[Ffmpeg-devel-irc] ffmpeg-devel.log.20170719

burek burek021 at gmail.com
Thu Jul 20 03:05:04 EEST 2017


[01:01:27 CEST] <cone-839> ffmpeg 03Jun Zhao 07master:5b8a708492f8: lavc/vaapi_encode_h265: Remove duplicate slice_segment_address.
[01:01:28 CEST] <cone-839> ffmpeg 03Mark Thompson 07master:8a442d7a8a68: pixdesc: Improve scoring for opaque/unknown pixel formats
[01:01:29 CEST] <cone-839> ffmpeg 03Mark Thompson 07master:3192821237f6: pixdesc: Add a test for av_find_best_pix_fmt_of_2()
[01:01:30 CEST] <cone-839> ffmpeg 03Mark Thompson 07master:e6b4fb95d76b: doc: Fix names of VAAPI encoders
[01:56:08 CEST] <cone-839> ffmpeg 03Carl Eugen Hoyos 07master:5c7e7ac6d450: lavc/codec_desc: Mark some uncompressed codecs as lossless.
[02:08:45 CEST] <cone-839> ffmpeg 03foo86 07master:f6b86b62f785: avcodec/dca: move some enumeration typedefs into headers
[02:08:46 CEST] <cone-839> ffmpeg 03foo86 07master:9e37cc11012f: avcodec/dca: move bits per sample array to dca.c
[02:08:47 CEST] <cone-839> ffmpeg 03foo86 07master:2123ddb4251b: avcodec: add avpriv_dca_parse_core_frame_header()
[02:08:48 CEST] <cone-839> ffmpeg 03foo86 07master:33fc6c17a6e7: avcodec/dca_core: switch to common frame header parsing function
[02:08:49 CEST] <cone-839> ffmpeg 03foo86 07master:e54b9be1ffd6: avcodec/dca_parser: switch to common frame header parsing function
[02:08:50 CEST] <cone-839> ffmpeg 03foo86 07master:3b7ec920af42: avformat/dtsdec: switch to common frame header parsing function
[02:08:51 CEST] <cone-839> ffmpeg 03foo86 07master:4b6c212e859f: avcodec/dca_parser: export profile information
[02:08:52 CEST] <cone-839> ffmpeg 03foo86 07master:07a7f08b1e4b: avcodec/dca_parser: avoid use of magic values
[02:08:53 CEST] <cone-839> ffmpeg 03foo86 07master:902cd30e9b72: avcodec/dca_core: probe extension headers directly
[02:08:54 CEST] <cone-839> ffmpeg 03foo86 07master:012620aa1b40: avcodec/dca: avoid using bitstream reader in a non-standard way
[03:22:42 CEST] <atomnuker> michaelni: is a shift of exactly the number of bits of a variable really undefined?
[03:23:04 CEST] <atomnuker> how did they fail to define something so trivial?
[03:32:02 CEST] <michaelni> atomnuker, 6.5.7 "The integer promotions are performed on each of the operands. The type of the result is that of the promoted left operand. If the value of the right operand is negative or is greater than or equal to the width of the promoted left operand, the behavior is undefined. "
[03:33:09 CEST] <atomnuker> damn, where's the logic in that?
[03:34:55 CEST] <michaelni> iam pretty sure there are reasons for this design but iam the wrong one to ask, i dont know the reasons, i could merly guess
[03:36:52 CEST] <atomnuker> even a guess would give me a hint to why this was done this way
[03:37:27 CEST] <atomnuker> wait, couldn't you just perform 2 shifts of half the type's width and it would be perfectly defined?
[03:38:29 CEST] <atomnuker> I really can't think of a reason why this was done this way
[03:45:53 CEST] <Gramner> because many cpus doesn't support that
[03:46:56 CEST] <Gramner> e.g. x86 only uses the lowest 5 bits for a 32-bit int so trying to shift by 32 is equal to shifting by 0
[03:48:50 CEST] <atomnuker> Gramner: does amd64 only use the bottom 6 bits too?
[03:49:00 CEST] <Gramner> yes
[03:49:23 CEST] <Gramner> this is actually useful for bit twiddling trickery
[03:49:55 CEST] <atomnuker> how so/
[03:49:58 CEST] <atomnuker> *?
