>>>>> "Markus W" == mw  <mw@eunet.ch> writes:

Markus W> Did you write this from memory, or is this the actual code
Markus W> you were compiling? This originally was moveml #0xC0C0,sp@-
Markus W> which symbolically would be moveml d0/d1/a0/a1,sp@- nothing
Markus W> going on with d2... Take care, you're later branching into
Markus W> the other interrupt-handler, which expects to find
Markus W> d0/d1/a0/a1 on the stack to restore... Store d2 after
Markus W> deciding that it's indeed a fake level 6.

Oops, you're right of course, and even more, the argument to hardclock
is relative to SP which was the cause for statclock to fail.  I now
save d0/d1/a0/a1 initially and later d2.  I have recomputed the
hardclock argument location and now statclock doesn't panic anymore.
But... the root partition still fails to mount :-)

Well, I'm going to compile the scsi subsystem with debugging on to
find out some more.

See ya'

Niklas