[Ffmpeg-devel-irc] ffmpeg-devel.log.20181028

[burek]

[00:25:55 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.0:82e796a4c9cc: avcodec/msrle: Check that the input is large enough to contain a end of picture code
[00:25:56 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.0:15296d64ca4f: avutil/integer: Fix integer overflow in av_mul_i()
[00:25:57 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.0:527e64d32c34: Changelog: Update
[02:18:28 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07n3.0.12:HEAD: avfilter/vf_pixdesctest: Use 32bit read/write
[02:57:17 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:7489a527f000: avutil/pixfmt: Document chroma plane size for odd resolutions
[02:57:18 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:5cbf4849e376: swresample/swresample: Fix input channel count in resample_first computation
[02:57:19 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:9da24737a3b2: avcodec/diracdec: Prevent integer overflow in intermediate in global_mv()
[02:57:20 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:330ed0772c6f: avcodec/dirac_dwt_template: Fix several integer overflows in horizontal_compose_daub97i()
[02:57:21 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:7068bcf58ad2: avcodec/diracdec: Change frame_number to 64bit as its a 32bit from the bitstream and we also have a -1 special case
[02:57:22 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:a594ce26ce38: avcodec/diracdec: Check slice numbers for overflows in relation to picture dimensions
[02:57:23 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:9abcade734cb: avcodec/diracdec: Check bytes count in else branch in decode_lowdelay() too
[02:57:24 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:7abc4445f911: avcodec/qtrle: Check remaining bytestream in qtrle_decode_XYbpp()
[02:57:25 CEST] <cone-511> ffmpeg 03Nikolas Bowe 07release/3.4:c90457a95ee8: lavc/svq3: Fix regression decoding some files.
[02:57:26 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:44e878d08674: avformat/flvenc: Check audio packet size
[02:57:27 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:b6098dd17fec: avcodec/aacpsdsp_template: Fix integer overflow in ps_stereo_interpolate_c()
[02:57:28 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:4df3a367df8e: avcodec/mpegaudio_parser: Initialize poutbuf*
[02:57:29 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:7f480bedd069: avcodec/shorten: Check verbatim length
[02:57:30 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:ec573bd2eb08: avcodec/shorten: Fix integer overflow in residual/LPC combination
[02:57:31 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:e3cc5e81ab00: avcodec/shorten: Fix signed 32bit overflow in shift in shorten_decode_frame()
[02:57:32 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:007da8396fed: avcodec/scpr: Check for min > max in decompress_p()
[02:57:33 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:17c034797791: avformat/rmdec: Fix EOF check in the stream loop in ivr_read_header()
[02:57:34 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:ff5196a98afb: avformat/mlvdec: read_string() received unsigned size, make the argument unsigned
[02:57:35 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:0ef49c0818f2: avformat/nsvdec: Do not parse multiple NSVf
[02:57:36 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:812f7fae356a: avcodec/snowdec: Fix integer overflow with motion vector residual
[02:57:37 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:88afcff2f553: avcodec/vb: Check for end of bytestream before reading blocktype
[02:57:38 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:b61b38766ee9: avcodec/hq_hqa: Check remaining input bits in hqa_decode_mb()
[02:57:39 CEST] <cone-511> ffmpeg 03Michael Bunk 07release/3.4:462edf5b9435: examples: Fix use of AV_CODEC_FLAG_GLOBAL_HEADER
[02:57:40 CEST] <cone-511> ffmpeg 03Dale Curtis 07release/3.4:eab5f6e419ec: avformat/mov: Error on too large stsd entry counts.