[03:51:11 CEST] <Gramner> https://git.videolan.org/?p=x264.git;a=blob;f=common/x86/mc-a2.asm;hb=ba24899b0bf23345921da022f7a51e0c57dbe73d#l1570 for example
[03:53:24 CEST] <Gramner> being able to shift by the integer width would certainly be useful in many other cases. but since it's commonly unsupported in hardware it's also unsupported in C
[03:54:45 CEST] <Gramner> it it was supported in C every single variable shift would be significantly slower on such hardware since it would have to explicitly check for that edge case every time
[03:55:55 CEST] <Gramner> so the solution is to let the programmer check for the edge case when it's needed
[03:56:01 CEST] <atomnuker> I see, this makes sense now
[03:57:09 CEST] <kierank> atomnuker: insane thunder
[03:58:16 CEST] <rcombs> yeah, that's where most UB and implementation-defined behavior comes from
[04:55:01 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:1de61c6b015f: avcodec/fic: Fix multiple runtime error: signed integer overflow: 5793 * 419752 cannot be represented in type 'int'
[04:55:02 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:053ca0356448: avcodec/mimic: Use ff_set_dimensions() to set the dimensions
[04:55:03 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:9532ad523ec5: avcodec/aacsbr_fixed: Fix multiple runtime error: shift exponent 150 is too large for 32-bit type 'int'
[04:55:04 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:f6fbc874d1e3: avcodec/mlpdec: Do not leave a invalid num_primitive_matrices in the context
[04:55:05 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:6bd5fcc463c5: avcodec/aacsbr_fixed: Fix multiple runtime error: shift exponent 170 is too large for 32-bit type 'int'
[04:55:06 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:9b7c194ba1d7: avcodec/sbrdsp_fixed: fix runtime error: left shift of 1 by 31 places cannot be represented in type 'int'
[04:55:07 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:fcc87a4676bc: avcodec/mlpdsp: Fix runtime error: signed integer overflow: -24419392 * 128 cannot be represented in type 'int'
[04:55:08 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:02b8d62a8c03: avcodec/takdec: Fix runtime error: left shift of negative value -63
[04:55:09 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:4cad001b0494: avcodec/aac_defines: Fix: runtime error: left shift of negative value -2
[04:55:10 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:d7b53e87b9e5: avcodec/takdec: Fix runtime error: signed integer overflow: 8192 * 524308 cannot be represented in type 'int'
[04:55:11 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:c06677aab1eb: avcodec/vmnc: Check location before use
[04:55:12 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:0e003452c040: avcodec/mpeg4videodec: Check for multiple VOL headers
[04:55:13 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:96f0b1de4959: avcodec/aacdec_fixed: Fix runtime error: shift exponent 34 is too large for 32-bit type 'int'
[04:55:14 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:9a993c5bc89e: avcodec/mjpegdec: Fix runtime error: signed integer overflow: -32767 * 130560 cannot be represented in type 'int'
[04:55:15 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:0169072409cf: avcodec/ivi_dsp: Fix multiple runtime error: left shift of negative value -71
[04:55:16 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:4e8405b347bf: avcodec/jpeglsdec: Check get_bits_left() before decoding a picture
[04:55:17 CEST] <cone-839> ffmpeg 03Max Justicz 07release/3.0:5753c06dd79f: avcodec/sanm: Fix uninitialized reference frames
[04:55:18 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:c071e74aad6a: avcodec/jpeg2000dec: Check tile offsets
[04:55:19 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:16f76d544db9: avcodec/jpeg2000dec: Fix copy and paste error
[04:55:21 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:6185b5e9bcbf: avcodec/smc: Check remaining input
[04:55:21 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:7fe0a0e9e73a: avcodec/aacdec_fixed: Fix runtime error: signed integer overflow: -2147483648 * -1 cannot be represented in type 'int'
[04:55:22 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:278bd748a8b6: avutil/internal: Do not enable CHECKED with DEBUG
[04:55:23 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:780c360e760b: avformat/mux: Fix copy an paste typo
[04:55:24 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:385642f8c86c: avcodec/ra144dec: Fix runtime error: left shift of negative value -17
[04:55:25 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:a3ba6a66a3ee: avcodec/mlpdec: Do not leave invalid values in matrix_out_ch[] on error
[04:55:26 CEST] <cone-839> ffmpeg 03Kevin Mark 07release/3.0:3ad48f9acb19: doc/filters: Clarify scale2ref example
[04:55:27 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:1a800e8ff670: avcodec/ivi_dsp: Fix runtime error: left shift of negative value -2
[04:55:28 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:ca3fab8c807d: avcodec/sbrdsp_template: Fix: runtime error: signed integer overflow: 849815297 + 1315389781 cannot be represented in type 'int'
[04:55:29 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:25442bd95fdf: avcodec/libfdk-aacdec: Correct buffer_size parameter
[04:55:30 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:bd4dcd873166: avcodec/wnv1: More strict buffer size check
[04:55:31 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:aaf104e525b4: avcodec/aacdec_fixed: Fix multiple runtime error: shift exponent 127 is too large for 32-bit type 'int'
[04:55:32 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:5c1c1e89de4e: avcodec/jpeg2000dec: Check tile offsets more completely
[04:55:33 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:9495075a7f86: avcodec/jpeg2000: Fix runtime error: signed integer overflow: 4185 + 2147483394 cannot be represented in type 'int'
[04:55:34 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:21cb305903c9: avcodec/snow: Fix runtime error: signed integer overflow: 1086573993 + 1086573994 cannot be represented in type 'int'
[04:55:35 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:a8625d11105f: avcodec/aacdec_fixed: Fix runtime error: left shift of 1 by 31 places cannot be represented in type 'int'
[04:55:36 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:62979de16865: avcodec/webp: Fixes null pointer dereference
[04:55:37 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:7f0c66b8cf0a: avcodec/aac_defines: Add missing () to AAC_HALF_SUM() macro
[04:55:38 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:7e7c25a48e59: avcodec/ra144: Fix runtime error: signed integer overflow: 11184810 * 404 cannot be represented in type 'int'
[04:55:39 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:3b3a3b7a5a7f: avcodec/ra144: Fix runtime error: signed integer overflow: -2449 * 1398101 cannot be represented in type 'int'
