X2 Hardfreezes x86_64

[dom_cyrus]

Hi@all

First I want to say, that I know, that there are lots of threads in this forum about this. But I don't see any solution for my problem. I get randomly some Hardfreezes. I have to say, that it never crash until I installed gnome. Before I had no X11-Server and it didn't crash for many hours. I have a logitech g5 and I don't know if this makes any problems.

My question are:
How can I find out where the problem is? (When I look at /var/log I can't find some usefull logs. I have installed syslog-ng.)
Which kernel configuration should be used? (It seems that ondemand governor makes some problems / ACPI)
Is there somewhere a good "How-To" for X2 users?

System:
2.6.12-gentoo-r10
NVIDIA 6800
MSI Neo k8
Ram: 3gb
proc: AMD64 X2 4400

ps: I had installed an amd64 System before I changed the CPU. So all the components are the same except for mouse and proc...
THX

[fangorn]

As X2 user you should definately disable "Preempive kernel" and Preempt the big kernel lock" in Processor features. Also the IOMMU switch is needed.

If you use make xconfig you can easily see the help texts for the options

[dom_cyrus]

I have disabled Preempt the big kernel lock and I have enabled the IOMMU, but the problem is still there. It seems it always freezes under heavy load... The problem is, that I don't get any kernel panic or other things. The system just freezes... I read somewhere that the nptl and nptlonly useflag can make some problems. I have both inabled, but I don't think that this can provide freezes like I have.

[fangorn]

How much memory do you own?

I had some problems getting a stable network with 4 GB RAM at work. If you have more than 3 GB of RAM you should try experimenting with the RAM mapping switches in the BIOS.

Look if you have the latest BIOS version. Test the failsafe settings of the bios. My BIOS has optimized standard settings. If yours has this too, test it, if you accidentaly set something strange. Try without networking. ...

I do not use nptlonly, but nptl never made me headaches

[robnotts]

What is wrong with PREEMPT? I have been running with PREEMPT for ages on an X2... so unless I am lucky, PREEMPT might not be the problem, although I can understand the reasons for turning it off.... Running with 2GB of memory.
Rob.
_________________
---

Gentoo Phenom][ X4 955 on AMD790 + Geforce 220GT 8GB/1.75TB (Desktop)
+ MythTV (3xFreeview,1xFreesat HD) on 1080p
Gentoo Turion64 X2 Geforce 6150 2GB/120GB (Laptop)

[fangorn]

Had some problems with nv_sata and two discs that dissapeared after deleting preempt out of my kernel config.

[dom_cyrus]

ARG still have this freezes. The thing is, that all things freeze, but I can still move mouse and sound is also still working. So maybe it isn't a full freeze. I have also added echo 0 > /proc/sys/kernel/randomize_va_space. But with no result and i turned off preempt. I don't know what it is. With Windoze I have no problems...

[robbyjo]

Well, I've got the same freezes like dom_cyrus, of which I posted a thread here. I tend to believe that we've got the same problem. But folks in this forum tend to blame the hardware...

[dom_cyrus]

Do you still have the problems? Ah yeah I made some tests. My problem seems to do nothing with heavy lead, because it can freeze when I just have open firefox. And I run some apps for many hoers (both cpus 100%) and nothing happend, but then after I closed all apps and waited maybe 30 min it freezes again...

When it freezes it locks the keyboard (no numlock switching anymore) but the mouspointer still moves also the sound is still playing and no kernel oop is coming. Can please someone say me how I can log such things?

[energyman76b]

Hi,

have you tried a more recent kernel?
Like 2.6.13?

2.6.12 is pretty old - and who knows all the potential bugs fixed since that?
_________________

[dom_cyrus]

I have tried a lot of different kernels including latest vanilla sources. Right now I have installed kernel-2.6.14-rc4 but still the same problems.

ps: This would still be nice

[dom_cyrus]

I searched a llittle on bentoo bugzilla and I found a lot of nvidia errors there, for example this one http://bugs.gentoo.org/show_bug.cgi?id=86322 The guy tells about excatly the same behavior of my system... Maybe the problem is not the cpu...

[energyman76b]

Hi,

maybe it is your PSU.
the X2 suck down a lot of power - and when X is running, the graphiccard needs a lot too - so this two combined just might overwhelm your PSU.

Weak PSUs are a common source of problems and big Watt numbers don't say anything about their quality or ability to deal with high loads.
_________________

[dom_cyrus]

I don't think it is the PSU, because it runs in windows for hours and I use there c4d rendering stuff and it never happend a freeze or something like this, even when I am rendering stuff more several hours (both CPU 100%). The things I do now is disable nvidia driver (nv) and I also will install another linux distro to see if the problem also appears there.

If someone knows anything else I can do to test it, it would be very nice.

