Description of problem: A rawhide upgrade in the last few days caused X to lockup after a few seconds of use when dual head is enabled on a Radeon 7000. It was working well only a few days ago. When X locks up, all key and mouse input fails at the local console, but ssh allows logins. This error repeats in Xorg.0.log many times: (EE) RADEON(0): Idle timed out, resetting engine... 01:00.0 VGA compatible controller: ATI Technologies Inc Radeon RV100 QY [Radeon 7000/VE] (prog-if 00 [VGA]) Subsystem: ATI Technologies Inc: Unknown device 1b8a Flags: stepping, 66Mhz, medium devsel, IRQ 11 Memory at f0000000 (32-bit, prefetchable) [size=128M] I/O ports at ec00 [size=256] Memory at ff9f0000 (32-bit, non-prefetchable) [size=64K] Expansion ROM at 80000000 [disabled] [size=128K] Capabilities: [58] AGP version 2.0 Capabilities: [50] Power Management version 2 Attaching complete Xorg.0.log and lspci -v will be attached. NOTE: Disabling dual head makes X stable again. Version-Release number of selected component (if applicable): xorg-x11-6.8.1-4
Created attachment 104769 [details] Xorg.0.log
Created attachment 104773 [details] lspci -v
Elayne, please downgrade to the xorg-x11 from FC3test2 to confirm that it works with that slightly older version. ftp://videl.ics.hawaii.edu/mirrors/fedora/linux/core/test/2.91/i386/os/Fedora/RPMS/ You can download it from here. Downgrading would require you to: 1) rpm -qa |grep xorg-x11 Get a list of xorg-x11 packages installed. 2) Download the matching list of packages of the older version. 3) rpm -Uvh xorg-x11*.rpm --oldpackage Then try both the non-dual and dual config files to see if the behavior is different.
(EE) RADEON(0): Idle timed out, resetting engine... This suggests it is a DRI/DRM interaction issue. Try disabling DRI by uncommenting the dri module load line in the config file. (yes, the log file indicates DRI was auto-disabled at run time, but please try this also)
I did try that. It didn't seem to help.
One more thing I should note.. The radeon driver in current rawhide has not changed in quite some time now, nor has the X server generally speaking. If new changes in rawhide caused this problem to surface, I would suspect the kernel perhaps. More information is needed to determine anything conclusively however.
I would suspect kernel too, but it was the same 541 kernel both before and after. Will do more testing tomorrow.
Ok, please try "noaccel" in the driver section. If this prevents the problem, we can experiment with XaaNo options to try to track down the problem point. It is possible that perhaps some DRI-only code isn't protected with appropriate conditionalization perhaps.
Warren: Can you also try a singlehead configuration, with DRI manually disabled? I'd like to see if we can rule out dualhead as being an issue and perhaps tie it to the non-DRI codepaths in the driver. TIA
Just wanted to state, I'm not quite as suspect of this being a kernel issue now, as it only seems to occur in the non-DRI path from what you mention above. Please attach the X server config also.
*** This bug has been marked as a duplicate of 134766 ***
Changed to 'CLOSED' state since 'RESOLVED' has been deprecated.