_frames_in_stream before encoding SPS/PPSMartin Storsjo2015-06-252-6/+6
| | | | | | | When header mode is enabled, the SPS is populated before this field was initialized in ih264e_pic_init. Change-Id: I7e38828e12bee15bd3a53562b4e4d0d65354326a
* Allow setting profile IV_PROFILE_MAINMartin Storsjo2015-06-251-1/+2
| | | | | | | | Now once that the encoder supports the main profile, allow it to be set in the IVE_CMD_CTL_SET_PROFILE_PARAMS request. The actual profile value set here is ignored in practice, but at least allow setting it to main.
* Added check for minimum output buffer size.Harinarayanan K K2015-06-255-4/+17
| | | | | | | | Changed macro specifying the minimum size required for output buffer. Added an error check on the size allocated for output buffer. Change-Id: I98e4f46e62ffc974df760f2633689de079ca3e5e
* Added code to handle cases with qp less than 10Harinarayanan K K2015-06-252-10/+35
| | | | | | | | | | | | When qp goes less than 10, all I16x16 mode evaluations are disabled and I4x4 evaluations are enabled irrespective of preset. This will ensure that the residual will not exceed the supported range of entropy Minimum Qp cannot be set less than 4 now. When QP goes lesser, the residual may exceed the supported range even with I4x4 enabled. Change-Id: I25b404fcd9c9e9dbdd77679280968635ee047eb3
* Reduced memory requirements.Harinarayanan K K2015-06-257-64/+69
| | | | | | | | | | | | | | | | Buffer allocation is based on minimum level required for the input resolution rather than the input max level. Number of maximum context sets can be set to 1 to reduce memory usage. Added a macro ENC_MIN_PU_SIZE for minimum size of inter prediction unit supported by encoder. Changed the maximum constraint on number of MBs for NMB processing to width in Mbs. Change-Id: I5a9255e93935d90c13262681aafc772aedf8ae81
* Fixed QP lockupsHarinarayanan K K2015-06-252-99/+103
| | | | | | | | | Fixed QP getting locked up at particluar values even when it is supposed to be changing by ensuring that Qp will have a swing of atleast +-2 from previous value and correcting the Qp translation tables Change-Id: Id6f0fa4f282c288b81cfe278f34957a0fa68ff8c
* Fixed issues with forcing I and IDR frames.Harinarayanan K K2015-06-253-25/+37
| | | | | | | | | | Setting I frame interval to 1 caused encoder to go into infinite loop. Fixed that by stopping RC update. Forcing of I and IDR frame did not cause the frames at expected time due to delayed encoding. This is fixed now. Change-Id: Ibf549fb335c053ee33de768455b20e46be3e7c1c
* Fix string handling for generating version stringsMartin Storsjo2015-06-251-14/+5
| | | | | | | | | | | | | | | Don't do concatenation with strncat; the length parameter in strncat is only for how many chars to append at most, not for the full output buffer size. To safely use strncat, one would have to do strncat(buf, str, sizeof(buf) - strlen(buf)). By using snprintf, we guarantee that the buffer is null terminated, and we don't need to use strnlen at all. (If compatibility with older MSVC versions that lack snprintf, one can use _snprintf instead and manually add the null termination.) Change-Id: I1c2322c7a406ddd5e6551a96c460da60deeffda1
* Remove a duplicate headerMartin Storsjo2015-06-251-101/+0
| | | | | | ithread.h is already in the common directory. Change-Id: I300fd9c07075e16793e7b645e483663d081ca0cc
* Remove the codec level stride fieldMartin Storsjo2015-06-253-17/+0
| | | | | | | | | This field isn't used at the moment at all. The public field ive_ctl_set_dimensions_ip_t.u4_strd is kept to avoid breaking other calling code, although it doesn't have any actual function any longer. Change-Id: Icfc88846fa63f209c67b8e1f8e0cc21bc4ddb73a
* Always copy data to a local buffer if padding is neededMartin Storsjo2015-06-251-12/+20
| | | | | | | | | | | | If horizontal padding is needed, copy the data to avoid having to try to write into the input buffer. If vertical padding is needed, we could get away with only copying the last MB row (as previously was done), but then one extra row of pixels are needed for intra pred. Thus always copy all of the picture, for simplicity, if padding is needed. Change-Id: I6b456a12956120d64e7c6b86d7b571e159f4566b
* Set the luma/chroma strides depending on source bufferMartin Storsjo2015-06-252-2/+12
| | | | | | | | | | | | | | | | | If the data is read directly from the input buffer, set the stride values to those of the input buffer. If the data is copied, set the stride to the value of the local buffer instead. This fixes handling of cases where the input buffer stride is (significantly) larger than the maxwidth of the video. This also effectively makes the parameter ive_ctl_set_dimensions_ip_t.u4_strd useless - nothing needs to know the stride until you actually encode a frame, and at that point, we can either use the stride of the input buffer, or of the local pu1_y_csc_buf_base where the copied content is stored. Change-Id: Icde400b4a0867d25855e621e143454e608748aa3
* Use a separate field for the chroma strideMartin Storsjo2015-06-255-13/+19
