Red Hat Bugzilla – Bug 182472
Video corruption, ATI Radeon 7000 and ES1000 - Display Engine stops issuing MC requests.
Last modified: 2007-11-30 17:07:23 EST
This problem was originally reported by Dell.
Occasionally, on startx, badly corrupted video is displayed completely
obscuring the desktop. Ctrl-Alt-Bksp does not resolve the problem - the
resulting text session remains corrupted. This has been seen at 1024x768 and
1280x1024 but it is more prevalent at higher resolutions - it occurs about 20%
of the time at 1920x1400,24bpp@60Hz when using ES1000 with 200MHz vRAM.
ATI has determined that the problem is caused by a FIFO which fills, stalling
the display controller. This occurs when the memory controller (MC) attempts
to send transactions to system memory that were intended for the video frame
buffer. There is a window of opportunity where the frame buffer address in
MC_FB_Location differs from the display engine registers.
There is now a WIP fix available at:
The fix will replace xorg-x11-6.8.1-ati-radeon-RV100-bus-master-fix.patch from
RH Bug# 165179. Disabling bus mastering is no longer necessary since the "bad
transactions" are now stopped closer to the source.
This fix is also mentioned in RH Bug 180150.
Can you provide a target release for Dell?
Created attachment 125070 [details]
Patch to eliminate window between MC update and display controller.
This patch is ready to apply to RHEL4 however the xorg patch may be more
up-to-date please review both.
I see there are some issues with this patch upstream. Can ATI investigate and
post a response with the correct patch ?
------- Additional comment #3 from Michel DÃ¤nzer on 2006-02-23 19:05 [reply]
The current patch has been reported to hang on some systems, see
https://bugzilla.novell.com/show_bug.cgi?id=151631 . This is being investigated.
Ugh, this bug was filed twice, and people are picking one at random when
*** This bug has been marked as a duplicate of 182471 ***