I've been poking at the sparc32 SMP code and I have it to the point
where it will run a minimal userland.  It is still far from being
stable though.

For reference, I started with the SMP patch [0] that wli sent to
the list last July.  To go along with the attached SMP patch, it
needs the spinlock fix [1] from Art Haas, and the attached
DEBUGSPINLOCK patch from Tom Callaway.

Here's a list of problems I have seen with it so far:
- the scsi driver must be compiled in or it has problems.
- when I compiled with gcc 3.3.4, it drops back to the prom with
     a level 15 interrupt.
- from a serial console, init hangs
- /proc/interrupts shows 0 interrupts for the 2nd cpu
- sun4d still needs fixing

Testing with gcc 3.2.2 and 2 150 MHz HyperSPARCs, if I compile
in scsi and leave out all of the networking, I am able to pass
"init=/bin/bash" and get a shell everytime.

I am sure there are still areas of bitrot and neglect that remain,
so if anyone has any ideas, feel free to share them.

Bob

[0] http://marc.theaimsgroup.com/?l=linux-sparc&m=109105275025491
[1] http://marc.theaimsgroup.com/?l=linux-sparc&m=110806229008323