On Thu, Nov 16, 2006 at 12:44:57PM -0800, Jason Thorpe wrote:
> 
> On Nov 16, 2006, at 10:15 AM, Joerg Sonnenberger wrote:
> 
> >On Thu, Nov 16, 2006 at 09:43:20AM -0800, Jason Thorpe wrote:
> >>m68k platforms based on the Hibler pmap (i.e. all of them but sun2/
> >>sun3/sun3x) have completely separate 32-bit address spaces for kernel
> >>and user space.
> >
> >Is the kernel space using the full 32bit address space or only  
> >31bit? In
> >the latter case, this would still catch the issues. Checking for
> >overflows in the src domain would make sense as well.
> 
> Full separate 32-bit address space.  The Motorola MMU supports this.

Hm. We could place a reservation at the 31bit boundary and check that
the input for copyout does not cross it or wrap around, which is a
generalised version of what i386 implicitly does with the userspace
limit check.

Joerg