Red Hat Bugzilla – Bug 186663
xorg R7 locks up system with PCI Radeon 9200
Last modified: 2007-11-30 17:11:28 EST
Description of problem:
Since the upgrade from FC4 to FC5, the head running on a Radeon 9200 stopped
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Start X
Computer locks up. The keyboard stops working and ping requests are not replied to.
X works as it did under FC4
The system also has a GeForce AGP card that runs the first head using the nv
driver. There were no crashes with FC4.
I am attaching a number of log files and configurations for the several attempts
I made. The last one was made updating to the latest driver rpm package from
xorg-x11-drv-ati-18.104.22.168-4.cvs20060322. I remounted / with -o sync, in order to
log as much as possible before the crash.
Created attachment 126694 [details]
Crash after coercing the vesa driver
Created attachment 126695 [details]
configuration with vesa driver
Created attachment 126696 [details]
Log of crash with radeon driver
Created attachment 126697 [details]
Configuration using Radeon driver
Created attachment 126698 [details]
Log of crash with int10 module renamed
Since the last log entry was about the int10 module, I tried to see what
happened if I prevented it from loading. No luck.
Created attachment 126699 [details]
Log of crash with int10 module restored, but rawhide radeon driver
Being glutton for punishment, I tried several things in order to track down the
I installed the fglrx rpm from Livna. Crash. Maybe it's a kernel problem?
I updated to the 2069 updates-testing kernel. Crash.
I updated to the new 2080 update kernel from today. Crash.
I booted with the latest 2.6.15 kernel from FC4. Crash.
Normally, I would blame this on the hardware, but prior to the FC5 update, the
system had uptimes measured literally in months.
do you have a SMP kernel running?
Is this system from Dell?
Yes, this is a Dell Optiplex GX270. BIOS version is A06 according to dmidecode.
I tried both the SMP and UP kernels (the system has HypeThreading enabled), even
today's freshly released updates.
Is this fixed in any way by https://bugs.freedesktop.org/show_bug.cgi?id=6750 or
https://bugs.freedesktop.org/show_bug.cgi?id=6751? Is there anything from Raw
Hide I can test on FC5 to confirm the bug was fixed?
Please uninstall all proprietary drivers from the system, then reinstall
all FC5 xorg and mesa packages to ensure all X server modules you are using
are the ones provided by Fedora Core, and there are no conflicts caused
from the 3rd party drivers.
Ensure that you are using the latest Fedora Core 5 update packages as well,
and nothing from rawhide.
It does appear that int10 is hanging for some reason, so that appears to
be the best place for diagnosis to proceed. Does the X server just hang
there, but stay in memory livelocked? Or does the whole system deadlock?
Can you reproduce this from within gdb as root and get a backtrace?
the crashes started occurring when I updated the system from FC4 to FC5. The
system always deadlocked, right away, as soon as X blanked the screen when
transitioning from the text console to vt7; there's never been a case when it
didn't freeze entirely. I doubt I can use gdb to debug this.
I didn't install the livna fglrx and the rawhide ati drivers until long after
the crashes occurred; I have since uninstalled them in order to pick up any new
FC5 updates, then I verified with RPM the contents of xorg-*/mesa-* and restored
the old configuration files. According to yum list extras, there's nothing from
Rawhide on the system anymore.
The reason I mentioned Rawhide again is that Dave Airlie made this comment:
"ajax: you might want to consider this for stable, dual-card gets really broken
by later Linux kernels, you have to use this interface on them."
It sounds to me as if there's no hope for me with the existing packages; I am
willing to test more recent ones to debug or confirm the problem has been fixed.
That patch is already applied for xserver 1.1 and later.
Anyone able to reproduce this with FC6 or FC7?
Reporter, could you please confirm whether this bug is still reproducable with
the current versions of either FC6 or FC7? Otherwise, I will have to close this
bug as INSUFFICIENT_DATA. Thanks.
Fedora Core 5 is no longer supported, please, could you reproduce this bug with
the updated version of the currently supported distribution (Fedora Core 6, or
Fedora 7, or Rawhide)? If this issue turns out to still be reproducible, please
let us know in this bug report. If after a month's time we have not heard back
from you, we will have to close this bug as CANTFIX/INSUFFICIENT_DATA.
Setting status to NEEDINFO, and awaiting information from the reporter.
Thanks in advance.
I no longer have access to the system, so I can't help with this anymore. Feel
free to close the bug.
Closing per reporter's request. Please, reopen if you are able to reproduce this
with currently supported distribution.