| | | | | | | | | | | | | | When both luma and chroma are copied to the local buffer (either due to yuv format conversion, or due to padding), they have got the same stride, but if chroma is copied while luma is used directly from the input buffer, they might have different strides. Therefore add a separate field for chroma stride. This commit only adds the field, while it still has got the same value as before. Change-Id: I0dce97ad4d91cd1d9aba4b4472c6a0de45a314bc
* armv8: Don't touch the x18 registerMartin Storsjo2015-06-251-1/+0
| | | | | | | | | The x18 register is platform specific, and using it in portable assembler code should be avoided. In this particular case, using it is not actually needed at all. Change-Id: I07016fdd3a264ed77eba37abcd1748b51cf7a49e
* arm: Properly match /* */ comments in the arm assemblyMartin Storsjo2015-06-258-14/+3
| | | | | | | | | | | | | Some versions of the clang built-in assembler interprets /* as a start of a multi-line comment, even within lines starting with a @. Therefore, make sure to terminate these properly. This fixes warnings about /* within a block comment when building using the clang built-in assembler (in some versions of clang), and fixes issues where part of the assembly source was effectively commented out, when building with this assembler. Change-Id: I4b520334d92740f8c60def94ee990561b2cb8b7c
* armv8: Don't accidentally write 8 bytes instead of 4Martin Storsjo2015-06-252-18/+18
| | | | | | | This fixes running when built with clang - by pure luck the out of bounds writes didn't cause any harm when the calling C code was built with gcc. Change-Id: I4e33dd403b785cbcd24fb80f4db7a86310a32225
* armv8: Remove redundant NEON element size declarationsMartin Storsjo2015-06-251-122/+122
| | | | | | | | | | When specifying one specific lane of the vector, the number of lanes don't need to be specified. The clang built-in assembler doesn't allow the redundant declarations, while binutils gas work fine with both forms. Change-Id: I86077ce0774d4594a1295b6860e4944df87dde2f
* Handle non-mod-16 widths in the NEON version of 422i->420sp conversionMartin Storsjo2015-06-251-6/+31
| | | | | | | This uses the same mechanism as in ih264e_fmt_conv_420p_to_420sp_a9q to do an overlapping copy of the last few pixels. Change-Id: Ic93a00d0683337fd0f6f540119000466685f64bd
* Added support for Main Profile toolsets in encoder.Harinarayanan K K2015-06-2554-1990/+8411
| | | | | | | | Added support for CABAC entropy coding. Added support for B slices. Fixed an issue in rate control constant QP mode. Change-Id: Ib759d35e8e943f941aa9b8bbff0362d92c619994
* Fixed an overread in YUV420 Semi-planar input usecaseHarish Mahendrakar2015-06-122-1/+56
| | | | | | | | Handled non-multiple of 16 dimensions for 420 semiplanar input Modified test code to remove alignment of width and height Bug: 21586373 Change-Id: I83ff8165364a863d577fcac81e711b07eec9c004
* Support levels 5.0 and 5.1 in the encoder properlyMartin Storsjo2015-06-031-0/+8
| | | | | | | | The function ih264e_get_lvl_idx lacked handling of them before, which meant that it returned index 0 (corresponding to level 1.0) if level 5.0 or 5.1 were selected. Change-Id: I482213e68e11fd091c72ec57a593392aa1a0a4ae
* Remove leftover printfs in x86 encoder initializationMartin Storsjo2015-06-032-2/+0
| | | | | | | A library shouldn't printf things to the common stdout, at least not in normal builds. Change-Id: Ifd875369c2347549f141fe0f9754e51814c0de96
* Only initialize u4_deblk_prev_row if it will be usedMartin Storsjo2015-06-031-14/+14
| | | | | | | This avoids reads out of bounds when encoding videos with a height of one single macroblock. Change-Id: Iec982cebb0995c8ecf160751eaa72df38604e8b0
* Fix an overread in the slice mapMartin Storsjo2015-06-031-0/+3
| | | | | | | | | After incrementing the macroblock position in ih264e_update_proc_ctxt, the slice index of the next macroblock is read. The value that is read isn't actually used, but make sure that the array is large enough to avoid overreads. Change-Id: I44ddb49773244c7d1e404199fc57a9e30a8959e0
* Include space for chroma in the padding allocated in ↵Martin Storsjo2015-06-031-1/+1
| | | | | | | | | | | | | | ih264e_get_total_pic_buf_size Chroma is already accounted for in e.g. num_samples by multiplying the number of luma samples by 3/2, so include it in the padding calculation as well. This fixes encoding 176x144 at level 1.0 - normally this isn't an issue since the padding is a bit overestimated, but for this particular resolution/level combination, it fails unless this is calculated properly. Change-Id: Id5df6e5aee4d3d78f4c28a3b7d5fed6fe42a1f2c
* Resolved warnings and fixed alignment of few assembliesHarish Mahendrakar2015-05-056-680/+685
| | | | | | | | | Resolved warnings seen in x86 modules Fixed alignment of few modules Updated comments in few arm modules for consistency Fixed warnings seen in clang build Change-Id: I0623169b5e84a6a6f09c3d2212e754101272f5e9
generated by cgit v1.2.3 (git 2.25.1) at 2024-04-23 13:27:39 +0000