[Ffmpeg-devel-irc] ffmpeg-devel.log.20150120

burek burek021 at gmail.com
Wed Jan 21 02:05:02 CET 2015


[02:05] <cone-663> ffmpeg.git 03Michael Niedermayer 07master:e069a4bf2082: avcodec/snowdec: set AVFrame.pict_type correctly
[02:05] <cone-663> ffmpeg.git 03Michael Niedermayer 07master:7873f9aa727b: avcodec/snowdec: AVMotionVector support
[02:08] <kierank> michaelni: can you revert fe439c20698fc4992b9373cb068efb438fe1a098
[02:46] <michaelni> kierank, why? have you implemented decoding of that sample so this wont be needed anymore ?
[02:46] <michaelni> or is there some other issue with that commit ?
[02:46] <kierank> it breaks normal samples
[02:46] <kierank> https://trac.ffmpeg.org/ticket/4274
[02:57] <k_sze[work]> How does frame-dropping work when dealing with DirecShow H.264 video sources?
[02:57] <k_sze[work]> *DirectShow
[03:11] <cone-663> ffmpeg.git 03Michael Niedermayer 07master:ff5b9a1cbe96: Revert "avcodec/h264: also show frames with missing fields when CODEC_FLAG2_SHOW_ALL is set"
[03:36] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:e9ddf726aa72: avformat/segment: Use av_freep() avoid leaving stale pointers in memory
[03:36] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:a6d59978a00e: avformat/cdxl: Fix integer overflow of image_size
[03:36] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:1c983ee2c10e: avformat/flvdec: do not inject dts=0 metadata packets which failed to be parsed into a new data stream
[03:36] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:ef33242c2ac5: avformat/flvdec: Increase string array size
[03:36] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:92a36a4e78e6: avfilter/vf_sab: fix filtering tiny images
[03:36] <cone-663> ffmpeg.git 03wm4 07release/2.2:0d481efb7b81: avcodec/dvdsubdec: fix out of bounds accesses
[03:36] <cone-663> ffmpeg.git 03Dale Curtis 07release/2.2:57710c3646d4: mov: Avoid overflow with mov_metadata_raw()
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:20a03d5c9323: avformat/mov: fix integer overflow in mov_read_udta_string()
[03:37] <cone-663> ffmpeg.git 03Dale Curtis 07release/2.2:7c270a5e3b60: mov: Fix negative size calculation in mov_read_default().
[03:37] <cone-663> ffmpeg.git 03Paul B Mahol 07release/2.2:bebe3d35f362: lavfi: check av_strdup() return value
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:8abe459ac6ce: avformat/matroskadec: Use av_freep() to avoid leaving stale pointers in memory
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:3b17c1e13e45: avformat/utils: Clear pointer in ff_alloc_extradata() to avoid leaving a stale pointer in memory
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:3e3193f03c7d: cmdutils: Use 64bit for file size/offset related variable in cmdutils_read_file()
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:81bdaacb6556: avformat/mov: Fix mixed declaration and statement warning
[03:37] <cone-663> ffmpeg.git 03James Almer 07release/2.2:6e2204b15292: configure: bump year
[03:37] <cone-663> ffmpeg.git 03wm4 07release/2.2:b76dc8b5b8f7: avcodec/dvdsubdec: error on bitmaps with size 0
[03:37] <cone-663> ffmpeg.git 03wm4 07release/2.2:a4a87a7efd8b: avcodec/dvdsubdec: fix accessing dangling pointers
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:0684cd5d8c92: ffmpeg: Clear error message array at init.
