On Wed, 25 Oct 2000, Guido Guenther wrote:
> On Wed, Oct 25, 2000 at 10:14:53AM -0700, Keith M Wesolowski wrote:
> > On Wed, Oct 25, 2000 at 07:01:29PM +0200, Guido Guenther wrote:
> > Not to sound defensive, but I'm fairly sure this isn't an
> > fdisk-related problem. The partitions that fdisk creates follow the
> > spec as far as I can see.
> I didn't want to blame fdisk for that(how could I, I want you to send
> the patches upstream ASAP :)
> >
> > > What puzzles me even more is that I get illegal instructions for almost
> > > all commands I execute afterwards. Any comments on this one?
> >
> > I rather suspect that this is the same problem that causes the request
> > for the out-of-bounds block in the first place: kernel memory
> > corruption. Unfortunately I have few ideas as to what the specific
> > problem is. I would start bug-hunting in the sgi disklabel kernel
> > parts. Make sure that it's compatible with what fdisk is doing.
> That's a starting point - thanks. Ian pointed out that it might be
> related to the fact that I use two harddisks which is interesting since
> I see the problems only when writing on sda, sdb seems to be o.k.
> Regards,
> -- Guido
This is a problem which I'm seeing for a loooooong time now.
I've been geeting this correuption a while back too - look at the mailing
list archives. The problem seems to be cache corruption as far as I can
see. Since Ralf checked in some small fixes for the cache handling it's a
bit better - but not completely fixed.
The Problem is also tracked on the buc tracker btw (#5) .
I have seen this only when copying from one harddisk to another.
Especialy if one wants to copy the kernel sources it will messup.
CU, Klaus
--
Full Name : Klaus Naumann | (http://www.mgnet.de/) (Germany)
Nickname : Spock | Org.: Mad Guys Network
Phone / FAX : ++49/177/7862964 | E-Mail: (spock@mgnet.de)
PGP Key : www.mgnet.de/keys/key_spock.txt
|