RAOFAug 22 17:59:50 Faye kernel: [  299.663018] [Hardware Error]: Machine check events logged.  Woo!09:44
RAOFThat's why she's called Faye :)09:44
=== ara_ is now known as ara
SarvattRAOF: every sandybridge system i've seen does that in 3.0+, and that one was named asuka before you named it faye :)15:14
jcristaui think #831336 should be reassigned somewhere that's not xterm17:43
jcristaulike unity or something17:43
brycehjcristau, did you look at GdmLog2.txt?18:00
bryceh*** glibc detected *** /usr/bin/Xorg: free(): corrupted unsorted chunks: 0x0000000002b844c0 ***18:00
brycehhowever I can't tell that he was actually using gdm at the time of the crash rather than lightdm18:02
jcristaubryceh: ah, no, didn't see that one, i only looked at the most recent logs18:07
brycehthat was the only one I could find with an actual error18:07
jcristaui assumed the error, if any, would be in .xsession-errors18:08
jcristauwhich isn't in there i think18:08
brycehnope; for xterm I should have that be automatically gathered18:08
jcristauit's more likely to have private data than the other logs i guess18:09
DinkI was looking through xserver-xorg bugs and not sure if what I am experiencing is related to some "fixed" bugs #778490, #774978, #740785,#775705. When I start FF6 everything crashed and goes back to gdm login.  Currently xserver-xorg-input-synaptics version 1.3.99+git20110116.0e27ce3a-0ubuntu12.118:43
DinkFF6 works fine when I run ubuntu in safe mode18:43
DinkThis is 64bit natty running in vm18:48
DinkAll I can get is18:54
Dinkconnect(4, {sa_family=AF_INET, sin_port=htons(6000),sin_addr=inet_addr("")}, 16) = -1 ECO18:54
DinkNNREFUSED (Connection refused)18:54
brycehDink, collect a full backtrace (see http://wiki.ubuntu.com/X/Backtracing for details)19:37
=== yofel_ is now known as yofel
Dinkbryceh: I tried but before I could type backtrace while FF6 was running it killed X21:55
brycehDink, heh, do it from console or ssh21:56
Dinkhmm so open up ff6 via gui while on a ssh/console session then try gdb ?21:57
Dinkstill not sure ff6 will stay open log enough but will try 21:57
brycehDink, no you need to run X within gdb.  See the url I gave you for directions how.21:59
Dinkcool got it22:02
DinkWhere do you want it ?22:03
brycehDink, in a bug report22:04
Dink(gdb) backtrace full22:04
Dink#0  0x00007f201f835335 in PrlCtlUpdateGLXSize ()22:04
Dink   from /usr/lib/xorg/modules/drivers/prlvideo_drv.so22:04
DinkIs that the stuff that should be outputted ?22:04
RAOFYeah, that's the sort of thing which should be outputted.  What's prlvideo? :)22:05
brycehRAOF, sounds like a driver we don't support ;-)22:05
DinkParallels 22:06
RAOFbryceh: It does indeed, yes :)22:06
DinkI am running this in Parallels 22:06
brycehvirtualized thingee22:06
brycehDink, ok you need to file a bug with Parallels22:06
RAOFDink: So, it seems likely that the Parallels X driver has a bug; Firefox is generally excellent at triggering such things.22:06
Dinkbryceh: ok thanks will do. 22:07
brycehsometimes with these virtualized video drivers, they need to rebuild their drivers to match what we're shipping in ubuntu22:07
DinkRAOF: thanks for the info. 22:07
brycehI don't know if that's what's happening here, but that's the most common issue we see with these drivers22:07
brycehe.g. GLXSize makes me wonder if they need to update the prl driver to support mesa 7.11 which RAOF just put in relatively recently22:08
brycehRAOF, heya22:09
RAOFbryceh: Good morning / evening :)22:09
RAOFHm.  Or maybe afternoon?22:09
brycehRAOF, speaking of drivers we don't support,  I wanted to get your thoughts on bug #81500022:09
brycehyep 3pm; High Naptime22:10
brycehRAOF, I'm thinking we should bump that over to the kernel, but do you think there's anything we would/could/should do on the X end?  22:12
brycehick... in more troubling news, looks like xvfb broke again... https://bugs.launchpad.net/~doko/+reportedbugs (bug #829470)22:13
RAOFI think you're right, and that should be a kernel bug.  I'm pretty sure there should be a KMS poulsbo driver, but it doesn't seem to be loading.22:13
brycehalright, I'll bump it22:13
RAOFHm.  Which of doko's bugs are you referring to?22:14
