X.org Lockups

MetatronX99 · n00b Joined: 24 Jul 2004 Posts: 70 Location: Warren, MI

I found that GCC 3.3.3-rX was what was causing Firefox to crash for me.

Firefox is both stable on GCC 3.3.4 and 3.4.1, for me at least.
_________________
GCC 3.4.2-r2
Glibc 20040808 +NPTL +Prelink
Linux 2.6.9-rc2-love4
~x86 System

GenKreton · Posted: Sat Aug 07, 2004 9:42 pm Post subject:

creyato · Tux's lil' helper Joined: 29 Jul 2004 Posts: 93

dudes, i found out why our comps have been freezing! in your kernel disable IO_APIC (i had to redit my config file because i couldnt see it in menuconfig). Im on a athlon64 system, and its been running for 19hrs now and it always wouldev stopped by now! the guy who said this worked for him was on an x86 system so this should work for everyone, good luck!!

GenKreton · Posted: Sun Aug 08, 2004 1:28 pm Post subject: Re: ...

creyato · Tux's lil' helper Joined: 29 Jul 2004 Posts: 93

weird wonko · Tux's lil' helper Joined: 05 Jan 2003 Posts: 141

I'm also on a Athlon-64 and had IO_APIC enabled. I turned it off now, let's see if this helps. I had no freezes during the last week, though, this error doesn't hapen too often at the moment.

snuffer · Apprentice Joined: 12 Jun 2004 Posts: 207 Location: Israel

Well it seems that IO_APIC is not teh problem on my computer because i have it disabled. im using gentoo-dev-sources-2.6.7-r11.
_________________
Registered Linux User #360735.

pandaxiongmao · Guru Joined: 29 Sep 2003 Posts: 478 Location: USA

This problem really makes me nervous every time I use my mouse to scroll the website.
_________________
CPU: Intel Core 2 Duo
GPU: nVidia GeForce 9800 GT
MB: Asus P5N-E SLI

creyato · Tux's lil' helper Joined: 29 Jul 2004 Posts: 93

CompNerd · Posted: Mon Aug 09, 2004 6:13 am Post subject:

Well, I am just a mess of the applications. Grellm2, Firefox, fluxbox, ati video card and yes, CONFIG_X86_IO_APIC=y in my kernel. This annoying bug seems to happen randomly (although, normally when compiling or doing another intensive task).

The IO APIC makes sense as being a possible culprit. IO APIC basically SMP's Interrupts. I am guessing that its loosing an interrupt, and thus seriously crashing an application, rendering the system useless (relatively).

One thing that I noticed was that my hdd would grind away when this happened. Was wondering if anyone else had that symptom. I suppose I will be disabling IO_APIC.

CompNerd

dju` · Retired Dev Joined: 11 Feb 2003 Posts: 370

I have the problem too, sometimes X freezes, can't move mouse or use keyboard, but my box still answers to ssh.
now I have rebooted, launched X, returned back to console with ALT+F1, and waited. I have 3 days uptime now, no random hang up appears. So I believe its a kernel/nvidia-driver related issue, maybe 4k stacks, because it seems only to happen when I'm on the X display. I'm running latest gentoo-dev-sources and nvidia kernel.

CompNerd · Posted: Wed Aug 11, 2004 12:52 pm Post subject:

Well the system so far has been behaving (knock on wood) after the removal of IO_APIC from the kernel. Should check that.

snuffer · Apprentice Joined: 12 Jun 2004 Posts: 207 Location: Israel

CompNerd are u sure theres supposed to be a line 'CONFIG_X86_IO_APIC' because i dont have it i have 'CONFIG_X86_UP_IOAPIC'
_________________
Registered Linux User #360735.

CompNerd · Posted: Thu Aug 12, 2004 8:36 pm Post subject:

Hmm, not sure anymore. I get CONFIG_X86_IO_APIC from one machine, and CONFIG_X86_UP_IOAPIC from another. I am assuming that you have it unset (in which case it means that the configuration utility just renames that option when its set/unset.

CompNerd

snuffer · Apprentice Joined: 12 Jun 2004 Posts: 207 Location: Israel

You are right it seems that it does rename it.
_________________
Registered Linux User #360735.

dwblas · Posted: Sat Aug 14, 2004 12:48 am Post subject:

I have a CONFIG_X86_UP_APIC in my . config. It is not set and I have had 2 freezes in the last 3 weeks. I'm going with Mozilla/Firefox being the possible cause and closing it whenever possible. The hang seems to happen after the computer/Firefox has been running for awhile which could mean some sort of memory corruption.

