Bug#407036: xserver-xorg: Server crashes each time after hibernate with suspend2
Yes I have tried disabling dri and my laptop still crashes on resume with the new version of xserver-xorg-core. I've confirmed again just now that it still crashes. If the kernel is crashing then perhaps we can trace it there somehow? Bruce On Sun, Jul 08, 2007 at 11:31:19AM +0200, Brice Goglin wrote: > Bruce MacDonald wrote: > > Dear Brice, I've been trying to collect some more useful information > > about this problem. I can't get a core file to be created. I've > > verified that X will dump a core for me, by sending it a quit signal > > with kill -3. The core is there and I can get a backtrace using the > > dbg package. But when I resume after suspending still no core file is > > created. So maybe my machine crashes before the core file can be made? > > If you have any suggestions I'd be happy to try them. Bruce > > > > I am sorry, but I am running out of ideas too :( > > The problem could be related to upstream bug [1], did you try without > DRI enabled? > (adding "Disable "dri"" to the module section of xorg.conf should do the > trick) > > Brice > > [1] https://bugs.freedesktop.org/show_bug.cgi?id=3272 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#407036: xserver-xorg: Server crashes each time after hibernate with suspend2
Dear Brice, I've been trying to collect some more useful information about this problem. I can't get a core file to be created. I've verified that X will dump a core for me, by sending it a quit signal with kill -3. The core is there and I can get a backtrace using the dbg package. But when I resume after suspending still no core file is created. So maybe my machine crashes before the core file can be made? If you have any suggestions I'd be happy to try them. Bruce -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#407036: xserver-xorg: Server crashes each time after hibernate with suspend2
Hi Brice, ssh from another machine and screen is a good idea thanks. I tried that and gdb attaches properly to X but when I suspend the laptop it crashes (ie the whole machine crashes, not just the X server; I had to force a power down then power up again). That is even with the old xserver that normally suspends ok. I also tried again for a core file (I didn't have ulimit set properly and now I do), but no core file is created when the resume crashes my laptop (using the latest xserver-xorg-core). I do have the NoTrapSignals option set and it is recognized in the Xorg log file. So run out of things to try again ... Bruce On Sun, Jun 17, 2007 at 01:31:38PM +0200, Brice Goglin wrote: > Bruce MacDonald wrote: > > Hi, Any ideas for debugging this problem? I can't login from a second > > machine since the resume script has to remove and reinstall the > > network driver for both wired and wireless. I couldn't get a core dump > > by setting the NoTrapSignals option in ServerFlags. I tried a script > > suggested on x.org for server debugging but X would not start at all, > > from > > > > http://wiki.x.org/wiki/Development/Documentation/ServerDebugging > > > > But I've never debugged the X server before. > > > > Bruce > > > > One idea would be to keep gdb in a screen that you attach/detach from ssh: > * login through ssh before suspend, launch "screen", and attach "gdb -p > $(pidof X)" from there > * suspend/resume, and make sure the network is back > * login again by ssh again, reacquire the screen with something like > "screen -rd" > * once X has crashed, work in gdb with 'bt full' or so > > Somebody put some Debian-specific hints about debugging (including with > a coredump) at http://wiki.debian.org/XStrikeForce/XserverDebugging by I > never actually tried it. > > Brice > > > -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#407036: xserver-xorg: Server crashes each time after hibernate with suspend2
Hi, Any ideas for debugging this problem? I can't login from a second machine since the resume script has to remove and reinstall the network driver for both wired and wireless. I couldn't get a core dump by setting the NoTrapSignals option in ServerFlags. I tried a script suggested on x.org for server debugging but X would not start at all, from http://wiki.x.org/wiki/Development/Documentation/ServerDebugging But I've never debugged the X server before. Bruce On Thu, May 31, 2007 at 06:53:32PM +1200, Bruce MacDonald wrote: > Hi Brice, thats right I have not applied the suspend2 patch. Just > using Debian kernel source and compiling my own kernel; no patches. > Bruce > > On Thu, May 31, 2007 at 08:46:05AM +0200, Brice Goglin wrote: > > Bruce MacDonald wrote: > > > Hi Brice, I use my own script. After unloading USB modules just do > > > > > > echo mem > /sys/power/state; > > > > > > Has worked really well for three years on my Dell Latitude D600 and > > > still does except with the new package. > > > > > > In the kernel > > > > > > # CONFIG_SOFTWARE_SUSPEND is not set > > > > > > Is that related to suspend2? > > > > > > > No, suspend2 is a patch that you have to apply to own-recompiled kernel. > > So looks like you're not using suspend2 and the bug is not > > suspend2-specific. > > > > Brice > > -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#407036: xserver-xorg: Server crashes each time after hibernate with suspend2
Hi Brice, thats right I have not applied the suspend2 patch. Just using Debian kernel source and compiling my own kernel; no patches. Bruce On Thu, May 31, 2007 at 08:46:05AM +0200, Brice Goglin wrote: > Bruce MacDonald wrote: > > Hi Brice, I use my own script. After unloading USB modules just do > > > > echo mem > /sys/power/state; > > > > Has worked really well for three years on my Dell Latitude D600 and > > still does except with the new package. > > > > In the kernel > > > > # CONFIG_SOFTWARE_SUSPEND is not set > > > > Is that related to suspend2? > > > > No, suspend2 is a patch that you have to apply to own-recompiled kernel. > So looks like you're not using suspend2 and the bug is not > suspend2-specific. > > Brice > -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#407036: xserver-xorg: Server crashes each time after hibernate with suspend2
Hi Brice, I use my own script. After unloading USB modules just do echo mem > /sys/power/state; Has worked really well for three years on my Dell Latitude D600 and still does except with the new package. In the kernel # CONFIG_SOFTWARE_SUSPEND is not set Is that related to suspend2? Bruce On Thu, May 31, 2007 at 08:25:56AM +0200, Brice Goglin wrote: > Lukas Ruf wrote: > >> Brice Goglin <[EMAIL PROTECTED]> [2007-05-30 00:38]: > >> > >> Hi guys, > >> > >> Does this crash of the X server after resume from hibernate with > >> suspend2 still occurs these days? Did you try with latest > >> xserver-xorg-core in unstable (2:1.3.0.0.dfsg-5) and your latest > >> driver? > >> > >> > > > > yes, I did. It crashed since then once. I.e. far more seldom than > > before. > > > > Ok. Did any of you try to hibernate with a plain vanilla kernel instead > of suspend2? > > Brice > -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#407036: xserver-xorg: Server crashes each time after hibernate with suspend2
Hi Brice, yes the X server still crashes every few times on resume from suspend to memory with that that xserver-xorg-core package. I have tried to attach gdb to the X server, in virtual console but can't get that to work. Killing xscreensaver helps but still doesn't work, so I can't get a crash report. Bruce On Wed, May 30, 2007 at 12:31:36AM +0200, Brice Goglin wrote: > Hi guys, > > Does this crash of the X server after resume from hibernate with > suspend2 still occurs these days? Did you try with latest > xserver-xorg-core in unstable (2:1.3.0.0.dfsg-5) and your latest driver? > > thanks, > Brice > -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#407036: xserver-xorg-core: dell d600 laptop crashes after resume from suspend to memory
Hi Brice, On Mon, May 14, 2007 at 10:52:16PM +0200, Brice Goglin wrote: > Bruce MacDonald wrote: > > Recently I noticed sometimes my laptop would crash on resuming from > > suspend to memory, which has worked well mostly for the last three > > years. The crashes became much more frequent with the latest upgrade > > to xserver-xorg-core (which is 2:1.3.0.0.dfsg-4), and I had to go back > > to 2:1.1.1-21 to get the crashes to pretty much stop. I need to try it > > for a few more days to be sure. I've also downgraded xorg and > > xserver-xorg from 1:7.2-3 to 1:7.1.0-18 and xserver-xorg-input-evdev > > from 1:1.1.2-6 to 1:1.1.5-2. Also have gone back to kernel 2.6.17 > > instead of 2.6.20; I haven't figured out whether the kernel is > > important; it takes at least a couple of days to evaluate a change > > because the problem is intermittent. > > > > Does it happen if you switch to a VT console before suspending? If not, > does X crash when you switch back from VT to X after resume? If so, > could you try to attach a gdb before that? Yes it seems like X crashes when I switch from the VT console to X after the resume. My laptop screen is always blank after resume in a VT console so its a little hard to tell, but I'm pretty sure the console comes up ok even when X crashes on the switch. However, before a suspend if I try to attach a gdb before switching from VT to X, gdb does not finish the attach; it never comes back and the X process takes all CPU. I've tried that before suspending so I am familar enough to do it blind after resume; I never did try it after a resume because gdb doesn't come back to the prompt. Is there something else I need to do to get the attach to work? > > > The laptop crashes and has to be rebooted. > > > > So, the laptop crashes, not only Xorg, right? If so, what makes you > think it is a Xorg problem, and not something else like the kernel. Even > if the problem occurs less often with xserver 1.1 than 1.3 with the same > kernel, the problem could be in the kernel and occur less often because > of what xserver does above. Hard to say... Good point; yes the laptop crashes and needs rebooting; I can't kill Xorg and I can't alt-ctl-del; I have to power it down. But after another day or so of testing it seems like the kernel isn't the problem. With the downgraded xorg packages, and upgraded (2.6.20) kernel, it seems just as fine as 2.6.17 kernel. And when I upgrade the xorg packages the intermittent crashing after resume comes back with either kernel, although it might be worse with 2.6.20 (its hard to be sure). Bruce > > Brice > -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