[03:37] <cone-663> ffmpeg.git 03wm4 07release/2.2:fc1fed62d9da: vp9: fix parser return values in error case
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:33a67961c8c6: avformat/movenc: workaround bug in "PathScale EKOPath(tm) Compiler Suite Version 4.0.12.1"
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:ecae610207e3: avcodec/hevc: Fix handling of skipped_bytes() reallocation failures
[03:37] <cone-663> ffmpeg.git 03wm4 07release/2.2:b1959b1719c8: qpeg: avoid pointless invalid memcpy()
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:855ae45c5a64: avdevice: Use av_format_get_control_message_cb()
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:e91df69cf204: avfilter/vf_framepack: Check and update frame_rate
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:8babbdc9b15f: avcodec/flac_parser: fix handling EOF if no headers are found
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:3e8b73e65bb0: avformat/rmdec: Check for overflow in ff_rm_read_mdpr_codecdata()
[03:37] <cone-663> ffmpeg.git 03Rob Sykes 07release/2.2:a40e6a214ec6: swresample/soxr_resample: fix error handling
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:f2fde86dae21: avformat/aviobuf: Check that avio_seek() target is non negative
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:30099bf0f014: avformat/utils: Do not update programs streams from program-less streams in update_wrap_reference()
[03:37] <cone-663> ffmpeg.git 03wm4 07release/2.2:b9510b3274cf: lavu/frame: fix malloc error path in av_frame_copy_props()
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:3b9cb8d7d819: configure: create the tests directory like the doc directory
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:de434423910e: avformat/hdsenc: Use av_freep() avoid leaving stale pointers in memory
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:f5fd937fc5df: avformat/flvdec: Use av_freep() avoid leaving stale pointers in memory
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:7279be7c75c3: avcodec/vmdvideo: Check len before using it in method 3
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:c351cd720a0c: avcodec/utvideodec: Fix handling of slice_height=0
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:a0f4f12b6273: avformat/mov: check atom nesting depth
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:b8021620e21a: avformat/mov: fix integer overflow of size
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:226727f08fb0: swscale: increase yuv2rgb table headroom
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:50e04b3f3ca9: avcodec/h264: make the first field of H264Context an AVClass
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:9ce4686bfef8: avcodec/indeo3: use signed variables to avoid underflow
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:133dc77da96f: avcodec/dcadec: Check that the added xch channel isnt already there
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:13838647ca63: avcodec/hevc: clear filter_slice_edges() on allocation
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:25dc978bb1b5: avcodec/h264: Clear delayed_pic on deallocation
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:85b2396265f0: avcodec/hevc_ps: Check diff_cu_qp_delta_depth
[03:37] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:f1d59a207fef: avcodec/h264: Check *log2_weight_denom
[03:38] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:2528468e204c: avcodec/indeo3: ensure offsets are non negative
[03:38] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:3769601fb6d8: Add FFMPEG_VERSION into the binary libs
[03:38] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:9789612689d7: Makefile: add dependencies which require ffversion.h
[03:45] <cone-663> ffmpeg.git 03Michael Niedermayer 07release/2.2:86a01362c0e4: Update for FFmpeg 2.2.12
[04:00] <cone-663> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n2.2.12': unknown revision or path not in the working tree.
[04:00] <cone-663> Use '--' to separate paths from revisions
[04:00] <cone-663> refs/tags/n2.2.12:HEAD: Revert "avcodec/h264: also show frames with missing fields when CODEC_FLAG2_SHOW_ALL is set"
[08:24] <ubitux> damn there are a lot of good things to take in these subtitles files
[08:24] <ubitux> maybe i should ask akira4 to work on these instead...
