WineHQ
Bug Tracking Database – Bug 28664

 Bugzilla

 

Last modified: 2013-11-15 13:40:43 UTC  

qedit/mediadet test crashes on FreeBSD9

Bug 28664 - qedit/mediadet test crashes on FreeBSD9
qedit/mediadet test crashes on FreeBSD9
Status: CLOSED FIXED
AppDB: Show Apps affected by this bug
Product: Wine
Classification: Unclassified
Component: winegstreamer
1.3.30
x86 Linux
: P2 normal
: ---
Assigned To: Mr. Bugs
: download, patch, source, testcase
Depends on:
Blocks:
  Show dependency tree
 
Reported: 2011-10-11 19:13 UTC by Austin English
Modified: 2013-11-15 13:40 UTC (History)
2 users (show)

See Also:
Regression SHA1:
Fixed by SHA1: deff27e23ee9dca1791501c3ba31a0e6cf042c64
Distribution: ---
Staged patchset:


Attachments
backtrace (10.65 KB, text/plain)
2011-10-11 19:13 UTC, Austin English
Details
Output from failed qedit:mediadet test on Mac OS (+relay) (299.55 KB, application/x-bzip2)
2013-07-09 20:46 UTC, Charles Davis
Details
Output from failed qedit:mediadet test on Mac OS (+strmbase,+quartz,+qtsplitter) (8.14 KB, application/x-bzip2)
2013-07-13 19:43 UTC, Charles Davis
Details
possible fix (1.06 KB, patch)
2013-07-15 10:10 UTC, Aric Stewart
Details | Diff
Corrected fix (1.10 KB, patch)
2013-07-15 10:23 UTC, Aric Stewart
Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Austin English 2011-10-11 19:13:46 UTC
Created attachment 36837 [details]
backtrace

The test was running without $DISPLAY set, but that passes fine for me on gentoo/debian.

Backtrace attached.
Comment 1 Charles Davis 2011-10-17 21:13:11 UTC
It also crashes on Mac with the same stack trace.
Comment 2 Austin English 2013-07-09 14:40:47 UTC
This is okay for me in PC-BSD9 (i386) and 1.6-rc4. It also passes in wine-1.2. I tested both versions with/without DISPLAY set.

@Charles, is it broken on OSX?
Comment 3 Charles Davis 2013-07-09 20:41:16 UTC
(In reply to comment #2)
> This is okay for me in PC-BSD9 (i386) and 1.6-rc4. It also passes in wine-1.2.
> I tested both versions with/without DISPLAY set.
> 
> @Charles, is it broken on OSX?
Well, it doesn't crash anymore, but it fails with:

mediadet.c:281: Test failed: IMediaDet_get_FrameRate: fps is inf

I haven't really looked at this. It might take me a while to come up with a fix. From the looks of this, we're dividing by zero on Mac--probably because the QuickTime component isn't even returning a valid AM_MEDIA_TYPE to begin with (cf. dlls/qedit/mediadet.c:605, the division in question). From what I've seen, it *should* be returning one, but it's not, and I can't figure out why at the moment.

I've CC'd Aric, and I'll attach a file containing the complete output from the test, including +relay output. (Or should I file a new bug?) In any case, maybe he can help me figure this out.
Comment 4 Charles Davis 2013-07-09 20:46:40 UTC
Created attachment 45178 [details]
Output from failed qedit:mediadet test on Mac OS (+relay)
Comment 5 Aric Stewart 2013-07-12 08:34:16 UTC
Could you generate a +strmbase,+quartz,+qtsplitter log?

On my mac I am seeing a crash in FilterGraph2_AddFilter
Comment 6 Charles Davis 2013-07-13 19:43:13 UTC
Created attachment 45257 [details]
Output from failed qedit:mediadet test on Mac OS (+strmbase,+quartz,+qtsplitter)

Done.

You're seeing a crash? Something must be different about my setup. I have Flip4Mac installed, which lets QT understand Windows Media formats. Could that account for the test not crashing (though still failing) on my system?
Comment 7 Aric Stewart 2013-07-15 10:01:47 UTC
(In reply to comment #6)
> Created attachment 45257 [details]
> Output from failed qedit:mediadet test on Mac OS
> (+strmbase,+quartz,+qtsplitter)
> 
> Done.
> 
> You're seeing a crash? Something must be different about my setup. I have
> Flip4Mac installed, which lets QT understand Windows Media formats. Could that
> account for the test not crashing (though still failing) on my system?

Ok that is likely.  My qtdecoder reports that it cannot handle the format and so I fall back to "ffdshow Video Decoder" which crashes when adding the filter.

Looking at the qtdecoder you are correct: AvgTimePerFrame is unset.

I can develop a patch to correct that for the qtdecoder without too much problem. The trick is is having a AvgTimePerFrame of 0 an error or is the way that qedit does that test incorrect? 

-aric
Comment 8 Aric Stewart 2013-07-15 10:10:49 UTC
Created attachment 45276 [details]
possible fix

Could you try this patch. It should properly set the AvgTimePerFrame for video streams using the qtdecoder.
Comment 9 Aric Stewart 2013-07-15 10:23:12 UTC
Created attachment 45277 [details]
Corrected fix

I found if i ran in a clean wine prefix I could get proper test behavior.

I found problems with the old fix.  This one appears to be correct.
Comment 10 Charles Davis 2013-07-15 13:30:28 UTC
(In reply to comment #8)
> Created attachment 45276 [details]
> possible fix
> 
> Could you try this patch. It should properly set the AvgTimePerFrame for video
> streams using the qtdecoder.
That fixes it, alright. Now there are no test failures.
Comment 12 Austin English 2013-07-29 18:45:18 UTC
Fixed for Mac and FreeBSD.
Comment 13 Alexandre Julliard 2013-08-02 13:19:12 UTC
Closing bugs fixed in 1.7.0.
Comment 14 Alexandre Julliard 2013-11-15 13:40:43 UTC
Removing 1.6.x milestone from bugs included in 1.6.1.


Privacy Policy
If you have a privacy inquiry regarding this site, please write to [email protected]

Hosted By CodeWeavers