Linux freezes sometimes after cpu upgrade[solved]

November Rain · Last edited by November Rain on Sun Dec 17, 2006 4:40 pm; edited 1 time in total

Hi,

I upgraded my Athlon 64 3000+ Single Core to a Athlon 64 4200+ Dual Core. Since this time my system freezes sometimes. At first i thought it could be the ram, so i tested it with memtest. After 8 hours there where no errors. I tried to install another linux(suse 10.1) to recompile my gentoo system but the suse linux freezes too.

I only changed the cpu, its the same board, ram and graphiccard as before. The first thing i did after installing the cpu was to recompile the kernel(2.6.18-r2 gentoo sources) but before it finished the pc freezed for the first time. It freezes mostly when its emerge something.

Sorry for my bad english. Whats wrong with my system? i hope you can help me.

Athlon 64 4200+ x2
Asus a8v deluxe
5* 512 mb GeiL ultra x RAM(ddr 400 dual channel)
nvidia gf 6800 gs 512 mb

bLUEbYTE84 · Posted: Mon Nov 20, 2006 8:31 pm Post subject:

freeze -> tough problem

Let's start by checking the cpu temperature and vcore voltages from the BIOS.

erik258 · Posted: Mon Nov 20, 2006 8:32 pm Post subject:

your system is overheating under heavy load, im almost sure of it.

[edit] do check that bios stuff, but remember that within the bios settings utility, your cpu certainly won't be anywhere near high load. so the temperature will probably go down to quite a lot lower than it would be if you were acutally running under load. if it looks high, here's some solutions for you [/edit]

for starters, try reapplying thermal paste:
Take your heatsink /cpu off and clean top of cpu and bottom of heatsink with isopropyl alcohol, getting all extra thermal paste off. the alcohol takes it off nicely. then, reapply a generous portion of cpu paste and put the heatsink back on. i assume you can find info about doing this online. just make sure to use enough and have it distributed over the entire surface of the cpu.

if the problem persists, try running with the side of the case off and seeing if it's your case temperature that's too high, causing lockups in chipset or elsewhere inside. that doesn't seem too likely, since it worked before, but remember your new cpu does have 2 cores.

if the problem isn't the case temperature, i think you may need a better heatsink. yours may not work well enough for 2 cores; you may need to upgrade your heatsink.
_________________
Configuring a Firewall? Try my iptables configuration
LinuxCommando.com is my blog for linux-related scraps and tidbits. Stop by for a visit!

November Rain · Posted: Mon Nov 20, 2006 8:42 pm Post subject:

The temperature after a freeze is about 30 - 40 °C. I can play hours HalfLife2 in Windows without any freezes. This happens only when i run linux. It just freezes again and i was only surfing the internet.

bLUEbYTE84 · Posted: Mon Nov 20, 2006 8:49 pm Post subject:

As it is %95 a kernel configuration problem, I suggest you to start with a rather default kernel configuration, possibly with a different kernel source (from kernel.org), and configuring step by step. That is, add some stuff & reboot & test in a cycle. But I know this is a tiresome process.

The most important place to mess with in kernel config is the Processor Type & ... menu.

November Rain · Posted: Mon Nov 20, 2006 8:52 pm Post subject:

yes, im using the gentoo-sources with genkernel (genkernel all). is there a chance to find the wrong parameter without checking the whole kernel? i only changed from single core athlon 64 to dual core athlon 64

Edit: i changed the cpu type from pentium pro to athlon64/opteron and the max of supported cpu's to 2(was

... is there a guide which tells me what options my athlon 64 x2 needs?

bLUEbYTE84 · Posted: Mon Nov 20, 2006 9:18 pm Post subject:

bLUEbYTE84 · Posted: Mon Nov 20, 2006 9:36 pm Post subject:

Look what I found for you

https://forums.gentoo.org/viewtopic-t-390499.html

hope you can solve it

November Rain · Posted: Tue Nov 21, 2006 6:48 pm Post subject:

Ok i started my machine with the new kernel and it freezed 3 times while booting. But it freezes not at the same state, 2 times freezed while loading udev and the third time it freezed after loading udev.

November Rain · Posted: Wed Nov 22, 2006 2:55 pm Post subject:

Ok, this is realy confusing. After i failed in finding a solution for the freezes while booting i compiled the new gentoo-sources and changed nothing in the configuration(only usb printersupport) and it worked for 2 hours and it completed a whole 'emerge world' before it freezes again. i tried zu restart my pc and after loading kde the system turned off, like i click on the power switch.

whats happening to my system?

erik258 · Posted: Wed Nov 22, 2006 5:25 pm Post subject:

you're not running development packages are you? ~amd64 is an option that's best to enable on a per-package basis, as necessary.
_________________
Configuring a Firewall? Try my iptables configuration
LinuxCommando.com is my blog for linux-related scraps and tidbits. Stop by for a visit!