[04:55:40 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:24ab34a3fcf3: avcodec/truemotion2: Fix runtime error: left shift of 1 by 31 places cannot be represented in type 'int'
[04:55:41 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:27fecc7a73db: avcodec/truemotion2: Fix passing null pointer to memset()
[04:55:42 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:759739e9964f: avcodec/jpeg2000dec: Use ff_set_dimensions()
[04:55:43 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:c92acd5ed48d: avcodec/ansi: Fix frame memleak
[04:55:44 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:d542185e0541: avcodec/wavpack: Fix runtime error: signed integer overflow: 24 * -2147483648 cannot be represented in type 'int'
[04:55:45 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:343c1d1d7d1f: avcodec/wavpack: Check float_shift
[04:55:46 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:c82db1441159: avcodec/acelp_pitch_delay: Fix runtime error: value 4.83233e+39 is outside the range of representable values of type 'float'
[04:55:47 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:794bcc6eca21: avformat/avidec: Limit formats in gab2 to srt and ass/ssa
[04:55:48 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:f5c5bec448ca: avcodec/cavsdec: Fix runtime error: signed integer overflow: 59 + 2147483600 cannot be represented in type 'int'
[04:55:49 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:309ec23130a1: avcodec/pnm: Use ff_set_dimensions()
[04:55:50 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:9589a7f39175: avcodec/ra144: Fixes runtime error: signed integer overflow: 7160 * 327138 cannot be represented in type 'int'
[04:55:51 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:a249250b3a88: avcodec/hevc_ps: Fix runtime error: signed integer overflow: 2147483628 + 256 cannot be represented in type 'int'
[04:55:52 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:9d8c59ce507a: avcodec/cinepak: Check input packet size before frame reallocation
[04:55:53 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:ecc3fe3859f3: avcodec/wavpack: Fix runtime error: signed integer overflow: 2013265955 - -134217694 cannot be represented in type 'int'
[04:55:54 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:2c3bcf6bf238: avcodec/wavpack: Fix runtime error: shift exponent 32 is too large for 32-bit type 'int'
[04:55:56 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:3764e3a2c528: avcodec/aacps: Fix runtime error: left shift of 1073741824 by 1 places cannot be represented in type 'INTFLOAT' (aka 'int')
[04:55:56 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:641b38c9c8a2: avformat/options: log filename on open
[04:55:57 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:06f69315293b: avcodec/ac3dec_fixed: Fix runtime error: left shift of 419 by 23 places cannot be represented in type 'int'
[04:55:58 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:ce4196561a8f: avcodec/pafvideo: Check packet size and frame code before ff_reget_buffer()
[04:55:59 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:e0d1be353979: avcodec/dxv: Check remaining bytes in dxv_decompress_raw()
[04:56:00 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:af85f35d2ce8: avcodec/hevc_ps: Fix runtime error: index 32 out of bounds for type 'uint8_t [32]'
[04:56:01 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:1065957f7b05: avutil/softfloat: Fix sign error in and improve documentation of av_int2sf()
[04:56:03 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:c4a3020c1ba3: avcodec/qdrw: Fix null pointer dereference
[04:56:03 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:bb1014279d86: avformat/hls: Check local file extensions
[04:56:04 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:49b744631768: avcodec/cavs: Fix runtime error: signed integer overflow: -12648062 * 256 cannot be represented in type 'int'
[04:56:05 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:d37559993fe5: avcodec/tiff: Avoid loosing allocated geotag values
[04:56:06 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:b6b70060ba5b: avcodec/mjpegdec: Check that reference frame matches the current frame
[04:56:07 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:0ed96055f424: avcodec/takdec: Fix multiple runtime error: signed integer overflow: 637072 * 4096 cannot be represented in type 'int'
[04:56:09 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:a3b9a5c4c42c: avcodec/pafvideo: Fix assertion failure
[04:56:09 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:3ca6d2645880: avcodec/mpeg4videodec: Fix runtime error: signed integer overflow: 53098 * 40448 cannot be represented in type 'int'
[04:56:10 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:968036dbf97f: avcodec/ac3dec_fixed: Fix multiple runtime error: signed integer overflow: -39271008 * 59 cannot be represented in type 'int'
[04:56:11 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:2dcae36de1b8: avcodec/indeo4: Check remaining data in Pic hdr extension parsing code
[04:56:12 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:72b8c6c645da: avcodec/cfhd: Check band parameters before storing them
[04:56:13 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:3784d34c3bdf: avcodec/flicvideo: Fix runtime error: signed integer overflow: 4864 * 459296 cannot be represented in type 'int'
[04:56:14 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:e7f8984f990b: avcodec/ra144: Fix runtime error: signed integer overflow: -2200 * 1033073 cannot be represented in type 'int'
[04:56:15 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:91da2f8ebb82: avcodec/tiff: Fix leak of geotags[].val
[04:56:17 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:f3be926246b6: avcodec/aacdec_fixed: Fix runtime error: left shift of negative value -1297616
[04:56:17 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:e8c87c21c0cc: avcodec/snowdec: Fix runtime error: left shift of negative value -1
[04:56:18 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:fef11b7c38bd: avcodec/wavpack: Fix runtime error: signed integer overflow: 1886191616 + 277872640 cannot be represented in type 'int'
[04:56:19 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:0f59d5a6e0a8: avcodec/jpeg2000dwt: Fix runtime error: left shift of negative value -123
[04:56:20 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:3db46e6447f1: avcodec/sbrdsp_fixed: Return an error from sbr_hf_apply_noise() if operations are impossible
[04:56:21 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:5a35ac54358d: avcodec/aacsbr_fixed: Check shift in sbr_hf_assemble()
[04:56:22 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:301791a2bd0f: avcodec/mpeg4videodec: Fix integer overflow in num_sprite_warping_points=2 case
[04:56:23 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:ea5603ff3a55: avcodec/mpeg4videodec: Check sprite delta upshift against overflowing.
