X.org Lockups (part 2)

Meaulnes · n00b Joined: 15 Mar 2004 Posts: 49

Thought that I would add my $.02.

I started compiling gnome before I left work last night and when I cam in, something (be it gnome, xorg, kernel) would lock up about the time I tried to do anything in Gnome. It was almost immediate.

In the past I have always config'd X manually, either via xf86config or xorgconfig, but this time I was feeling particularly lazy, so I ran X -configure to config X. Since TWM (or whatever it is) started up, I figured I was good to go. So I was surprised when I had 3 hard reboots this morning.

Enough having been enough, I decided to try manually configuring X (using xorgconfig) and I am now happily computing in Gnome. Granted I have only been in Gnome for about an hour, but my lock ups were coming within about 30 seconds.

I will grant you that this is probably unrelated to the issues that most of you are having, but if it helps some poor soul who was in the same situation as I was ....

Nerdanel · Posted: Thu Aug 11, 2005 4:21 pm Post subject:

That looks more like our problem than a normal X configuration error. I had a lot of the latter kind back when I had to configure X for the first time.

In any case, I'd suggest you backup your current config and run the autoconfiguring again. The difference between the configs might be informative and worth posting here, even if it's nothing more than GLX turned on/off.

asg719 · Posted: Fri Aug 12, 2005 3:00 pm Post subject:

I just want to add my 2 cents to this discussion since I've been lurking in this thread hoping it may find an answer to my issue. I've been having Xorg issues but they aren't hard locks like described in this thread. For me things will be going great but if I goto bed and wake up X will be consuming 100% of the resources, I can usually just restart X and things will be fine but it still is a pain. There is a bug report for this issue here https://bugs.freedesktop.org/show_bug.cgi?id=2769 Other than that though my system is fairly stable. Here is my config for the record

Kernel 2.6.12-ck5
xorg 6.8.2-r2
nvidia-kernel-6629-r6

xorg.conf below

Lasker · Guru Joined: 17 Jul 2002 Posts: 445

So, my system is running absolutely stable now! :-)

My last lock up was on Jul 31 and even under heavy load all the issues are gone (even the lose connection to keyboard/mouse).

Some days ago I had a very strange accident with my box (nothing related to our problem) after which I had to reset and reconfigure my BIOS. During this process I had several issues, from totally being unable of starting X to symptoms looking quite similar to the ones described here (and I had before). Sorry for being so unspecific, but I had to fiddle around with some IRQ-watching settings (in the APM section) which are not very well documented in my BIOS manual. And since my BIOS is surely different from most of yours I don't see much sense in posting my new settings here. Nevertheless, it might give some vague hints about where to search, particularly due to the fact that all those issues didn't occur under XP (which is running on the same box). Meanwhile it looks to me as if Linux (or xorg-x11) would be more picky about wrong BIOS settings...

localghost · Posted: Sun Aug 14, 2005 2:51 pm Post subject:

I would like to add my experience. I had an "accident" with emerge depclean without --pretend. Oops!
Result: when I start X with gdm or kdm, the screen goes black and the keyboard freezes. I can't get back to the console. However I can webmin into the machine, and probably also ssh (didn't try that, but if webmin works I see no reason why ssh wouldn't). I noticed that X uses 99% cpu.

revdep-rebuild didn't help, so now I'm doing an emerge -eN world

For the reference:

/etc/X11/xorg.conf

HermanR · Posted: Sun Aug 14, 2005 3:01 pm Post subject:

Just to get things clear for myself (and hopefully some others):

This thread discusses (at least) two different kind of lockups:

(1) X crashes under some heavy duty (often including a web browser such as Firefox), leaving the kernel running. You will still be able to move the mouse; sometimes there will be an error message posted to the logs. Magic-SysRq or getting in through ssh will enable you to restart X. This is related to the inner workings of X and is not hardware (motherboard/graphics card/chipset) specific. The frequency with which this crash occurs can sometimes be diminished by changing AGP-settings and the like.

(2) The computer suddenly stops responding, without any warning or error message in the logs. This occurs even without any system stress. (In fact, it looks like it occurs less frequently when the system is busy, but that may be just my impression.) There is no way to revive the system, not even with ssh. The power button is the only way out...

My system has the type (2) system lockups. There is a slight change that there is something wrong with my (BIOS?) setup, since I have only Linux installed and it is a only a few months old. My memory has been thoroughly tested, however. Next week I will install XP, and see if it eXPeriences the same occasional crashes too or not. My guess is that it will work fine (as far as any Windows OS can ever work fine, but you get my meaning) and that my lockups are the same as those of many others.

