Bug 38964 - [GT2 2D] Video rendering freezes in certain apps after C3 suspend-resume
Summary: [GT2 2D] Video rendering freezes in certain apps after C3 suspend-resume
Status: RESOLVED DUPLICATE of bug 33487
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: unspecified
Hardware: Other All
: medium normal
Assignee: Chris Wilson
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-07-04 23:15 UTC by dave
Modified: 2011-07-05 11:56 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
dmesg output (123.07 KB, text/plain)
2011-07-04 23:15 UTC, dave
no flags Details
Xorg log (28.65 KB, text/plain)
2011-07-04 23:16 UTC, dave
no flags Details
xrandr --verbose output (4.54 KB, text/plain)
2011-07-04 23:17 UTC, dave
no flags Details
VBIOS dump (64.00 KB, application/octet-stream)
2011-07-04 23:18 UTC, dave
no flags Details
intel_reg_dumper output (11.06 KB, text/plain)
2011-07-04 23:19 UTC, dave
no flags Details

Description dave 2011-07-04 23:15:16 UTC
Created attachment 48755 [details]
dmesg output

Note that I am not using Compiz or other accelerated/fancy window managers, just an old-fashioned xmonad.

The issue is not a hard lockup of the entire machine, rather frequent multi-second lockups of video output in certain kind of 2d apps. I've noted this so far while trying to play videos with Adobe Flash Player in a browser, and the game Battle for Wesnoth.

Upon starting these programs, video rendering is normal and smooth for 1-2s, then video rendering freezes. The rest of the system works normally (audio, cpu work, disk I/O...), and video outside the windows of these apps continues to render normally. 3-10s, the frozen app resumes rendering, with the expected content, for another 1-2s, and the cycle repeats.

While investigating, I upgraded to the xorg-edgers Ubuntu PPA (those are the versions I'm reporting below). With those X drivers, the lockup bug does not seem to occur after a cold boot or reboot, but appears after the first C3 (ram) suspend-resume cycle, and persists until the next full reboot. With the stable Ubuntu drivers, this behavior starts even before the first suspend cycle.

Environment:
- Chipset: Intel(R) Sandybridge Mobile (GT2)
- Architecture: i686
- libdrm version: 2.4.26
- mesa version: 2.1 Mesa 7.10.2
- intel driver version: 2.15.0
- xorg version: X.Org X Server 1.10.2.901 (1.10.3 RC 1)
- kernel version: 2.6.38-8-generic
- Distribution: Ubuntu 11.04
- Machine: Lenovo Thinkpad X220 i5
- Display: LVDS

Reproduction:
- On the above hardware, install Ubuntu 11.04. Install the xorg-edgers PPA to get bleeding edge video drivers.
- Reboot.
- Run Battle for Wesnoth. Muck around in the UI for ~1min, notice rendering does not freeze.
- C3 suspend and resume.
- Run Battle for Wesnoth. Muck around in the UI, notice rendering is now choppy and freezy. Notice the surrounding stuff doesn't seem affected.
- Reboot the machine.
- Run Battle for Wesnoth. Notice rendering works again.

All logs attached per the guide
Comment 1 dave 2011-07-04 23:16:10 UTC
Created attachment 48756 [details]
Xorg log
Comment 2 dave 2011-07-04 23:17:01 UTC
Created attachment 48757 [details]
xrandr --verbose output
Comment 3 dave 2011-07-04 23:18:13 UTC
Created attachment 48758 [details]
VBIOS dump
Comment 4 dave 2011-07-04 23:19:36 UTC
Created attachment 48759 [details]
intel_reg_dumper output
Comment 5 Chris Wilson 2011-07-05 00:23:47 UTC
From the description and lack of the usual hangchecks in dmesg, I suspect this is a FBC issue. Can you please try a 3.0 kernel which has FBC disabled by default? Or set i915.enable_fbc=0 (though I think that only came in during 2.6.39+).
Comment 6 dave 2011-07-05 11:18:17 UTC
Updated to 3.0-rc6 from the kernel PPA. With the combination of 3.0-rc6 and bleeding edge Xorg drivers, I'm unable to reproduce the bug.

While I have my X220 in a configuration amenable to sacrifice, is there any further data I can collect (in any configuration of kernel+X drivers) to help development?
Comment 7 Chris Wilson 2011-07-05 11:20:38 UTC
About the only thing useful to check is whether the bug comes back with i915.enable_fbc=1 (append to your kernel boot parameters in grub).
Comment 8 dave 2011-07-05 11:33:38 UTC
With my kernel version, the argument to grub is i915.i915_enable_fbc=1 (otherwise the driver fails to modprobe). Booting with FBC enabled brings back the bug.
Comment 9 Chris Wilson 2011-07-05 11:56:40 UTC

*** This bug has been marked as a duplicate of bug 33487 ***


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.