AVDA: ../../third_party/ffmpeg/libavformat/utils.c:594
AVDA: ../../third_party/ffmpeg/libavformat/mov.c:5697
AVDA: mov_read_default: 17 ftyp
AVDA: mov_read_default: 22 mdat
AVDA: mov_read_default: 22 mdat
AVDA: mov_read_default: 28 moov
AVDA: mov_read_default: 30 mvhd
AVDA: ../../third_party/ffmpeg/libavformat/mov.c:5189 -1094995529
AVDA: ../../third_party/ffmpeg/libavformat/mov.c:5189 -1094995529
AVDA: ../../third_party/ffmpeg/libavformat/mov.c:5719
AVDA: ../../third_party/ffmpeg/libavformat/utils.c:596
../../media/ffmpeg/ffmpeg_regression_tests.cc:402: Failure

ffmpeg got smarter about catching errors.  12 is correct.

commit a398f054fdb9b0f0b5a91c231fba6ce014143f71
Author: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>
Date:   Sun Oct 23 16:38:52 2016 +0200

    mov: validate time_scale
    
    A negative timescale doesn't make sense and triggers assertions in
    av_rescale_rnd.
    
    Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

TODO: update ffmpeg_regression_test MV4_17 (note: not 12 as written elsewhere -- oops!) to expect return code 12.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/acc729512852c66d25b86ebf3533c1531d331983

commit acc729512852c66d25b86ebf3533c1531d331983
Author: liberato <liberato@chromium.org>
Date: Wed Jan 18 18:54:29 2017

Test fixes for M57 ffmpeg roll.

Fixed Cr449958/FFmpegRegressionTest.BasicPlayback/0 .  Now OnOpacityChanged may
be called any number of times.  449958 does really switch opacity twice.

ffmpeg regression tests MP4_17 now fails (correctly) with
DEMUXER_ERROR_COULD_NOT_OPEN, rather than PIPELINE_OK

BUG= 677211 , 677214 

Review-Url: https://codereview.chromium.org/2639333002
Cr-Commit-Position: refs/heads/master@{#444428}

[modify] https://crrev.com/acc729512852c66d25b86ebf3533c1531d331983/media/ffmpeg/ffmpeg_regression_tests.cc
[modify] https://crrev.com/acc729512852c66d25b86ebf3533c1531d331983/media/test/pipeline_integration_test_base.cc