[04:56:24 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:7c82082ac8d4: avcodec/hevc_refs: Check nb_refs in add_candidate_ref()
[04:56:25 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:f19152559f5f: avcodec/hevcdec: Check nb_sps
[04:56:27 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:23d02b446557: avcodec/jpeg2000: Fixes integer overflow in ff_jpeg2000_ceildivpow2()
[04:56:27 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:a5feb7e9bd47: avcodec/truemotion2: Move skip computation after checks
[04:56:28 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:79393f4fb140: avcodec/shorten: Sanity check maxnlpc
[04:56:29 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:e8555274b93a: avcodec/jpeg2000dec: Check nonzerobits more completely
[04:56:30 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:59d40e2d6802: avcodec/hevcdec: Fix signed integer overflow in decode_lt_rps()
[04:56:31 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:fce334274bf5: avcodec/hevcpred_template: Fix left shift of negative value
[04:56:32 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:e46f8c96d0c2: avcodec/jpeg2000dsp: Reorder operations in ict_int() to avoid 2 integer overflows
[04:56:34 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:d5abcaacb396: avcodec/takdec: Fixes: integer overflow in AV_SAMPLE_FMT_U8P output
[04:56:34 CEST] <cone-839> ffmpeg 03Anton Mitrofanov 07release/3.0:48ab5a4dc003: avcodec/h264_cabac: Fix CABAC+8x8dct in 4:4:4
[04:56:35 CEST] <cone-839> ffmpeg 03Anton Mitrofanov 07release/3.0:273e87be3b24: avcodec/h264_mb: Fix 8x8dct in lossless for new versions of x264
[04:56:36 CEST] <cone-839> ffmpeg 03Anton Mitrofanov 07release/3.0:85e0f61b41fc: avcodec/h264: Fix mix of lossless and lossy MBs decoding
[04:56:38 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:9b346f12fa44: avcodec/htmlsubtitles: Replace very slow redundant sscanf() calls by cleaner and faster code
[04:56:39 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:784d57bb62de: avcodec/aacdec_fixed: Check s for being too small
[04:56:40 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:628e2b2dd34a: avcodec/wavpack: Fix undefined integer negation
[04:56:41 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:fae49f28e657: avcodec/lpc: signed integer overflow in compute_lpc_coefs() (aacdec_fixed)
[04:56:42 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:6dca67f3a648: avcodec/mpeg4videodec: Fix overflow in virtual_ref computation
[04:56:43 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:eb7ad560754d: avcodec/hevc_filter: Fix invalid shift
[04:56:44 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:605bc45295e9: avcodec/cfhd: Fix undefined shift
[04:56:45 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:df11b45644e0: avcodec/cfhd: Check bpc before setting bpc in context
[04:56:46 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:0879ecd4a357: avcodec/tiff: Update pointer only when the result is used
[04:56:47 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:addf70e4928a: avcodec/takdec: Fix integer overflow
[04:56:48 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:bb02e5bc0918: avcodec/wavpack: Fix integer overflow
[04:56:49 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:8b9e522e08d9: avcodec/mpeg4videodec: Fix GMC with videos of dimension 1
[04:56:50 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:0fe174fa9c80: avcodec/wavpack: Fix integer overflow in wv_unpack_stereo()
[04:56:51 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:654ff561032a: Update for 3.0.9
[04:56:52 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:3cae97b090e1: avcodec/apedec: Fix integer overflow
[04:56:53 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:34535941ae94: avcodec/jpeg2000dwt: Fix integer overflow in dwt_decode97_int()
[04:56:54 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:cc9082dce19c: avcodec/vb: Check vertical GMC component before multiply
[04:56:55 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:7807d2478bfd: avcodec/cfhd: Fix invalid left shift of negative value
[04:56:56 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:665885bfc189: avcodec/hevc_ps: Fix integer overflow with beta/tc offsets
[04:56:57 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:2170bdb56d68: avcodec/wavpack: Fix invalid shift
[04:56:58 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:0ddff40bf888: avcodec/sbrdsp_fixed: Fix integer overflow in sbr_hf_apply_noise()
[04:56:59 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:f1143f5dc0e2: avcodec/aacps (fixed point): Fix multiple signed integer overflows
[04:57:00 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:9a2ca3cce2cf: avcodec/mjpegdec: Clip DC also on the negative side.
