Screen corruption when changing focus xserver-xorg-video-ati

[Jeremy Wilkins]

Binary package hint: xserver-xorg-video-ati

There is an error where the display will randomly corrupt to a negative like image when switching focus between windows. Switching the focus to another window will temporarily fix the problem, switching the focus back initiates the problem again. This problem increases in frequency the more windows that are opened. Opening glxgears seems to initate the problem faster, but the problem happens less often when running in 3D compositing mode. Running on Karmic with all updates installed.

ProblemType: Bug
Architecture: amd64
Date: Wed Jul 29 23:18:18 2009
DistroRelease: Ubuntu 9.10
Lsusb:
 Bus 002 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
 Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
MachineType: Gateway MX6433
Package: xserver-xorg-video-ati 1:6.12.99+git20090629.f39cafc5-0ubuntu1
ProcCmdLine: root=UUID=af17c779-4b6b-407a-a2fe-5ffc00d6b536 ro nomodeset quiet splash
ProcEnviron:
 SHELL=/bin/bash
 PATH=(custom, user)
 LANG=en_US
 LANGUAGE=en_US:en
ProcVersionSignature: Ubuntu 2.6.31-4.23-generic
RelatedPackageVersions:
 xserver-xorg 1:7.4+3ubuntu5
 libgl1-mesa-glx 7.5-1ubuntu1
 libdrm2 2.4.12-1ubuntu1
 xserver-xorg-video-intel 2:2.8.0-0ubuntu1
 xserver-xorg-video-ati 1:6.12.99+git20090629.f39cafc5-0ubuntu1
SourcePackage: xserver-xorg-video-ati
Uname: Linux 2.6.31-4-generic x86_64
dmi.bios.date: 04/16/07
dmi.bios.vendor: Phoenix Technologies LTD
dmi.bios.version: 68.05
dmi.board.name: MX6433
dmi.board.vendor: Gateway
dmi.board.version: 68.04
dmi.chassis.asset.tag: No Asset Tag
dmi.chassis.type: 8
dmi.chassis.vendor: Gateway
dmi.chassis.version: Rev.1
dmi.modalias: dmi:bvnPhoenixTechnologiesLTD:bvr68.05:bd04/16/07:svnGateway:pnMX6433:pvr1MA30001044:rvnGateway:rnMX6433:rvr68.04:cvnGateway:ct8:cvrRev.1:
dmi.product.name: MX6433
dmi.product.version: 1MA30001044
dmi.sys.vendor: Gateway
fglrx: Not loaded
system:
 distro: Ubuntu
 architecture: x86_64kernel: 2.6.31-4-generic

[Jeremy Wilkins (wjeremy)]

I am not sure if this is also related to this bug, but some programs that draw upon a window surface like freeciv also seem to have troubles refreshing their content which ends up with a blank section or completely blank window. This makes freeciv unplayable.

[Bryce Harrington (bryce)]

Hi Jeremy,

I think I'll send this one upstream, and you've provided most of the information required, but could you also take a photo or two of the screen when it's corrupted? Often that gives upstream clues as to where in graphics memory the error is occurring.

[Jeremy Wilkins (wjeremy)]

Ok, I will give a couple screen shots. One of the corruption and another of the failure of screen refreshes unless you think I should file them separately.

[Jeremy Wilkins (wjeremy)]

Hmm, the framebuffer must not be corrupted, because the screen shot with ksnapshot shows everything perfectly with the correct colors. This leads me to think it is a display overlay or something of the sort. I will take a camera shot of my laptop display for an example instead.

[Jeremy Wilkins (wjeremy)]

Here is the picture of what freeciv looks like. This was taken by ksnapshot since this is a rendering problem.

[Jeremy Wilkins (wjeremy)]

This one was taken by ksnapshot (framebuffer grab) while the screen corruption was occuring, but as you see it looks like a normal screen. The program shown here is virtualbox as it consistently exposes this bug once you open a vm in running state. There are a few other programs that do that as well. glxgears can also initiate it I have noticed, so it is most likely 3D overlay related,

[Jeremy Wilkins (wjeremy)]

This one is taken from my camera phone (my actual camera's battery was dead), since I know of no other way to get this image.

[Jeremy Wilkins (wjeremy)]

Bryce: Since the latest libdrm update today that pulled in the new libradeon library I have not experienced the inverted display problem anymore. Prior to that update the 3.0.4 update of virtualbox no longer initiated the problem, though glxgears was still causing it until today. glxgears still causes another problem: the screen goes black until exited. This may be a side effect of this bug. It is hard to say. I'm not saying the problem is gone, just that I can no longer reproduce it.

[Jeremy Wilkins (wjeremy)]

Ok, another update. As I suspected, the problem still isn't gone, it's just a lot harder to reproduce than before. In order to reproduce it, it required a lot of tinkering with programs like virtualbox and glxgears, etc.

