[FFmpeg-devel] [PATCH 1/2] avcodec/h264_ps: Check for truncation at fixed_frame_rate_flag

[Carl Eugen Hoyos]

Am Mo., 16. Dez. 2019 um 01:07 Uhr schrieb James Almer <jamrial at gmail.com>:
>
> On 12/15/2019 9:03 PM, Carl Eugen Hoyos wrote:
> > Am Mo., 16. Dez. 2019 um 01:00 Uhr schrieb James Almer <jamrial at gmail.com>:
> >>
> >> On 12/15/2019 8:50 PM, Carl Eugen Hoyos wrote:
> >>> Am So., 15. Dez. 2019 um 23:25 Uhr schrieb James Almer <jamrial at gmail.com>:
> >>>>
> >>>> On 12/15/2019 7:00 PM, Michael Niedermayer wrote:
> >>>>> Fixes: Ticket7249 (No longer displaying a scary red message)
> >>>>>
> >>>>> Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
> >>>>> ---
> >>>>>  libavcodec/h264_ps.c | 4 ++++
> >>>>>  1 file changed, 4 insertions(+)
> >>>>>
> >>>>> diff --git a/libavcodec/h264_ps.c b/libavcodec/h264_ps.c
> >>>>> index e8738d8502..74f12f8979 100644
> >>>>> --- a/libavcodec/h264_ps.c
> >>>>> +++ b/libavcodec/h264_ps.c
> >>>>> @@ -203,6 +203,10 @@ static inline int decode_vui_parameters(GetBitContext *gb, AVCodecContext *avctx
> >>>>>              sps->num_units_in_tick = num_units_in_tick;
> >>>>>              sps->time_scale = time_scale;
> >>>>>          }
> >>>>> +        if (get_bits_left(gb) == 0) {
> >>>>
> >>>> Doing this check here sounds like it's tailored specifically for the
> >>>> sample in the ticket. What if another file is truncated one bit after
> >>>> this point? Or one bit before? You will get the scary red message for
> >>>> it. Will you also add a check for it?
> >>>> We can't clutter this file with new get_bits_left() checks every time a
> >>>> user shares their broken samples with us just to get a less scary output
> >>>> in their terminals.
> >>>
> >>> Do you know of another proprietary or open-source encoder that encodes
> >>> streams with such a truncation?
> >>
> >> There was an old hevc encoder that did something like that. We ended up
> >> adding a workaround for it as well.
> >
> >> And i can imagine capturing a
> >> transport stream with bad network conditions could result in all kinds
> >> randomly placed truncation.
> >
> > I don't think a "random" truncation as produced by reception or
> > network issues can lead to similar messages as in the ticket.
>
> Can you elaborate why? If a sample stored in some container is truncated
> in the VUI one bit after this point, will it not still print the
> overread error level log message?

But not continuously.

Carl Eugen