Re: How to debug X lockup (advice from gurus wanted)
On 09/06/2011 11:14 PM, Roberto Ragusa wrote: Anyway, the problem happened again one hour ago, so I just decided to upgrade only the kernel to the one from F15 and hope for the best. If it happens again, I will then upgrade Xorg and Mesa. Replying to myself in this old thread just to let google-searchers know that after upgrading my F14 installation to the F15 kernel 2.6.40.4-5.fc15.i686.PAE (including module-init-tools as a dep), the laptop has now an uptime of 15 days, no more X lockups. There were also more than 40 suspend/resume cycles without any failure, so it seems like the new kernel has fixed that issue too. As a regression, the ethernet driver sometimes fails to work after resume; it is a known bug for 3.0, easily fixed with modprobe -r e1000e; modprobe e1000e so I can live with it. (Of course I'm expecting a lockup minutes after sending this mail...) -- Roberto Ragusamail at robertoragusa.it -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: How to debug X lockup (advice from gurus wanted)
On 9/1/11 5:05 PM, Roberto Ragusa wrote: Hmmm, turning off SMP is not realistic, as this laptop has a Core 2 Duo. Sure it is. Boot with maxcpus=1 on the kernel command line. - ajax -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: How to debug X lockup (advice from gurus wanted)
Am 06.09.2011 13:10, schrieb Adam Jackson: On 9/1/11 5:05 PM, Roberto Ragusa wrote: Hmmm, turning off SMP is not realistic, as this laptop has a Core 2 Duo. Sure it is. Boot with maxcpus=1 on the kernel command line i would recommend nosmp because it really disables the kernel-smp code signature.asc Description: OpenPGP digital signature -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: How to debug X lockup (advice from gurus wanted)
On 09/06/2011 01:18 PM, Reindl Harald wrote: Am 06.09.2011 13:10, schrieb Adam Jackson: On 9/1/11 5:05 PM, Roberto Ragusa wrote: Hmmm, turning off SMP is not realistic, as this laptop has a Core 2 Duo. Sure it is. Boot with maxcpus=1 on the kernel command line i would recommend nosmp because it really disables the kernel-smp code Ahem... what I meant is that it is not realistic for me to sacrifice one of the CPUs just to hope to get more stability. I intended the suggestion was turn off SMP support if you have only one CPU. Thinking about it, the kernel is already no-SMP on systems with one CPU nowadays, right? Anyway, the problem happened again one hour ago, so I just decided to upgrade only the kernel to the one from F15 and hope for the best. If it happens again, I will then upgrade Xorg and Mesa. -- Roberto Ragusamail at robertoragusa.it -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
Re: How to debug X lockup (advice from gurus wanted)
On 08/30/2011 09:24 PM, stan wrote: On Mon, 29 Aug 2011 21:06:29 +0200 Roberto Ragusa m...@robertoragusa.it wrote: Hi, it sometimes happens to me that X completely locks up, while the machine is still alive on the network. This is on F14, untainted kernel, nouveau driver, no 3D used, KDE desktop on a 32-bit machine with 8GiB RAM and PAE kernel. It typically happens when something is going to be drawn on the screen (a window pops up or virtual desktop change). I would like to open a bug, but I'm not able to attach any kind of usable log; dmesg says nothing, all I can say is that the screen remains frozen (including the pointer), the X server and the kernel keep doing some SIG ALRM stuff and any attempt to access the X server stalls the command (xrandr or xset, for example), in a Ctrl-C responsive way. Any idea? I had this problem in F14 while using the stock kernel. When I compiled a custom kernel it went away. While I made *many* changes, I think the one that mattered was turning off SMP on my single core system. I suspect, without proof, that the SMP code was occasionally causing a race condition, and a deadlock. Other possible causes for fix: moved to pre-emptable desktop, reserved 128 K low memory for kernel, moved to deadline scheduler. You could open a bug against the kernel, but it has moved on so far that it will probably languish. [now crossposting fedora, fedora-devel] Hmmm, turning off SMP is not realistic, as this laptop has a Core 2 Duo. I had been compiling my kernels until many years ago, and I would like to stay with the distro's kernel. I hope that upgrading to F15/F16 will mix things up and solve this in some way. Opening a bug seems a waste of time. I'm cross-posting to fedora-devel, in case someone can suggest how to collect useful info to open a kernel bug. Thanks. -- Roberto Ragusamail at robertoragusa.it -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel