lkml.org 
[lkml]   [1998]   [Nov]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: 2.1.125 filesystem corruption under extreme load
Arjan van de Ven wrote:
>
> Hi,
>
> I just found that the 2.1.125 kernel corrupted my ext2 filesystem under
> extreme load.
> WHat I did:
>
> tar -zcvf ~/linux2130.tar.gz linux
> Create a new tarfile for the new kernel
> cd linux
> make oldconfig
> make bzImage
> make -j modules
> ^^^
> forgot the "2" here.
>
> This caused the load to go to (give or take a few) 124, and eventually
> filled all 128 megs of ram and 50 megs of swapspace. Then the make crashed
> down but several files in the tree were totally corrupted (I had to untar
> the recently tar'd file).
>
> I am using kernel 2.1.125 (I know it is old, but if this bug isn't solved
> yet, it is important anyhow), on a 300 Mhz K6-2 system with 128 Mb ram.
> The disc in question is a 6.4 gig UDMA WDCAC364, _not_ using DMA-mode.
>

This sounds exactly like what I did under 2.1.108, compiling 2.1.129.
(Perhaps it doesn't happen under 2.1.130 but I'm not in a position to
find out yet). I lost a bunch of files from the kernel tree too. I
posted a msg about this maybe 10 days ago, subject was "Weird: loss of
files during kernel build (!)".

Linus, I'm CCing you to make sure you know about this (!) since 2.2 is
imminent - perhaps you can organise someone to do some tests under the
newest and greatest kernel with a non-production machine?

My configuration is radically different to yours Arjan: Dell Wstn 410,
2xPII-400, 512MB RAM, Aic7xxx 7890 (2xQuantum Viking II 4GB) and 7880
(1xQuantum Viking 4GB, CDROM, DAT), kernel at the time was 2.1.108 with
aic7xxx driver version 5.1.0pre10, and it was SMP. Distrib is Redhat
5.1.

I think we can exonerate the scsi layer since it happens on IDE too.
Are you using an SMP kernel Arjan? If not then that's out too.

The bit that bothers me is that I didn't even let the machine hit the
process limit, I ctrl-c'ed it as it passed through the 430 process mark
(and it did stop very quickly after that). Perhaps top just didn't spot
this happening but I really don't think it happened. The system didn't
go crazy with swap, due to the half-gig of RAM, though it may have
swapped a little.

Neil

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

\
 
 \ /
  Last update: 2005-03-22 13:45    [W:1.409 / U:0.088 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site