[04:57:01 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:5e78e477fa69: avcodec/aacdec_template: Fix undefined integer overflow in apply_tns()
[04:57:02 CEST] <cone-839> ffmpeg 03Michael Niedermayer 07release/3.0:4e7ddba5949a: avcodec/aacdec_template (fixed point): Check gain in decode_cce() to avoid undefined shifts later
[04:57:42 CEST] <jamrial> ?
[04:58:04 CEST] <jamrial> none of these commits are in the repo
[05:01:21 CEST] <jamrial> michaelni: the above push doesn't show up in http://git.ffmpeg.org but does in https://git.videolan.org
[05:01:24 CEST] <jamrial> any idea why?
[05:02:40 CEST] <jamrial> oh, there it is
[05:02:43 CEST] <jamrial> that was pretty slow
[05:15:03 CEST] <michaelni> yes, it can be a few minutes behind
[07:12:20 CEST] <thebombzen> Is there a technical reason NUT doesn't support mkv-like ASS subtitles with attachments, or has it just not been added to the spec?
[07:12:55 CEST] <thebombzen> For example, with Opus and HEVC, it just needed to be added to the spec so it's on the list of supported codecs, and once that happened, it was very easy to allow HEVC and Opus inside NUT
[07:13:57 CEST] <thebombzen> Because if there *isn't* a technical reason, I'd be happy to submit the patch to add ASS+TTF-Attachment support to nut by simply adding it to the spec's list of codecs, and then adding it to the code after that
[08:28:19 CEST] <ldts> jkqxz:  typically you would use the v4L2 API to export the mmaped buffer provided by V4L2 as dmabuf (VIDIOC_EXPBUF); when the buffer is dequed -ie after decode has completed- you can use this dmabuf instead of pass it to drm for display for instance.
[10:04:16 CEST] <durandal_1707> i gonna apply dolbye decoder,  anybody against?
[10:06:06 CEST] <j-b> no.
[10:10:56 CEST] <JEEB> if nobody objected on the ml so far... :D
[10:31:21 CEST] <jkqxz> ldts:  What I was suggesting was, suppose you /only/ pass DRM objects around inside libavcodec.  That would seem to remove the need for the extra external API to propagate buffer virtual pointers between components, and also make it easier to consume the output in other contexts (scanout, EGL import, etc.).
[10:31:25 CEST] <cone-962> ffmpeg 03Jan Sebechlebsky 07master:5c9dcd5c484f: libavformat/fifo: Fix initialization of underlying AVFormatContext
[10:31:49 CEST] <jkqxz> Hence the question as to whether that causes problems anywhere else.
[10:50:04 CEST] <ldts> jkqxz: hi, are those DRM objects dmabuf file descriptors? if so, some v4l2 drivers do support importing externally generated dmabuffs into v4l2 buffers; but not all v4l2 drivers support V4L2_MEMORY_DMABUF..
[11:10:54 CEST] <cone-962> ffmpeg 03Muhammad Faiz 07master:a3d3cfa65c01: avutil: merge slice threading implementation from avcodec and avfilter
[11:10:55 CEST] <cone-962> ffmpeg 03Muhammad Faiz 07master:b505f15b1530: avcodec/pthread_slice: use slice threading from avutil
[11:10:56 CEST] <cone-962> ffmpeg 03Muhammad Faiz 07master:22b72de04bfd: avfilter/pthread: use slice threading from avutil
[12:48:15 CEST] <cone-962> ffmpeg 03foo86 07master:930fe4b1f75d: avcodec: add Dolby E decoder
[12:48:16 CEST] <cone-962> ffmpeg 03foo86 07master:f04ef268164f: avformat: add SMPTE 337M demuxer
[12:50:27 CEST] <durandal_1707> paras2052: can you write fits parser?
[14:00:50 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.0:0df61711cfb9: Changelog: update
[14:46:58 CEST] <BtbN> philipl, can you have a look at the 3 nvidia patches on the ml? I'm on vacation for the next 2 weeks. They are fine with me if there is no issue with P016 I forgot about.
[15:00:57 CEST] <nevcairiel> wasnt there some reason we didnt support 444p16
[15:22:31 CEST] <cone-962> ffmpeg 03foo86 07n3.0.9:HEAD: avformat: add SMPTE 337M demuxer
[15:26:29 CEST] <BBB> n3.0.9:HEAD?
