February 2011 - ffmpeg-devel - lists.ffmpeg.org

[PATCH] rtpdec: make the NTP time values unsigned.
by flameeyes＠gmail.com 09 May '14

09 May '14

As per RFC3550, section 4, the NTP time is provided as 64-bit unsigned integer, so follow the same logic here. --- libavformat/rtpdec.h | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/libavformat/rtpdec.h b/libavformat/rtpdec.h index 86af2b9..87fd419 100644 --- a/libavformat/rtpdec.h +++ b/libavformat/rtpdec.h @@ -148,8 +148,8 @@ struct RTPDemuxContext { RTPStatistics statistics; ///< Statistics for this stream (used by RTCP receiver reports) /* rtcp sender statistics receive */ - int64_t last_rtcp_ntp_time; // TODO: move into statistics - int64_t first_rtcp_ntp_time; // TODO: move into statistics + uint64_t last_rtcp_ntp_time; // TODO: move into statistics + uint64_t first_rtcp_ntp_time; // TODO: move into statistics uint32_t last_rtcp_timestamp; // TODO: move into statistics /* rtcp sender statistics */ -- 1.7.1.1

4 3

[WIP] rotate filter(s)
by stefano.sabatini-lala＠poste.it 12 Jun '13

12 Jun '13

Hi, in attachment a first stub at the rotate filter. I don't know if it is a good idea to keep the float variant, while the rotate_ss may be useful (and the rotation angle could be made parametric). The integer-only rotating code is taken from tests/rotozoom.c, I don't know if we have a corresponding version in the libs. fill_line_with_color() and draw_rectangle() are duplicated from vf_pad.c, so maybe we should make them public (or at least ff_ them), drawutils.[ch] may be a good place where to put them. Note that ffplay/SDL doesn't work with odd width/height sizes, so you may need to rescale to an even wxh size (this could be done with a parametric scale as discussed some months ago). What this rotate filter still lacks, apart a resolution to the refactoring problems mentioned above, is a YUV planar version of the rotating function, and possibly the extension to other RGB variants, which should be pretty easy to add. Everyone feel free to continue to work on this and post an updated patch. Regards. -- FFmpeg = Fostering and Frightening Multimedia Portable Emblematic Guru

9 23

Yamaha SMAF 'CNTI' chunk is wrong.
by yoshihisa.uchida＠gmail.com 07 Jan '13

07 Jan '13

Hi, ffmpeg developers, For Yamaha SMAF, Contents Info chunk ('CNTI') includes only following members: - Contents Class - Contents Type - Contents Code Type - Copy Status - Copy Counts (see http://smaf-yamaha.com/what/s52.html) But, in the current ffmpeg, 'CNTI' includes metadata ("VN:libavcodec,"). According to SMAF Specifications Optional Data Chunk (http://smaf-yamaha.com/what/s52.html) I corrected that metadata include Optional Data Chunk ('OPDA'). Regards, uchan

5 5

[PATCH] OpenEXR decoder
by jimmy＠ghost.dk 06 Apr '12

06 Apr '12

Hi, Here is my stab at a OpenEXR decoder. It only supports 16-bit and 32-bit float. 8-bit integer is in theory supported aswell by the format, but I have yet to find software which supports writing it, but it should be quite straigthforward to implement aswell, but didn't want to touch it since I had nothing to test with. The 32-bit float path is very slow (about 60x slower than the 16-bit path) and not optimized in way yet. -- Best Regards Jimmy Christensen Developer Ghost A/S

