On 26 Feb 2002, Wolfgang Rupprecht wrote:

: It certainly is an interesting idea and I stopped short of suggesting
: exactly this on an mmap() thread on gnu.emacs.bug last week.
:
: The fly in the ointment is that emacs uses actual pointers in the lisp
: cells.

That's a bug, from the standpoint of loading the data using mmap() at a
later time.  It should use offsets from a base address (or even "offset from
this value's address", which doesn't require a base address variable) rather
than direct pointers.  That's position-independent data, similar to how
position-independent code works.

With that said:

: Will NetBSD always honor requests to mmap() things into the same area that
: they were previously mmapped into?  Under what conditions will NetBSD
: decline the request to preserve the old mmapping?

NetBSD will allow a MAP_FIXED request if that memory location/block is
actually available.

-- 
-- Todd Vierling <tv@wasabisystems.com>  *  Wasabi & NetBSD:  Run with it.
-- CDs, Integration, Embedding, Support -- http://www.wasabisystems.com/