September 2011 - FFmpeg-trac - lists.ffmpeg.org

#441(avcodec:new): MPEG audio decoder prints "incorrect frame size" error for valid code
by FFmpeg 08 Sep '11

08 Sep '11

#441: MPEG audio decoder prints "incorrect frame size" error for valid code ---------------------------------+---------------------------------- Reporter: chrisr | Type: defect Status: new | Priority: normal Component: avcodec | Version: 0.7.3 Keywords: | Blocked By: Blocking: | Reproduced by developer: 0 Analyzed by developer: 0 | ---------------------------------+---------------------------------- Decoding an MP3 audio stream (CODEC_ID_MP3) using libavcodec and libavutil is creating lots of messages like these: [mp3 @ 0xad218660] incorrect frame size [mp3 @ 0xad218660] incorrect frame size [mp3 @ 0xad218660] incorrect frame size [mp3 @ 0xad218660] incorrect frame size [mp3 @ 0xad218660] incorrect frame size However, the audio is playing fine, so there can't really be an error here. The application is funnelling an audio stream into avcodec_decode_audio3(), and is just expecting the function to consume as many bytes as it needs each time: In simplified form: {{{ uint8_t *buffer_offset = buffer; while (buffer_size > 0) { av_init_packet(&avpkt); avpkt.data = buffer_offset; avpkt.size = buffer_size; bytes_consumed = avcoded_decode_audio3(context, output_buffer, &output_size, &avpkt); if (bytes_consumed < 0) { // Error handling return; } /* Send decoded output to device */ buffer_size -= bytes_consumed; buffer_offset += bytes_consumed; } }}} The spurious error message is coming from decode_frame() in libavcodec/mpegaudiodec.c: line 1977 {{{ }else if(s->frame_size < buf_size){ av_log(avctx, AV_LOG_ERROR, "incorrect frame size\n"); buf_size= s->frame_size; } }}} At very worst, this should only be AV_LOG_WARNING because it's internally correctable and doesn't affect normal operation. But I would actually argue that this message could be removed completely instead, because (to be blunt) who cares if the input buffer is too big, so long as the entire frame is present and we return the correct number of bytes consumed to the caller? This message is just noise. -- Ticket URL: <https://ffmpeg.org/trac/ffmpeg/ticket/441> FFmpeg <http://ffmpeg.org> FFmpeg issue tracker

1 5

#421(FFmpeg:new): Online Doxygen docs mixes up ffmpeg.c docs with avconv.c docs.
by FFmpeg 08 Sep '11

08 Sep '11

#421: Online Doxygen docs mixes up ffmpeg.c docs with avconv.c docs. -------------------------+--------------------- Reporter: MattE | Owner: michael Type: defect | Status: new Priority: normal | Component: FFmpeg Version: git-master | Keywords: Doxygen Blocked By: | Blocking: Reproduced: 0 | Analyzed: 0 -------------------------+--------------------- In the doxygen page for the ffmpeg.c source code when you click on a function name to be taken to the docs for that function, if that function is defined in ffmpeg.c it will take you to the avconv.c docs instead of the ffmpeg docs. For example, if you click on the function name write_frame here: http://www.ffmpeg.org/doxygen/trunk/ffmpeg_8c-source.html#l00783 it takes you to http://www.ffmpeg.org/doxygen/trunk/avconv_8c.html#ec8a41459c0bd4c4d4e84b51… instead of to http://www.ffmpeg.org/doxygen/trunk/ffmpeg_8c.html#ec8a41459c0bd4c4d4e84b51… -- Ticket URL: <https://ffmpeg.org/trac/ffmpeg/ticket/421> FFmpeg <http://ffmpeg.org> FFmpeg issue tracker

1 1

#350(FFmpeg:new): Invalid value 'strict' for option 'strict'
by FFmpeg 08 Sep '11

08 Sep '11

#350: Invalid value 'strict' for option 'strict' ------------------------+--------------------- Reporter: KSHawkEye | Owner: michael Type: defect | Status: new Priority: important | Component: FFmpeg Version: git | Keywords: -strict Blocked By: | Blocking: Reproduced: 0 | Analyzed: 0 ------------------------+--------------------- When trying to use strict for option -strict on a Windows FFmpeg build, I'm getting: {{{ [NULL @ 00000000004A5A60] [Eval @ 000000000022F800] Undefined constant or missin g '(' in 'strict' [NULL @ 00000000004A5A60] Unable to parse option value "strict" Invalid value 'strict' for option 'strict' }}} The input file does not matter, but I tested with h264, aac, and flac to name a few. This bug should be repeatable with a simple command line: {{{ ffmpeg -i "infile.mp4" -vcodec mpeg4 -an -strict strict "outfile.avi" }}} using the git-371888c build found here: http://ffmpeg.zeranoe.com/builds/ I believe {{{ Undefined constant or missing '(' in 'strict' }}} is the key to this issue, but I could be wrong. Other values '''do''' work, such as "very". -- Ticket URL: <https://ffmpeg.org/trac/ffmpeg/ticket/350> FFmpeg <http://ffmpeg.org> FFmpeg issue tracker

1 2

#331(avcodec:new): Endless loop when reading jpeg-ls sample
by FFmpeg 08 Sep '11

08 Sep '11

#331: Endless loop when reading jpeg-ls sample -------------------------+--------------------- Reporter: cehoyos | Owner: Type: defect | Status: new Priority: normal | Component: avcodec Version: git-master | Keywords: Blocked By: | Blocking: Reproduced: 0 | Analyzed: 0 -------------------------+--------------------- When reading one of the samples provided with the source of David A. Clunie's jpeg-ls implementation, ffmpeg loops endlessly. {{{ (gdb) r -i CT2.noruns.jls Starting program: ffmpeg_g -i CT2.noruns.jls ffmpeg version N-31266-g3950376, Copyright (c) 2000-2011 the FFmpeg developers built on Jul 6 2011 10:20:08 with gcc 4.5.3 configuration: --cc=/usr/local/gcc-4.5.3/bin/gcc libavutil 51. 11. 0 / 51. 11. 0 libavcodec 53. 7. 0 / 53. 7. 0 libavformat 53. 5. 0 / 53. 5. 0 libavdevice 53. 2. 0 / 53. 2. 0 libavfilter 2. 24. 3 / 2. 24. 3 libswscale 2. 0. 0 / 2. 0. 0 ^C Program received signal SIGINT, Interrupt. 0x000000000065f133 in ls_get_code_regular (Q=<value optimized out>, state=0x1282a40, gb=0x1277cc8) at libavcodec/jpeglsdec.c:93 93 for(k = 0; (state->N[Q] << k) < state->A[Q]; k++); #0 0x000000000065f133 in ls_get_code_regular (Q=<value optimized out>, state=0x1282a40, gb=0x1277cc8) at libavcodec/jpeglsdec.c:93 #1 ls_decode_line (Q=<value optimized out>, state=0x1282a40, gb=0x1277cc8) at libavcodec/jpeglsdec.c:239 #2 0x00000000006602ab in ff_jpegls_decode_picture (s=0x1277cc0, near=<value optimized out>, point_transform=<value optimized out>, ilv=<value optimized out>) at libavcodec/jpeglsdec.c:298 #3 0x0000000000675fba in ff_mjpeg_decode_sos (s=0x1277cc0, mb_bitmask=<value optimized out>, reference=<value optimized out>) at libavcodec/mjpegdec.c:1035 #4 ff_mjpeg_decode_frame (s=0x1277cc0, mb_bitmask=<value optimized out>, reference=<value optimized out>) at libavcodec/mjpegdec.c:1533 #5 0x000000000078e208 in avcodec_decode_video2 (avctx=0x1277800, picture=0x7fffffffd7c0, got_picture_ptr=0x7fffffffdaec, avpkt=0x127a340) at libavcodec/utils.c:747 #6 0x00000000004c1e3d in try_decode_frame (avpkt=<value optimized out>, st=<value optimized out>) at libavformat/utils.c:2136 #7 av_find_stream_info (avpkt=<value optimized out>, st=<value optimized out>) at libavformat/utils.c:2426 #8 0x000000000040cec9 in opt_input_file (opt=<value optimized out>, filename=0x7fffffffe277 "CT2.noruns.jls") at ffmpeg.c:3368 #9 0x0000000000411641 in parse_options (argc=3, argv=0x7fffffffddf8, options=0x94f360, parse_arg_function=0x40f4c0 <opt_output_file>) at cmdutils.c:283 #10 0x0000000000410896 in main (argc=3, argv=0x7fffffffddf8) at ffmpeg.c:4563 }}} -- Ticket URL: <https://avcodec.org/trac/ffmpeg/ticket/331> FFmpeg <http://ffmpeg.org> FFmpeg issue tracker