[02:57:41 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:073a65aefc2e: avcodec/indeo4: Check dimensions in decode_pic_hdr()
[02:57:42 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:ee8b4c16d792: avcodec/ra144: Fix undefined integer overflow in add_wav()
[02:57:43 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:ab5d93076238: avcodec/h264_refs: Document last if() in ff_h264_execute_ref_pic_marking()
[02:57:44 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:1a106752f37c: avcodec/dvdsubdec: Avoid branch in decode_run_8bit()
[02:57:45 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:98709a124482: avcodec/shorten: Fix bitstream end check in read_header()
[02:57:46 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:80af29f32e54: avcodec/zmbv: Update decomp_len in raw frames
[02:57:47 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:127ec77e8cc5: avcodec/zmbv: Check that the decompressed data size is correct
[02:57:48 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:f80da843b219: avcodec/mpeg4videodec: Fix undefined shift in get_amv()
[02:57:49 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:50aa132f4d34: avcodec/dvdsubdec: Sanity check len in decode_rle()
[02:57:50 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:848726afc642: avcodec/gdv: Replace divisions by shifts in rescale()
[02:57:51 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:0cbd4fb9955e: avcodec/unary: Improve get_unary() docs
[02:57:52 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:e9975d1b5118: avformat/utils: Fix integer overflow in discontinuity check
[02:57:53 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:d17d08035cd4: avformat/utils: Never store negative values in last_IP_duration
[02:57:54 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:a3ef90a73cd9: avcodec/ra144: Fix integer overflow in add_wav()
[02:57:55 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:ced37ef52c24: avcodec/h264_cavlc: Check mb_skip_run
[02:57:56 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:6763ff890e41: avcodec/mpeg4videodec: Fix typo in sprite delta check
[02:57:57 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:1bcc79db44ed: avcodec/jpeg2000dec: Fix off by 1 error in JPEG2000_PGOD_CPRL handling
[02:57:58 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:707ec3cfc00b: avcodec/msrle: Check that the input is large enough to contain a end of picture code
[02:57:59 CEST] <cone-511> ffmpeg 03Michael Niedermayer 07release/3.4:0e853b61e65c: avutil/integer: Fix integer overflow in av_mul_i()
[02:02:51 CET] <J_Darnley> Oh hey.  I can watch an hour more now that I've gone back in time
[08:19:37 CET] <cone-726> ffmpeg 03hwren 07master:9c190ad39481: lavc/libxavs2: unified naming style
[08:19:37 CET] <cone-726> ffmpeg 03hwren 07master:4c2326281106: lavc/libxavs2: fix intra period meaning conflict
[08:19:37 CET] <cone-726> ffmpeg 03hwren 07master:c00ed8d0e7bd: lavc/libxavs2: enable OpenGop
[08:19:37 CET] <cone-726> ffmpeg 03Jun Zhao 07master:6885fa749996: lavc/libdavs2: Remove yuv420p10le from fromat list.
[08:19:37 CET] <cone-726> ffmpeg 03Jun Zhao 07master:bdfd2e3c79dd: lavc/libxavs2: Remove yuv420p10le from fromat list.
[12:46:10 CET] <BtbN> philipl, I'll look at the series tomorrow, or at least this week if I don't find time tomorrow.
[14:32:58 CET] <cone-726> ffmpeg 03Paul B Mahol 07master:bb54c0ae719e: avfilter/af_afftdn: switch to activate
[15:02:42 CET] <cone-726> ffmpeg 03Paul B Mahol 07master:0c8b5cb369fc: avfilter/af_afftdn: add alias for sample_noise end
[15:15:20 CET] <philipl> BtbN: much appreciated.
[17:42:57 CET] <cone-726> ffmpeg 03Mark Thompson 07master:7070955d43dd: libaomenc: Add support for tiles
[18:02:05 CET] <durandal_1707> what to not do next ... ?
[18:03:48 CET] <atomnuker> an ldac decoder?
[18:05:57 CET] <durandal_1707> no more sony
[18:51:54 CET] <durandal_1707> libavcodec/libaomenc.c:545:9: warning: unused variable 'pict_type' [-Wunused-variable]
[19:00:10 CET] <jkqxz> durandal_1707:  Upgrade libaom.  Is it useful to fix it for the old version?
[19:00:56 CET] <durandal_1707> jkqxz: right, ignore it
[19:02:33 CET] <jamrial> jkqxz: no, not worth it. as soon as they tag a new release i'll make it the minimum required version anyway
[19:03:04 CET] <jamrial> libaom 1.0.0 is missing some essential API, a crapload of optimizations, and a lot of fixes
[21:55:40 CET] <durandal_1707> is there a way after idct to reduce all blocks[x] by 1?
