Re: 2.6.8-rc1: Possible SCSI-related problem on dual Opteron w/ NUMA

From: R. J. Wysocki
Date: Sun Jul 18 2004 - 07:40:10 EST


Andi,

On Saturday 17 of July 2004 21:09, R. J. Wysocki wrote:
> On Saturday 17 of July 2004 20:12, Andi Kleen wrote:
> > On Sat, Jul 17, 2004 at 06:26:03PM +0200, R. J. Wysocki wrote:
> > > I got this on a dual Opteron system on 2.6.8-rc1 with the latest x86-64
> > > patchset from Andi:
> >
> > Does it happen with x86_64-2.6.8-1 too ?
>
> It did not happen when I was running that kernel, but I had only run it for
> a couple of times. It generally does not happen with this one either.
> It's happened only once and I reported it immediately. But ...
>
[snip]

I had this problem again this morning. I was unpacking the kernel tarball to
/dev/sda8 and it went south (the tarball had been partially unpacked before
the partition was remounted r-o). Then, I got back to 2.6.7 and ran fsck -
now it found some errors (obviously) and fixed them. Next (on 2.6.7), I
unpacked the kernel to /dev/sda8 (again) and compiled the 2.6.8-rc2. I ran
it, unpacked the kernel to /dev/sda8 (again) and compiled it - everything
worked. Then, I applied your patch on top of the newly created 2.6.8-rc2
tree and compiled the kernel. After installing and running it I tried to
unpack the kernel to /dev/sda8 (again) and it went south, so I got back to
the "plain" 2.6.8-rc2, ran fsck and fixed the partition, unpacked the kernel
to /dev/sda8 - and it all worked.

So, it seems, there's something in your patch that causes this misbehavior.

rjw

--
Rafael J. Wysocki
----------------------------
For a successful technology, reality must take precedence over public
relations, for nature cannot be fooled.
-- Richard P. Feynman
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/