Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 27 Nov 1997 14:10:21 +0100
From:      Holger Bauer <bauer@itsm.uni-stuttgart.de>
To:        Jakob Curdes <jc@info-systems.com>
Cc:        aic7xxx@FreeBSD.ORG
Subject:   Re: Problems with 7880 and smp system
Message-ID:  <199711271347.OAA00411@kirk.lampoldshausen.de>

next in thread | raw e-mail | index | archive | help
    Jakob> most 2.0 kernels seem to have deadlock problems with smp, I
    Jakob> decided to give it a try and switch to the latest
    Jakob> development kernel. I assume that there these problems are
    Jakob> fixed ?! But, running on 2.1.66 my system is rather less
    Jakob> stable than before.  I suspect that the scsi driver for my

Is your kerneld running? Have you really applied make modules and make
install_modules in the /usr/src/linux directory? If you haven't this
might explain the missing logs and the occuring deadlocks.

I am running a Gigabyte DX586 with kernel version 2.1.23, 2.1.63 and
2.1.65 and have problems if I compile the 2.1.65 kernel with verbose
/proc/scsi/aic7xxx/0 reporting mechanism. The first time I apply 
cat /proc/scsi/aic7xxx/0 I'll get all my setting, everyting is
fine. However it might happen that in the middle of another 'cat'
browsing the system gets in some deadlock state. Well, I thought I
might find out why the MHz-report in the startup process (up to kernel
2.1.2X) disappeared and if I really run my SOOOOOOO Slow Fireball
TM3200S in the appropriate mode.

I won't try my cat command before I've send out this message :-)
though.

Anyway, does anybody know, what

Nov 27 08:27:16 kirk kernel: aic7xxx: possible overflow at loop 6:8

means? This only occured in kernel 2.1.65 during the startup message
and will only be reported to the /var/log/messages file.

To come back to the original question. NO, besides the cat problem
I've never had any problems with my Dual-Pentium Gigabyte 586 DX
SMP setup. Even some people claimed that the two processors should come
out of the same series. This is not true in my case. One Intel Pentium
133 in the box and the second plain one bought about 1 year later from a
different seller.

I'm running our parallel flow simulation package (MPI) on it.

Holger




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199711271347.OAA00411>