[FFmpeg-devel] [PATCHv2] avformat/mp3dec, rmdec: check return value of ffio_ensure_seekback

[Michael Niedermayer]

On Tue, Jan 05, 2016 at 09:51:15PM -0800, Ganesh Ajjanagadde wrote:
> On Tue, Jan 5, 2016 at 4:57 PM, Michael Niedermayer
> <michael at niedermayer.cc> wrote:
> > On Tue, Jan 05, 2016 at 01:57:09PM -0800, Ganesh Ajjanagadde wrote:
> >> On Tue, Jan 5, 2016 at 11:01 AM, wm4 <nfxjfg at googlemail.com> wrote:
> >> > On Tue, 5 Jan 2016 08:32:02 -0800
> >> > Ganesh Ajjanagadde <gajjanag at mit.edu> wrote:
> >> >
> >> >> On Tue, Jan 5, 2016 at 5:29 AM, wm4 <nfxjfg at googlemail.com> wrote:
> >> >> > On Mon,  4 Jan 2016 17:50:01 -0800
> >> >> > Ganesh Ajjanagadde <gajjanagadde at gmail.com> wrote:
> >> >> >
> >> >> >> ffio_ensure_seekback can fail due to e.g ENOMEM. This return value is
> >> >> >> checked here and a diagnostic is logged.
> >> >> >>
> >> >> >> All usage of ffio_ensure_seekback in the codebase now has the return value checked.
> >> >> >>
> >> >> >> Reviewed-by: wm4 <nfxjfg at googlemail.com>
> >> >> >> Reviewed-by: Ronald S. Bultje <rsbultje at gmail.com>
> >> >> >> Reviewed-by: Michael Niedermayer <michael at niedermayer.cc>
> >> >> >> Signed-off-by: Ganesh Ajjanagadde <gajjanagadde at gmail.com>
> >> >> >> ---
> >> >> >>  libavformat/mp3dec.c | 13 +++++++++++--
> >> >> >>  libavformat/rmdec.c  |  3 ++-
> >> >> >>  2 files changed, 13 insertions(+), 3 deletions(-)
> >> >> >>
> >> >> >> diff --git a/libavformat/mp3dec.c b/libavformat/mp3dec.c
> >> >> >> index c76b21e..57ebcc8 100644
> >> >> >> --- a/libavformat/mp3dec.c
> >> >> >> +++ b/libavformat/mp3dec.c
> >> >> >> @@ -372,11 +372,19 @@ static int mp3_read_header(AVFormatContext *s)
> >> >> >>          uint32_t header, header2;
> >> >> >>          int frame_size;
> >> >> >>          if (!(i&1023))
> >> >> >> -            ffio_ensure_seekback(s->pb, i + 1024 + 4);
> >> >> >> +            if ((ret = ffio_ensure_seekback(s->pb, i + 1024 + 4)) < 0) {
> >> >> >> +                av_log(s, AV_LOG_WARNING,
> >> >> >> +                       "initial junk detection and skipping impossible due to: %s\n", av_err2str(ret));
> >> >> >> +                goto skip_fail;
> >> >> >> +            }
> >> >> >>          frame_size = check(s->pb, off + i, &header);
> >> >> >>          if (frame_size > 0) {
> >> >> >>              avio_seek(s->pb, off, SEEK_SET);
> >> >> >> -            ffio_ensure_seekback(s->pb, i + 1024 + frame_size + 4);
> >> >> >> +            if ((ret = ffio_ensure_seekback(s->pb, i + 1024 + frame_size + 4)) < 0) {
> >> >> >> +                av_log(s, AV_LOG_WARNING,
> >> >> >> +                       "initial junk detection and skipping impossible due to: %s\n", av_err2str(ret));
> >> >> >> +                goto skip_fail;
> >> >> >> +            }
> >> >> >>              if (check(s->pb, off + i + frame_size, &header2) >= 0 &&
> >> >> >>                  (header & SAME_HEADER_MASK) == (header2 & SAME_HEADER_MASK))
> >> >> >>              {
> >> >> >> @@ -387,6 +395,7 @@ static int mp3_read_header(AVFormatContext *s)
> >> >> >>          }
> >> >> >>          avio_seek(s->pb, off, SEEK_SET);
> >> >> >>      }
> >> >> >> +skip_fail:
> >> >> >>
> >> >> >>      // the seek index is relative to the end of the xing vbr headers
> >> >> >>      for (i = 0; i < st->nb_index_entries; i++)
> >> >> >> diff --git a/libavformat/rmdec.c b/libavformat/rmdec.c
> >> >> >> index 4e46a3d..470e227 100644
> >> >> >> --- a/libavformat/rmdec.c
> >> >> >> +++ b/libavformat/rmdec.c
> >> >> >> @@ -618,7 +618,8 @@ static int rm_read_header(AVFormatContext *s)
> >> >> >>              size = avio_rb32(pb);
> >> >> >>              codec_pos = avio_tell(pb);
> >> >> >>
> >> >> >> -            ffio_ensure_seekback(pb, 4);
> >> >> >> +            if ((ret = ffio_ensure_seekback(pb, 4)) < 0)
> >> >> >> +                av_log(s, AV_LOG_WARNING, "seeking back impossible due to: %s\n", av_err2str(ret));
> >> >> >>              v = avio_rb32(pb);
> >> >> >>              if (v == MKBETAG('M', 'L', 'T', 'I')) {
> >> >> >>                  ret = rm_read_multi(s, s->pb, st, mime);
> >> >> >
> >> >> > I maintain that this should not be done, because it makes the code
> >> >> > verbose for no reason.
> >> >>
> >> >> Michael has pointed out that when seekback fails, one should really
> >> >> break out of the junk detection loop in mp3dec. Your idea fails to
> >> >> achieve it.
> >> >
> >> > Why? It's not really harmful. You could even argue that this will make
> >> > the common case (skipping jumk and finding a valid frame in a seekable
> >> > file) work in low memory situations, while your patch breaks it.
> >
> > ffio_ensure_seekback() never fails for seekable files
> > it checks s->seekable and return 0 if true before attempting to
> > allocate anything
> >
> >
> >>
> >> I assumed Michael had a good reason for it, but from what you say here
> >> and my examination just now, I would at a first glance agree with you.
> >> Ronald had concerns about repetition of messages, so if he (and
> >> Michael) are fine with logging within ffio_ensure_seekback, then I
> >> will log it there.
> >
> > as the code is written currently, if ffio_ensure_seekback() fails
> > and that being ignored, the url_seek() to return to the begin will
> > fail too quite likely while forward seeks will succeed.
> > this would result in random amounts of data to be lost at the begin
> > of the file
> > it may print errors or warnings if such are added but nothing (like
> > the applications return code) would indicate a problem, the user though
> > could end with a output file that is missing the first few seconds of
> > audio.
> >
> > if we break out on a ffio_ensure_seekback() failure then problems
> > might still occur on files that have junk at te begin but files that
> > do not have junk (which is what a valid file should look like)
> > should work fine
> 
> Maybe, but many mp3's streamed over http have junk at the beginning,
> so such broken files are not rare.

what kind of junk is that ?
the mp3 code should resync on its own very quickly if its started
from a wrong byte offset
OTOH failure to seekback would loose several seconds of audio i
suspect

do you have an example file where (if the backward url_seek is
commented out the) the results are better with the loop executed
than without ?


[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Democracy is the form of government in which you can choose your dictator
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20160106/d07b55ed/attachment.sig>