1 1

#92(FFmpeg:new): Extract all key frames from a video using ffmpeg tool
by FFmpeg 07 Sep '11

07 Sep '11

#92: Extract all key frames from a video using ffmpeg tool --------------------------+--------------------- Reporter: benoit | Owner: michael Type: enhancement | Status: new Priority: wish | Component: FFmpeg Version: git | Keywords: Blocked By: | Blocking: Reproduced: 0 | Analyzed: 0 --------------------------+--------------------- I've been asked several times if we could make the ffmpeg command line tool able to extract all key frames from a video. I don't think it's possible today, but may be a nice feature to have. -- Ticket URL: <https://ffmpeg.org/trac/ffmpeg/ticket/92> FFmpeg <http://ffmpeg.org> FFmpeg issue tracker

1 3

#395(undetermined:new): ffmpeg: failed to set value 'XXX' for option 'passlogfile'
by FFmpeg 07 Sep '11

07 Sep '11

#395: ffmpeg: failed to set value 'XXX' for option 'passlogfile' --------------------------+-------------------------- Reporter: j | Type: defect Status: new | Priority: normal Component: undetermined | Version: unspecified Keywords: | Blocked By: Blocking: | Reproduced: 0 Analyzed: 0 | --------------------------+-------------------------- in both git branches release/0.7 and release/0.8 {{{ ffmpeg -i input.mov -pass 1 -passlogfile /tmp/2pass /tmp/test.webm }}} fails with: {{{ ffmpeg: failed to set value '/tmp/2pass' for option 'passlogfile' }}} this happens with this configuration: {{{ ./configure \ --enable-gpl \ --enable-version3 \ --enable-libvorbis \ --enable-pthreads \ --disable-shared \ --disable-ffserver \ --disable-ffprobe \ --enable-postproc \ --enable-libtheora \ --enable-libvpx \ --disable-encoders \ --disable-muxers \ --enable-encoder=libvorbis,libvpx \ --enable-muxer=webm }}} -- Ticket URL: <http://ffmpeg.org/trac/ffmpeg/ticket/395> FFmpeg <http://ffmpeg.org> FFmpeg issue tracker

1 6

#429(FFmpeg:new): ffmpeg is killed by kernel (out of Memmry)
by FFmpeg 07 Sep '11

07 Sep '11

#429: ffmpeg is killed by kernel (out of Memmry) -------------------------------------+------------------------------------- Reporter: mus | Owner: michael Type: defect | Status: new Priority: important | Component: FFmpeg Version: git | Keywords: memory x264 Blocked By: | killed kernel Reproduced by developer: 0 | Blocking: | Analyzed by developer: 0 -------------------------------------+------------------------------------- I originally created a bug report for this for libav. Since ffmpeg (compiled from git) has exactly the same issue and no one has reacted to the libav bug report yet, I thought I should add it here. ffmpeg is killed by the kernel when encoding some HD videos with x264 because it uses too much memory. See the original bug report here, this should include all needed information, including the used ffmpeg command line: http://bugzilla.libav.org/show_bug.cgi?id=31 -- Ticket URL: <https://ffmpeg.org/trac/ffmpeg/ticket/429> FFmpeg <http://ffmpeg.org> FFmpeg issue tracker

1 13

#436(FFplay:new): ffplay disables DPMS timeouts; no way to re-enable them during play
by FFmpeg 06 Sep '11

06 Sep '11

#436: ffplay disables DPMS timeouts; no way to re-enable them during play --------------------------------------+--------------------------------- Reporter: tpavlic | Owner: michael Type: defect | Status: new Priority: normal | Component: FFplay Version: git-oldabi | Keywords: DPMS Blocked By: | Blocking: Reproduced by developer: 0 | Analyzed by developer: 0 --------------------------------------+--------------------------------- ffplay (and mplayer) appears to disable DPMS timeouts while it runs. I can verify that "xset q" shows that DPMS is enabled and timeouts are set correctly; however, so long as ffplay (or mplayer) are open, monitors refuse to sleep. With mplayer, I can at least issue an "xset dpms force off" and the monitors will turn off and stay off, but standby and suspend are impossible (monitors pop back on). With ffplay, even "xset dpms force off" causes monitors to pop back on. I can understand why this is a desirable behavior for playing movies. However, some applications (e.g., the electricsheep screen saver) are not deteriorated by allowing monitors to sleep. So there should be a way to turn off this DPMS-dodging feature of ffplay (and mplayer). -- Ticket URL: <https://ffmpeg.org/trac/ffmpeg/ticket/436> FFmpeg <http://ffmpeg.org> FFmpeg issue tracker

