Bug 24379

Summary: Random system freezes with RV350
Product: xorg Reporter: Alessio Gaeta <alga777>
Component: Driver/RadeonAssignee: xf86-video-ati maintainers <xorg-driver-ati>
Status: RESOLVED FIXED QA Contact: Xorg Project Team <xorg-team>
Severity: critical    
Priority: medium CC: alga777, hramrach
Version: git   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
Present (working) xorg.conf
none
Present (working) Xorg.0.log
none
Xorg log with virtual set to 2048x2048 none

Description Alessio Gaeta 2009-10-07 10:39:24 UTC
Hello, I get random system freezes using:

Ubuntu 9.10 Karmic (Beta)
xserver-xorg-video-radeon from Tormod Volden: 6.12.99+git20091002.cc45856a

Toshiba M30X with an ATI Radeon 9700 (M10) - RV350 - 64 MB VRAM

The freezes seem GPU hardlocks, mouse pointer is frozen too and keyboard stops responding (no Caps Lock, no SysReq magic keys); ATM I cannot if system is accessible via ssh (I suspect yes, though). As said, system hardlocks randomly; anyway, more often the freeze is triggered by:

  * window maximize/minimize (more often with compiz enabled)
  * evince pdf reading (compiz enabled and disabled)

Increasing virtual size makes crashes more frequent, using multiple monitors with xrandr is a guaranteed crash. This makes me think that the issue can be related to VRAM amount and/or GART transfers (I don't know the matter so well, actually...).

The only way I found to stop freezes is setting RenderAccel to false, but then I get screen corruption increasing virtual size beyond the default (greater the virtual is, greater is the corruption). And obviously the system becomes noticeably slower...

Options like AccelDFS, AGPMode=[PCI|auto|1] (set via xorg.conf and via modprobe), ClockGating, DynamicPM, dynclks  seem to not make difference.
Comment 1 Alessio Gaeta 2009-10-07 10:41:13 UTC
Forgot to mention that there is no errors in Xorg.0.log* nor other system logs.
Comment 2 Alex Deucher 2009-10-07 11:24:03 UTC
Please attach your xorg log and config.
Comment 3 Alessio Gaeta 2009-10-07 15:05:37 UTC
Created attachment 30148 [details]
Present (working) xorg.conf

This is the my present xorg.conf; it works because the RenderAccel=false.

Virtual, AGP and GARTSize are commentend out (leaving defaults).

/sys/module/radeon/parameters are:
agpmode          0
benchmark        0
connector_table  0
dynclks          -1
gartsize         512
modeset          0
no_wb            0
r4xx_atom        0
test             0
vramlimit        0

(all defaults)
Comment 4 Alessio Gaeta 2009-10-07 15:06:43 UTC
Created attachment 30149 [details]
Present (working) Xorg.0.log

This is the Xorg log with present (working) configuration.
Comment 5 Alessio Gaeta 2009-10-07 15:18:01 UTC
Created attachment 30153 [details]
Xorg log with virtual set to 2048x2048

This is a previous Xorg log with virtual set to 2048x2048 (set by the Ubuntu display configurator).
The xorg.conf was a little different from previous (as you can infer from the log): RenderAccel was enabled (and system crashed...) and AGP was 1 (non influential, due to a quirk I suppose).

I noted the diminished amount of texture RAM. Maybe the system crashes when attempts to allocate central memory via GART? Less texture memory, more chance to crash... But then, why disabling RenderAccel workarounds the problem (but leaves screen corruption)?
Comment 6 dominik 2009-10-25 02:38:42 UTC
I can confirm this bug on my laptop as well.

Ubuntu 9.10 Karmic Koala (beta)

Acer Aspire 1680wlmi with an ATI Radeon 9700 (M10) - RV350

xserver-xorg-video-radeon from the x-edgers repository. 
(1:6.12.99+git20091024.f0d9d80f-0ubuntu0tormod 
Comment 7 Alex Deucher 2012-04-25 05:58:04 UTC
Is this still an issue with a newer driver/kernel?
Comment 8 Alessio Gaeta 2012-05-26 08:07:22 UTC
Sorry for the late reply.

At the moment on that laptop is running fine Linux Mint 12, so I think this particular problem can be considered solved.

BTW, still the laptop hard freezes (with symptoms similar to those descripted) when the OS is booted without an USB mouse plugged in. But maybe that's another issue...

Thanks

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.