[gurke]

i had the same problem evry 2 or 3 days. i didnt have the problem again, since i upgraded to 2.6.13-gentoo-r3. this could also mean that the problem just didnt come up again. i had it once while updating glibc and once while surfing with opera (opera also randomly just closes itself - no crashreport by kde).

system:
Linux ghost 2.6.13-gentoo-r3 #1 SMP Tue Oct 11 14:59:58 CEST 2005 x86_64 AMD Athlon(tm) 64 X2 Dual Core Processor 4200+ AuthenticAMD GNU/Linux
Gigabyte GA K8N Ultra-9
Nvidia GForce 6600 GT PCIx
_________________
lesen gefährdet die dummheit.

[dom_cyrus]

I also have compiled glibc recently. Do you have set the useflag nptl and nptl only? May this useflags can be responsible for the hardfreezes?

ADD: Another problem is the nvidia-kernel module which has some problems --> http://www.nvnews.net/vbulletin/showthread.php?t=57990 Is this update included in ebuild (emerge nvidia-kernel)?

[formel1hund]

Try this:

remove some of the memory, and run memtest,
disable everything in the bios you don't use, to free up irq's.
if you have a soundblaster (emu10k1) ensure it dosen't share irq with anything.

[PrakashP]

It is Nvidia related and nothing X2 specific. You might want to take a look at nvnews,net in Linux section to readmore about it...

[gargola]

Hi, I am having these problems too. I am not sure how to solve this one, since there are no error entries in any of the logs.

Sympton: Computer freezes every couple of minutes. It seems to be cpu load related, however I am not sure. Usually happens when running KDE, however it is not limited to, since it also happens running fluxbox. (could x11-base/xorg-x11-6.8.2-r6 be an issue?)

I do not think it has anything to do with the power supply, as windows works perfectly. A memtest86 didn't show any memory error at all,
At the beginning I thought it was the nvidia SATA controller, However I exchanged the disks to the SIL controller and the problem was still there.
I also rolled back to previous kernels: 2.6.12 and tryed newer vanillas too: 2.6.14

Any ideas? Any help is welcome!!!!!!!

PD.: Are there any X2 specific HOWTO / Tipps and Tricks?
PD.: I know people have adviced against it: I my kernel is preemtible due to the fact that the ati-drivers need a preemtible kernel in order to work

[Avuton Olrich]

I'm not sure this is the issue, but I have been locking up at least once a day, and I caught the error in console once. It is caused by an MCE:

http://marc.theaimsgroup.com/?l=linux-kernel&m=113243366321261&w=2

[gargola]

I found the cause! --> at least to my problem.

I am using an emu10k1 sound card, which was sharing an irq with one of my sata controllers. The fact is that after taking out the emu10k1 controller, and enabling the ac97 sound device on the mother board, my system stopped hard freezing.
(I think this is a known problem for amd64 )

I still have to try connecting my emu10k1 to other slots, since I do wish to use this device. Any other ideas how to figure thisone out?

Does anyone know what the irqpoll kernel option does?

[gargola]

Yep, that was the problem! I tryed out placing my SATA card and emu10k1 in different slots, util one of the configurations worked. I placed the emu10k1 in the fisrt slot, and the sata in the last one. My interrupts read as follows:

CPU0 CPU1
0: 65489 1384076 IO-APIC-edge timer
1: 419 11896 IO-APIC-edge i8042
8: 0 0 IO-APIC-edge rtc
9: 0 0 IO-APIC-level acpi
12: 2913 65710 IO-APIC-edge i8042
14: 16407 377723 IO-APIC-edge ide0
15: 25 583 IO-APIC-edge ide1
16: 251868 4970498 IO-APIC-level eth0
17: 29939 241163 IO-APIC-level libata
18: 289027 19342232 IO-APIC-level libata
19: 6 469 IO-APIC-level ehci_hcd:usb1
20: 0 106 IO-APIC-level ohci_hcd:usb2
21: 3007 56840 IO-APIC-level EMU10K1
22: 2643 60720 IO-APIC-level SysKonnect SK-98xx
NMI: 2251 3441
LOC: 1449594 1449593
ERR: 0
MIS: 0

[gigli]

I have had the same problem when i have installed different distros (eg. Fedora, Ubuntu, Debian (amd64)), it just hardlocked. In my case it was my graphic card (nvidia 6800gt) which locked the computer. Xorg nv driver didn't work, i used vesa until i could install the nvidia driver and that did it. I never had any issues with the cpu (amd64 x2 4400).

[deepspace9]

What memory do you have?

At first I had freezes too, but after I raised my memory voltage to 2.8V, everything worked just fine!
_________________
Athlon 64 X2 3800+
MSI K8N Neo4 FI
Geil Ultra-X PC3200 400MHz CL2 5-2-2 DDR 2x512 GB
Asus 6600 Silence GFX
Watercooling (I just love the silence )

[fudge]

I've been having similar keyboard and mouse click lockups, but I noticed a pattern. The lockups occurred when using firefox, eclipse and wireshark. They all use GTK+. Yesterday, I added