[15:26:33 CEST] <BBB> what is that
[15:26:46 CEST] <kierank> yeah
[15:26:47 CEST] <nevcairiel> the bot just gets confused when someone tags
[15:27:01 CEST] <nevcairiel> it shows the latest master commit for some reason
[15:27:19 CEST] <BBB> aha, yes indeed
[15:27:34 CEST] <BBB> (I checked the git manually to see what that was about, its just michaelni tagging a new release)
[15:27:41 CEST] <BBB> confusing
[15:29:41 CEST] <BBB> ubitux: do you have experience with fixing [failed to restore the stack] in tsan?
[15:30:00 CEST] <BBB> ubitux: I see some old chromium posts about workarounds but they dont appear to work anymore :-/
[15:31:01 CEST] <ubitux> no idea what this is about
[16:50:21 CEST] <jamrial> BBB: if fixing vf_ssim asm requires you to make changes that would also need to be done to the XOP version of ssim_4x4_line, poke me
[16:50:32 CEST] <jamrial> or just disable it and i'll take a look later
[16:50:52 CEST] <BBB> does lavfi allow us to require frame padding?
[16:57:33 CEST] <durandal_1707> BBB: same as lavc
[16:57:53 CEST] <durandal_1707> linesize tells the padding
[16:59:35 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:50c2ef91d341: avcodec/ac3dec_fixed: Fix multiple runtime error: signed integer overflow: -39271008 * 59 cannot be represented in type 'int'
[16:59:36 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:ad9ce1fa1d96: avcodec/indeo4: Check remaining data in Pic hdr extension parsing code
[16:59:37 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:741c34196883: avcodec/h264_parse: Check picture structure when initializig weight table
[16:59:38 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:460abcd67188: avcodec/cfhd: Check band parameters before storing them
[16:59:39 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:311f2f5abaa6: avcodec/flicvideo: Fix runtime error: signed integer overflow: 4864 * 459296 cannot be represented in type 'int'
[16:59:40 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:5217145824bf: avcodec/ra144: Fix runtime error: signed integer overflow: -2200 * 1033073 cannot be represented in type 'int'
[16:59:41 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:f3ac7e40d6bf: avcodec/tiff: Fix leak of geotags[].val
[16:59:42 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:53a32fdf0a5f: avcodec/aacdec_fixed: Fix runtime error: left shift of negative value -1297616
[16:59:43 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:e1b6d78bf706: avcodec/snowdec: Fix runtime error: left shift of negative value -1
[16:59:44 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:34282abc57f4: avcodec/wavpack: Fix runtime error: signed integer overflow: 1886191616 + 277872640 cannot be represented in type 'int'
[16:59:45 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:cce9471373cf: avcodec/jpeg2000dwt: Fix runtime error: left shift of negative value -123
[16:59:46 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:452c78a09cb5: avcodec/libvpxdec: Check that display dimensions fit in the storage dimensions
[16:59:47 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:d2567caea9b1: avcodec/sbrdsp_fixed: Return an error from sbr_hf_apply_noise() if operations are impossible
[16:59:48 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:3c98e4be896e: avcodec/aacsbr_fixed: Check shift in sbr_hf_assemble()
[16:59:49 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:7a86581afd68: avcodec/mpeg4videodec: Fix integer overflow in num_sprite_warping_points=2 case
[16:59:50 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:f33c3ccbe747: avcodec/mpeg4videodec: Check sprite delta upshift against overflowing.
[16:59:51 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:f8a331598efe: avcodec/hevc_refs: Check nb_refs in add_candidate_ref()
[16:59:52 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:10d821309b02: avcodec/hevcdec: Check nb_sps
[16:59:53 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:0a709e2a10b8: avcodec/dnxhd_parser: Do not return invalid value from dnxhd_find_frame_end() on error
[16:59:54 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:15cc15170909: avcodec/jpeg2000: Fixes integer overflow in ff_jpeg2000_ceildivpow2()
[16:59:55 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:1729101c44c0: avcodec/truemotion2: Move skip computation after checks
[16:59:56 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:01ed8d93b2b9: avcodec/shorten: Sanity check maxnlpc
[16:59:57 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:ee2396cefd7e: avcodec/jpeg2000dec: Check nonzerobits more completely
[16:59:58 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:f17443cdcdfa: avcodec/hevcdec: Fix signed integer overflow in decode_lt_rps()
[16:59:59 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:7119574f4831: avcodec/hevcpred_template: Fix left shift of negative value
[17:00:00 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:347cc89daf90: avcodec/jpeg2000dsp: Reorder operations in ict_int() to avoid 2 integer overflows
[17:00:01 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:4384481fbcb6: avcodec/takdec: Fixes: integer overflow in AV_SAMPLE_FMT_U8P output
[17:00:02 CEST] <cone-962> ffmpeg 03Anton Mitrofanov 07release/3.3:b8d0d7674066: avcodec/h264_cabac: Fix CABAC+8x8dct in 4:4:4
[17:00:03 CEST] <cone-962> ffmpeg 03Anton Mitrofanov 07release/3.3:e8558abeaf70: avcodec/h264_mb: Fix 8x8dct in lossless for new versions of x264
