[FFmpeg-devel] [PATCH 1/2] avformat/nutenc: fix duration estimation by writing EOR packets
Michael Niedermayer
michael at niedermayer.cc
Wed Dec 19 17:53:42 EET 2018
On Tue, Dec 18, 2018 at 10:55:16PM +0100, Paul B Mahol wrote:
> On 12/18/18, Michael Niedermayer <michael at niedermayer.cc> wrote:
> > On Tue, Dec 18, 2018 at 02:56:54PM +0100, Paul B Mahol wrote:
> >> Signed-off-by: Paul B Mahol <onemda at gmail.com>
> >> ---
> >> libavformat/nut.h | 3 +
> >> libavformat/nutenc.c | 130 +++++++++++++++++++++++++++++++++++++++++--
> >> 2 files changed, 129 insertions(+), 4 deletions(-)
> >>
> >> diff --git a/libavformat/nut.h b/libavformat/nut.h
> >> index a4409ee23d..4b15dca1ea 100644
> >> --- a/libavformat/nut.h
> >> +++ b/libavformat/nut.h
> >> @@ -76,6 +76,9 @@ typedef struct StreamContext {
> >> int last_flags;
> >> int skip_until_key_frame;
> >> int64_t last_pts;
> >> + int64_t eor_pts;
> >> + int64_t eor_dts;
> >> + int eor_flushed;
> >> int time_base_id;
> >> AVRational *time_base;
> >> int msb_pts_shift;
> >> diff --git a/libavformat/nutenc.c b/libavformat/nutenc.c
> >> index e9a3bb49db..9d1c540a9d 100644
> >> --- a/libavformat/nutenc.c
> >> +++ b/libavformat/nutenc.c
> >> @@ -940,6 +940,80 @@ fail:
> >> return ret;
> >> }
> >>
> >> +static void nut_update_max_pts(AVFormatContext *s, int stream_index,
> >> int64_t pts)
> >> +{
> >> + NUTContext *nut = s->priv_data;
> >> + StreamContext *nus = &nut->stream[stream_index];
> >> +
> >> + if (!nut->max_pts_tb || av_compare_ts(nut->max_pts, *nut->max_pts_tb,
> >> pts, *nus->time_base) < 0) {
> >> + nut->max_pts = pts;
> >> + nut->max_pts_tb = nus->time_base;
> >> + }
> >> +}
> >> +
> >> +static int nut_write_eor(AVFormatContext *s, AVPacket *pkt)
> >> +{
> >> + AVIOContext *bc = s->pb, *dyn_bc;
> >> + NUTContext *nut = s->priv_data;
> >> + StreamContext *nus = &nut->stream[pkt->stream_index];
> >> + int ret, frame_code, flags, needed_flags;
> >> + int64_t pts = pkt->pts;
> >> + int64_t coded_pts;
> >> + FrameCode *fc;
> >> +
> >> + coded_pts = pts & ((1 << nus->msb_pts_shift) - 1);
> >> + if (ff_lsb2full(nus, coded_pts) != pts)
> >> + coded_pts = pts + (1 << nus->msb_pts_shift);
> >> +
> >> + nut->last_syncpoint_pos = avio_tell(bc);
> >> + ret = avio_open_dyn_buf(&dyn_bc);
> >> + if (ret < 0)
> >> + return ret;
> >> + put_tt(nut, nus->time_base, dyn_bc, pkt->dts);
> >> + ff_put_v(dyn_bc, 0);
> >> +
> >> + if (nut->flags & NUT_BROADCAST) {
> >> + put_tt(nut, nus->time_base, dyn_bc,
> >> + av_rescale_q(av_gettime(), AV_TIME_BASE_Q,
> >> *nus->time_base));
> >> + }
> >> + put_packet(nut, bc, dyn_bc, 1, SYNCPOINT_STARTCODE);
> >> +
> >> + frame_code = -1;
> >> + for (int i = 0; i < 256; i++) {
> >> + FrameCode *fc = &nut->frame_code[i];
> >> + int flags = fc->flags;
> >> +
> >> + if (flags & FLAG_INVALID)
> >> + continue;
> >> +
> >> + if (!(flags & FLAG_CODED)) {
> >> + continue;
> >> + }
> >> +
> >> + frame_code = i;
> >> + break;
> >> + }
> >> + av_assert0(frame_code != -1);
> >> +
> >> + fc = &nut->frame_code[frame_code];
> >> + flags = fc->flags;
> >> + needed_flags = FLAG_KEY | FLAG_EOR | FLAG_CODED_PTS |
> >> FLAG_STREAM_ID;
> >> +
> >> + ffio_init_checksum(bc, ff_crc04C11DB7_update, 0);
> >> + avio_w8(bc, frame_code);
> >> + if (flags & FLAG_CODED) {
> >> + ff_put_v(bc, (flags ^ needed_flags) & ~(FLAG_CODED));
> >> + flags = needed_flags;
> >> + }
> >> + if (flags & FLAG_STREAM_ID) ff_put_v(bc, pkt->stream_index);
> >> + if (flags & FLAG_CODED_PTS) ff_put_v(bc, coded_pts);
> >> + ffio_get_checksum(bc);
> >> +
> >> + nut_update_max_pts(s, pkt->stream_index, pkt->pts);
> >> +
> >> + return 0;
> >> +}
> >> +
> >> static int nut_write_packet(AVFormatContext *s, AVPacket *pkt)
> >> {
> >> NUTContext *nut = s->priv_data;
> >
> >> @@ -956,6 +1030,9 @@ static int nut_write_packet(AVFormatContext *s,
> >> AVPacket *pkt)
> >> int data_size = pkt->size;
> >> uint8_t *sm_buf = NULL;
> >>
> >> + if (!data_size)
> >> + return nut_write_eor(s, pkt);
> >
> > that could happen for non EOR packets too
>
> How would you solve it?
