Let me go straight to th point:

I woke up today and switched on my monitor (I leave my computer on 24hs/day) and had nothing but a tty-like screen with something like this (the transcription my not be accurate......the computer was completely locked and there was nothing i could do besides pushing the reset button, so I copied most of it to a piece of paper and am copying it again in here):

I have recently compiled kernel 2.6.9 and am using ReiserFS so I think the all those things about "pages" are related to the Reiser paging system.

My CPU (amd athlon 3000) is really hard to keep cool; I have already three cooler installed in my box and have removed the side of the box, but still, when I rebooted it after the lockdown the CPU temperate was 68.5C / 155F.

I really think this is the reason of the kernel exception, but thought i should post just in case it is not.

any thoughts??

ricardo birmann

The paging refers to the virtual memory system. Basically, a kernel module has requested an address from a virtual page in user-space that isn't mapped to a physical memory page or a page of your swap file.

(We know it's a kernel module because if it was a userspace program, it would SIGSEGV and that would be that; sadly, the kernel tends to die when problems happen).

The most common cause of errors like this (aside from programmer error, i.e. bugs) is when a memory location in RAM or on the CPU decides to flip a bit of a memory pointer, making it point to an invalid address.

This can happen because of a number of reasons, including overheating a CPU register, so your hypothesis is probably correct.

yes, I think it is....

I was recompiling my kernel and, for each time i tried to run make I got an error on a different function... (during compilation cpu time goes as high as 100%)

I think the REAL problem is the brazilian summer (I live in Sao Paulo) that is just started; the air temperature is around 32C during the day.... I will just have to buy a better cooler,

thanks for your repĺy,

ricardo

Random errors during compilation that are not repeatable are often caused by this same fault, because gcc uses lots of pointers internally. (One problem is the amount of cheap audio-grade memory floating around inside computers, which was never tested to hold exact numbers).

An interesting RAM test is to take a complex project (like GCC) and try and compile it; if it gets all the way through in one go then your RAM is probably OK. Overheating (or worse, overclocking) still tends to make problems appear, though.