Mysterious Freeze

Message

iarwain · Post by **iarwain** » Tue Jan 25, 2005 3:22 pm

A question for those with the "Type 1" problem:

Are you having the lock-ups with ext3?

I'm just curious. I've used reiserfs and xfs the last years. But now I'm testing another computer with nvidia and ext3 and I haven't had any freeze.

joehni · Post by **joehni** » Tue Jan 25, 2005 9:42 pm

kmare wrote:... but I'm really starting to think that it is (mostly) a problem with reiserfs (or a journaling problem). I've heard from some ppl that while they download a lot of stuff (the HD usage gets high) they see system lockup, corrupted file system and such... mostly with bittorrent... I don't know what else to check. Hopefully the problem will suddenly disappear...

Nope. Definately not with my system (using XFS). I have sporadic freezes, but can force them by setting the preemptive flag in the kernel and stop any OpenGL app.

joehni · Post by **joehni** » Tue Jan 25, 2005 9:43 pm

kmare wrote:yours is clearly a problem with the ati opengl 3d drivers. The problem that happens to most of us is that the freeze happens randomly (even without 3d accelleration enabled...)

Well, I am running NVidia with reproducable OpenGL freezes ...

lookinin · Post by **lookinin** » Tue Jan 25, 2005 9:54 pm

joehni wrote: Well, I am running NVidia with reproducable OpenGL freezes ...

Glad it's not just me - this started happening not long after I got direct rendering working properly... I was quite happy with my glxgears framerates, and xscreensaver was working great, but now it's useless junk taking up space on my hd.

spacesun · Post by **spacesun** » Wed Jan 26, 2005 7:42 am

in this therat i didnt found any solution
but i found a log entry in /var/log/messages

/var/log/messages:Jan 18 21:24:36 spacesun NVRM: Xid: 6, PE0000 05fc 80808080 00000000 80808080 80808080
/var/log/messages:Jan 19 11:35:05 spacesun NVRM: Xid: 6, PE0000 0f7c ef7cef7c 00000000 ef7cef7c ef7cef7c
/var/log/messages:Jan 19 20:16:12 spacesun NVRM: Xid: 6, PE0000 0400 ffff7bef 00000000 ffff7bef bdf7bdf7
/var/log/messages:Jan 20 19:59:18 spacesun NVRM: Xid: 6, PE0000 0400 ffffffff 00000000 ffffffff ff77ffff
/var/log/messages:Jan 25 08:25:38 spacesun NVRM: Xid: 6, PE0000 0400 ffffffff 00000000 ffffffff fef7f6d6
/var/log/messages:Jan 25 14:03:14 spacesun NVRM: Xid: 6, PE0000 047c 817e817e 017e817c 00000000 817e817e
/var/log/messages:Jan 25 22:05:41 spacesun NVRM: Xid: 6, PE0000 0400 ffffffff 00000000 ffffffff ffb7ff96
this say something so someone?
and after that the system hangs!

nvidia-kernel-1.0.6629-r1
nvidia-glx-1.0.6629-r1
kernel 2.6.9

i just installed a Gf4 Tii4200 of a frend of mine!
with the Gf4 mx 440 i never had problems!!

thx

davecs · Post by **davecs** » Wed Jan 26, 2005 11:46 am

I thought I may have seen the last of this one. Since I raised the RAM voltage by 0.1v, changed my keyboard and mouse, and upgraded kernel to 2.6.10, I had not had a Type 1 freeze for weeks. Then I got two last night. I found that by pressing ALT-SYSRQ-I I stop X and get a text login, I can even restart X by typing "kdm". However, a few things are different (a very verbose response to the text login) and one or two daemons seem to have stopped.

Maybe I should try moving to ext3. I've tried everything else! I don't think any one thing is to blame, I think the problem is an interaction between different things, and changing something, no matter how irrelevant it may seem, could be the answer. Who knows?

Dolio · Post by **Dolio** » Thu Jan 27, 2005 12:03 am

I changed to all ext3 a while ago, and it didn't help anything.

lookinin · Post by **lookinin** » Thu Jan 27, 2005 12:07 am

Has anyone been able to determine any other similarities? Are we all running console frame buffer? I haven't had time to completely go through this whole thread yet - is there a list to check my system against posted?

bunder · Post by **bunder** » Thu Jan 27, 2005 12:45 am

i made one earlier in the posting, but i am not 100% sure its accurate.

lookinin · Post by **lookinin** » Thu Jan 27, 2005 12:54 am

beugh wrote:i made one earlier in the posting, but i am not 100% sure its accurate.

