[11:39] <alkisg_web> Hi, my xorg has hanged with 100% CPU usage. With strace -p `pidof X` from an ssh connection I see it's on a clock_gettime/settimer loop: http://alkisg.pastebin.com/hGGVYdyt
[11:39] <alkisg_web> Any hints before I start trying to kill apps in case it recovers?
[11:43] <tjaalton> strace isn't useful, see https://wiki.ubuntu.com/X/Debugging instead
[11:44] <alkisg_web> Thank you, looking..
[11:44] <tjaalton> backtracing in particular
[11:52] <alkisg_web> "Now do what you need to make the X server crash. Or, if the problem is that the X server is locked up and doesn't react, stop it with ctrl-C. Now get a backtrace: "
[11:52] <alkisg_web> What if Ctrl+C doesn't stop it?
[11:53] <alkisg_web> Ctrl+Z suspends the task, but Ctrl+C doesn't do anything...
[11:55] <tjaalton> hmm, dunno
[12:01]  * alkisg_web even killed -9 Xorg but the gdb process still wouldn't accept ctrl+C, only ctrl+Z... I'll try again on the next crash (it's too often with this openchrome driver :()
[16:45] <Azelphur> Sarvatt: no crashes in 2 days, your patch is definitely working :)
[16:45] <Azelphur> you should commit it :D
[16:46] <jitaki> hello
[16:49] <Sarvatt> Azelphur: it needs more review first, your case is rather special and it has a high probability of breaking other things :)
[16:49] <Azelphur> hehe ok
[16:49] <Azelphur> Sarvatt: it's getting pretty heavy testing by me, as this is the houses MythTV box
[16:49] <Azelphur> so 24/7 on with multiple users, it gets a lot of use, if I see any problems I'll let you know :)
[17:08] <hyperair> Sarvatt: update: when compiz hangs on XGetSomethingOrOtherThatShouldNotHang, Xorg uses 100% CPU.
[17:08] <hyperair> or near there anyway
[17:08] <hyperair> Sarvatt: gdb shows that the bottom of the stack trace always shows xxvi, and the top is always in libpixman
[21:53] <alkisg> The stock Lucid openchrome driver, 1:0.2.904+svn827-1, crashes my PC. If I boot it with Debian Squeeze (openchrome 1:0.2.904+svn842-2), it doesn't crash. Is there any way to update only that specific driver in Lucid and test if that is indeed the cause/fix of my problem?
[21:54] <alkisg> (it crashes after a few minutes of usage, but I was unable to get a backtrace, gdb wouldn't accept ctrl+c)
[22:20] <bjsnider> alkisg, the openchrome driver has not had any code changes upstream (if there is such a thing in regards to that driver) since may 2k10
[22:21] <alkisg> bjsnider: in the changelog I see some fixes that relate to hangs, which may be what I experience
[22:21] <alkisg> From svn827, the lucid version, to svn842, the squeeze version
[22:21] <bjsnider> ok, so debian has a newer version
[22:22] <alkisg> I looked at xorg-edgers ppa but I don't see a new version to try out there..
[22:22] <alkisg> The maverick or natty version would also do, but the packages are not compatible - can I just recompile those?
[22:23] <alkisg> E.g. send them to my ppa on launchpad and have them compiled there, and then install the .deb?
[22:25] <bjsnider> you could do that with the squeeze package if you think it will help
[22:25] <bjsnider> lucid would have an older version because it came out before may 2k10
[22:25] <bjsnider> you could upgrade the whole system to maverick
[22:27] <alkisg> It's a thin client so I'd have to upgrade my server to maverick too :(
[22:27] <alkisg> The squeeze package has a dependency on xorg abi 6.0 or something like that, which in not satisfiable on lucid, do I have to change some constant in the code before compiling?
[22:31] <alkisg> ...anyway the most important question is this, suppose I do verify that the lucid openchrome package has some serious bug and an it hangs for some models, would it be possible to get a backport for the package?
[22:44] <alkisg> Thank you, bb