X crash after security updates

gustavoc · 02-13-2014, 10:42 AM

Hi All,

Last week (6 feb) I applied the latest patches using slackpkg. After the weekend, I suffered several X crashes after session locks and also in a "random" manner. The logs (/var/log/messages) showed the following:

Feb 13 11:34:33 dragon kernel: [ 6287.760161] nouveau E[ PGRAPH] 0000:02:00.0] ILLEGAL_MTHD
Feb 13 11:34:33 dragon kernel: [ 6287.760166] nouveau E[ PGRAPH][0000:02:00.0] ch 2 [0x001fb34000 X[1896]] subc 0 class 0
x0030 mthd 0x0208 data 0x00000080
Feb 13 11:34:33 dragon kernel: [ 6287.760181] nouveau E[ PGRAPH][0000:02:00.0] ILLEGAL_MTHD
Feb 13 11:34:33 dragon kernel: [ 6287.760185] nouveau E[ PGRAPH][0000:02:00.0] ch 2 [0x001fb34000 X[1896]] subc 0 class 0
x0030 mthd 0x020c data 0x00000016
Feb 13 11:34:33 dragon kernel: [ 6287.760201] nouveau E[ PGRAPH][0000:02:00.0] ILLEGAL_MTHD
Feb 13 11:34:33 dragon kernel: [ 6287.760205] nouveau E[ PGRAPH][0000:02:00.0] ch 2 [0x001fb34000 X[1896]] subc 0 class 0
x0030 mthd 0x0210 data 0x00000001

Looking at the new packages, I suspected the libXfont 1.4.7 update was the one causing problems. Downgraded to 1.4.6 (original 14.1) and the problem is solved.
Don't know if anyone faced this, please comment.

Regards.

Gustavo

(my system: AMD AthlonX2 3800+, 2 GB Ram, nvidia G210 graphics card (running nouveau drivers), asus M2N68 main board, Slackware 14.1)

gustavoc · 02-13-2014, 02:04 PM

Sorry! After about 30 mins the problema appeared again... will keep investigating.

Thanks!

metaschima · 02-13-2014, 02:11 PM

I think the problem is nouveau, not Xorg, but I can't be 100% sure. Anything in Xorg.0.log ?

You could also try updating the nouveau drivers and see if it helps.

colorpurple21859 · 02-13-2014, 02:53 PM

It wouldn't happen to be happening when using firefox? If so use seamonkey. Have a similar problem, but haven't had time to figure it out.

BratPit · 02-14-2014, 02:16 AM

Hi

Check cairo update.

After upgrade on my 14.1 multilib /Nvidia/ I have 3 times frozen system /X/, so I downgrade it the problem has disappeared now.

This update fixes only some issues in gimp . Old cairo works smoothly too /no issues on gimp/. So no problem downgrading .

metaschima · 02-14-2014, 12:10 PM

The newer cairo is compiled with gl support, so it could be cairo.

gustavoc · 02-14-2014, 12:36 PM

Hi All,

Thanks for the replies. I also tried downgrading cairo to 1.12.14 and the machine worked ok all day .... till now, again screen garbaged, X crashed, rebooted from console.
I also think that there's something wrong with nouveau driver buy since my system never had this problem before, I suspected of the recent patches applied.
Attached syslog and Xorg.log. Nothing suspicious in messages.

Thanks!

Gustavo.

gustavoc · 02-14-2014, 12:41 PM

By the time it took to reboot and rewrite my reply, metaschima and BratPit posted their replies about cairo. The difference I saw between yesterday (newer cairo) and today is that yesterday my machine crashed every hour or so, today only once. Didn't made any research about cairo. I hope this weekend i have some time. (at work now ...)
Thanks!

Gustavo.

metaschima · 02-14-2014, 01:32 PM

I would try newer nouveau drivers. It could still be Xorg.

phenixia2003 · 02-14-2014, 01:40 PM

Hello,

Quote:

Originally Posted by gustavoc

Hi All,

Thanks for the replies. I also tried downgrading cairo to 1.12.14 and the machine worked ok all day .... till now, again screen garbaged, X crashed, rebooted from console.
I also think that there's something wrong with nouveau driver buy since my system never had this problem before, I suspected of the recent patches applied.
Attached syslog and Xorg.log. Nothing suspicious in messages.

Thanks!

Gustavo.