[08:34] <ubitux> thx
[08:34] <ubitux> oups
[11:51] <ubitux> anshul_: ping
[11:53] <ubitux> i might make akira4 (the subtitles opw student) work on the patchset from the Shan guy to extract the different features (i saw many sub text codecs improvements)
[11:54] <ubitux> and i might make her work on the cc thing
[11:54] <ubitux> please tell me if you have pending work or plans on this 
[12:12] <anshul_> ubitux: I am very busy these day, thanks to take load from me
[12:13] <anshul_> I have completed those roullup as you said in mail
[12:14] <anshul_> but I could not make rollup and reap_output apart, they were tightly coupled
[12:15] <anshul_> just not I am working ISDB closed caption, I will put that later in our closed caption
[12:17] <anshul_> I have also done, webVtt in hls if someone want to take it over, I would be happy to transfer it
[13:56] <kierank> anshul_: your patch doesn't solve 1778
[14:41] <anshul_> kierank: reopened it, thanks for reminding
[14:45] <anshul_> kierank: I looked at the specs of other codec, none of the codec data sheet specify anything about nut muxer, most of them do not specify  anything other then ts
[14:45] <anshul_> its the container people decide where they will put the codec data
[14:45] <kierank> sigh
[14:45] <kierank> you fail to understand
[14:45] <anshul_> In Mp4 same closed caption is there in different stream
[14:45] <kierank> i am vlc and i see captions in nut and in the elementary stream
[14:46] <kierank> not standardised afaik
[14:46] <kierank> apple proprietary
[14:47] <anshul_> ok I got your point, do you know of any opensource closed caption encoder
[14:47] <kierank> no
[15:07] <cone-305> ffmpeg.git 03Anshul Maheshwari 07master:012559f2bb59: Cosmetic changes in cc_caption
[15:07] <cone-305> ffmpeg.git 03Anshul Maheshwari 07master:88ba88e8877f: CHAR_DEBUG to av_dlog
[15:07] <cone-305> ffmpeg.git 03Anshul Maheshwari 07master:d6d354254f68: Initilaize array to known size
[15:07] <cone-305> ffmpeg.git 03Anshul Maheshwari 07master:1013c15f72f9: changed name of erase_display_memory to screen_changed
[16:29] <LConstantin> Hi. Does anyone know if these vulnerabilities were patched in FFmpeg? http://www.binarysniper.net/2015/01/vlc-media-player-215-memory-corruption.html
[16:30] <LConstantin> VLC devs said the vulnerabilities are in libavcodec
[16:30] <LConstantin> https://trac.videolan.org/vlc/ticket/13390
[16:30] <LConstantin> https://trac.videolan.org/vlc/ticket/13389
[16:32] <wm4> LConstantin: they're ancient, so yes
[16:33] <nevcairiel> who the fuck does security research on a OS for which official security patches have ceased
[16:34] <LConstantin> do you happen to know the corresponding CVEs in FFmpeg advisories?
[16:34] <nevcairiel> the issues could never be reproduced in ffmpeg
[16:34] <LConstantin> just trying to determine exactly when they were patched and whether other projects that use FFmpeg are vulnerable because they haven't updated FFmpeg
[16:34] <nevcairiel> if you can make it crash, please report a bug
[16:35] <nevcairiel> apparently vlc also claims that updated versions of avcodec are not even affected
[16:37] <wm4> LConstantin: well, take the samples (or ask for them), and try to reproduce?
[16:37] <wm4> and no reason to distrust the word of vlc here
[16:37] <thardin> don't this depends on how various programs are packaged?
[16:39] <LConstantin> I'm a security reporter with the IDG News Service (PCWorld, Computerworld, etc.). Since FFmpeg has no official media contact I thought I'd ask here. I understand that the latest version of FFmpeg is not vulnerable. I was trying to determine when exactly it was patched (in which version) so that I can determine if other players that use older versions of FFmpeg might also be vulnerable, not just VLC.
[16:41] <kierank> j-b: ^
[16:41] <thardin> you could use git-bisect to figure that out automagically
[16:41] <nevcairiel> I stick to my opinion that a security researcher working on Windows XP nearly a year after official security patches for this OS have ceased should be discredited and his work tripple checked before making any claims.
[16:41] <wm4> thardin: he doesn't/can't do any real work
[16:41] <thardin> oh
[16:41] <wm4> thardin: he's just here to annoy everyone and to make questionable statements about open source security
[16:41] <kierank> wm4: ?
[16:42] <wm4> that's what it sounds like
[16:42] <kierank> wm4: right, because reporters generally go on irc to check their facts
[16:45] <j-b> kierank: sorry, no, it's just bullshit.
[16:45] <j-b> it doesnot even crash vLC
[16:45] <kierank> I mean you should respond to the guy
[16:45] <LConstantin> I didn't make any statements. I asked questions.
