lkml.org 
[lkml]   [2001]   [Mar]   [19]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [OT] how to catch HW fault
On Sun, Mar 18, 2001 at 09:11:46PM +0100, you [kees] claimed:
> Hi,
>
> I tried memtest86 for 24 hours also and that didn't gave a clue. When bad
> ram was really involved I'd expected to find things like:
> failing fsck's, failing kernel compiles and such. But none of them
> the system runs perfect if it doesn't freeze(lockup).
>
> So yes, only the CPU's and the mobo are at question. What I was looking
> for was a tool like memtest86 but now for motherboards.....

You really cannot say that bad memory is involved ONLY when fsck's fail and
kernel compiled fail. No way.

Think about it: that failing bit can well be in a place that kernel never
touches, and gcc usually does not touch. Moreover the bit flip usually does
not happen every time; you have to stress the memory for hours and sometimes
use a specific bit pattern to trigger the problem.

I had one machine that compiled kernel just fine, and ran pretty smoothly
overall, but experienced weird hickups like dying apps, failing oracle
install etc. Not too much though, I was attributing them to buggy software.
Then I tried to take a large backup. Bzip failed (internal error) one third
of the time, and once produced a different result. I quickly hacked up an
user space memory tester, and sure enough it reported an error after five
hours. (The machine was already in production, so I couldn't just take it
down and launch memtest86.) I verified the problem with memtest86 during the
next night, and applied the marvellous badmem patch to the kernel. After
marking the problematic place unuable, all problem disappeared. I just lost
2MB out of 256.

What I learned is that spotting meory error can be difficult, and the
symptoms can be stealthy.


-- v --

v@iki.fi
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:29    [W:0.079 / U:0.032 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site