Bug 78827 - [uxa hsw] xv video corrupted
Summary: [uxa hsw] xv video corrupted
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: unspecified
Hardware: Other All
: medium normal
Assignee: Chris Wilson
QA Contact: Intel GFX Bugs mailing list
Depends on:
Reported: 2014-05-17 12:58 UTC by Tom Horsley
Modified: 2014-06-05 10:03 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:

photo of corrupted mplayer window (437.23 KB, image/jpeg)
2014-05-17 17:15 UTC, Tom Horsley
no flags Details
Xorg.0.log (36.74 KB, text/plain)
2014-05-17 17:17 UTC, Tom Horsley
no flags Details
Xorg.0.log with SNA enabled (22.88 KB, text/plain)
2014-05-17 22:35 UTC, Tom Horsley
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Tom Horsley 2014-05-17 12:58:08 UTC
I see several bugs against various linux distros about the video corruption problem that appeared on intel graphics when the 3.14 kernel arrived, but I don't see an upstream bug here, so I'm making this one. Details of the problem can be found in various other bug reports:




Only seems to affect "xv" video, gl rendering is OK (and a good work around), but xv seems to be the default for most players, so it gets noticed.
Comment 1 Chris Wilson 2014-05-17 16:09:31 UTC
Please attach your Xorg.0.log.
Comment 2 Chris Wilson 2014-05-17 16:09:46 UTC
And a screenshot or photograph.
Comment 3 Tom Horsley 2014-05-17 17:15:35 UTC
Created attachment 99223 [details]
photo of corrupted mplayer window

If I take a screenshot, the image provided in the screenshot is not corrupt (which probably says something about where the problem is). This is a photo I took of the screen while playing the video for the Doctor Who episode "Blink" with mplayer using the -vo xv option.

Horizontal bands seem to be leftover from old frames trying to catch up. In fact, if I pause the video, in a few seconds they will manage to catch up and everything will look normal till I start playing again.
Comment 4 Tom Horsley 2014-05-17 17:17:14 UTC
Created attachment 99224 [details]

There are other copies of the log and dmesg and wot-not in the bugzillas pointed at. Perhaps they have something in common.
Comment 5 Chris Wilson 2014-05-17 19:39:03 UTC
You are all using hsw + uxa. UXA Xv uses textured video, which should mean that there is no discrepancy between the GetImage screenshot and the image.

Are you using any form of compositing?

Have you tried a more recent driver with the switch to SNA? (That at least would provide us with more diagnostics.)
Comment 6 Tom Horsley 2014-05-17 22:20:46 UTC
I don't know what the ksnapshot tool uses to take a screenshot, but it definitely shows an image without the banding (using gimp for screenshot has the same results - no banding).

I'm running a plain fvwm session with no compositing or any kind of fancy eye candy.

I see the debian bug says that adding an xorg.conf.d file to force SNA did fix the problem. I haven't tried that yet.
Comment 7 Tom Horsley 2014-05-17 22:35:47 UTC
Created attachment 99253 [details]
Xorg.0.log with SNA enabled

OK I created the file /etc/X11/xorg.conf.d/intel.conf with this content:

Section "Device"
   Identifier  "Intel Graphics"
   Driver      "intel"
   Option "AccelMethod" "sna"

Then I rebooted and I can indeed play video now with mplayer -vo xv without the banding effect. Attached is the new Xorg.0.log file from booting with intel.conf.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct.