2.1.8X boot hangs on 2940UW controller...

[Robert G. Brown]

On Tue, 10 Mar 1998, Steve Tinney wrote:

> >My point being, I don't think it's the bios on the motherboard or scsi
> >card.  Rather it's a change with the smp code, I understand it's been
> >changed to make it much faster in 2.1.85.
> 
> There may be problems with such changes, but I have had chronic problems
> with a P6DNF configured with only one PPro 200 and a 2940.  I am about
> to take the SCSI out and put in an IDE drive to see if it cures the problems.

Try two other things as well:  Move the card to a different PCI slot
(actually, move all your PCI cards around) and make sure that your BIOS
is correctly configured in irq-space for the PCI slots:-).  Also check
carefully for irq and dma conflicts; I know that it isn't supposed to
matter, but I believe that the adaptec is only really "happy" (under
2.0.X) when it holds irq 10.  I also know that it (used to, at least)
flake out and crash even DOS/Windows, not just linux, when it didn't
hold irq 10.  Probably a bios bug, I agree, but one that it is easier to
accomodate than fight.

I've actually had REALLY nasty problems with IDE drives and dual PPro's
-- I'm still totally convinced that the drives can somehow be crashed in
software by 2.0.(X<31) on a fast enough SMP system, because I have lost
a stunning 8 out of 9 ide drives of various sizes and from two different
manufacturers (WD and Seagate) with identical errors over two years.
Since I put in 2.0.32 and 2.0.33, I've had no further crashes; I really
suspect an occult race/deadlock in the 2.0.30 kernel (which, from some
of Leonard Zubkoff's comments, may have affected P6DNF's particularly
nastily) was causing a write timing error to overwrite formatting
information over a very long time period.  Something like this was known
to be possible for 850 MH WD IDE drives at boot time, but I saw it in
2 GB Seagates and larger WD drives as well.  

The SCSI drives seemed more tolerant of this particular error; I've had
zero SCSI drive failures over two years and only a very few
kernel/controller problems (like the one I'm working on now:-(

   rgb

Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy.duke.edu




To Unsubscribe: send mail to majordomo at FreeBSD.org
with "unsubscribe aic7xxx" in the body of the message