Although I have had lockups when my system was idle (and no screensaver running), it has never crashed when it had no user process running, like in kdm waiting for me to login. I do not know if that is relevant or not. The solution then would be simple: never login. But how do I get my work done?

As far as I can tell, the type (1) crashes are not software or hardware specific to any meaningful degree. Does that also hold true for the type (2) crashes? My system: VIA chipset (K8T800Pro), nVidia FX5200 graphics card, kernel 2.6.12, Kde 3.4.1, etc.
_________________
Gentoo (64 bits) on AMD64 X2 4400+, MSI K8T Neo2-FIR, 2 Gb RAM, 320 + 160 Gb SATA, nVidia Geforce 6600

spitefulcrow · n00b Joined: 18 Jul 2004 Posts: 44

Bobbie · Tux's lil' helper Joined: 21 Jul 2002 Posts: 77

localghost · Posted: Tue Aug 16, 2005 12:49 pm Post subject:

nokernel · n00b Joined: 19 Jul 2004 Posts: 18 Location: Québec

So I have the lockup number 2. A hard system lock with no responces and logs nor ssh.

But I have this system locks only when I use kde not when I simply start xterm. And as of now I dont have those lockups when I am connecting to my machine with nomachine software ( NX ) with KDE. So that's something that kde asks to the display driver that makes this hang.

For your informatio nI have an asus A8V-DELUXE mb with an ati 9800 pro agp, 1 gig of ram (tested with memtest86), cpu AMD ATHLON 64 3000+

spitefulcrow · n00b Joined: 18 Jul 2004 Posts: 44

Gergan Penkov · Posted: Tue Aug 16, 2005 7:00 pm Post subject:

If you have mostly problems with mozilla, mozilla-firefox or application, which somehow link against them, as gaim-encryption, evolution, evolution-data-server, you could try to build the things as described in this thread https://forums.gentoo.org/viewtopic-t-370844.html and see if this help.
I am not sure that there is some connection to the lockups, but incorrect runt-time library resolutions to the mozilla-libraries could be the possible culprit to some of the problems, but not to all of them (pure kde-lockups and in game-lockups could not be solved in this way :

).
I am not sure that this will solve the firefox-lockups, but one could try:)
_________________
"I knew when an angel whispered into my ear,
You gotta get him away, yeah
Hey little bitch!
Be glad you finally walked away or you may have not lived another day."
Godsmack

HermanR · Posted: Wed Aug 17, 2005 8:45 pm Post subject:

An update to my earlier post: When my system freezes, I can use the 'Magic SysRq' to reboot the computer (Alt-SysRq-b), as I found out after compiling support for this in my kernel. The other 'Magic SysRq'-key combinations do not seem to work. A reboot is somewhat more 'gentle' than pushing the power button, but I'm not happy yet.
_________________
Gentoo (64 bits) on AMD64 X2 4400+, MSI K8T Neo2-FIR, 2 Gb RAM, 320 + 160 Gb SATA, nVidia Geforce 6600

converter · Apprentice Joined: 24 Dec 2002 Posts: 163

feliperal · Apprentice Joined: 09 Mar 2003 Posts: 168

I was having problems getting the darn Nvidia 7676 drivers working. Everytime I start X, I get the "Black Screen of Death." . While reading the Nvidia docs, I came upon this revelation:

localghost · Posted: Thu Aug 18, 2005 4:16 pm Post subject:

I solved my problem with replacing bad ram AND emerge -eN world
_________________
411 /0µr 84$3 4r3 8310n9 70 µ$.

RaZoR1394 · Guru Joined: 09 Jan 2005 Posts: 356

I've now tried vanilla-2.6.13-rc6 and gentoo-2.6.12-r9. They still lockup but significantly less.
_________________
http://en.wikipedia.org/wiki/Cell_(microprocessor)

danielsbrewer · Posted: Sat Aug 20, 2005 9:35 am Post subject:

There seems to be a new set of NVidia drivers in portage. Does anyone know if these help?

I have been watching this thread for some time hoping for a fix. I think this is a serious problem. I have the problem where it freezes but you can still move the mouse, ssh in and stuff.

Is anything being done about this?

kmare · Posted: Sat Aug 20, 2005 11:53 am Post subject:

it still lockups for me on xorg 6.8.2 and kernel 2.6.12.x.. hope to see 2.6.13 released soon to see if it fixes anything...

andrewwalker27 · l33t Joined: 27 Jun 2005 Posts: 660

The newer nvidia drivers are no different, it still locks up the display.

converter · Apprentice Joined: 24 Dec 2002 Posts: 163

About a week ago I started experiencing these "lockups" (NVRM errors) more frequently, and for the first time while running Quake III Arena.