9 77

dvd/vob "decoder" part 1: the system itself
by erik＠arawix.com 03 Feb '12

03 Feb '12

Hello; Included are the files, and patches needed for the dvd decoder. I threw away my original system on trying to insert packets into the main stream (as the "attachment/data" type stream). The reason for that is that the dvd data stream, that contains commands for the client can change/insert packets "in the past". Example; a user seeks back. The DVD ifo system will insert packets containing "change the subtitle color" so it works ok for the displayed cell/chapter. LibAV doesn't expect new packets inserted "backwards" in the stream, and the buffer system gets all messed up because packet sizes are off. Changing to a new system means my "hack" on decoding the packet header isn't needed any more, and except for stream flagging the mpegps demuxer should be unchanged. My attachment decoder is not needed anymore (for those who looked at it, only dvd_protocol.c/h are still needed). The new system uses a callback, that the user registers. This is, after much experimenting, the best way I could come up with. The callback though is registered per protocol, and thus changes are needed in "gasp" URLContext. Reason for keeping it per context? Well, a user can be reading from several dvd devices at the same time if he/she uses the avlibs. Or some end devices (like the ps3) send several request to open a file at the same time. If one callback is registered for the whole library, this messes up the system. Whats send in the callback? Basically AVDVDParseContext packets. They contain a type (the enum effdvd_Type ) that tells the end-user app what should be done, and a pts timestamp in dvd time (90000/tick). The packets should be handled at that time. There are some packets which are obvious (like: new subtitle palette) and some that are not. The most important one is effdvd_Reset_IFO. effdvd_Reset_IFO asks the end-user app: clear your streams, flag all streams you are using as invalid (more later on that), and be prepared to re-init all your codec decoders. Why does one need to do that? Because a vob file 1 might contain audio data in ac3 with 2 channels, while a vob 2 might have it in 6 channels. Or the resolution of the video stream might change (from dvd pal to dvd ntsc widescreen for example). First; DVD's may contain up to 64 streams. One of the patches changes the MAX_STREAMS constant to match that. Second; Stream re-init. I implemented this by adding a flag to AVStream (at the end, discard_flags). Set this to one in end-user app, and this will tell the demuxer that this stream has been discarded by user (user closed all codecs, etc). It works by telling the demuxer that once it dedects a certain packet type, and normally does a av_new_stream; don't re-add the stream to the end of the new list (pushing more and more streams on the list). Example; suppose the discard_flag is set to 1 for the video stream (and we only have one). When a new packet is received by the demuxer, its loop will not find a valid video stream anymore. It thus will create a new one, but will fill the slot for the "old" video stream with the new video stream. Don't know if this text makes sense, bit a quick look at the code should be helpful. Why not do this in the protocol reader? Basically 1) because the protocol doesn't and shouldn't know about the streams and 2) because the end-user needs to close and free his/her decoders. Once that's done, the demuxer doesn't even know. Next; chapter seeking. I added a new flag for that: #define AVSEEK_FLAG_DIRECT 8 ///< seek directly, skipping any stamps and translation #define AVSEEK_FLAG_CHAPTER 16 ///< seek to chapter, if supported Using that will basically use av_seek_frame() to override a search like it does for AVSEEK_FLAG_BYTE. The arguments timestamp will thus position chapters, instead of timestamps. Summary; whats the total change; 1) The whole setup adds a protocol handler called ffdvd. User inits his open file with av_open_input_file ("dvdread:e:/") or av_open_input_file("dvdread:/cdrom1/") 2) User registers a callback by using url_fset_protocol_callback(), a new function in avio.c. I tried to make it generic. 3) User polls streams as before for audio/video 4) User can get callback events from the protocol, with timestamps. 5) Events get examined for the timestamps on the protocol packet, and handled upon. 6) When needed, these events are executed, and for some functions a notify is send back (example, when user selects a button dvd_protocol_select_button() is called). 7) That's it. How well does this all work? I build a example app separate from ffmpeg, on which rather big changes are going to be needed. Note that these changes have almost nothing to do with the dvd protocol handler. They are mostly implementing the "we should look at this/this is wrong/etc" comments (example; discarding new streams). I have ffplay ok, but will need to work with someone on ffmpeg. Source is. Offcourse available. I am attaching the main files here, and the most important patches. Since I am not sure on how the continue on this, please tell me how this needs to be handled (ege do patches first, what order,. Etc. ) If the system is accepted, at least one patch needs a major version bump. I should probably split up some patches too. Note 2; I need to talk to the people for libdvdread, since I fixed a few problems in their code (and added one function to read attributes that its not available now dvdnav_get_video_attributes()). To have this working, we would offcourse also need to depend on libdvdread (if the user enabled it by ./configure, changes for makesfiles and that needed). Note 3: we will probably need some commandline extensions to specify seeking to a chapter, etc. for ffmpeg I have worked on and of on this for the better part of 2 months figuring out the best and least intrusive way. If this all gets rejected, I am fine with that. If it doesn't, also. Just let me know. Erik

