Re: [sheflug] Oops

[Marc Kelly <marc [at] xxxxxxxxxxxxxxxxxxxxx>]

lesleyb wrote:

> I have been experiencing some CPU overheating problems.  I have an 
> Athlon XP 2K Palomino on an Asus A7V333 motherboard and, ever since I 
> built it I have never been able to run it stably at 1667MHz.  The board 
> is *not* in jumperfree mode and so I can only choose between 1250 MHz 
> and 1667MHz.  

I have a similar story actually. I have an Athlon XP Palomino, which I
have to down clock the FSB from 133 (266DDR) to 100 (200DDR) and run
with a 12.5 multiplier, giving around a 1250. It will throw segfaults
or cause kernel oops when run at higher speeds.

It used to work at full speed, but the motherboard it was on had
"issues" and fried it own northbridge in some way, it killed one of
the IDE channels etc and became very unstable. I have a horrible
feeling the board was overvolting the CPU. Either by design, or
failure and things finally gave up.

On the current board it runs fine at this slower speed, but does the
same things you mention if I crank it up to the design speed. If you
can, I'd check the voltages either in the bios, or if needed with a
multimeter.. and compare to something like
http://www.amdboard.com/amdid.html for your CPU just to be sure the
bios isn't doing something funny with it all. Its a "normal" cpu? No
funny messing with the links on the top to changes its mode/model or
anything?

Another possible is that there is some weird issue with voltages to
the RAM, since these can be controlled on some boards as well the ram
might be getting hot and then throwing spurious corrupt data out, but
passing memory tests is kind of annoying.. makes it hard to decide.

You do get OOPS's from somethings, I have a web cam on a machine the
sometimes causes them for unknown reason, bad drivers. But to be
heating up to shut down levels is kind of scary, and is quite likely a
very good reason for oops's since things go a tad crazy internally
when temps get too hot.

> I suspect today's modprobe oops (http://pastebin.com/565562) occurred 
> when I tried combining 1x1Gb Kingston RAM and 1x256Mb Samsung, both 
> PC2700 DDR 333MHz.

You could try fiddling with the memory timings to slow things down a
little, if your board allows it. Or even reordering the sockets the
memory is in to see if its being silly and taking the settings from
the first stick, which might be too fast for the second stick.

> I did check my syslog. The first oops (http://pastebin.com/565290)
> occurs on 12th Feb and seems to be an alloc problem while in
> screensaver mode because the process is lisa.

if this "lisa" is the same as i think it is, its some networking tool
for doing network-neighborhood stuff. Still it probably isn't related.

> The second oops (http://pastebin.com/565562) seems to have occurred at 
> boot time because it's modprobe that's suffering here.

Your not doing anything weird like trying to force powernow style
frequency changing on a non mobile chip or anything like that? Or a
mobile chip being used in a desktop?


-- 
/\/\arc Kelly
..Just your average physicist trying to get by in a world full of
normal people...

___________________________________________________________________

Sheffield Linux User's Group -
http://www.sheflug.co.uk/mailfaq.html

  GNU the choice of a complete generation.