brycehbut I'd bet most of those are outcomes of broken xvfb22:15
brycehso far not able to repro on my hw22:16
RAOFI'll see if it does here.22:18
brycehxvfb in general seems not to fail (launching xterm or whatever).  Going to try the build test in pygtk next22:19
brycehRAOF, do you remember when we talked about maybe adding xvfb to the xorg-server build scripts, to catch these types of issues ourselves?  Know if that idea went anywhere?22:20
RAOFYeah.  It has not yet gone anywhere.22:21
brycehXlib:  extension "RANDR" missing on display ":99".22:33
brycehhmm perhaps that's innocuous22:41
RAOFxvfb does not support RANDR, and I'm fairly sure that GTK no longer blows up if it's running on a server which does not support RANDR.22:43
brycehyeah running the test locally I'm noticing the fatal error is actually  TypeError: Unsupported type: <class 'gtk._gtk.WindowType'>22:44
RAOFSo this doesn't seem to be an xvfb bug, but rather a failing test that's not reported well.22:44
brycehprobably some gnome3 transition problem.22:50
RAOFYeah, may well be.22:50
brycehRAOF, we're still getting a lot of those false gpu hang bugs; thankfully real gpu hangs seem to be extremely rare22:53
brycehon one of my upstream reports someone said they see it on ubuntu but not mandriva (or whatever that distro is called these days).22:54
brycehRAOF, otoh apw seems not sure what else to look at on our end.  we're running short on ideas22:56
RAOFbryceh: As in: we're running out of bugs for the kernel team to fix?!22:56
RAOFOh, or that the GPU lockup doesn't seem to occur on Mandriva, and they're running the same kernel?22:57
brycehhowever as it's (usually) a symptomless-hang, you'd not notice it on a non-Ubuntu system unless you were specifically looking for it in dmesg22:58
RAOFHah.  Right.22:58
RAOFI presume apw has got the testers to try an upstream / drm-intel-next kernel?23:00
brycehwe've done that before, although not recently (if there's a particular fix you're thinking of)23:01
brycehwhen he'd investigated before, he felt it wasn't the kernel code that was bad but rather an issue with how the drivers were being loaded23:02
brycehickle says the bug itself is an invalid memory issue, which makes me wonder if it's some sort of race condition during driver initialization23:03
brycehthen by the time the driver has faulted and reset, the memory is kosher (or gets re-initted)23:03
brycehI'm also wondering if it has to do with something done during the great bootspeed optimization campaign23:04
brycehunfortunately the gpu hang tools didn't really capture this properly until well after that work was done, so hard to correlate to that23:05
brycehof course, I can't reproduce it on any of my machines23:06
brycehsuch is the curse of being an X maintainer ;-)23:06
brycehI've been trying to look for commonalities between different reporters, like if they have specific proprietary kernel modules loaded, or similar hardware or something.  But from the recently reported bugs there's quite a bit of variance from person to person23:07
RAOFIs it the same GPU at least?23:09
brycehcurrent reports are against i915, 1945, 196523:11
brycehI don't see it on my i94523:11
=== techmik67 is now known as techmik
RAOFWell, that's a nice wide range at least. :)23:15
bryceheven the arrandale chris van hoof gave me doesn't show it; and it's positively thick with bugs at the moment23:16
bryceh#830276 looks like a false gpu hang but appears to actually hang23:18
brycehalright, enough bug work for the day... food, then xdiagnose.  bbiab23:21
SarvattIPEHR: 0x7xxxxxxx have always been mesa problems if i'm remembering right, IPEHR: 0x018xxxxx were always hangs during dpms cycles23:36
brycehSarvatt, ooh good to know; would that happen to be documented anywhere that you know?23:37
Sarvattnope just trends i noticed going through hundreds of bugs, we get a rash of 0x018xxxxx dpms bugs right after release every single release since lucid23:38
Sarvattwell its documented in the gpu docs23:38
RAOFAs in the register specifications?23:38
Sarvattyeah, why am I not finding it again..23:41
RAOFThis might be good to be documented on a wiki page?23:45

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!