Bug 12029

Summary: Xorg uses lot of CPU and doesn't update the display after upgrade to 6.6.193
Product: xorg Reporter: Sebastien Bacher <seb128>
Component: Driver/RadeonAssignee: xf86-video-ati maintainers <xorg-driver-ati>
Status: RESOLVED INVALID QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium    
Version: 7.2 (2007.02)   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
the xorg configuration
none
6.6.3 driver log
none
6.6.193 driver log
none
6.6.3 driver log none

Description Sebastien Bacher 2007-08-16 10:30:18 UTC
Using the new version of the ati driver (6.6.193) on gutsy the display freezes quickly when using epiphany-browser. The mouse pointer is still moving but nothing else on the display change, switching to a VT doesn't work. Connecting by ssh to the box work, Xorg is using all the CPU and there is no way to stop it. Downgrading to 6.6.3 fixes the bug.
Comment 1 Sebastien Bacher 2007-08-16 10:31:41 UTC
The video card used is an ATI Radeon 9600 pro and the driver is "ati"
Comment 2 Michel Dänzer 2007-08-17 00:40:44 UTC
Sounds like a GPU lockup. Please attach the config file and log files for both driver versions.
Comment 3 Sebastien Bacher 2007-08-17 14:53:50 UTC
Created attachment 11164 [details]
the xorg configuration
Comment 4 Sebastien Bacher 2007-08-17 14:54:39 UTC
Created attachment 11165 [details]
6.6.3 driver log
Comment 5 Sebastien Bacher 2007-08-17 14:55:12 UTC
Created attachment 11166 [details]
6.6.193 driver log
Comment 6 Sebastien Bacher 2007-08-17 14:55:36 UTC
configuration and log attached, reopening
Comment 7 Sebastien Bacher 2007-08-17 15:00:50 UTC
Created attachment 11167 [details]
6.6.3 driver log
Comment 8 Michel Dänzer 2007-08-18 03:44:06 UTC
What is the AGP transfer mode set to in the BIOS setup? Which mode does the kernel end up setting with each driver version (dmesg | grep AGP)? Does Option "AGPMode" "4" help?
Comment 9 Sebastien Bacher 2007-08-18 05:03:17 UTC
The bios has "AGP Fast Writing" and "AGP 3.0" in auto

$ dmesg | grep AGP
[   33.153134] ACPI: PCI Interrupt Routing Table [\_SB_.PCI0.AGPB._PRT]
[   47.263749] agpgart: Detected AGP bridge 0
[   47.263767] agpgart: Setting up Nforce3 AGP.
[   47.268145] agpgart: AGP aperture is 128M @ 0xe0000000
[   75.526133] agpgart: Found an AGP 3.0 compliant device at 0000:00:00.0.
[   75.526331] agpgart: Putting AGP V3 device at 0000:00:00.0 into 4x mode
[   75.526355] agpgart: Putting AGP V3 device at 0000:01:00.0 into 4x mode
Comment 10 Sebastien Bacher 2007-08-19 13:58:58 UTC
the AGPMode option fixes the issue
Comment 11 Michel Dänzer 2007-08-20 00:45:30 UTC
(In reply to comment #9)
> The bios has "AGP Fast Writing" and "AGP 3.0" in auto

Does any of the following work as well without Option "AGPMode"?

* Option "AGPFastWrite" or disabling Fast Writes in the BIOS
* Setting the transfer mode to 4x or 8x explicitly in the BIOS
Comment 12 Sebastien Bacher 2007-08-25 04:11:17 UTC
disabling the fast writing in the bios doesn't make a difference, forcing the agp 3.0 speed to 4x rather than auto seems to work correctly
Comment 13 Michel Dänzer 2007-08-25 06:54:23 UTC
(In reply to comment #12)
> forcing the agp 3.0 speed to 4x rather than auto seems to work correctly

What about 8x?

I'm afraid this is looking more and more like WONTFIX, as defaulting to 4x caused problems in other setups, so we're currently relying on the BIOS to set up the transfer rate it chooses to be reliable.
Comment 14 Sebastien Bacher 2007-08-25 08:54:40 UTC
the choices are "auto", "4x", "4x8x". What is creating the issue? The video card or the bios? Could the driver default to 4x on boggy cards?
Comment 15 Michel Dänzer 2007-08-29 01:31:33 UTC
(In reply to comment #14)
> the choices are "auto", "4x", "4x8x". 

So if you choose "4x8x", what transfer rate does it end up using then? Does it work without Option "AGPMode" in that case?

> What is creating the issue? The video card or the bios? Could the driver default
> to 4x on boggy cards?

The issue could be specific to the combination of BIOS, card, AGP bridge and possibly the phase of the moon, so detecting it automatically could be tricky.
Comment 16 Alex Deucher 2008-12-03 00:08:33 UTC
If this is still an issue with the ati driver from git, we can add an agp quirk for your card/gpu combo.
Comment 17 Alex Deucher 2010-10-19 15:51:06 UTC
closing due to lack of feedback.

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.