6 10

[PATCH] Port tinterlace filter from MPlayer.
by stefano.sabatini-lala＠poste.it 01 Jan '12

01 Jan '12

--- configure | 1 + doc/filters.texi | 26 +++++ libavfilter/Makefile | 1 + libavfilter/allfilters.c | 1 + libavfilter/avfilter.c | 2 +- libavfilter/vf_tinterlace.c | 260 +++++++++++++++++++++++++++++++++++++++++++ 6 files changed, 290 insertions(+), 1 deletions(-) create mode 100644 libavfilter/vf_tinterlace.c diff --git a/configure b/configure index 0538201..d3e2c14 100755 --- a/configure +++ b/configure @@ -1402,6 +1402,7 @@ udp_protocol_deps="network" # filters ocv_smooth_filter_deps="libopencv" +tinterlace_filter_deps="gpl" yadif_filter_deps="gpl" # libraries diff --git a/doc/filters.texi b/doc/filters.texi index d000276..3397989 100644 --- a/doc/filters.texi +++ b/doc/filters.texi @@ -390,6 +390,32 @@ not specified it will use the default value of 16. Adding this in the beginning of filter chains should make filtering faster due to better use of the memory cache. + at section tinterlace + +Perform temporal field interlacing - merge pairs of frames into an +interlaced frame, halving the framerate. Even frames are moved into +the upper field, odd frames to the lower field. + +The frames are counted starting from 1, so the first input frame is +considered odd. + +Accepts a parameters specifying the mode. Available modes are: + +0 Move odd frames into the upper field of the new frame, even into the + lower field, generating a double height frame at half framerate. + +1 Only output odd frames, even frames are dropped; height + unchanged, half framerate + +2 Only output even frames, odd frames are dropped; height + unchanged, half framerate + +3 Expand each frame to full height, but pad alternate lines with + black; framerate unchanged and double height. + +4 Interleave even lines from even frames with odd lines from odd + frames. Height unchanged at half framerate. + @section unsharp Sharpen or blur the input video. diff --git a/libavfilter/Makefile b/libavfilter/Makefile index 174b83e..431fe33 100644 --- a/libavfilter/Makefile +++ b/libavfilter/Makefile @@ -34,6 +34,7 @@ OBJS-$(CONFIG_PIXDESCTEST_FILTER) += vf_pixdesctest.o OBJS-$(CONFIG_PIXELASPECT_FILTER) += vf_aspect.o OBJS-$(CONFIG_SCALE_FILTER) += vf_scale.o OBJS-$(CONFIG_SLICIFY_FILTER) += vf_slicify.o +OBJS-$(CONFIG_TINTERLACE_FILTER) += vf_tinterlace.o OBJS-$(CONFIG_UNSHARP_FILTER) += vf_unsharp.o OBJS-$(CONFIG_VFLIP_FILTER) += vf_vflip.o OBJS-$(CONFIG_YADIF_FILTER) += vf_yadif.o diff --git a/libavfilter/allfilters.c b/libavfilter/allfilters.c index ec7d933..5224fde 100644 --- a/libavfilter/allfilters.c +++ b/libavfilter/allfilters.c @@ -54,6 +54,7 @@ void avfilter_register_all(void) REGISTER_FILTER (PIXELASPECT, pixelaspect, vf); REGISTER_FILTER (SCALE, scale, vf); REGISTER_FILTER (SLICIFY, slicify, vf); + REGISTER_FILTER (TINTERLACE, tinterlace, vf); REGISTER_FILTER (UNSHARP, unsharp, vf); REGISTER_FILTER (VFLIP, vflip, vf); REGISTER_FILTER (YADIF, yadif, vf); diff --git a/libavfilter/avfilter.c b/libavfilter/avfilter.c index 221dfaa..72566e5 100644 --- a/libavfilter/avfilter.c +++ b/libavfilter/avfilter.c @@ -19,7 +19,7 @@ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ -/* #define DEBUG */ +#define DEBUG #include "libavcodec/audioconvert.c" #include "libavutil/pixdesc.h" diff --git a/libavfilter/vf_tinterlace.c b/libavfilter/vf_tinterlace.c new file mode 100644 index 0000000..771c6be --- /dev/null +++ b/libavfilter/vf_tinterlace.c @@ -0,0 +1,260 @@ +/* + * Copyright (C) 2010 Stefano Sabatini + * Copyright (C) 2003 Michael Zucchi <notzed at ximian.com> + * + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License along + * with FFmpeg if not, write to the Free Software Foundation, Inc., + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. + */ + +#include "libavcore/imgutils.h" +#include "avfilter.h" + +typedef struct { + /** + * 0: Move odd frames into the upper field of the new