Re: pmap vmlocking patch for m68k ports...

[Greg Oster <oster%cs.usask.ca@localhost>]

Andrew Doran writes:
> On Tue, Jan 29, 2008 at 12:41:59AM +0900, Izumi Tsutsui wrote:
> > oster%cs.usask.ca@localhost wrote:
> > 
> > > Digging a bit further, it seems that pmap_motorola.c for m68k archs 
> > > hasn't been updated to the New World vmlocking.  With help from 
> > > bjs@, I have the following patch which gets me over the above 
> > > diagnostic panic...  
> > > 
> > > Any comments on these patches?  
> > 
> > It seems uvm_kernel_object->vmobjlock should be a mutex,
> > but I doubt pmap->pm_lock is actually required.
> 
> If we ever want to do kernel preemption on m68k it's needed but that's
> very unlikely.

IMO it's needed if we intend to keep sprinking KASSERT()s regarding 
those locks.... otherwise DIAGNOSTIC kernels are not going to work on 
m68k arches...

Later...

Greg Oster