November Rain · Posted: Wed Nov 22, 2006 6:32 pm Post subject:

I'm using x86 and not amd64. Ok i read something about problems with the emu10k1 and the amd64 processors, after i put the card out of my pc the system boot up again and i can use it for a while until it freezes again. I changed the bios setting "pnp os" so that linux can set the irq's for non boot devices. i hope this can help me.

Edit: No it freezes again, it often freeze when im compiling the kernel.

erik258 · Posted: Wed Nov 22, 2006 6:43 pm Post subject:

i am running x86 on an amd64 box too.

what setting do you have these?

November Rain · Posted: Wed Nov 22, 2006 6:49 pm Post subject:

erik258 · Posted: Thu Nov 23, 2006 7:30 pm Post subject: ok, i'm surprised by this...

November Rain,
you say

November Rain · Posted: Thu Nov 23, 2006 8:28 pm Post subject:

im using march=athlon64 in my cflag.

November Rain · Posted: Sat Nov 25, 2006 1:15 pm Post subject:

I changed to Vanilla Sources 2.6.17.13 and i got no hardfreezes any more. but i have still freezes in which i can still move the mouse but the keyboard is dead. if i plug a usb device while a freeze then it get logged in /var/log/message. is seems like x is waiting for something. any ideas?
_________________
Gentoo Linux - Intel Core 2 Quad Q6600 - 32 Bit
4 GB G.Skill PC6400 DDR 2
MSI P35 Neo 2 FR
MSI GeForce 8800 GTS 512 MB G92 OC & Club3D 8400 GS
Samsung 245B plus & HP w2408 Vivid Color

erik258 · Posted: Sat Nov 25, 2006 4:25 pm Post subject:

it does sound like X is your problem.

are you running an unstable development X?
_________________
Configuring a Firewall? Try my iptables configuration
LinuxCommando.com is my blog for linux-related scraps and tidbits. Stop by for a visit!

November Rain · Posted: Sat Nov 25, 2006 5:34 pm Post subject:

* x11-base/xorg-x11
Available versions: [M]6.8.2-r8 [M]6.9.0-r3 7.0-r1 7.1
Installed: 7.1
Homepage: http://xorg.freedesktop.org
Description: An X11 implementation maintained by the X.Org Foundation (meta package)

Ok, now when it doesn't freeze anymore, i just cant use the keyboard and mouse. the system is still running, playing mp3's, showing instant messages and so on. Im using the new Beta Driver from Nvidia:

Horst michel # eix nvidia-drivers
* x11-drivers/nvidia-drivers
Available versions: 1.0.8774 1.0.8776 1.0.9629 1.0.9742
Installed: 1.0.9742
Homepage: http://www.nvidia.com/
Description: NVIDIA X11 driver and GLX libraries

If i try ALT-PRINT-K then i can restart the X server and can log in again.
_________________
Gentoo Linux - Intel Core 2 Quad Q6600 - 32 Bit
4 GB G.Skill PC6400 DDR 2
MSI P35 Neo 2 FR
MSI GeForce 8800 GTS 512 MB G92 OC & Club3D 8400 GS
Samsung 245B plus & HP w2408 Vivid Color

erik258 · Posted: Sat Nov 25, 2006 6:14 pm Post subject:

ok, i looked over this forum and the one listed before.

i think your problem is probably the combination of SMP (a complicated beast) and other kernel options.

therefore i still think you should take out everything in the kernel you don't absolutely need. you can always add more later.
some things that come to mind as stuff you could disable

CONFIG_X86_32=y
what is this and where is it set? you are running 64-bit and i couldn't find this option anywhere.

#
# CPU Frequency scaling
#
CONFIG_CPU_FREQ=y

CONFIG_PCCARD=m
isn't this a desktop? probably won't need pcmcia unless you have such an adaptor
CONFIG_X86_POWERNOW_K7=m
CONFIG_X86_POWERNOW_K7_ACPI=y
CONFIG_X86_POWERNOW_K8=m
CONFIG_X86_POWERNOW_K8_ACPI=y

you have both K7 ACPI and K8 ACPI , is that what you want? if you don't have specific reasons for doing so, you may want to remove the K7 option.

