one cpu core L2 cache is failing

[canadaist]

If I have a root console open, I am starting to get

kernel: [Hardware Error]: CPU:1 (f:4b:2) MC1_STATUS[-|CE|-|-|-]: 0x9000000000000171

And it crashes the computer eventually,
my limited knowledge and searching suggests this is cpu 1, the second cpu, and the use of

Code: Select all

maxcpus=1

in boot will force the second cpu to be idle, not cpu 0,
and will allow me to use the computer until I get a new cpu or box

but I need help as to where and how to edit the boot command

Thanks in advance

This just started, crashing will happen immediately if I use flash, quickly if I use a browser
(which may be flash, but it happened as I first tried to post this, and there is no flash
on this site) but eventually it will crash with little running, overnight for example.

Extra credit: I have a HP Pavilion a1626n and recommendations on cpu upgrade would be welcomed

[canadaist]

I have turned one cpu off by appending maxcpus=1 to grub in mcc.

I dont know if its the right one... will wait to see

on extra credit question,

AMD says:

HP seems to restrict potential upgrades to socket AM2 (darn) running at between 4000 to 5000 (designated, not actual) speeds

which indicates, unless I shop ebay, (grrr) my choices are limited to few, and far between

ALSO:
I dont know how to check for which cpu L2 cache was failing, though the original message said cpu1, which probably means the second one (ie NOT cpu0)
and I am not sure that maxcpus=1 turns off a specific one, or if that specific one turned off is the "next one" ie cpu1

[eor2004]

Hi, from experience cpu1 means the second or #2 cpu, cpu0 is the first or #1 cpu, I said from experience because I have a second pc who had an LGA775 intel celeron dual core 2.80 Ghz with LinuxMint and Windows 7 which I use occasionally and was giving me problems like in linux kernel panics and in windows blue screen of death, crashes and freezes, ect... So one day I was using LM and suddenly a kernel panic, when I started to read the info I saw a line that said something like this: "cpu1 is locked up 'cannot unlock', that's when I realized the cpu1 was damaged, lucky me I had a spare LGA775 intel celeron 1.8 Ghz from an ASRock damaged motherboard, I installed it and the machine worked (and is still working) like a charm, no more BSOD, crashes or kernel panic, so the same problem maybe happening to you, so my advice is to get another cpu.

[ITA84]

Looking around I couldn't find a kernel option to turn off a specific CPU core, but maybe there's automatic fallback when the one in use stops working. It might be possible to disable a specific core after boot, by using the command

Code: Select all

echo 0 > /sys/devices/system/cpuX/online

as root (X being the core/CPU number). You'd have to put this inside a boot-time service and have it run as soon as the /sys filesystem is available.

[zeebra]

I think you need to compile another kernel. This can help you get a variety of debugging and error output from your CPU if you build the correct modules into the kernel. You probably need to build in CPU hotplugging as well.

If you search google for "disable cpu core kernel" you can use Grub to set maxcpus like you described. But that does not specify which one. But cpu1= core2

You can disable CPU cores with userland tools that change the properties of the files in /sys/devices/system/cpu/x. probably something like CPUpower would do. or cpufreqd or other CPU frequency tool. Or you can do it manuall unless your boot overrides it. If it overries it you can change that or write your own script do disable the cpu core after each boot.

https://www.kernel.org/doc/Documentatio ... otplug.txt
http://www.cyberciti.biz/faq/debian-rhe ... tplug-cpu/
http://www.netfibre.com/?p=619