> > 
> > On Thursday, October 16, 2003, at 03:25  AM, Richard Earnshaw wrote:
> > 
> > > True, but in that case the worst that would happen is that we would 
> > > take a
> > > second fault after adding the page read-only on the first fault.
> > 
> > Except that we specifically changed the pmap_enter() interface in order 
> > to avoid double faults like these.  It would really suck to reintroduce 
> > double faults for this (pretty common) case.
> > 
> >          -- Jason R. Thorpe <thorpej@wasabisystems.com>
> > 
> 
> But then, so would having to decode every Thumb instruction.
> 
> R.
> 

And even if we don't do this always, we could use it to speed up modified 
emulation: there's no need to decode the instruction if it's a permission 
fault, if we know that can only happen for a write.

R.