image, even into the lower field, generating a double-height video at half framerate. + * 1: Only output odd frames, even frames are dropped; height unchanged, half framerate + * 2: Only output even frames, odd frames are dropped; height unchanged, half framerate + * 3: Expand each frame to double height, but pad alternate lines with black; framerate unchanged. + * 4: Interleave even lines from even frames with odd lines from odd frames. Height unchanged at half framerate. + */ + int mode; + + int vsub; ///< chroma vertical subsampling + int frame; ///< frame number + int frame_pending; ///< tell if there is a pending frame ready to be sent + AVFilterBufferRef *picref; ///< previously stored picref +} TInterlaceContext; + +static int query_formats(AVFilterContext *ctx) +{ + static const enum PixelFormat pix_fmts[] = { + PIX_FMT_YUV420P, PIX_FMT_YUV422P, PIX_FMT_YUV410P, + PIX_FMT_YUV444P, PIX_FMT_GRAY8, PIX_FMT_NONE + }; + + avfilter_set_common_formats(ctx, avfilter_make_format_list(pix_fmts)); + + return 0; +} + +static av_cold int init(AVFilterContext *ctx, const char *args, void *opaque) +{ + TInterlaceContext *tinterlace = ctx->priv; + tinterlace->mode = 0; + tinterlace->picref = NULL; + tinterlace->frame = 1; + tinterlace->frame_pending = 0; + + if (args) sscanf(args, "%d", &tinterlace->mode); + + if (tinterlace->mode < 0 || tinterlace->mode > 4) { + av_log(ctx, AV_LOG_ERROR, + "Unknown mode %d, use a value between 0 and 4\n", tinterlace->mode); + return AVERROR(EINVAL); + } + + return 0; +} + +static av_cold void uninit(AVFilterContext *ctx) +{ + TInterlaceContext *tinterlace = ctx->priv; + + if (tinterlace->picref) avfilter_unref_buffer(tinterlace->picref); +} + +static int config_out_props(AVFilterLink *outlink) +{ + AVFilterContext *ctx = outlink->src; + AVFilterLink *inlink = outlink->src->inputs[0]; + TInterlaceContext *tinterlace = ctx->priv; + + tinterlace->vsub = av_pix_fmt_descriptors[outlink->format].log2_chroma_h; + outlink->w = inlink->w; + outlink->h = (tinterlace->mode == 0 || tinterlace->mode == 3) ? inlink->h*2 : inlink->h; + + av_log(ctx, AV_LOG_INFO, "mode:%d in_h:%d out_h:%d\n", + tinterlace->mode, inlink->h, outlink->h); + + return 0; +} + +static void copy_picture_fields(uint8_t *dst[4], int dst_linesize[4], + const uint8_t *src[4], int src_linesize[4], + enum PixelFormat format, int w, int + src_h, int lower_fields, + int only_half_src_fields, int src_lower_fields) +{ + int i, k = 1, vsub = av_pix_fmt_descriptors[format].log2_chroma_h; + if (only_half_src_fields) { + src_h /= 2; + k = 2; + } + + for (i = 0; i < 3; i++) { + int h = i == 1 || i == 2 ? src_h >> vsub : src_h; + int linesize = av_image_get_linesize(format, w, i); + + if (lower_fields) + dst[i] += dst_linesize[i]; + if (only_half_src_fields && src_lower_fields) + src[i] += src_linesize[i]; + av_image_copy_plane(dst[i], dst_linesize[i]*2, + src[i], src_linesize[i]*k, linesize, h); + } +} + +static void end_frame(AVFilterLink *inlink) +{ + AVFilterContext *ctx = inlink->dst; + AVFilterLink *outlink = ctx->outputs[0]; + TInterlaceContext *tinterlace = ctx->priv; + AVFilterBufferRef *outpicref; + AVFilterBufferRef *picref = inlink->cur_buf; + + if (tinterlace->mode == 0) { + if (!tinterlace->picref) { + outpicref = avfilter_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h); + avfilter_copy_buffer_ref_props(outpicref, picref); + outpicref->video->w = outlink->w; + outpicref->video->h = outlink->h; + outpicref->video->interlaced = 1; + outpicref->video->top_field_first = 1; + + /* write odd frame fields into the upper fields of the new frame */ + copy_picture_fields(outpicref->data, outpicref->linesize, + picref ->data, picref ->linesize, + outlink->format, inlink->w, inlink->h, 0, 0, 0); + tinterlace->picref = outpicref; + } else { + outpicref = tinterlace->picref; + + /* write even frame fields into the lower fields of the new frame */ + copy_picture_fields(outpicref->data, outpicref->linesize, + picref ->data, picref ->linesize, + outlink->format, inlink->w, inlink->h, 1, 0, 0); + tinterlace->frame_pending = 1; + } + } else if ((tinterlace->mode == 1 && tinterlace->frame & 1) || /* send only odd frames */ + (tinterlace->mode == 2 && !