the flags of the packet should be the easiest way to mark EOR packets
>
> >
> >
> >> +
> >> if (pkt->pts < 0) {
> >> av_log(s, AV_LOG_ERROR,
> >> "Negative pts not supported stream %d, pts %"PRId64"\n",
> >> @@ -1150,10 +1227,7 @@ static int nut_write_packet(AVFormatContext *s,
> >> AVPacket *pkt)
> >> nus->keyframe_pts[nut->sp_count] = pkt->pts;
> >> }
> >>
> >> - if (!nut->max_pts_tb || av_compare_ts(nut->max_pts, *nut->max_pts_tb,
> >> pkt->pts, *nus->time_base) < 0) {
> >> - nut->max_pts = pkt->pts;
> >> - nut->max_pts_tb = nus->time_base;
> >> - }
> >> + nut_update_max_pts(s, pkt->stream_index, pkt->pts);
> >>
> >> fail:
> >> av_freep(&sm_buf);
> >> @@ -1161,6 +1235,53 @@ fail:
> >> return ret;
> >> }
> >>
> >> +static int nut_interleave_packet(AVFormatContext *s, AVPacket *out,
> >> + AVPacket *pkt, int flush)
> >> +{
> >> + NUTContext *nut = s->priv_data;
> >> + int ret;
> >> +
> >> + if (pkt) {
> >> + StreamContext *nus = &nut->stream[pkt->stream_index];
> >> +
> >> + nus->eor_pts = FFMAX(pkt->pts + pkt->duration, nus->eor_pts);
> >> + nus->eor_dts = FFMAX(pkt->dts + pkt->duration, nus->eor_dts);
> >> + } else if (flush) {
> >> + StreamContext *nus;
> >> + int i;
> >> +
> >> + for (i = 0; i < s->nb_streams; i++) {
> >> + nus = &nut->stream[i];
> >> +
> >> + if (nus->eor_flushed)
> >> + continue;
> >> +
> >> + if (s->streams[i]->last_in_packet_buffer)
> >> + continue;
> >> +
> >> + break;
> >> + }
> >> +
> >> + if (i < s->nb_streams) {
> >> + pkt = av_packet_alloc();
> >> + if (!pkt)
> >> + return AVERROR(ENOMEM);
> >> +
> >> + ret = av_new_packet(pkt, 0);
> >> + if (ret < 0)
> >> + return ret;
> >> + pkt->stream_index = i;
> >> + pkt->pts = nus->eor_pts;
> >> + pkt->dts = nus->eor_dts;
> >> + pkt->duration = 0;
> >> + nus->eor_flushed = 1;
> >
> >> + ret = ff_interleave_packet_per_dts(s, out, pkt, 0);
> >> + pkt = NULL;
> >> + }
> >> + }
> >> + return ff_interleave_packet_per_dts(s, out, pkt, flush);
> >
> > this looks a bit odd, shouldnt there be a return after the first call
> > to ff_interleave_packet_per_dts() ?
> > as it would assign the ret variable but then ignore it, would
> > potentially output something in "out" but then overwrite it
> > in the 2nd call
>
> So what you propose?
to add a return in there, so all the EOR get added and it then resumes
with the flushing but doing one per call
or does this not work ?
thanks
[...]
--
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
If you think the mosad wants you dead since a long time then you are either
wrong or dead since a long time.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20181219/7a4780c0/attachment.sig>
More information about the ffmpeg-devel
mailing list