On Sun, Feb 24, 2008 at 11:56:52PM +0000, Chris Gilbert wrote:
Perhaps I'm just being overly paranoid, and worried that something
somewhere will allow a memory access it shouldn't rather than
causing a
trap or access fault.
I think the idea is to do several _remove() or _add calls and then
synchronize the hardware with the expected state just once, for
expensive
MMU hardware accesses.
I think you should update as long as you have control of the
program flow - something like within the same system call, unless you
depend on the access control yourself.
PMAP experts may correct me of course.