Tsubai Masanari wrote:

> But why is the cache flush needed in setregs() and pmap_procwr() is
> not enough?

Ok -- thanks for bugging me again about this.  You are right.  The
way this came up is our internal tree was synced a few weeks before
the 11/29/99 check-in to kern_exec.c, so the code was needed for
that codebase.

I will double test the change with our tree now and make the fix.

I would however like to keep the semantics of the regular and
_2way cache index cache ops the same.

jeffs