1 3

#426(undetermined:new): Clip buggy playback in ffmpeg but ok in vlc or QT
by FFmpeg 06 Sep '11

06 Sep '11

#426: Clip buggy playback in ffmpeg but ok in vlc or QT ----------------------------+--------------------- Reporter: ehsmeng | Type: defect Status: new | Priority: normal Component: undetermined | Version: git Keywords: mjpeg yuvj422p | Blocked By: Blocking: | Reproduced: 0 Analyzed: 0 | ----------------------------+--------------------- Buggy playback in ffmpeg. Output when transcoding: [mjpeg @ 0x101820400] error count: 75 [mjpeg @ 0x101820400] error y=52 x=59 [mjpeg @ 0x101820400] mjpeg_decode_dc: bad vlc: 0:0 (0x1019a3a30)te=9115.1kbits/s [mjpeg @ 0x101820400] error dc [mjpeg @ 0x101820400] error y=50 x=57 [mjpeg @ 0x101820400] error count: 70 [mjpeg @ 0x101820400] error y=45 x=58 [mjpeg @ 0x101820400] error count: 65 [mjpeg @ 0x101820400] error y=45 x=58 [mjpeg @ 0x101820400] error count: 66 [mjpeg @ 0x101820400] error y=45 x=58 [mjpeg @ 0x101820400] error count: 65 [mjpeg @ 0x101820400] error y=52 x=58 [mjpeg @ 0x101820400] error count: 68 [mjpeg @ 0x101820400] error y=46 x=60 [mjpeg @ 0x101820400] error count: 68417kB time=00:00:00.72 bitrate=4746.5kbits/s [mjpeg @ 0x101820400] error y=46 x=58 [mjpeg @ 0x101820400] error count: 64 [mjpeg @ 0x101820400] error y=46 x=58 frame= 25 fps= 16 q=24.8 Lsize= 464kB time=00:00:01.00 bitrate=3804.3kbits/s video:463kB audio:0kB global headers:0kB muxing overhead 0.206299% Playback works fine in Quicktime and VLC. Attaching example file. -- Ticket URL: <https://ffmpeg.org/trac/ffmpeg/ticket/426> FFmpeg <http://ffmpeg.org> FFmpeg issue tracker

1 5

#56(avcodec:new): ffmpeg fails to build on ppc64 with 32-bit userland in fft_altivec_s.S
by FFmpeg 01 Sep '11

01 Sep '11

#56: ffmpeg fails to build on ppc64 with 32-bit userland in fft_altivec_s.S ------------------------+--------------------- Reporter: gnu_andrew | Type: defect Status: new | Priority: normal Component: avcodec | Version: git Keywords: ppc | Blocked By: Blocking: | Reproduced: 0 Analyzed: 0 | ------------------------+--------------------- powerpc-unknown-linux-gnu-gcc -I. -I"/var/tmp/portage/media- video/ffmpeg-9999/work/ffmpeg-9999" -D_ISOC99_SOURCE -D_POSIX_C_SOURCE=200112 -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -DHAVE_AV_CONFIG_H -mcpu=970 -mpowerpc-gfxopt -mpowerpc64 -MMD -MF libavcodec/ppc/fft_altivec_s.d -MT libavcodec/ppc/fft_altivec_s.o -c -o libavcodec/ppc/fft_altivec_s.o libavcodec/ppc/fft_altivec_s.S libavcodec/ppc/fft_altivec_s.S: Assembler messages: libavcodec/ppc/fft_altivec_s.S:447: Error: junk at end of line, first unrecognized character is `@' libavcodec/ppc/fft_altivec_s.S:448: Error: junk at end of line, first unrecognized character is `@' make: *** [libavcodec/ppc/fft_altivec_s.o] Error 1 Tip of checkout is: commit c104a6e90226eaa253aeba471c8859e0da922da7 Author: Michael Niedermayer <michaelni(a)gmx.at> Date: Mon Apr 11 11:49:07 2011 +0200 oggdec: use av_freep() instead of av_free() Signed-off-by: Michael Niedermayer <michaelni(a)gmx.at> -- Ticket URL: <https://ffmpeg.org/trac/ffmpeg/ticket/56> FFmpeg <http://ffmpeg.org> FFmpeg issue tracker

1 1