180.* graphical corruption and freezes system

[Rudd-O]

I spoke too soon. Eventually, after a few hours of use, I get the NVRM messages and the repaint errors and then the GUI lock.

[Rudd-O]

Originally Posted by olivn

I'm now running 2.6.28 with volontary preempt and without cgroups[/code]

I experienced the same errors again. Same voluntary preempt and no cgroups. All I had to do is play unreal with software rendering for a few hours, then launch compiz and glxgears. BOOM. NVRM crap.

[helmotz]

Originally Posted by psychok9

Strange: today I've got corruption on Wine + WOW (OpenGL) also with 177.80...
Same 180.xx.

I'm experiencing similar freezes (without corruption) with both 180 and 178. I reverted yesterday to 173.14.15 and I still don't see any freezes.
Did you try reverting back to 173.14.15?

F.

[Rudd-O]

Originally Posted by helmotz

I'm experiencing similar freezes (without corruption) with both 180 and 178. I reverted yesterday to 173.14.15 and I still don't see any freezes.
Did you try reverting back to 173.14.15?

F.

I cannot compile 178.xx with 2.6.28, and 173 does not support my GPU (8800 GT). Thus, so far I have only used 180.18.

I am running out of options here.

[Rudd-O]

Just running glxgears alone (no compositing manager loaded) does not produce GUI freeze, but horizontal "stripes" of black color seem to flash on toppa the gears. This is even AFTER restarting the X server when the GUI lockup happened.

Jan 4 07:17:06 karen kernel: NVRM: Xid (0003:00): 25, 0004 0000502d 0000047e 00000472 00000180 00000462
Jan 4 07:17:06 karen kernel: NVRM: Xid (0003:00): 25, 0004 0000502d 0000047e 00000472 00000180 00000462
Jan 4 07:17:09 karen kernel: NVRM: Xid (0003:00): 25, 0004 0000502d 0000047e 00000472 00000180 00000462
Jan 4 07:17:09 karen kernel: NVRM: Xid (0003:00): 25, 0004 0000502d 0000047e 00000472 00000180 00000462
Jan 4 07:17:22 karen kernel: NVRM: Xid (0003:00): 25, 0004 0000502d 0000047e 00000472 00000180 00000462
Jan 4 07:17:22 karen kernel: NVRM: Xid (0003:00): 25, 0004 0000502d 0000047e 00000472 00000180 00000462

My observations lead me to conclude there is a problem in the 3D part of the driver that, when coupled with compositors, breaks the compositor and causes it to freeze.

Unreal tournament over wine, using the opengl driver, usually works fine, accelerated, for up to half an hour, then exits with an UnrealEngine error dialog box.

Come on guys, I am trying to use my computer and play a game from 1999 here, and I got myself a $250 card to NOT be able to do it?

[ledoc]

Originally Posted by Rudd-O

I cannot compile 178.xx with 2.6.28, and 173 does not support my GPU (8800 GT). Thus, so far I have only used 180.18.

I am running out of options here.

There is a patch somewhere in the forum to make both 177.xx (I suppose you mean those) compile with 2.6.28.

[spectred]

Originally Posted by ledoc

There is a patch somewhere in the forum to make both 177.xx (I suppose you mean those) compile with 2.6.28.

http://www.nvnews.net/vbulletin/show...52&postcount=3

[Rudd-O]

Originally Posted by spectred

http://www.nvnews.net/vbulletin/show...52&postcount=3

Whoooo :-) Thanks man. Lemme give it a shot.

[Rudd-O]

OK, maybe I'm speaking too soon, but after installing the patched 117.82 driver, 3D is finally stable, no NVRM messages, even when running glxgears along with Compiz. Note that my kernel 2.6.28 is vanilla, is without cgroups and with only voluntary preemption.