A couple days ago my video card (Gainward G4TI4800-SE) died. When I pulled the card, the fan was very difficult to spin, so I'd imagine it was overheating more as the fan became less and less efficient and finally melted down when the fan died. By the way, since when did motherboards stop giving the four tones for a dead video card? The lack of feedback from the system made it difficult to find the problem.

I replaced the card temporarily with a new MSI FX5200, and have not been able to reproduce the NVRM errors since.
_________________
converter

rvalles · Tux's lil' helper Joined: 19 Feb 2003 Posts: 121

Here, I'm having the mouse-moves-everything-else-is-frozen kind of hang. KDE starts fine, and then looks up sometimes while changing firefox tabs. xfce4 causes the hang on start (not xfce4 composite support related, btw; I have that well disabled.).
Nvidia 6xxx work fine, and _all_ 7xxx I've tried (I try every new version) does have the same problem. Sickens me it's been half a year or so, and they still haven't fixed it, so I can't use the 7xxx ones

Damn nvidia and their closed drivers! Lucky ATI r300 owners have the r300.sf.net thing. If only if somebody reverse-engineered the nvidia drivers too...

Miker · n00b Joined: 20 Jun 2002 Posts: 65 Location: Belgium

Here's my report of X trouble:

I just installed my new laptop, and X lockups are frequent. They're the 'no mouse, no keyboard' variant, so a total reboot is needed.

I'm using the unstable ati-drivers (the stable ones don't compile with the latest stable gentoo-sources kernel).

Has anyone tried switching kernels yet to solve this?

I tried disabling agpgart in the kernel but this did not help. Any pointers appreciated!

the_eye · n00b Joined: 25 Sep 2003 Posts: 6

I'd like to add my experience to the pile ...

I am running with an old-ish nvidia card (GeForce 2MX-400).

Used to have no lockups at all.

Did some updates in late July and promptly started seeing type A lockups (caused by firefox and scrolling with wheelmouse) after a few seconds of usage.

Tried some solutions, in the end downgraded nvidia-drivers back to 1.0.6629-rsomething and that was the end of type A. But this only caused the surfacing of type B lockups after about 3-4 days. I mostly noticed these when I was away someplace else and tried to log in via ssh which would stop working at some stage.

In these cases, sysrq only helps halfway, because the only key-combo that works is alt-sysrq-b, alt-sysrq-s and alt-sysrq-u don't do anything.

These lockups happened without any message in any log.
Tiniest hint: I sometimes do unattended video and audio-recordings (started with atd) and on thursday night, when one such recording was running (using the command-line: ecasound -b:2048 -z:intbuf -z:db,200000 -r -t:4200 -f:s16_le,2,44100 -i:alsa,default -o:somefile.raw) a lockup occured about halfway through, but I still received an email from atd someplace in between the recording stopping and the computer locking up alltogether. At least that's the way it seems. At about the same time, regular log-entries (like some cron stuff) stop occuring. the email I got from atd read:
<some normal output from ecasound>
sh: line 70: 29774 Killed ecasound -b:2048 -z:intbuf -z:db,200000 -r -t:4200 -f:s16_le,2,44100 -i:alsa,default -o somefile.raw

dunno if that helps in any way.

so I thought that since the only higher function I am using on this nvidia-card is xv (i.e. mplayer and tvtime), I might as well try the open-source drivers. So I did that and now, haha, X doesn't lock up completely but apparently sometimes receives as Signal 11 (no further info in the logs) and crashes. But at least the computer stays usable. OTOH I only have been using the nv drivers for 3 days now, so I might be seeing a lockup later.

the_eye · n00b Joined: 25 Sep 2003 Posts: 6

as a followup to my own post from a few hours ago:

i had been happy to early. Have been sshed into my home-box up untill about 30 minutes ago when suddenly it stopped responding. ping, traceroute and tcptraceroute suggest that it _is_ actually my home-box that is down and not some network-problem.

So it seems using the nv drivers doesn't help matters with type B lockups, at least for me. Next step would be to not run X for a while and see what that does.

I'm kinda suspecting hardware-trouble on my end, after all, this box isn't the newest any more, but OTOH memtest86 completed more than a whole pass yesterday without finding any errors in RAM. d'oh!