[FFmpeg-trac] #764(avcodec:new): Increasing robustness of runlength decoding for scantable access in mpeg12.c
FFmpeg
trac at avcodec.org
Sat Dec 17 16:47:05 CET 2011
#764: Increasing robustness of runlength decoding for scantable access in
mpeg12.c
-------------------------------------+-------------------------------------
Reporter: erik | Owner:
Type: defect | Status: new
Priority: normal | Component: avcodec
Version: unspecified | Resolution:
Keywords: mpeg | Blocked By:
robustness | Reproduced by developer: 0
Blocking: |
Analyzed by developer: 0 |
-------------------------------------+-------------------------------------
Comment (by reimar):
Do you have any numbers on how much "fast" helps? MPEG-2 decoding should
be really cheap by today's standards and I wouldn't expect it to really
help much.
Worse, with errors as in your case it will even process a large amount of
data for no purpose.
My recommendation would be
1) try without fast flag. You get error concealment and it might not have
a relevant cost. (note: only for decoding, if you are also encoding you
should keep it for that part).
2) Should it be relevantly slower, try leaving the fast flag in but use
the non-fast functions for this case.
3) Add the > 63 check as in the other cases instead. It should be only
little to not at all slower than your change for the normal decoding case
and it should be _much_ faster for the error case (if you reached values
like 7000 that would indicate that you looped around that loop about 100
times more than necessary).
And I don't want you to apologize, and I don't want you to give up because
I sometimes get a bit annoyed. But it is a really good idea for you to be
aware of any non-default configuration you have chosen and to mention it
in bug reports. And also to make sure you have a solid justification for
it, since they might cause you a significant maintenance cost in the long
term. Apart from that there clearly is a lack of documentation we are to
blame for.
--
Ticket URL: <https://ffmpeg.org/trac/ffmpeg/ticket/764#comment:15>
FFmpeg <http://ffmpeg.org>
FFmpeg issue tracker
More information about the FFmpeg-trac
mailing list