http://bugs.freedesktop.org/show_bug.cgi?id=20954


Kevin Locke <k...@cornell.edu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |k...@cornell.edu
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |




--- Comment #9 from Kevin Locke <k...@cornell.edu>  2009-09-05 09:32:01 PST ---
Appologies for reopening this bug if it truly is fixed in the KMS branch.  I
have been unable to get KMS setup on my machine, but have tried my best to
analyze the changes in linux-next to check if this is truly fixed but I can not
see how it would be.  Also note, I am not an advanced kernel or DRI hacker, so
this analysis may be completely bogus.  My analysis of the problem is as
follows:

During X shutdown, RADEONDRICloseScreen (radeon_dri.c:1915) in the X driver
calls the radeon_cp_init (radeon_cp.c:1313) ioctl with RADEON_CLEANUP_CP. 
radeon_cp_init then calls radeon_do_cleanup_cp (radeon_cp.c:1221) which sets
mmio to NULL when it memset's most of dev_priv to 0.

Later at module unload time (or pci unplug time in the KMS branch) drm_unload
(drm_drv.c:358) (drm_put_dev (drm_stub.c:490) in KMS) calls drm_vblank_cleanup
(drm_irq.c:97) which calls vblank_disable_fn (drm_irq.c:74) which calls
radeon_get_vblank_counter (radeon_irq.c:284) and causes an Oops trying to read
from mmio, which is NULL.

I'm not sure if the solution is to not zero mmio during cleanup or protect
against it being NULL in get_vblank_count or not call get_vblank_count during
vblank cleanup (presumably by saving the last vblank during driver cleanup).  I
don't know enough about the code to make that determination.

As best I can tell, this sequence happens in both the current git master as
well as the linux-next KMS branch, but perhaps I am overlooking something... 
my apologies if this really is properly fixed.

Thanks for reconsidering it.


-- 
Configure bugmail: http://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
--
_______________________________________________
Dri-devel mailing list
Dri-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dri-devel

Reply via email to