(tinterlace->frame & 1))) { /* send only even frames */ + tinterlace->picref = avfilter_ref_buffer(picref, AV_PERM_READ); + tinterlace->frame_pending = 1; + } else if (tinterlace->mode == 3) { + outpicref = avfilter_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h); + avfilter_copy_buffer_ref_props(outpicref, picref); + outpicref->video->w = outlink->w; + outpicref->video->h = outlink->h; + outpicref->video->interlaced = 1; + outpicref->video->top_field_first = tinterlace->frame & 1; + + /* cleanup the destination picture */ + memset(outpicref->data[0], 0, outpicref->linesize[0] * outlink->h); + if (outpicref->data[1]) + memset(outpicref->data[1], 128, outpicref->linesize[1] * (outlink->h >> tinterlace->vsub)); + if (outpicref->data[2]) + memset(outpicref->data[2], 128, outpicref->linesize[2] * (outlink->h >> tinterlace->vsub)); + + /* copy fields */ + copy_picture_fields(outpicref->data, outpicref->linesize, + picref ->data, picref ->linesize, + outlink->format, inlink->w, inlink->h, tinterlace->frame & 1, 0, 0); + tinterlace->picref = outpicref; + tinterlace->frame_pending = 1; + } else if (tinterlace->mode == 4) { + /* Interleave even lines from even frames with odd lines from odd frames. Height unchanged at half framerate. */ + if (!tinterlace->picref) { + outpicref = avfilter_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h); + avfilter_copy_buffer_ref_props(outpicref, picref); + outpicref->video->w = outlink->w; + outpicref->video->h = outlink->h; + outpicref->video->interlaced = 1; + outpicref->video->top_field_first = 1; + + /* write odd frame odd fields into the upper fields of the new frame */ + copy_picture_fields(outpicref->data, outpicref->linesize, + picref ->data, picref ->linesize, + outlink->format, inlink->w, inlink->h, 0, 1, 0); + tinterlace->picref = outpicref; + } else { + outpicref = tinterlace->picref; + + /* write even frame even fields into the lower fields of the new frame */ + copy_picture_fields(outpicref->data, outpicref->linesize, + picref ->data, picref ->linesize, + outlink->format, inlink->w, inlink->h, 1, 1, 1); + tinterlace->frame_pending = 1; + } + } + + tinterlace->frame++; + avfilter_unref_buffer(inlink->cur_buf); +} + +static void null_start_frame(AVFilterLink *inlink, AVFilterBufferRef *picref) { } + +static int request_frame(AVFilterLink *outlink) +{ + AVFilterContext *ctx = outlink->src; + TInterlaceContext *tinterlace = ctx->priv; + + while (!tinterlace->frame_pending) { + int ret; + + if ((ret = avfilter_request_frame(outlink->src->inputs[0]))) + return ret; + } + + avfilter_start_frame(outlink, tinterlace->picref); + avfilter_draw_slice(outlink, 0, outlink->h, 1); + avfilter_end_frame(outlink); + tinterlace->picref = NULL; + tinterlace->frame_pending = 0; + return 0; +} + +static int poll_frame(AVFilterLink *outlink) +{ + TInterlaceContext *tinterlace = outlink->src->priv; + + return tinterlace->frame_pending; +} + +static void null_draw_slice(AVFilterLink *link, int y, int h, int slice_dir) { } + +AVFilter avfilter_vf_tinterlace = { + .name = "tinterlace", + .description = NULL_IF_CONFIG_SMALL("Perform temporal field interlacing."), + + .priv_size = sizeof(TInterlaceContext), + .init = init, + .uninit = uninit, + .query_formats = query_formats, + + .inputs = (AVFilterPad[]) {{ .name = "default", + .type = AVMEDIA_TYPE_VIDEO, + .start_frame = null_start_frame, + .draw_slice = null_draw_slice, + .end_frame = end_frame, }, + { .name = NULL}}, + + .outputs = (AVFilterPad[]) {{ .name = "default", + .type = AVMEDIA_TYPE_VIDEO, + .config_props = config_out_props, + .poll_frame = poll_frame, + .request_frame = request_frame, }, + + { .name = NULL}}, +}; -- 1.7.1