CompNerd · Posted: Sat Aug 14, 2004 2:57 am Post subject:

Hmm, I just suffered another crash :-(

. However, this time I noticed something. I swtiched to TTY12 (ctrl-alt-f12) and had a glance at the kernel messages. The oddest thing was listed:

Out of memory. Killing {application}

I am not sure what exactly is the cause of this error (and please no YOU ARE OUT OF MEMORY!! flames :twisted:

). It takes a while for the system to respond to the switch (although, it appears in /var/log/messages) Can someone else please check this?

tail -n 1000 /var/log/messages | grep Out

CompNerd

CompNerd · Posted: Sat Aug 14, 2004 3:41 am Post subject:

Well, in response to the out of memory errors, I have decided to re-compile glibc with NPTL. Native Posix Thread Libraries should help reduce the load on the system, and hopefully preventing the out of memory errors. This is getting annoying. I wish I could find some cause for the error.

CompNerd

dwblas · Posted: Sat Aug 14, 2004 6:26 pm Post subject:

It's pretty unlikely that all of us ran out of memory at the same time. It has to be some app that isn't behaving properly. I'm shutting down Firefox for most of the day. If someone else wants to shut down gkrellm, and we have someone already who has upgraded gcc, we can see who crashes and who doesn't and at least isolate the problem and file a bug report.

CompNerd · Posted: Sun Aug 15, 2004 12:17 am Post subject:

Well, I finally figured out the reason for the out of memory error -- I was out of memory :-p I did not have any swap setup (dont know how though). I did suffer a crash (yes, again :-(

). This was after prelinking and switching to NPTL. The good thing though was that this crash did not crash the system, just locked up the keyboard nicely. The mouse worked, and new apps could be launched even

The one thing that didnt respond though was gkrellm. Stupid thing froze up. I am willing to keep it off for a few days, however, I do believe that it is grellm that is the culprit.

Maybe someone else would try a different WM or DE.

CompNerd

snuffer · Apprentice Joined: 12 Jun 2004 Posts: 207 Location: Israel

I dont think that it is gkrellm, because i have crashed without using gkrellm. When using Opera to download multiple files this seems to happen. I have swap setup though. Can people please post there ethernet cards?
I have an RT8139.
_________________
Registered Linux User #360735.

val · Apprentice Joined: 30 Jun 2004 Posts: 234

I have this problem on both of my machines which have very different hardware. My older machine is much more vulnerable. I've even got it to the point where it's basically reproducible: I open many many tabs in mozilla and it's bound to happen.

My laptop is a dell with 1.5GHz PM with 1 Gig ram. I have an IPW2200 wireless. I never use ethernet and forget what I have. I've stopped having the problem on this machine since I stopped using gdesklets. This machine is running gentoo kernel 2.5.7-r8 and whatever the last xfree was. I'm now running gkrellm2 and while that periodically segfaults, the system hasn't froze for weeks. (It's not on all the time though)

My older system is a desktop with a PIII 550MHz, 384 ram. I have a sis900 ethernet card. I'm running xorg 6.7.0 and I've tried various kernels ranging from the latest gentoo-dev kernel to a vanilla-dev kernel back to a 2.4 kernel and I always have the problem.

Right now, I'm hoping that once the new x.org comes out, that will fix it. It is supposed to be much faster and hopefully also more efficient: and won't cause 100% cpu usage.

Let me know if I can provide any more information to help narrow down this problem.

snuffer · Apprentice Joined: 12 Jun 2004 Posts: 207 Location: Israel

Has happened to me again, 100% sure no gkrellm running. I might just switch back to xfree cuz thsi is really annoying.

I have a GA-SQ800 whihc has a sis655 if that helps.
_________________
Registered Linux User #360735.

cprior · n00b Joined: 03 Jun 2004 Posts: 57 Location: #gentoo ;)

I am running 2.6.7 and 2.6.8-rc8 with fluxbox most of the time. I too get those lockups. I am on an old PII laptop and I very well understand the mentioned "fear to scroll".

I tried to remedy the situation by removing all fb stuff, both apps like qingy, splash-utils, etc as well as all fb-related things in the kernel.
I still got lockups.

Then I installed the new xorg and my fluxbox wouldn't start. So I did the long planned, unwillingly done kde emerge (mostly as a showcase)

Now in kde I do _not_ get any lockups, I even have the taskbar sliding in and out!

Before removing the fb-stuff I got "overlapping write-combining ranges" when start(ing)x. I have yet to confirm if the new xorg still doesn't detect this reserved memory for graphic display (what the error is about, as per Google)

I don't use either firefox nor gkrell*-stuff.

So I don't have a solution but at least some remarks to throw in.
Some of you might test another wm or upgrade xorg to see if that helps.

I am just too glad that it does not seem to be ahardware prob, the poor student that I am! ;-)

snuffer · Apprentice Joined: 12 Jun 2004 Posts: 207 Location: Israel

Thats odd, i am using xorg-6.8 but i get more crashes in KDE then in fluxbox. Odd
_________________
Registered Linux User #360735.