[This has also been sent to the linux-smp@vger.kernel.org list, but I
thought the debian community might have some insight on the dselect
error.]
Hi,
I have recently had a pile of problems with my Tyan 1832DL based
machine. Before adding the second processor, I began having a number of
BSoD crashes under Windows 2000 when I had only had one BSoD in the
previous year. That was when I decided to move to Linux permanently,
combining parts from an older secondary box that was running Linux.
Currently, the machine is running:
two P2 450's
one generic 256MB 10ns ECC Registered SDRAM
Voodoo 3000 PCI
both on-board IDE buses running, with a IBM 10GB and a Plextor
CD-R/W
a Adaptech 2940UW SCSI controller with a 4GB and a 10GB IBM drive
two 3Com 3C905B PCI-based NICs
Creative Lab SoundBlaster AWE32 PnP
both serial ports and the parallel ports have been disbaled
and I'm running Debian off of the Testing pool with a 2.2.18pre21 kernel.
Originally, I didn't think this was a SMP problem because of the Windows
crashes and becuase Linux crashed many times before I added the SMP kernel
(but after I installed the secnd processor). When I did finally get
around to installing SMP and reading the HowTo, I began using the
"mem=255M" kernel parameter which vastly improved the situation.
I will mention that the most common activity just prior to a
freeze/reboot, is playing with dselect, debian's console based front-end
to the package manager. One thought I had was that the problem was
related to /dev/sdb, where /var was located (and thus dselect had lots of
activity), but I have since moved /var to my root at /dev/sda and have had
two crashes since then.
Now I am down to one or two reboots/freezes a week, but I don't know
where to look next. I have double checked my BIOS for allt he
recommendations listed in the HowTo, and I haven't found any error
messages in any of the logs during any freeze or reboot. Where can I look
for clues?
Anm
