[FFmpeg-trac] #7945(undetermined:reopened): regression in interplay MVE decoding

FFmpeg trac at avcodec.org
Thu Oct 24 14:23:21 EEST 2019

#7945: regression in interplay MVE decoding
             Reporter:  criezy       |                    Owner:
                 Type:  defect       |                   Status:  reopened
             Priority:  important    |                Component:
                                     |  undetermined
              Version:  git-master   |               Resolution:
             Keywords:  regression   |               Blocked By:
             Blocking:               |  Reproduced by developer:  0
Analyzed by developer:  0            |

Comment (by criezy):

 I am not sure how you define valid, but this file is present in both the
 demo and full game version of the Interplay game Kingdom: The Far Reaches.
 I have also downloaded the file from this ticket just now to check that it
 had not been corrupted by the upload and I can confirm it is still
 identical to the original file.

 This file is properly displayed in the game, and I can also play it
 without issue using ffplay from commit e2218ed8ce. The next commit however
 (8f86e66, avcodec: add stride alignment needed for AVX-512) breaks
 playback of this file. The same happens with may other files from that

 I have uploaded some of those to my dropbox:

 It may be worth noting that all these videos have in common that support
 for them was added through a series of patches by hpvp in June and July
 2017. You can see the patches there:
 So playback for those only worked for a few months until 8f86e66 came
 along in November 2017.

 Note that when switching from e2218ed8ce to 8f86e66 a simple recompile is
 not sufficient and I need to clean and run configure again for the issue
 to appear. If I just recompile without running configure then the video
 still play properly in 8f86e66.

Ticket URL: <https://trac.ffmpeg.org/ticket/7945#comment:5>
FFmpeg <https://ffmpeg.org>
FFmpeg issue tracker

More information about the FFmpeg-trac mailing list