[16:45] <thardin> hanlon's razor
[16:46] <cone-305> ffmpeg.git 03Paul B Mahol 07master:fd33658f866d: lavfi: remove mp=ilpack
[16:48] <kierank> durandal_1707: !!
[16:49] <cone-305> ffmpeg.git 03Paul B Mahol 07master:be1b8b4da5e4: fix Makefile: remove ilpack
[16:50] <ubitux> what about soft pulldown?
[16:50] <wm4> doesn't it do nothing?
[16:50] <durandal_1707> it should be ported
[16:51] <iive> it turns soft telecine into hard telecine
[16:51] <kepstin-laptop> soft pulldown is kind of useful, to fix dvds that were incorrectly soft-telecined :/
[16:51] <LConstantin> j-b I understand that you're probably upset with the media coverage of this issue reflecting badly on VLC. If it doesn't even crash VLC and the report is not valid why not take the opportunity to formulate a response and set the record straight.
[16:52] <kepstin-laptop> (or rather, were incorrectly detelecined before being soft-telecined)
[16:52] <wm4> durandal_1707: didn't you say the mp filter was broken? or was that in mpv only?
[16:53] <durandal_1707> wm4: it was broken in mp but i belive i fixed that
[16:53] <LConstantin> If it is a valid issue, just say what users need to do and that's it. Software has flaws, people accept that. It's how you respond to them that matters.
[16:53] <wm4> durandal_1707: oh I see, field flags were not passed
[16:54] <wm4> LConstantin: if it's really a libavcodec issue, I either expect it was reported to ffmpeg (and fixed), or vlc used an old version of libavcodec, and it was already fixed in a newer version
[16:57] <j-b> LConstantin: because I have better things to do.
[16:58] <LConstantin> wm4 Thanks.
[17:06] <cone-305> ffmpeg.git 03Michael Niedermayer 07master:48895a23fc03: avdevice/alsa-audio-common: Use *alloc_array()
[17:06] <cone-305> ffmpeg.git 03Michael Niedermayer 07master:081e117bfee0: avfilter/af_ashowinfo: Use av_realloc_array()
[17:46] <cone-305> ffmpeg.git 03Paul B Mahol 07master:357c6a92bc94: avfilter/af_join: Use av_mallocz_array()
[17:48] <ubitux> durandal_1707: please swap the args
[17:49] <durandal_1707> why?
[17:49] <ubitux> for consistency with the prototype
[17:49] <ubitux> libavutil/mem.h:av_alloc_size(1, 2) static inline void *av_mallocz_array(size_t nmemb, size_t size)
[17:55] <cone-305> ffmpeg.git 03Paul B Mahol 07master:f191c98e4d55: avfilter/af_join: swap av_mallocz_array() args for consistency with the prototype
[18:00] <wm4> not like it matters...?
[18:11] <akira4> ubitux, could you explain the WebVTT and CEA-608 work in a little more detail? 
[18:12] <ubitux> akira4: yep sure
[18:12] <ubitux> akira4: did you see the 3 .c files? do you have them opened?
[18:12] <akira4> just a sec.
[18:13] <akira4> yep
[18:14] <ubitux> look into srtenc.c
[18:14] <ubitux> and webvtt thing
[18:15] <ubitux> iirc there are some color tag support and stuff like
[18:15] <ubitux> you need to extract that code and put it into libavcodec/webvttenc.c
[18:15] <ubitux> which is currently a very basic encoder not supporting much tags
[18:15] <ubitux> basically, try extracting all the webvtt stuff for now
[18:16] <ubitux> the goal will be to extract all the features from these files
[18:16] <ubitux> it seems there are quite a bunch of things, like ttml, cc, etc
[18:16] <ubitux> for the cc (CEA-608 thing), you need to edit the cc decoder recently added from Anshul
[18:17] <ubitux> so same thing, you need to find how they constructed the ASS markup into these .c
[18:17] <ubitux> and add them in the appropriate place (in our cc decoder)
[18:17] <ubitux> but start with the webvtt, it's easier
[18:17] <akira4> hmm. alright.
