Red Hat Bugzilla – Bug 20013
X 4.0.1 crashes with ATI Mach64 card
Last modified: 2007-04-18 12:29:34 EDT
I'm using v.4.0.1 with an All-In-Wonder Pro AGP under kernel 2.4.0-test8
(not using /dev/agppart though). When the system's under heavy load and
some combination of X, Netscape 4.75, Konqueror (from standard KDE2 RPMS),
and StarOffice 5.2 are running, X will randomly die with signal 11. My
memory's fine as far as I can tell...X was solid as a rock when I was
running a pgcc-optimized 3.3.6 under RH6.1 a few weeks ago. The URL above
points to a server log file from a successful start. I don't have an
unsuccessful start laying around the computer, but the only difference is
that the last line includes something like:
Caught signal 11: exiting...
The one additional change I've made to the X config is I've got TrueType
fonts loaded in.
I have been getting a similar crash. I seem to get it when running VMWare or
wine, and it happens the most when switching VTs. It may have something to do
with bug number 18651.
I have the following configuration:
RedHat 7.0, patched GLIBC 2.2 (always the latest)
PII-400 (Compaq Armada M700)
ATI Rage Mobility-P
Kernel 2.2.16 stock or 2.4.0-test12 (happens in both)
I've obtained the SRPM for XFree86-4.0.2-1 from Rawhide, and compiled it myself
using the "GCC 2.96" RH7 provides, and using the optflags "-O6 -march=i686" for
speed optimization. Now, the only time I've been able to crash X is running
Konqueror (kde2.0) with Java (JRE 1.3 from java.sun.com) and using Yahoo Chat
(chat.yahoo.com). I don't have any actual console logs, but right before X
crashes, there's some output to console that Qt's complaining about null
pixmaps. If I can write down the message next time I see it, I'll append it to
Ok, I need you to replicate any problem you are having (email@example.com)
using an official Red Hat kernel that has been released for Guinness, and
using an official Red Hat XFree86. I need specific gory details, X server
logs, X config files, yada yada.
Just for the record: VMware related bugs, must be duplicated without VMware and
without loading it's modules at boot time. Home brew kernels, and home built
XFree86 packages are not supported.
So, please provide me with the info requested above, and details on how to
replicate and I'll gladly look into this for you.
I haven't been able to replicate the problem in the past month or two under
XF-4.0.2 with the rawhide RPMS I described above. I'm running under a non-RH
kernel, and a Rawhide-based XFree, so I can't get that info. Unless someone
else can come up with it, go ahead and close out the report.
Ok, it looks like it is fixed then. If it works for you, it doesn't matter as
much what you're using. ;o)