9 25

About H.264 SVC implementation in ffmpeg
by abiaozwz＠gmail.com 08 Dec '11

08 Dec '11

FFmpeg currently does not support SVC? am I right? Who have any information about the SVC implementation in ffmpeg.

5 4

[WIP] life source
by stefano.sabatini-lala＠poste.it 04 Dec '11

04 Dec '11

Hi, proof of concept life source based on Conway's game of life. I'm using the overlay filter to test it. Enjoy. -- FFmpeg = Foolish & Fast Minimalistic Picky Ecletic Guru

7 12

[PATCH] lavf: inspect more frames when the time base is unreliable
by anssi.hannula＠iki.fi 06 Oct '11

06 Oct '11

23.976 fps H.264 Matroska files are usually detected as 24 fps (r_frame_rate) by av_find_stream_info(). Fix that by inspecting the timestamps of 30 frames instead of just 20 for video codecs with unreliable time base. --- This seems the most straight-forward and safe fix for this issue, with the unfortunate cost of the 10 additional frames that have to be demuxed. Another (or additional) option could be to improve the heuristics themselves, if someone more experienced with this timing stuff wants to give it a try (or to provide hints). I guess one such way would be to fall back to codec time base early if it looks accurate based on the first few timestamps, but indeed I have no idea if it would break something else. Input welcome. For the record, here you can see the progression of the cumulative error of the two competing rates when detecting the frame rate: 0.042000 for 24.000000, at 0.042000, error 0.000333, cumulative 0.000000111 for 23.976024, at 0.042000, error 0.000292, cumulative 0.000000085 0.041000 for 24.000000, at 0.041000, error -0.000667, cumulative 0.000000444 for 23.976024, at 0.041000, error -0.000708, cumulative 0.000000502 0.083000 for 24.000000, at 0.083000, error -0.000333, cumulative 0.000000556 for 23.976024, at 0.083000, error -0.000417, cumulative 0.000000675 0.125000 for 24.000000, at 0.125000, error 0.000000, cumulative 0.000000556 for 23.976024, at 0.125000, error -0.000125, cumulative 0.000000691 0.167000 for 24.000000, at 0.167000, error 0.000333, cumulative 0.000000667 for 23.976024, at 0.167000, error 0.000167, cumulative 0.000000719 0.208000 for 24.000000, at 0.208000, error -0.000333, cumulative 0.000000778 for 23.976024, at 0.208000, error -0.000542, cumulative 0.000001012 0.250000 for 24.000000, at 0.250000, error 0.000000, cumulative 0.000000778 for 23.976024, at 0.250000, error -0.000250, cumulative 0.000001075 0.292000 for 24.000000, at 0.292000, error 0.000333, cumulative 0.000000889 for 23.976024, at 0.292000, error 0.000042, cumulative 0.000001076 0.333000 for 24.000000, at 0.333000, error -0.000333, cumulative 0.000001000 for 23.976024, at 0.333000, error -0.000667, cumulative 0.000001521 0.375000 for 24.000000, at 0.375000, error 0.000000, cumulative 0.000001000 for 23.976024, at 0.375000, error -0.000375, cumulative 0.000001661 0.417000 for 24.000000, at 0.417000, error 0.000333, cumulative 0.000001111 for 23.976024, at 0.417000, error -0.000083, cumulative 0.000001668 0.459000 for 24.000000, at 0.459000, error 0.000667, cumulative 0.000001556 for 23.976024, at 0.459000, error 0.000208, cumulative 0.000001712 0.500000 for 24.000000, at 0.500000, error 0.000000, cumulative 0.000001556 for 23.976024, at 0.500000, error -0.000500, cumulative 0.000001962 0.542000 for 24.000000, at 