In article <20030124160704.GA9348@yeah-baby.shagadelic.org>,
Jason R Thorpe <thorpej@wasabisystems.com> wrote:
>On Fri, Jan 24, 2003 at 06:30:02PM +1100, matthew green wrote:
>
> > will this let, eg, the sysvipc calls (shm?) in uvm become hooks
> > instead?  that means sysvipc can become an LKM far more easily!
>
>Possibly, yes.

Not that easy... Because the exechook() is called too late [after
the uvm space is switched to the one of the new image and we need
to destroy the shm space before that is gone].

Also there is the issue of when to destroy the shm space... I don't
think that the rest of the kernel has enough knowledge to do this.

christos