[17:00:04 CEST] <cone-962> ffmpeg 03Anton Mitrofanov 07release/3.3:b2522f35ecd1: avcodec/h264: Fix mix of lossless and lossy MBs decoding
[17:00:05 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:93456ca3eacc: avcodec/htmlsubtitles: Replace very slow redundant sscanf() calls by cleaner and faster code
[17:00:06 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:cf61bf81073e: avcodec/aacdec_fixed: Check s for being too small
[17:00:07 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:eaf2bacca1c4: avcodec/wavpack: Fix undefined integer negation
[17:00:08 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:72403ba2b9d6: avcodec/lpc: signed integer overflow in compute_lpc_coefs() (aacdec_fixed)
[17:00:09 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:ada21bca5551: avcodec/mpeg4videodec: Fix overflow in virtual_ref computation
[17:00:10 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:5b8a97d000ef: avcodec/hevc_filter: Fix invalid shift
[17:00:11 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:fe8960ab8623: avcodec/cfhd: Fix undefined shift
[17:00:12 CEST] <Shiz> im 12 and what is this
[17:00:12 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:0749384f0a7d: avcodec/cfhd: Check bpc before setting bpc in context
[17:00:13 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:f1116294aa3e: avcodec/tiff: Update pointer only when the result is used
[17:00:14 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:3ed986522ad5: avcodec/takdec: Fix integer overflow
[17:00:15 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:94f3c06678c7: avcodec/wavpack: Fix integer overflow
[17:00:16 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:f21e96109dc9: avcodec/mpeg4videodec: Fix GMC with videos of dimension 1
[17:00:17 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:ea153eb52cba: avcodec/hevc_ps: Fix max_dec_buffer check
[17:00:18 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:165b2ee6920a: avcodec/wavpack: Fix integer overflow in wv_unpack_stereo()
[17:00:19 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:0272afe70d62: avcodec/apedec: Fix integer overflow
[17:00:20 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:736c73a24345: avcodec/jpeg2000dwt: Fix integer overflow in dwt_decode97_int()
[17:00:21 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:f3cb2eedebbc: avcodec/hevcdec: do basic validity check on delta_chroma_weight and offset
[17:00:22 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:132037ad5b6e: avcodec/vb: Check vertical GMC component before multiply
[17:00:23 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:94c8e53034bc: avcodec/cfhd: Fix invalid left shift of negative value
[17:00:24 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:1a21edf7b885: avcodec/hevc_ps: Fix integer overflow with beta/tc offsets
[17:00:25 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:0a36341e967d: avcodec/h264_slice: Fix signed integer overflow
[17:00:26 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:fb5562036984: avcodec/wavpack: Fix invalid shift
[17:00:27 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:ba3a4a94bced: avcodec/hevcdec: do not let updated extradata corrupt state
[17:00:28 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:8d3ac812ffdd: avcodec/sbrdsp_fixed: Fix integer overflow in sbr_hf_apply_noise()
[17:00:29 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:20d4514f2578: avcodec/ylc: Fix vlc of 31 bits
[17:00:30 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:1667b3ea0fae: avcodec/aacps (fixed point): Fix multiple signed integer overflows
[17:00:31 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:dbb121688cec: avcodec/mjpegdec: Clip DC also on the negative side.
[17:00:32 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:6ae1b70cb42b: avcodec/magicyuv: Check that vlc len is not too large
[17:00:33 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:9ce4350c48d5: avcodec/aacdec_template: Fix undefined integer overflow in apply_tns()
[17:00:34 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:064d0c646282: avcodec/ffv1dec_template: Fix signed integer overflow
[17:00:35 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:2f2904030f22: avcodec/aacdec_template (fixed point): Check gain in decode_cce() to avoid undefined shifts later
[17:00:36 CEST] <cone-962> ffmpeg 03Michael Niedermayer 07release/3.3:ab81ea10350c: Update for 3.3.3
[18:00:12 CEST] <J_Darnley> Does anyone have tips about how to track down a problem on a piece of multithreaded code?
[18:01:04 CEST] <J_Darnley> Output is good with -threads 1 but not with more.
[18:01:28 CEST] <nevcairiel> you can try with thread-sanitizer, but its not very fun
[18:01:31 CEST] <J_Darnley> With more, some dark artifacts flicker about the picture.
[18:01:37 CEST] <nevcairiel> ask BBB about his experiences
[18:01:41 CEST] <BBB> ?
[18:01:52 CEST] <BBB> oh
[18:01:57 CEST] Last message repeated 1 time(s).
[18:01:57 CEST] <BBB> oooh
[18:02:00 CEST] <BBB> oooooooooh
[18:02:13 CEST] <J_Darnley> We've given him War Flashbacks
[18:02:18 CEST] <BBB> yeah
[18:02:32 CEST] <nevcairiel> feels more like he is happy to torture someone else with this
[18:02:38 CEST] <BBB> I was gonna ask if you have frog tears or garlic to bewitch the vampires
[18:02:51 CEST] <atomnuker> mutex everything till you make it single core and start removing them one at a time and looking with tsan
[18:02:56 CEST] <atomnuker> *thread
[18:03:56 CEST] <BBB> J_Darnley: do you have liberty to be more detailed? or is it private code?
