[Bug 10211] Possible correction for r300_pacify' s R300_RB3D_DSTCACHE_CTLSTAT write

[<bug...@fr...>]

http://bugs.freedesktop.org/show_bug.cgi?id=10211


Jerome Glisse <gl...@fr...> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #4 from Jerome Glisse <gl...@fr...>  2008-09-08 05:50:35 PST ---
The real problem with lockup doesn't come from flushing but from bad command
stream like the vertex uploading Nicolai fixed few weeks ago. We also need to
put some more delay when moving buffer and properly waiting for idle. Another
things that might help is to upload wptr of the ring every 32 dwords or some
others power of 2 value as it seems the ring fetcher can go crazy for unknown
reasons on some update. The new command stream ioctl is basically trying to
address all this in one shot as after a try to fix things from kernel side
proved me that this was a bad idea.

So i set this bug as invalid, meanwhile i would like to make ddx fallback to
software rendering on GPU lockup thus X should be responsive again even if dead
slow.


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