Description of problem: xorg segfaulting. This happens, when there is the following line in the log: (WW) ****INVALID IO ALLOCATION**** b: 0xe400 e: 0xe407 correcting Version-Release number of selected component (if applicable): It happens using the kernel-smp-2.6.15-1.1826.2.4_FC5, and the 2.6.15 vanilla kernel, but works without any problems using the 2.6.15-rc5n vanilla kernel. How reproducible: always using mentioned above kernels. Steps to Reproduce: 1. boot the system using the kernel-smp-2.6.15-1.1826.2.4_FC5 (level 5)
Created attachment 122909 [details] xserver log
Created attachment 122910 [details] rpm -qa output
lspci output the same as in bug #176724, attachment #122651 [details].
Created attachment 122911 [details] gdm output from the /var/log/messages
Created attachment 122939 [details] /usr/bin/X core dump
From the description above, it sounds like this might perhaps be a kernel bug causing the problem. Does the problem still occur with your system fully updated to current Fedora development? Davej: Any idea what kernel changes might have caused such a problem to appear?
no idea. if it is still a problem with the latest packages, the output of /proc/iomem from both the good and bad kernels would be interesting to see.
Damian: Could you provide the info from comment #7?
kernel-2.6.15-1.1909_FC5 is the current kernel in rawhide. Please upgrade to that kernel, and do a full update of all the latest Fedora devel packages. Once you've updated, reboot into the new kernel and test X, then update the bug report with your results. Please do this as soon as you can, as the FC5test3 deadline is quickly approaching. TIA
Created attachment 124282 [details] iomem - 2.6.15n - previously X server crashed on this kernel
Created attachment 124283 [details] ioports - 2.6.15n - previously X server crashed on this kernel
Created attachment 124285 [details] iomem - 2.6.15-1.1909_FC5
Created attachment 124286 [details] ioports- 2.6.15-1.1909_FC5
(In reply to comment #7) > no idea. if it is still a problem with the latest packages, Now using the latest devel packages and the 2.6.15-1.1909_FC5 kernel version the problem is not reproducible anymore, even using the vanilla 2.6.15n kernel (exactly the same binary files) on which the issue was previously reported. (In reply to comment #6) > From the description above, it sounds like this might perhaps be a kernel > bug causing the problem. IMHO, it didn't look like a kernel problem.
Ok, looks like something in rawhide fixes it for you, so we'll consider it resolved for now. If it recurs, just reopen the report and we'll reinvestigate. Thanks for testing. Setting status to "RAWHIDE"