[18:04:09 CEST] <J_Darnley> I can.  It is ffmpeg
[18:04:18 CEST] <J_Darnley> vc2enc though
[18:04:33 CEST] <BBB> oh this is your current wip right?
[18:04:34 CEST] <nevcairiel> the work on that is going to drive him insane
[18:05:11 CEST] <BBB> J_Darnley: have you ever heard the story of the man that went through the twilight zone portal? he came back, but his life was never the same again
[18:05:26 CEST] <BBB> J_Darnley: its kind of like the red pill vs blue pill question. are you ready to take the pill?
[18:05:35 CEST] <J_Darnley> Oh, I have given up on a particular aspect of the problem.  No other software supports that corner case so I will just make it a run-time error
[18:05:49 CEST] <BBB> anyway
[18:06:14 CEST] <BBB> Im not totally familiar with vc2 so its hard to say in practice whats going on :(
[18:06:19 CEST] <BBB> I assume its slice threading?
[18:06:23 CEST] <J_Darnley> Yes.
[18:07:16 CEST] <J_Darnley> Should I try thread-sanitizer first?
[18:07:20 CEST] <BBB> maybe
[18:07:21 CEST] <BBB> yes
[18:07:27 CEST] <BBB> is it reproducible with 1 single frame?
[18:07:33 CEST] <J_Darnley> Yes
[18:07:37 CEST] <BBB> cool
[18:07:41 CEST] <BBB> ok, so start debugging that with tsan
[18:07:44 CEST] <BBB> should be fairly easy
[18:15:40 CEST] <iive> J_Darnley: is the output generated by theaded encoder supposed to be binary identical to the one generated by single thread one?
[18:16:20 CEST] <J_Darnley> Good question.  I think so.
[18:17:33 CEST] <nevcairiel> that usually depends on if you use prediction on slice-borders, which with slice threading some may skip to actually be able to parallize
[18:17:53 CEST] <J_Darnley> I don't think vc2 has any prediction
[18:18:16 CEST] <nevcairiel> or motion vectors across slice borders, for that matter
[18:18:58 CEST] <atomnuker> J_Darnley: it has DC prediction in the core profile
[18:19:49 CEST] <atomnuker> it even has entropy coding rather than expgolomb, which makes it difficult to decode in real time at such high bitrates
[18:20:07 CEST] <J_Darnley> Dammit!  I get a great wad of relocation errors when linking.
[18:22:31 CEST] <nevcairiel> two questions - did you use the toolchain option in configure to enable tsan? and, did you do a full clean build? tsan gets unhappy with partial re-builds
[18:22:53 CEST] <J_Darnley> yes and yes
[18:23:09 CEST] <nevcairiel> no help then :(
[18:23:22 CEST] <J_Darnley> I'll use the compiler flags directly then
[18:41:15 CEST] <J_Darnley> nevcairiel: I apologize, I didn't do a clean build.  :)
[18:43:13 CEST] <J_Darnley> And now I have some messages from tsan to read
[18:43:17 CEST] <J_Darnley> Thanks all
[18:47:01 CEST] <J_Darnley> Hm.  Looks like the transform is writing somewhere it shouldn't
[19:06:08 CEST] <BBB> J_Darnley: you need ldflags+=-fsanitize=thread also
[19:06:19 CEST] <BBB> as for the threading issue, is dc prediction disabled across slice boundaries?
[19:06:31 CEST] <BBB> (and the data still initialized properly etc.)
[19:06:57 CEST] <atomnuker> (there is no DC prediction or any prediction in the encoder)
[19:22:20 CEST] <kierank> J_Darnley: must be the edges then
[19:22:26 CEST] <kierank> you can overwrite any of the edges
[19:26:41 CEST] <kierank> can't
[19:57:42 CEST] <cone-962> ffmpeg 03George Boyle 07master:6e156075b090: avfilter: Add blue and violet noise generation filters
[20:14:01 CEST] <cone-962> ffmpeg 03Paul B Mahol 07master:a17fcd8e0473: avfilter/af_astats: do not overwrite already set values when doing reset
[20:22:50 CEST] <cone-962> ffmpeg 03James Almer 07master:cb13f4483183: avcodec/dolby_e: use av_clip_uintp2()
[21:08:13 CEST] <cone-962> ffmpeg 03Paul B Mahol 07master:12791ec5b0d3: avfilter/af_astats: measure dynamic range
[22:37:36 CEST] <J_Darnley> I thought I had a brain wave
[22:38:13 CEST] <J_Darnley> My transform was using the same buffer for all 3 planes
[22:38:45 CEST] <J_Darnley> Changing that didn't fix it
[22:39:09 CEST] <J_Darnley> plus the original transform appears to be using the same buffer on all 3 plane transforms
[22:39:24 CEST] <J_Darnley> atomnuker: do you have a comment about that ^
[00:00:00 CEST] --- Thu Jul 20 2017



More information about the Ffmpeg-devel-irc mailing list