On Aug 27, 2007, at 13:03, Martin Husemann wrote:
> That is due to the missing tick interupt on cpu1 (I'm working on it)
>
>> 'halt' hung after stopping
>> cpu0, didn't drop back to OF.
>
> I see that too, seems we need to bounce cpu_reboot() to run on cpu0  
> always,
> so we exit to OF with the bootsrap cpu (currently we halt all other  
> cpus
> and exit to OF with whatever cpu happens to run cpu_reboot).

   Similar to matthew green, I note that the sparc port is able to  
halt and
drop with whatever the running CPU is.  I've had my quad-sparc drop
to OBP with a variety of CPUs being the active one.

   Perhaps this is something worth not worrying about until the ticks
are working properly on the non-0 CPUs, as that might be related to the
problem?

                                 - Chris