On Tue, 16 Jul 1996, The Great Mr. Kurtz [David A. Gatwood] wrote:

> Okay, so I decide to try the sbc-1 kernel (the one with a patch to let it
> work on more systems).

Note that the patch was very specific (and small), and only handles FPU
traps somewhat differently.  (This is for those folks that haven't looked
at these kernels yet.)

> First time I mount, it hoses the fs very badly
> and drops into the debugger.

>From what you mentioned in the other message, this sounds like a general
timeout problem in the driver.  In my experience, if you experience a
crash involving the SCSI driver, you should _NOT_ allow the system to
attempt to write a core file, but instead hit the power button.  I can't
even tell you how many filesystems I've lost because of SCSI lossage (both
ncrscsi and sbc).

> I boot from the EZ135, my backup drive, but
> didn't think to remove the sbc kernel from it... and hosed the superblock
> backup at 32, and apparently every other superblock backup.  There's a
> magic number problem....

"Oops!" :-(  Did you try any other backups?  You can find them with `newfs
-N'.

> As for testing the sbc kernel on the IIsi, I'm now having second thoughts.

The nsrscsi kernels (GENERIC and GENERIC-1) are going to be very close to
your working kernel.  I generally don't recommend that folks use sbc
unless they are having specific problems with the ncrscsi driver.

Sounds like you've had a _wonderful_ week, and I'm very sorry to hear that
the sbc driver was a part of it.

--scott