In article <BURLEY.92Aug13153840@geech.gnu.ai.mit.edu> burley@geech.gnu.ai.mit.edu (Craig Burley) writes:
quoted text
>In article <1992Aug13.163854.21617@midway.uchicago.edu> ace3@quads.uchicago.edu (Tony 'LLama' Acero) writes:
>
>   I have no idea what's going on and would appreciate any input! :-)
>   (The smiley is to indicate I'm not complaining and half-expecting
>   that I've done something bone-headed)
>
>I'm not sure about your problem or the person's to whose post you followed up,
>but...
>
>...I believe there is a bug in Linux that has the following behavior:
>
>-  causes Linux to "misread" one 1024KB chunk of data from a disk-based file
>   so that what your app ends up with is some _other_ 1024KB chunk
>   (apparently from the same file)
>
>-  occurs only during very heavy disk access, such as megabytes accessed
>   continually
>
>-  is intermittent, but happens enough to reproduce fairly easily
>
>-  might be SCSI-related (I have a SCSI system) but, based on responses I've
>   gotten from others saying they've seen the same behavior, probably isn't
>
>-  is still in 0.97 and perhaps happens somewhat more often there (though of
>   course it's hard to measure this)

Yep. I've seen this too. My system is running off SCSI disks. I'd agree
with everthing listed above and add:

        - It can happen on WRITES as well.

Occasionally when I'm rebuilding the kernel, make will die with some
horrid error like it can no longer find the makefile it was working from.
Shorty thereafter, anything which accesses the disk seg. faults and dies.

When I bring the system back up, the filesystem is hosed. Usually the
superblock is scrambled so fsck just refuses to look at the drive.

Not good.

-Chris

-- 
---- Chris Newbold ---- > "If you fool around with a thing for very long you <
   Intermetrics, Inc.   >                 will screw it up."                 <
  Opinions and statements expressed are MINE and do NOT represent my employer
                        newbie@inmet.camb.inmet.com