According to matthew green:
>
>   
>   raise_ipi_wait_and_unlock(cpu0): couldn't ping cpu1
>
>OK, so IPI's don't work on hypersparc cpus it seems.
>

Well, I can confirm they _do_ work (I put a printf in the nmi handler
and got output for the other two processors I have) but the master
processor does not seem to see the flag change state in the cpu_info
structure when the other cpu updates it which means either the slave
cpu is not flushing it's cache correctly or the master's cache is not
snooping the update.

-- 
===============================================================================
Brett Lymn, Computer Systems Administrator, BAE SYSTEMS
===============================================================================