Bug 15019

Summary:	Googleearth locks up radeon R300 card
Product:	Mesa	Reporter:	David Hagood <david.hagood>
Component:	Drivers/DRI/r300	Assignee:	Default DRI bug account <dri-devel>
Status:	RESOLVED DUPLICATE	QA Contact:
Severity:	normal
Priority:	medium	CC:	pedretti.fabio
Version:	unspecified
Hardware:	x86 (IA32)
OS:	Linux (All)
Whiteboard:
i915 platform:		i915 features:
Attachments:	xorg configuration file X log file

Description David Hagood 2008-03-14 19:20:09 UTC

I can 100% repeatably lock up my system to the point a reboot is required by running Googleearth Plus.

The card is identified by xorg as:
(--) Chipset ATI Radeon 9600XT AR (AGP) found

The lockup has Googleearth running at 100% CPU load, with ALL X programs (including Googleearth) unable to update the display (e.g. the clock won't update, the load average applet in the panel won't update, etc.). HOWEVER: the mouse pointer will still move (hardware mouse cursor), and the system can be SSH'ed into.

Killing Googleearth with a SIGTERM at this point does nothing. Attempting to attach to Googleearth with gdb will cause gdb to become unresponsive (e.g. you cannot ctrl-C to break into Googleearth to back-trace it, however you CAN ctrl-Z to suspend gdb).

Killing Googleearth with a SIGKILL will terminate it, but then X goes to 100% CPU load, and cannot be SIGTERM'ed, only SIGKILLed.

Nothing additional appears in /var/log/Xorg.0.log after starting Googleearth.

Comment 1 Michel Dänzer 2008-03-17 11:00:27 UTC

Which version of Mesa are you using? Also, please attach (as opposed to paste) the full xorg.conf and Xorg.0.log files.

Comment 2 David Hagood 2008-03-17 15:57:41 UTC

Created attachment 15240 [details]
xorg configuration file

Per request, here's the config file. Xorg log follows.

Comment 3 David Hagood 2008-03-17 15:58:10 UTC

Created attachment 15241 [details]
X log file

Here's the X log file

Comment 4 David Hagood 2008-03-17 16:00:37 UTC

dpkg -l  libgl1-mesa-gl\*
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Installed/Config-f/Unpacked/Failed-cfg/Half-inst/t-aWait/T-pend
|/ Err?=(none)/Hold/Reinst-required/X=both-problems (Status,Err: uppercase=bad)
||/ Name           Version        Description
+++-==============-==============-============================================
ii  libgl1-mesa-gl 7.0.3~rc2-1ubu A free implementation of the OpenGL API -- G

Comment 5 Michel Dänzer 2008-03-18 07:28:48 UTC

Does it still happen if you comment out all options in the driver section? If so, what if you use Option "AGPMode" "1" or "8" or Option "BusType" "PCI"?

Comment 6 David Hagood 2008-03-18 16:33:59 UTC

It seems to be the page flip option that locks things up - I removed all the options, and all was well. Adding them back one at a time worked until I added the page flip, and adding all options other than enable page flip worked, so there seems to be something about Google Earth and page flip that kills things.

NOTE: Google Earth is the only OpenGL app I have found that exhibits this behavior - other apps (e.g. skyrocket from the rss screensavers) work fine.

Comment 7 Michel Dänzer 2008-03-19 04:56:22 UTC

Does the problem also occur if you enable page flipping but leave all other options disabled?

Comment 8 David Hagood 2008-03-19 05:12:34 UTC

Yes, it is page flipping, and only page flipping, that causes the problem to manifest itself - I'd verified that enabling only page flipping would lock it up.

Comment 9 Roland Scheidegger 2008-03-19 06:40:02 UTC

does it also crash if you disable the lowimpact fallbacks (disable_lowimpact_fallback=true or use driconf)? These fallbacks have a tendency to crash, though I fail to see right now why it would make a difference with pageflipping enabled...

Comment 10 David Hagood 2008-03-19 10:29:51 UTC

Yes, I have the fallbacks disabled.

Comment 11 Fabio Pedretti 2009-11-14 05:10:55 UTC

Since it appears you are using Ubuntu, can you try to reproduce the problem with Ubuntu 9.10 (which include linux 2.6.31 and mesa 7.6)?

Comment 12 David Hagood 2009-11-14 06:58:21 UTC

Unfortunately, I no longer have the system that was showing the symptoms, so I can no longer test this bug.

Comment 13 Fabio Pedretti 2009-11-17 00:14:42 UTC

(In reply to comment #12)
> Unfortunately, I no longer have the system that was showing the symptoms, so I
> can no longer test this bug.

OK, there is anyway a similar bug for this problem. Marking as duplicate.

*** This bug has been marked as a duplicate of bug 21472 ***

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.