Don't know if this is related to your issue, but there's something weird in the Xorg.log you attached :

Code:

(EE) [mi] EQ overflowing.  Additional events will be discarded until existing events are processed.
(EE) 
(EE) Backtrace:
(EE) 0: /usr/bin/X (xorg_backtrace+0x49) [0x81d6099]
(EE) 1: /usr/bin/X (mieqEnqueue+0x21b) [0x81b616b]
(EE) 2: /usr/bin/X (QueuePointerEvents+0x6d) [0x80911cd]
(EE) 3: /usr/bin/X (xf86PostMotionEventM+0x24b) [0x80ca34b]
(EE) 4: /usr/lib/xorg/modules/input/evdev_drv.so (0xb6002000+0x4b7f) [0xb6006b7f]
(EE) 5: /usr/bin/X (0x8048000+0x71cb5) [0x80b9cb5]
(EE) 6: /usr/bin/X (0x8048000+0x9c4a7) [0x80e44a7]
(EE) 7: linux-gate.so.1 (__kernel_sigreturn+0x0) [0xffffe400]
(EE) 8: /lib/libc.so.6 (ioctl+0x14) [0xb7455064]
(EE) 9: /usr/lib/libdrm.so.2 (drmIoctl+0x40) [0xb76f8c60]
(EE) 10: /usr/lib/libdrm.so.2 (drmCommandWrite+0x3c) [0xb76fb5dc]
(EE) 11: /usr/lib/libdrm_nouveau.so.2 (nouveau_bo_wait+0xa5) [0xb726e145]
(EE) 12: /usr/lib/libdrm_nouveau.so.2 (nouveau_bo_map+0x33) [0xb726e1a3]
(EE) 13: /usr/lib/xorg/modules/drivers/nouveau_drv.so (0xb7776000+0x4f1d) [0xb777af1d]
(EE) 14: /usr/lib/xorg/modules/libexa.so (0xb7226000+0x5316) [0xb722b316]
(EE) 15: /usr/lib/xorg/modules/libexa.so (0xb7226000+0x57f6) [0xb722b7f6]
(EE) 16: /usr/lib/xorg/modules/libexa.so (0xb7226000+0x773f) [0xb722d73f]
(EE) 17: /usr/lib/xorg/modules/libexa.so (0xb7226000+0x3f8f) [0xb7229f8f]
(EE) 18: /usr/lib/xorg/modules/libexa.so (0xb7226000+0x780b) [0xb722d80b]
(EE) 19: /usr/lib/xorg/modules/libexa.so (0xb7226000+0x2ac7) [0xb7228ac7]
(EE) 20: /usr/lib/xorg/modules/drivers/nouveau_drv.so (0xb7776000+0xe1c0) [0xb77841c0]
(EE) 21: /usr/bin/X (BlockHandler+0x49) [0x807d919]
(EE) 22: /usr/bin/X (WaitForSomething+0xf8) [0x81d3598]
(EE) 23: /usr/bin/X (0x8048000+0x3129e) [0x807929e]
(EE) 24: /usr/bin/X (0x8048000+0x1f69a) [0x806769a]
(EE) 25: /lib/libc.so.6 (__libc_start_main+0xf3) [0xb73907a3]
(EE) 26: /usr/bin/X (0x8048000+0x1fa69) [0x8067a69]
(EE) 
(EE) [mi] These backtraces from mieqEnqueue may point to a culprit higher up the stack.
(EE) [mi] mieq is *NOT* the cause.  It is a victim.
[ 21191.004] [mi] Increasing EQ size to 512 to prevent dropped events.
[ 21191.004] [mi] EQ processing has resumed after 67 dropped events.
[ 21191.005] [mi] This may be caused my a misbehaving driver monopolizing the server's resources.
[ 21191.006] nouveau_exa_upload_to_screen:384 - falling back to memcpy ignores tiling

Furthermore, in the syslog you attached, there's a bunch of errors reported :

Code:

Feb 14 13:57:55 dragon-bak kernel: [21020.822380] nouveau E[   PFIFO][0000:02:00.0] CACHE_ERROR - ch 2 [X[946]] subc 3 mthd 0x1a00 data 0x00001111
Feb 14 13:57:55 dragon-bak kernel: [21020.822404] nouveau E[   PFIFO][0000:02:00.0] CACHE_ERROR - ch 2 [X[946]] subc 3 mthd 0x153c data 0x00000000
Feb 14 13:57:55 dragon-bak kernel: [21020.822421] nouveau E[   PFIFO][0000:02:00.0] CACHE_ERROR - ch 2 [X[946]] subc 3 mthd 0x1650 data 0x000000ff
Feb 14 13:57:55 dragon-bak kernel: [21020.822433] nouveau E[   PFIFO][0000:02:00.0] CACHE_ERROR - ch 2 [X[946]] subc 3 mthd 0x1654 data 0x00000000
Feb 14 13:57:55 dragon-bak kernel: [21020.822445] nouveau E[   PFIFO][0000:02:00.0] CACHE_ERROR - ch 2 [X[946]] subc 3 mthd 0x16b8 data 0x00000020

You should try a newer version of nouveau, or the proprietary nvidia driver.

--
SeB