0.542000, error 0.000333, cumulative 0.000001667 for 23.976024, at 0.542000, error -0.000208, cumulative 0.000002005 0.584000 for 24.000000, at 0.584000, error 0.000667, cumulative 0.000002111 for 23.976024, at 0.584000, error 0.000083, cumulative 0.000002012 0.625000 for 24.000000, at 0.625000, error 0.000000, cumulative 0.000002111 for 23.976024, at 0.625000, error -0.000625, cumulative 0.000002403 0.667000 for 24.000000, at 0.667000, error 0.000333, cumulative 0.000002222 for 23.976024, at 0.667000, error -0.000333, cumulative 0.000002514 0.709000 for 24.000000, at 0.709000, error 0.000667, cumulative 0.000002667 for 23.976024, at 0.709000, error -0.000042, cumulative 0.000002516 0.750000 for 24.000000, at 0.750000, error 0.000000, cumulative 0.000002667 for 23.976024, at 0.750000, error -0.000750, cumulative 0.000003078 0.792000 for 24.000000, at 0.792000, error 0.000333, cumulative 0.000002778 for 23.976024, at 0.792000, error -0.000458, cumulative 0.000003288 Without the patch the detection stops at this point. Error for 24.000 is still smaller and it is wrongly selected. Raising the frame count to 30 allows the cumulative error for 24.000 to raise significantly when compared to 23.976, therefore ensuring the correct selection: 0.834000 for 24.000000, at 0.834000, error 0.000667, cumulative 0.000003222 for 23.976024, at 0.834000, error -0.000167, cumulative 0.000003316 0.876000 for 24.000000, at 0.876000, error 0.001000, cumulative 0.000004222 for 23.976024, at 0.876000, error 0.000125, cumulative 0.000003332 0.917000 for 24.000000, at 0.917000, error 0.000333, cumulative 0.000004333 for 23.976024, at 0.917000, error -0.000583, cumulative 0.000003672 0.959000 for 24.000000, at 0.959000, error 0.000667, cumulative 0.000004778 for 23.976024, at 0.959000, error -0.000292, cumulative 0.000003757 1.001000 for 24.000000, at 1.001000, error 0.001000, cumulative 0.000005778 for 23.976024, at 1.001000, error 0.000000, cumulative 0.000003757 1.042000 for 24.000000, at 1.042000, error 0.000333, cumulative 0.000005889 for 23.976024, at 1.042000, error -0.000708, cumulative 0.000004259 1.084000 for 24.000000, at 1.084000, error 0.000667, cumulative 0.000006333 for 23.976024, at 1.084000, error -0.000417, cumulative 0.000004432 1.126000 for 24.000000, at 1.126000, error 0.001000, cumulative 0.000007333 for 23.976024, at 1.126000, error -0.000125, cumulative 0.000004448 1.168000 for 24.000000, at 1.168000, error 0.001333, cumulative 0.000009111 for 23.976024, at 1.168000, error 0.000167, cumulative 0.000004476 1.209000 for 24.000000, at 1.209000, error 0.000667, cumulative 0.000009556 for 23.976024, at 1.209000, error -0.000542, cumulative 0.000004769 libavformat/utils.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/libavformat/utils.c b/libavformat/utils.c index 4f51c26..8227c91 100644 --- a/libavformat/utils.c +++ b/libavformat/utils.c @@ -2257,7 +2257,7 @@ int av_find_stream_info(AVFormatContext *ic) break; /* variable fps and no guess at the real fps */ if( tb_unreliable(st->codec) && !(st->r_frame_rate.num && st->avg_frame_rate.num) - && st->info->duration_count<20 && st->codec->codec_type == AVMEDIA_TYPE_VIDEO) + && st->info->duration_count<30 && st->codec->codec_type == AVMEDIA_TYPE_VIDEO) break; if(st->parser && st->parser->parser->split && !st->codec->extradata) break; -- 1.7.3