Edit: I spoke too soon. glxgears simply aborts after a few minutes running (the NVRM Xid messages in the syslog), and I don't have to tell you what happens if Compiz is running concomitantly.

gdb backtrace of glxgears:

Program received signal SIGABRT, Aborted.
0x00007ffff7819f05 in raise () from /lib64/libc.so.6
(gdb) bt
#0 0x00007ffff7819f05 in raise () from /lib64/libc.so.6
#1 0x00007ffff781ba73 in abort () from /lib64/libc.so.6
#2 0x00007ffff7e4dce9 in _nv000519gl () from /usr/lib64/libGL.so.1
#3 0x00007ffff6d9f09e in ?? () from /usr/lib64/libGLcore.so.1
#4 0x00007ffff6b11286 in _nv000131gl () from /usr/lib64/libGLcore.so.1
#5 0x00007ffff7e52685 in _nv000519gl () from /usr/lib64/libGL.so.1
#6 0x00007ffff7e6b321 in glXSwapBuffers () from /usr/lib64/libGL.so.1
#7 0x0000000000402f90 in XParseGeometry ()
#8 0x00007ffff7805576 in __libc_start_main () from /lib64/libc.so.6
#9 0x0000000000401589 in XParseGeometry ()
#10 0x00007fffffffdef8 in ?? ()
#11 0x000000000000001c in ?? ()
#12 0x0000000000000001 in ?? ()
#13 0x00007fffffffe274 in ?? ()
#14 0x0000000000000000 in ?? ()

This is not funny or challenging anymore.

Originally Posted by Rudd-O

Whoooo :-) Thanks man. Lemme give it a shot.

[FranciscoB]

Dear NVidia's Devs,

I am here to add my own 'I am a customer, and suffering from this problem as well'.

I have had problems with both Gnome, and KDE. KDE will suffer graphics corruption, and require a hard reset.

I bought a system with a nvidia card due to good experiences at work, but right now my 3 year old laptop with (very cheap) Intel graphics run better than my brand new computer.

[Trail]

Funny thing, I just had a freeze when trying to post this message.

I myself don't get any xorg.log message, but I can find these in the syslog:

Jan 5 14:56:22 linux-b3sy kernel: NVRM: Xid (0001:00): 6, PE007f
Jan 5 14:56:22 linux-b3sy kernel: NVRM: Xid (0001:00): 62, CMDre 00000000 0000089c 0100cb18 00000007 00000000
Jan 5 14:56:33 linux-b3sy kernel: NVRM: Xid (0001:00): 8, Channel 00000001




# not sure if first 2 lines are relevant
Jan 5 15:01:56 linux-b3sy kernel: CE: hpet increasing min_delta_ns to 22500 nsec
Jan 5 15:21:56 linux-b3sy -- MARK --
Jan 5 15:26:31 linux-b3sy kernel: NVRM: Xid (0001:00): 13, 0004 00000000 00008297 00001310 00000000 00000040
Jan 5 15:39:19 linux-b3sy kernel: NVRM: Xid (0001:00): 6, PE007f
Jan 5 15:39:19 linux-b3sy kernel: NVRM: Xid (0001:00): 6, PE0001
Jan 5 15:39:49 linux-b3sy kernel: NVRM: Xid (0001:00): 62, CMDre 00000000 0000089c 0100cb24 00000007 00000000
Jan 5 15:39:49 linux-b3sy kernel: NVRM: Xid (0001:00): 6, PE0002
Jan 5 15:39:49 linux-b3sy kernel: NVRM: Xid (0001:00): 6, PE0003
Jan 5 15:39:49 linux-b3sy kernel: NVRM: Xid (0001:00): 6, PE0004
Jan 5 15:40:19 linux-b3sy kernel: NVRM: Xid (0001:00): 6, PE007f
Jan 5 15:40:20 linux-b3sy kernel: CE: hpet increasing min_delta_ns to 33750 nsec
Jan 5 15:40:20 linux-b3sy kernel: CE: hpet increasing min_delta_ns to 50624 nsec

After both freezes, magic sysrq works, but nothing else.

I tried setting compositing to disabled in xorg configuration for now.

[Rudd-O]

Originally Posted by Trail

I tried setting compositing to disabled in xorg configuration for now.

Did turning off compositing solve your stability problems?

Please note that the problems I experienced present themselves without compositing as well, but compositing accelerates their presentation.