[21:56:04 CET] <atomnuker> reduce?
[21:56:16 CET] <durandal_1707> minus 1
[21:56:22 CET] <atomnuker> as in pixels[] -= 1?
[21:56:26 CET] <durandal_1707> yes
[21:56:55 CET] <atomnuker> can you alter the DC before transform?
[21:57:01 CET] <jkqxz> Subtract N*N from the DC coefficient.
[22:04:51 CET] <durandal_1707> hmm, doesnt help, for some pixels i get fade to 0 each next frame - perhaps this decoder needs new idct?
[22:06:56 CET] <atomnuker> just write some SIMD, it'll be very quick
[22:09:35 CET] <durandal_1707> why would i write SIMD?
[22:14:52 CET] <atomnuker> well you can choose not to, the compiler will do it for you, in this case certainly
[22:20:40 CET] <durandal_1707> atomnuker: but why for certain quality/qscale/quant_matrix it would give good output and for others it would give wrong one?
[22:31:28 CET] <atomnuker> the -1 solution?
[22:32:21 CET] <atomnuker> I thought they had that to handle fades, but no, especially not in yuv
[22:32:30 CET] <durandal_1707> atomnuker: no, general case, the -1 solution makes some pixels stay as they should be, but others fade
[22:33:55 CET] <cone-118> ffmpeg 03Paul B Mahol 07master:e95987f6ca8e: avfilter/af_afftdn: fix memory leaks reported by coverity
[22:34:34 CET] <atomnuker> durandal_1707: is it signalled? is there something common for pixels which stay/fade?
[22:35:34 CET] <durandal_1707> nothing is signalled, except flag for different idct in reference dll
[22:36:24 CET] <durandal_1707> it probably does one line something different, it uses 255 constant - the other idct
[22:37:46 CET] <atomnuker> what's the blocksize?
[22:44:24 CET] <durandal_1707> 8x8
[22:46:01 CET] <durandal_1707> in both cases
[22:46:25 CET] <atomnuker> do the pixels which don't fade have the same transform?
[22:49:04 CET] <durandal_1707> everything fades with certain quantization, when its exactly 1
[22:49:21 CET] <durandal_1707> and it fade to 255
[22:50:01 CET] <durandal_1707> changing dc just fixes some pixels, but others fade to other side...
[22:50:31 CET] <atomnuker> oh, II see
[22:51:09 CET] <durandal_1707> for other, smaller quality if still fades somehow but much slower
[22:56:36 CET] <atomnuker> I suppose they're using scalar quantization and in some ways they maybe have some primitive temporal AQ
[23:02:31 CET] <jshanab> I am trying to learn the code as I want to consider replacing live555 in a large project but need to add some bits.  But I am having a bit of trouble figuring out how the pieces go together. For example in the libswscale directory there is a rgb2rgb.h and .c but then there is also an rgb2rgb.c rgb2rgb_template.c and a rgb_2_rgb.asm. I do see the _template.c is directly included. Is that how...
[23:02:32 CET] <jshanab> ...the code interfaces with the C code? 
[23:03:49 CET] <JEEB> usually SIMD is done so that the functions are defined as asm, and then there's function pointers that are set to the correctly optimized functions according to available instruction sets
[23:04:01 CET] <JEEB> that said, my condolences for first of all looking into swscale
[23:04:43 CET] <nevcairiel> the template file is used to create various versions of code for different  instruction sets
[23:05:09 CET] <JEEB> also isn't live555 just the rtsp parts?
[23:05:20 CET] <nevcairiel> it probably has all sorts of stuff now
[23:05:30 CET] <jshanab> I though it was easier than avcodec. I am trying to figure out the makefiles.
[23:06:57 CET] <jshanab> live555 is rtsp.rtp,rtcp and with my changes rtsp-over-https and rtp backchannel. It is server and client with a lot of muxers and demux but NO decoding or file. That is what we use ffmpeg for
[23:09:45 CET] <jshanab> So how do the c files of the same name and the assembly fit together.
[00:00:00 CET] --- Mon Oct 29 2018