9 36

[patch] libpostproc: mmx code uses stack below %esp
by yumkam＠mail.ru 23 Sep '11

23 Sep '11

Hello! While trying to catch unrelated bug (that finally was not bug, but misconfiguration [-lavdopts fast was too fragile for broken mpeg-2 stream]), I've run mplayer under valgrind, and got bunch of warnings: ==32414== ==32414== Invalid write of size 8 ==32414== at 0x874CF44: postProcess_MMX2 (in /path/to/mplayer) ==32414== Address 0xbeffa9d0 is just below the stack ptr. To suppress, use: --workaround-gcc296-bugs=yes I, of course, don't use gcc-2.96 ;-) I've looked into libpostproc/postprocess_template.c, and, indeed, it uses memory below %esp: === cut === static inline void RENAME(doVertDefFilter)(uint8_t src[], int stride, PPContext *c) { [...] __asm__ volatile( "pxor %%mm7, %%mm7 \n\t" "lea -40(%%"REG_SP"), %%"REG_c" \n\t" // make space for 4 8-byte vars "and "ALIGN_MASK", %%"REG_c" \n\t" // align ... } [...] static inline void RENAME(dering)(uint8_t src[], int stride, PPContext *c) [...same...] static av_always_inline void RENAME(do_a_deblock)(uint8_t *src, int step, int stride, PPContext *c){ [...same...] === cut === Not sure if this *must* be fixed, but it feels unsafe, so... Patch attached; doVertDefFilter and do_a_deblock changes should not affect speed, not sure about dering one.

3 9