Re: 2.1.42

Swen Thuemmler (swen@uni-paderborn.de)
Fri, 6 Jun 1997 11:14:32 +0200 (MET DST)


On Thu, 5 Jun 1997, Richard B. Johnson wrote:

[...]
> I tared his stuff to a tape, built an entirely new root-file system
> with all the new tools, etc. I used his same Adaptec SCSI Controller
> AHA1542, and Ethernet card. The only thing that has changed is the
> operating system. The hardware worked for many years.

If the Adaptec SCSI controller is a 1542B with BIOS version less than
3.10, you will get into trouble, since there seems to be a bug with
scatter-gather in this version. I had a lot of problems with BIOS 3.08,
and had to patch the kernel (in drivers/scsi/aha1542.h, change
#define AHA1542_SCATTER 16
to
#define AHA1542_SCATTER 8
)
Yesterday I made new EPROMS, and the problems are gone...

[...]

> The Operating System, with the new disk and his Adaptec controller and
> Ethernet card work fine on my Pentium, where I built the kernel. I
> built it as a '386 with math emulation of course. I have done this
> at least 10 times to upgrade other machines, but I never used a new
> kernel on anything less than a '486 before.

Hmm, if it was the same controller, there must be another problem.

> The machine boots (takes forever), then lets me log in. I can even
> ping it on the network. However, If I do anything using the file-system
> the machine just stops.
>
> If I don't do anything, just wait at the login prompt, the machine
> will eventually stop working.
>
> If I reduce interrupt activity, but disconnecting the network line,
> the machine will "instantly" hang. It seems to need a lot of interrupt
> activity to keep it alive.

Hmm, did you try the no-hlt option on booting (just wild guessing, don't
know if this will help...)

Greetings, Swen