# CONFIG_HPET_TIMER is not set
i usually enable this, but it shouldn't cause problems like this

CONFIG_NR_CPUS=8
i doubt you'll ever have more than 2 ; )
# CONFIG_X86_MCE is not set
turn Machine Check Exceptions on, i suggest
CONFIG_TOSHIBA=m
CONFIG_I8K=m
CONFIG_X86_REBOOTFIXUPS=y
you don't need it or want it

CONFIG_HIGHMEM4G=y
turn off this support unless you expect to have
CONFIG_HZ=100
I reccommend 1000 for such a fast cpu
CONFIG_APM=m
apm isn't compatible with smp.

CONFIG_CPU_FREQ=y
disable this for now, unless you think your cpu will overheat. i don't use it at all.

CONFIG_X86_SPEEDSTEP_CENTRINO=m
you don't have a centrino... that's made by intel

====================
in short, you apparently turned on everything you thought you might ever use or didn't recognize. this is a good way to have conflicts; go through and turn all that stuff back off and try looking online for a kernel compilation howto to get this working.

for example -- token ring ?? you have this enabled. you probably haven't ever, ever owned a computer with token ring. your amd64 box will never be on a token ring network. you did not need this support. the same is true of so much of your kernel config...

another example: WAN. you aren't on a a fiber line are you? you don't need wan support.
=============================

CONFIG_FB=y
better disable your framebuffer for diagnostic purposes.

============================
go through and research everything, at least with the 'help' option, and last resort go with what the 'help' opption suggests if you aren't sure.
_________________
Configuring a Firewall? Try my iptables configuration
LinuxCommando.com is my blog for linux-related scraps and tidbits. Stop by for a visit!

November Rain · Posted: Sat Nov 25, 2006 7:34 pm Post subject:

I changed many Options in my Kernel(vanilla sources 2.6.17.13) and i think its now stable. the problem now is that beryl seems to deactivate my keyboard and mouse after a while. the system is still running and everything seems to be fine, but i cant do anything. i can change the position of the mouspointer but i cant do anything with it, not clicking or moving windows around - nothing. when i press alt - print - k then the x-session restarts and i can log in again. i looked in /var/log/messages but there is nothing about this.
_________________
Gentoo Linux - Intel Core 2 Quad Q6600 - 32 Bit
4 GB G.Skill PC6400 DDR 2
MSI P35 Neo 2 FR
MSI GeForce 8800 GTS 512 MB G92 OC & Club3D 8400 GS
Samsung 245B plus & HP w2408 Vivid Color

erik258 · Posted: Sat Nov 25, 2006 7:44 pm Post subject:

i bet it's beryl's fault then; i bet beryl itself is locking up even, maybe.

i guess you try running a different window manager, just to diagnose
_________________
Configuring a Firewall? Try my iptables configuration
LinuxCommando.com is my blog for linux-related scraps and tidbits. Stop by for a visit!

November Rain · Posted: Sat Nov 25, 2006 10:59 pm Post subject:

so, i tried metacity and the same problem appears. but metacity runs much longer than beryl. i read something about the nvidia drivers and xorg could making trouble. i also tried to start kwin for testing but nothin happens, the windows decorations all disappears, i have to enable beryl or metacity.
_________________
Gentoo Linux - Intel Core 2 Quad Q6600 - 32 Bit
4 GB G.Skill PC6400 DDR 2
MSI P35 Neo 2 FR
MSI GeForce 8800 GTS 512 MB G92 OC & Club3D 8400 GS
Samsung 245B plus & HP w2408 Vivid Color

erik258 · Posted: Sun Nov 26, 2006 5:43 am Post subject:

i'm sorry you haven't yet resolved your problem fully. Beryl is a pretty new project; I think gentoo support can be slow too sometimes on the complicated projects because, concerning configurations, 'anything goes' in gentoo and that makes the job difficult for developers.

maybe when i get done installing my new amd64 system on SATA i can put beryl on it and see the prettiness and bugginess for myself.

until then, i fear it will be hard for me to assist you, since i have neither a dualcore (you lucky s.o.b ! ) nor beryl, but i wish you the best of luck and will be watching.
_________________
Configuring a Firewall? Try my iptables configuration
LinuxCommando.com is my blog for linux-related scraps and tidbits. Stop by for a visit!

November Rain · Posted: Sun Nov 26, 2006 11:56 am Post subject:

thanks for your help, many people seems to have problem with dual core, nvidia and xorg. i used the same hardware as before and it become unstable. i think smp makes it much more compilicated.
_________________
Gentoo Linux - Intel Core 2 Quad Q6600 - 32 Bit
4 GB G.Skill PC6400 DDR 2
MSI P35 Neo 2 FR
MSI GeForce 8800 GTS 512 MB G92 OC & Club3D 8400 GS
Samsung 245B plus & HP w2408 Vivid Color