Bug 24415

Summary: Radeon-driver init hangs machine in latest git updates
Product: xorg Reporter: Øyvind Stegard <oyvind>
Component: Driver/RadeonAssignee: xf86-video-ati maintainers <xorg-driver-ati>
Status: RESOLVED FIXED QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium    
Version: 7.4 (2008.09)   
Hardware: x86 (IA32)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
Xorg log from successful init (up to commit e59ae08270711512e64b70f79b6476cc2c52d230) none

Description Øyvind Stegard 2009-10-09 01:31:54 UTC 
Created attachment 30208 [details]
Xorg log from successful init (up to commit e59ae08270711512e64b70f79b6476cc2c52d230)

The Radeon-driver hangs the machine on init, after updating to the latest git as of today. It has worked fine up to and including commit 	«e59ae08270711512e64b70f79b6476cc2c52d230 Don't call radeon_cs_flush_indirect() without KMS. Fixes http://bugs.freedesktop.org/show_bug.cgi?id=24316».

I don't know which of the newer commits might be the culprit. It fails with black screen and hard hang at Xorg start. I'm having problems capturing a sane log, but it looks like it freezes right after DRI init (log output stopped there). I'll try to get some more details later.

Hardware:
ATI X1400 Mobile Radeon. Will attach Xorg-log from successful boot with driver built from git-commit e59ae08270711512e64b70f79b6476cc2c52d230 (just to provide some details about hardware).

Running Ubuntu Jaunty x86, kernel 2.6.28-15-generic.
libgl1-mesa-dri is version 7.4-0ubuntu3.3 (Ubuntu packaged).
I use '--disable-kms' when compiling the radeon-driver.
Comment 1 Øyvind Stegard 2009-10-09 02:07:05 UTC 
It's either commit:
  5f846360c46f5a989f5d0fde6d251cdbd61d4968 (radeon: clamp the internal FB map to the aperture size)
or commit:
  186bf719c394116341abff471fb4b4803650da9a (radeon: fix rotation since mixed pixmaps.)

that's the culprit. After stepping back to commit 	e08411af1aa8c7d4233ba593b84360397cdbb307 (radeon: Fix duplicated bit settings), the driver started working again.
Comment 2 Øyvind Stegard 2009-10-09 02:10:20 UTC 
Eh, sorry, it must be commit 186bf719c394116341abff471fb4b4803650da9a (radeon: fix rotation since mixed pixmaps.). If I build the driver up and including this commit, it fails to init and hangs the machine. If I build the driver up the previous commit, it works.
Comment 3 Dragos Delcea 2009-10-09 03:10:26 UTC 
(In reply to comment #2)
> Eh, sorry, it must be commit 186bf719c394116341abff471fb4b4803650da9a (radeon:
> fix rotation since mixed pixmaps.). If I build the driver up and including this
> commit, it fails to init and hangs the machine. If I build the driver up the
> previous commit, it works.
> 

It fails the same way here: black screen and hard hang when gdm tries to start (OS doesn't respond to ping)

Env: (Lenovo T60, Radeon Mobility X1400, gentoo stable - 2.6.30-r8, xorg-server 1.6.3.901-r2, plain DRI1 setup, only radeon built from git sources)

It's not 186bf719c394116341abff471fb4b4803650da9a though, the driver build from this commit works (I'm using it right now), it must be one of the subsequent ones.

So, for me up to 186bf719c394116341abff471fb4b4803650da9a it works, while 186bf719c394116341abff471fb4b4803650da9a fails as described above. Haven't tried anything in between.
Comment 4 Michel Dänzer 2009-10-09 03:17:01 UTC 
Please try again with current Git, Dave just pushed some fixes.
Comment 5 Øyvind Stegard 2009-10-09 03:22:26 UTC 
(In reply to comment #4)
> Please try again with current Git, Dave just pushed some fixes.
> 

Works in current HEAD, Comment #3 was right, I got it wrong on the last patch. The culprit was indeed 5f846360c46f5a989f5d0fde6d251cdbd61d4968 (radeon: clamp the internal FB mapto the aperture size).

This bug can be closed.
Comment 6 Dragos Delcea 2009-10-09 04:05:00 UTC 
1b25bad2d8ebe481ffb4b0fd3a85ab636bd4fec2 works for me too. Thanks for the quick response.

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.