nV News Forums


nV News Forums (http://www.nvnews.net/vbulletin/index.php)
-   NVIDIA Linux (http://www.nvnews.net/vbulletin/forumdisplay.php?f=14)
-   -   Hard lockups - AGP not guilty (http://www.nvnews.net/vbulletin/showthread.php?t=1431)

milo 09-02-02 06:49 AM

Hard lockups - AGP not guilty
I've been experiencing the pleasure of hard lockups on a 2.2GHz SMP Xeon system, with a GF4 PCI card. These lockups occur apparently at random, whilst carrying out normal activities (not running OpenGL applications). Obviously, the usual NvAgp=0 is not going to work, as there is no AGP slot on the board.
The kernel is a vanilla 2.4.19, which appears to have AMD's speculative cacheing patch in it (from reading the appropriate setup.c file).
Can anyone suggest the cause and a possible solution?

obiwan 09-05-02 04:57 PM

What kind of activities cause it to hang? Are you in a gui? websurfing?
Playing sound? etc.

also: is your box still accesible by ssh?

milo 09-06-02 03:32 AM


What kind of activities cause it to hang? Are you in a gui? websurfing?
Various, but local file browsing with Nautilus or altering GNOME or KDE settings has caused this effect. It appears to be random.


also: is your box still accesible by ssh?
No - this is what makes me suspect the NVdriver.

dion_starfire 09-08-02 04:42 AM

I have the same problem
I have had the same problem for quite a while now. Despite many many attempts to figure out what is causing the lockups, I still cannot find a pattern. I have seen lockups under KDE 2.x, KDE 3.x, and Enlightenment. At first I suspected AGP bug since they seemed to happen when I was using up a significant amount of my system resources, but now I get lockups even when my computer is idle (go to sleep, wake up, find funny vertical line pattern on screen, sys doesn't even respond to SysRq). I am running the latest nvidia drivers (build 2960), but I have had these lockups for several versions of the driver, and no-one can help...they all blame the AGP bug (I tried several bugfixes just in case...a few actually made the problem worse).

milo 09-08-02 08:00 AM

Do you have ACPI or any other form of power saving enabled in the BIOS? I have heard that this may cause lockups on SMP systems, and have disabled it on mine to see if there is any effect. I've not noticed a lockup yet, but then the machine has not really been used much.

dion_starfire 09-08-02 03:17 PM

I generally keep APM (not ACPI) enabled in the kernel, but since I don't have a SMP system, I doubt disabling it would help much.

FYI, sys specs are 1.0GHz AMD Thunderbird (single processor), 1024MB RAM, GeForce2 GTS, kernel 2.4.19 (but I've seen this bug as far back as 2.2.19, so I'm pretty sure it's not the kernel), XFree86 4.2.0 (also seen bug in 3.6) nvidia driver build 2960 (seen bug for I dunno how many versions)

steve9000 10-06-02 12:36 PM

Hard lockups
Try disabling the render acceleration

Option "NoRenderAccel" "True"


McIntosh 10-07-02 07:45 AM

I'm having a similar problem (see: http://www.nvnews.net/vbulletin/show...5527#post25527)

on a dual PIII with both 2960 & 3123, agp disabled etc.

lockups are erratic and dont seem to be linked to specific application activity.

testing with the NoRenderAccel option now (might I ask what does this option do?)


idangazit 10-08-02 09:31 AM

Welcome to the club. Make noise about this bug so Nvidia allocates developers to stability, not just performance.

I've been experiencing the same bug since, well, pre-2880. APM enabled, but you aren't suspending (and know that suspending is a cause of system death). Random hard, hard freezes. AGP is disabled (or not present, as is the case with one poster). Occurs when doing nothing specific in X, (I only run gnome) -- scrolling, moving the mouse, opening a menu, etc -- nothing specific at all.

Do you get somewhat "repeatable" state crashes? E.G. - Crash when setting a specific preference in Mozilla, then reboot, fsck, etc, load Mozilla, go through the same steps to get to the Mozilla pref, set it again the same way, crash? This isn't a moz-specific problem, happens all the time with random apps. However, change the order of what you do and the crash might dissapear, and then you can't get it to happen again so you can post something to the nvidia techs.

My solution: pass "apm=off" at boot time to the kernel. It isn't pretty, and I use a laptop, so it sucks, but for desktops the only penalty is increased power consumption, as the kernel won't issue HLT instructions to idle processors to slow them down or conserve power.


McIntosh 10-08-02 10:14 AM

I dont have apm support in the kernel and everywhere I could apm/acpi is turned off since it is useless for my application (server).

regarding reproducability, the freezes seem pretty random although the following activities:

- a starting kde (especially with 2950)
- starting a GL app
- running a GL app, xterm & mwm for several days

very often (the last one always) result in a freeze.


Juan Bill 11-08-02 11:14 PM

Mandrake 9.0 lock up
I think this is where i could be heard then.
I have a dell inspirion 4100 with geforce2go 32 ddr ram. and actiontec modem.
I'm quiet a newbie with linux. The current installation linux is Mandrake 9.0.
I have equiped the system with the 3123 nvidia drivers and made the correct changes to the XF86Config-4 file as suggested and download the drivers for my hsflinmodem from conexant.
My pattern of lockup is whenever i'm online and activated an opengl program/game the laptop locks up. I have no idea which fault it is. the modems or the graphics card?
Anyway IF i'm not online and activated the opengl program it does not lock up.

There is one exception is that if i activated an opengl game that does not require the desktop resolution change. And THEN go on the internet. No lockup.

I know some ppl have the same specifications laptop within this forum as me. So please speak up if you experience similarities or HAVE a solution that i eager for .
else... cheerio

Arthur 11-12-02 09:20 AM

Try checking your memory.
Might be a long shot, but try checking your memory. I've been having some really random crashs and instability on my machine lately.

Turned out to be the memory at fault. Go to www.memtest86.com, they have a free tool, that you boot up, that'll test all of your ram.


All times are GMT -5. The time now is 10:37 AM.

Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2015, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.