We have a few similarities. VESA_TNG, FRAMEBUFFER_CONSOLE. FB is making me really curious. Anyone having OpenGL freezes without console FB?

firephoto · Post by **firephoto** » Thu Jan 27, 2005 1:04 am

No framebuffer or related items here, just the once in a blue moon freeze.

lookinin · Post by **lookinin** » Thu Jan 27, 2005 1:10 am

Well, I'll let my xscreensaver and stuff run and see what happens, but a 1600x1200 xmms OpenGL spectrum analyzer window running as I type this isn't locking up anymore.

I didn't completely remove console FB from my kernel, but I changed my lilo.conf's append= line from:

Code: Select all

append="gentoo=nodevfs init=/linuxrc apm=off acpi=on video=vesafb:ywrap,mtrr,1280x1024-16@60 splash=verbose,theme:emergence resume2=swap:/dev/hda11"

to:

Code: Select all

gentoo=nodevfs init=/linuxrc apm=off acpi=on resume2=swap:/dev/hda11

dibblethewrecker · Post by **dibblethewrecker** » Thu Jan 27, 2005 9:35 am

i recompiled without fbsplash - and i still get the crash reproduced using xscreensaver!

that means the other two changes i made: adding reiser4 and changing the tick have resulted in the problem returning but i agree that it is almost certainly a combination of factors that are different in everycase and probably no one app or feature is to blame - i even have a kernel with reiser4 and fb-tng running fine - it is the most stable of my recent disasters - i think will have to go back to nitro4 on 2.6.9 - that has been my only rock kernel so far! except of course Arch stock kernels, which make me a bit sleepy

additionally i checked and tuxracer does not seem to trigger the crash tho and i still get the xscreensaver crash when i disable glx in xorg.conf

davecs · Post by **davecs** » Thu Jan 27, 2005 2:30 pm

I decided to change something, so I recompiled my kernel last night without pre-emptible. Let's see how that goes.

One thing I noticed immediately is that glxgears ran a little faster but as soon as I ran something else it slowed right down. Who knows?

lookinin · Post by **lookinin** » Thu Jan 27, 2005 2:34 pm

I am not using preempt. After my last post, I thought maybe I'd better test things again with console FB re-enabled. And everything has been running fine since ... I'll keep playing with it, but so far, so good. *fingers crossed* maybe it mysteriously fixed itself with my last kernel upgrade or something. I'm running kernel 2.6.10-r5

Cimbom · Post by **Cimbom** » Sun Jan 30, 2005 7:38 pm

Hi

The same problems apear on my Machine with FC3 and Gentoo 2004 as I used nvidia-6629 Drivers with Kernel 2.6.9. I try everything but nothing works fine.Now i go back to 6111 and the live goes easy. Thats the only way to solve this Problem,my gues is NVIDIA-Drivers 6629 are bugy, we stay patient and hope for the next build.

davecs · Post by **davecs** » Mon Jan 31, 2005 8:00 pm

I had another Type 1 (mouse moving, everything else frozen, unpredicatable cause) freeze last night with my non-pre-empted kernel, and what's more, prism2_usb won't load, even though I rebuilt it!

Dolio · Post by **Dolio** » Mon Jan 31, 2005 9:35 pm

I must say I'm getting pretty convinced that type 1 crashes are glx related, even if you're not running a glx program at the time.

I've not experienced a type 1 crash in a while with the latest nvidia drivers, provided I have the composite extension enabled, which automatically disables glx. However, the one time I disabled composite recently, I got a type 1 freeze. glx seems to be the culprit, and it doesn't matter whether or not you're actually using it. Merely having it enabled seems to cause problems.

The oddity with this is: nvidia's proprietary drivers, and xorgs open-source drivers use different glx implementations, different xservers (I think) and different kernel interfaces. So what do they have in common (glx related) that's causing these freezes? I guess this should get escalated up to the xorg people, as they'd more likely know what common components could be failing.

As for my type 2 crashes, I'm beginning to think they're somehow heat related. I'm in a fairly small college dorm room, with lots of stuff that causes heat. Most of the time I get hard freezes when compiling something, which generates even more heat. However, leaving the window open and lowering the room temperature by 20 - 30 degrees seems to allow me to compile whatever. It's pretty tenuous, but it seems to correlate pretty well.

