Very weird hardware issue

Ok, here's the plot:
I've got a box with a K6-III/500MHz on a DFI mainboard with VIA VT82C598MVP chipset. Installed debian and everything went well. Then I compiled the kernel which worked well the first time. But I forgot support for some hardware and recompiled it. Bang. Got sig11. Recompiled again and it just stops after a couple of minutes. No error, no reaction on any further input (even sys-request didn't work).

Ok, I think, this has got to be a hardware thing. Probably memory related. So I took a 128MB module from my main box and tried the kernel compile again. Same thing as before: Random errors at random places. And now the spooky thing: When I put the memory back into my main box it did the same. Segfaults all over the place, sig11 during kernel compile.

So apparently that damn computer ate my memory chip. How can that happen? I am absolutely sure that I didn't damage it when I mounted the module as I was really really careful. Now what? Replace the mobo? The power supply? Any pointers on where to start trouble shooting?

Thanks.

[pbharris]

demian my friend (and fellow fvwm user!)
the MB can indeed eat memory, it is wierd that it starts suddenly like this though, i ma guessing it has been working good for well over a year. i would check for something which may cause shorts, but this is unlikey. certainly compileing the kernel would not cause this, but perhaps a BIOS setting be screwed up. I know the sig11 is indicative of a bad memory location, so you may wish to try out a app called tufftest which checks this stuff out, when i had it you get a free copy for personal use.

demian my friend (and fellow fvwm user!)

;D

it is wierd that it starts suddenly like this though, i ma guessing it has been working good for well over a year.

Actually it didn't really start suddenly. I got the box from a friend who doesn't have any use for it. He said he had a lot of trouble with it and got random system crashes and the such. Well he uses windows and so obviously I just thought: Ahh put a proper OS on the drives and this will turn it into a very useful machine. Thanks for the link. I will give it a try.

[pbharris]

ah.... so this may be an old probelm. okee dokee.

check your pm

so you may wish to try out a app called tufftest which checks this stuff out.

Seems the tufftest is not so tough after all. The computer passed all tests without any promplem :-/

[Schotty]

Check to see if it isnt the PSU. That can cause some wierd effects (christ I jsut got done with a PSU issue). Easy way to test it is to swap it out and run the replacement PSU for a day or so doing something that would be intense enough for a power glitch to REALLY matter. Like for me, I did a Q3 test as I was ftping all kinds of stuff.

Remember without electricity the PC is useless :)

[pbharris]

are both of those memory modules still bad?

Phew, I'm starting to get confused about this myself. Good news is that I've got my main box up and running again. I got a new 256MB module (got timing now that prices are almost twice of what they used to be a couple of months ago but, oh well :-/ ) Now I'm running with 384MB meaning that I use one of the 128MB modules that I thought were fubar'd. I teased the system quite a bit (compiling three kernel in a row while running xmms with 3 visual plugins and have xine decode a divx file, surf the web with four mozilla and two galeon windows). Works great as far as I can tell.

Also I swapped the PSU of the 'broken' box (230 W-> 300 W) to no avail. Right now I have that box compile a kernel with an old 64MB module of PC100 RAM. I tweaked the BIOS settings a bit but I wasn't too sure of what I was doing Let's see how that goes.

So I guess the damage is restricted to one of the 128MB modules (which I might sell on EBay ;D ...naa j/k )