[Bryce Harrington (bryce)]

[This is an automatic notification.]

A new version of the -ati driver is now available in Karmic.

This is a significant update to -ati which brings in kernel mode-setting
(currently disabled) and scores of fixes for DRI2, EXA, etc.

I've posted the new version of this driver to the following PPA,
would you mind testing it and seeing if it resolves the bug you
reported?

  https://edge.launchpad.net/~bryceharrington/+archive/ppa/+sourcepub/709908/+listing-archive-extra

If you're not running this release of Ubuntu, you can try booting the Karmic
LiveCD and loading the PPA onto it, and then log out/in to restart X.
ISOs are available at http://cdimages.ubuntu.com/releases/

After testing Karmic, report back here whether it's still an issue or not,
and if it is please post a fresh Xorg.0.log and 'dmesg' output.

Note there could be new bugs... please file these as new reports using
the command 'ubuntu-bug xorg'.

[Jeremy Wilkins (wjeremy)]

Ok, this driver is amazing. All I can say is WOW! And it takes a LOT to impress me.

I did have it crash the desktop once while loading virtualbox minimized and unminimizing it, but that may have been virtualbox that did that since I think that may have happened with the old driver too.

Everything feels 40% faster, and even glxgears shows a higher fps (I know that doesn't mean much, but it does correlate). glxgears doesn't cause the black blanking effect it did before, and I haven't been able to reproduce the bug mentioned above. It appears that freeciv still suffers from poor redraw, but I haven't tested it on a different graphics driver to see if the driver is really the cause. glxgears shows impressive graphics management where it runs at 30% cpu while showing in the foreground at 3000+ fps I might add (the old driver barely made 2200), but idles to 4% when obscured by other windows. There is a negative side effect to this reduction in that the fps results goes down, but if a window is obscured, that behaviour is acceptable to me since the window is not visible anyways and it frees cpu resources for other processes.

I also required nomodeset on the kernel line before to boot into a graphical environment without crash. Now I can remove it without harm. All in all, I vote to move this driver into Karmic as the main driver for release. Of course, more testing is needed to ensure everyone has as positive a result as I am having.

[Jeremy Wilkins (wjeremy)]

I am noticing more and more opengl bugs I ran into before which are also fixed with this driver. There remains only one left, but I don't think it will ever be resolved. epsxe runs opengl fullscreen in a semi windowed mode and it flickers badly when desktop effects are on, but suspending Desktop Effects is sufficient to resolve it. There was an overlay bug previously with epsxe which required me to overlay a window on it to see the opengl rendered screen clearly. This is also resolved.

[Bryce Harrington (bryce)]

Jeremy, good to hear how well its working. Thanks for indicating that a lot of OpenGL bugs get resolved, perhaps this will fix some of the OpenGL issues people have reported previously.

[Launchpad Janitor (janitor)]

This bug was fixed in the package xserver-xorg-video-ati - 1:6.12.99+git20090825.fc74e119-0ubuntu1

---------------
xserver-xorg-video-ati (1:6.12.99+git20090825.fc74e119-0ubuntu1) karmic; urgency=low

  * Checkout from git 20090825 (master branch) up to commit
    fc74e1194c980d978667e02c60a29a761a694bde
    + Adds DRI2 / redirected direct rendering
      (LP: #96991)
    + Fix freeze on opengl games (bzflag, alienarena) (when KMS on)
      (LP: #348450)
    + Fix intermittent short screen blanking behavior
      (LP: #310864)
    + Fix faulty wine screen updates with EXA Compositing
      (LP: #314205)
    + Fix screen corruption issues when switching windows
      (LP: #406731)
    + Fix SIGSEGV crash in drmCommandNone()
      (LP: #352567)
    + Fix smeared/tearing of display on reboot
      (LP: #367741

  [Tormod Volden]
  * 199_add_git_version_to_log.diff: Log git commit id in RadeonPreInit()
  * sed -i s/DRI2BufferPtr/DRI2Buffer2Ptr/ src/radeon_dri2.c
      so it compiles against xserver 1.6.2.+

  [Bryce Harrington]
  * Restore patch system and quilt to build-depends
  * Drop patches already present upstream
    + 107_check_unsupported_composit_ops.patch
    + 108_quirk_agpmode_m6_ali.patch: AGPMode quirk.
    + 109_quirk_agpmode_m7_intel.patch: AGPMode quirk.
    + 110_quirk_agpmode_r420_sis.patch: AGPMode quirk.
    + 111_use_xaa_for_lowmem_or_nodri.patch

 -- Bryce Harrington <email address hidden> Sat, 29 Aug 2009 12:16:53 -0700