The only gap in my theory is that I don't know whether Windows freezes in hot situations. Of course, I don't compile anything on Windows, typically, so I guess I should go play some processor intensive games or something and see if it freezes. That would really point to hardware problems.

tlexii · Post by **tlexii** » Mon Jan 31, 2005 10:24 pm

I have had this recently start on a 2nd machine now after the first one mysteriously stopped crashing. I say mysteriously however I believe it may have been fixed when I recompiled glibc with nptl support (which I only did to fix GLX for doom3...so there's a possible glx connection).

Anyway, I recompiled glibc, cleaned and rebuilt the kernel, rebooted, removed all trace of nvidia and then rebuilt 6629. so far so good. I can't try the xscreensaver as this is a myth machine so its not installed.

Both are gcc-3.3.5, kernel 2.6.10-r1, agpgart, latest 6629 drivers.

Hopefully this might help some people but of course there are no guarantees

[edit] p.s. kernel headers 2.6.8.1

c0ns0le · Post by **c0ns0le** » Thu Feb 03, 2005 4:35 pm

Have you guys checked to make sure your using the correct gcc profile ie doing a gcc-config -c.

i just realized i was using the old 3.3 gcc profiles with a gcc-3.4. just a thought.

jrock2468 · Post by **jrock2468** » Thu Feb 03, 2005 7:43 pm

Wow, this problem has gone on a LONG time.

Just recently hit me after emerging world. Question for y'all:

What user is starting your X session? For me, gdm is starting Xorg as root (bad, I know). Just trying to get more info for troubleshooting this.

For the record, I'm using 2.6.7 w/ nvidia drivers on dual head display.

ikaro · Post by **ikaro** » Thu Feb 03, 2005 10:01 pm

I think some of you are getting lost to what the thread is about.
Running Experimental kernels with a bunch of patches can only result in disaster, and In no way can the problems some of you are experiencing be related with the other described by users which systems are running on software marked as stable, and with vanilla kernels.

Adding to my previous posts in this thread, today ( just some minutes ago ) i installed xfree-4.3.99.902-r2.
Until now its working ok, but lets see how it goes ... usually i cant use the browser for more than 10 minutes without the computer hardlocking.

yaneurabeya · Post by **yaneurabeya** » Fri Feb 04, 2005 6:30 am

Ok, as for some of the issues, it appears for my bug (Console freezes on logout), has been discovered to be caused by power support (APM in my case). I tried the latest kernel sources-2.6.11-r3-and the latest nvidia and Xorg sources and I still come up with a hard freeze when I exit XFCE. When I don't compile APM support it, it works just fine.

Trouble settings:
-Enable APM at boot time
-Enable console blanking
-Use real mode APM BIOS call to power off

I would try not compiling the console blanking in, but I've thrashed my fs'es enough for 1 day (4 times now...). nv drivers don't yield any different results than the nvidia drivers do for me, so those drivers aren't the issue and nvidia wasn't loadable under 2.6.11 (haha... because it 'couldn't compile'), so it's not nvidia related at all. It appears to more or less be a power management issue-like I said before-than anything else.

ikaro · Post by **ikaro** » Fri Feb 04, 2005 3:16 pm

Still running Xfree since yesterday and Its been working great.
Ive been using the scrool and clicking on alot of links on all the website i visit, but still no crash, which is good

I submitet a bug yesterday just to give some more awareness about these problems.
http://bugs.gentoo.org/show_bug.cgi?id=80642
In it are links to some forums threads(this included).

davecs · Post by **davecs** » Sun Feb 06, 2005 7:49 pm

Dolio wrote:I must say I'm getting pretty convinced that type 1 crashes are glx related, even if you're not running a glx program at the time.

I've not experienced a type 1 crash in a while with the latest nvidia drivers, provided I have the composite extension enabled, which automatically disables glx. However, the one time I disabled composite recently, I got a type 1 freeze. glx seems to be the culprit, and it doesn't matter whether or not you're actually using it. Merely having it enabled seems to cause problems.

The oddity with this is: nvidia's proprietary drivers, and xorgs open-source drivers use different glx implementations, different xservers (I think) and different kernel interfaces. So what do they have in common (glx related) that's causing these freezes? I guess this should get escalated up to the xorg people, as they'd more likely know what common components could be failing.

That's what I suspect, too. Definitely GLX is the common factor I have had through different computers and systems.

I got a type 1 lockup about an hour ago, first for about a week, so obviously removing pre-emptible from the kernel does not help, and just prevents the prism2_usb modules from loading. (Even recompiling it against the changed kernel) Not that I can get my wlan working anyway, but that's another subject!

Mysterious Freeze

Did somone solve the hang problem?

GCC version