the following text was included in the 2.6.15 kernel changelog
The following change was found from the Adaptec developers. Note the
"2.6.13+" reference, the word "deadlock", and the final "sign offs"
commit e5508c13ac25b07585229b144a45cf64a990171e
Author: Salyzyn, Mark <[EMAIL PROTECTED]>
Date: Sat Dec 17 19:26:30 2005 -0800
[PATCH] dpt_i2o fix for deadlock condition
Miquel van Smoorenburg <[EMAIL PROTECTED]> forwarded me this fix to
resolve a deadlock condition that occurs due to the API change in
2.6.13+ kernels dropping the host locking when entering the error
handling. They all end up calling adpt_i2o_post_wait(), which if you
call it unlocked, might return with host_lock locked anyway and that
causes a deadlock.
Signed-off-by: Mark Salyzyn <[EMAIL PROTECTED]>
Cc: James Bottomley <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>
Signed-off-by: Andrew Morton <[EMAIL PROTECTED]>
Signed-off-by: Linus Torvalds <[EMAIL PROTECTED]>
I can definitely say that kernel 2.6.15 fixed my problems. I also can
definitely say that the prior kernel to work was 2.6.12.6. My guess is
that the 2.6.8 also would work.
On Wed, 2006-01-11 at 21:27 +0100, Stefan G. Weichinger wrote:
> Stefan G. Weichinger wrote:
> > Freels, James D. wrote:
> >
> >> The problem I had started at kernels greater than 2.6.12.x (starting
> >> at 2.6.13.0) and was finally cleared at 2.6.15.0.
> >
> >
> > Interesting ... I am using that module with kernel 2.6.13 and have no
> > problems. Maybe this related to the old hardware I use, maybe to the
> > fact that the Suse-guys have patched that already (assumption: NO
> > research done on this by me ...)
>
> Some bell rang ....
>
> Could someone point mo to any related bugreport on this?
> Maybe this has to do with some strange symptoms I see at a customers
> site. They use aic7xx with linux-2.6.8 on a Suse-9.2 ...
>
> No complete lockups, but strange tape-errors all over the place.
> Everything swapped already, the only thing that helped so far was
> putting the drive out of the box and laying it on the top of the case
> with the SCSI-cables through the air ;) Smooth backups since then.
>
> Might be temperature, we'll try an external case for that drive.
>
> But maybe it has to with some module/kernel-problems as well ...
>
> Stefan
>
------------------------------
James D. Freels, Ph.D.
Oak Ridge National Laboratory
[EMAIL PROTECTED]
http://www.comsol.com/stories/hfir/