[18:18] <akira4> oh and about the merge that I did :( do you think I should get a new copy of the ffmpeg repo or work with the same thing?
[18:19] <ubitux> akira4: branch out your work (git checkout -b dvdwhatever)
[18:20] <ubitux> then reset your master on origin/master
[18:20] <ubitux> your master must stay clean all the time
[18:20] <ubitux> then branch out that clean master as a new base
[18:20] <ubitux> it will look like this:
[18:20] <ubitux> git checkout -b dvd
[18:20] <ubitux> git checkout master
[18:21] <ubitux> git reset --hard origin/master
[18:21] <ubitux> git pull
[18:21] <ubitux> git checkout -b webvtt
[18:21] <akira4> got it. Thanks.
[18:21] <ubitux> origin must point on the correct remote (the git @ source.ffmpeg.org)
[18:21] <akira4> okay.
[18:49] <akira4> ubitux, how would I test the encoder after making changes? Use a srt file and encode it into webvtt something like that?
[18:49] <ubitux> yes, in fate you have a .srt file with colors and stuff like that
[18:49] <ubitux> convert it to webvtt
[18:50] <akira4> alright.
[18:50] <ubitux> this will probably change the webvtt test btw
[18:50] <akira4> yeah.
[18:50] <ubitux> run a make fate-sub-webvttenc GEN=1
[18:50] <ubitux> and look at the diff
[18:50] <ubitux> (after your modification)
[18:51] <akira4> okay.
[20:53] <wm4> how does ffmpeg handle locales? what does av_asprintf("%f", 1.5) return?
[21:01] <rcombs> av_asprintf uses vsnprintf
[21:01] <wm4> yeah
[21:01] <wm4> but vsnprintf uses the current locale
[21:01] <wm4> so it might result in "1,5"
[21:02] <wm4> I guess libav* doesn't convert floats to strings that often as that it could be a problem...
[21:03] <rcombs> something something C locales considered harmful
[21:04] <wm4> certainly
[21:04] <wm4> some frameworks (Qt) even set it on program start, and the recommended way to fix it is setting it back to C
[21:05] <rcombs> libmediainfo fucks with it too
[21:07] <rcombs> hey maybe C1Y will have locale-independent versions of everything
[21:07] <wm4> yeah, let's add more shit, instead of fixing it
[21:08] <wm4> standards committees are all about adding shit to existing shit
[21:09] <rcombs> well making everything locale-independent would be nice but probably even less likely than new standardized locale-independent variants
[21:09] <wm4> I want to see who the fuck even "needs" the locale shit (especially LC_NUMERIC, which breaks snprintf/strtod)
[21:14] <cone-305> ffmpeg.git 03Michael Niedermayer 07master:ea3fc9fe6875: smoothstreamingenc: Add a missing "goto fail"
[21:14] <cone-305> ffmpeg.git 03Michael Niedermayer 07master:9cbaf5849b21: Merge commit 'ea3fc9fe68752eb1b9078b28bea18a1e781d1c2c'
[21:29] <cone-305> ffmpeg.git 03Martin Storsjö 07master:bba02479260d: libopenh264enc: Remove a workaround for silencing warnings about unused variables in the OpenH264 header
[21:29] <cone-305> ffmpeg.git 03Michael Niedermayer 07master:87c4a80e563a: Merge commit 'bba02479260d0e7dec8c530a7e75a1c7aa53c06e'
[21:55] <cone-305> ffmpeg.git 03Martin Storsjö 07master:440119b18836: libopenh264enc: Move a declaration of a variable into an ifdef
[21:55] <cone-305> ffmpeg.git 03Michael Niedermayer 07master:6c9dfa5a1b45: Merge commit '440119b18836887d98c9e337c5911563bb43588c'
[00:00] --- Wed Jan 21 2015


More information about the Ffmpeg-devel-irc mailing list