Subject: Re: PROPOSAL: removal of brk()/sbrk().
To: Todd Vierling <tv@wasabisystems.com>
From: Christos Zoulas <christos@zoulas.com>
List: tech-kern
Date: 02/28/2002 11:57:21
On Feb 28, 11:46am, tv@wasabisystems.com (Todd Vierling) wrote:
-- Subject: Re: PROPOSAL: removal of brk()/sbrk().

| On Thu, 28 Feb 2002, Christos Zoulas wrote:
| 
| : >Emacs was developed under ITS and also runs under TOPS-20.  Ditto MIT's
| : >MIDAS assembler which AFAICT uses the same dumping and undumping scheme.
| : >TeX was developed (I believe) under WAITS and also runs under TOPS-10
| : >and TOPS-20.
| : >
| : >I wonder if there are any UNIX programs, not ported from one of the PDP-10
| : >OSs, that undump themselves?  (Someone already mentioned sendmail.)
| :
| : monop(6).
| 
| Not the version we have, at least.  Our monop(6) has an "initdeck" which
| builds a position-independent binary data blob.

| The undumping mentioned in this thread has very specifically to do with
| position-dependent data:  either dumping the memory space of a program and
| creating an iffy binary image that can later be re-executed directly, or
| dumping a section of memory whilst expecting to be able to mmap(...,
| MAP_FIXED, ...) the blob back into place.


Well, I am talking about save and restore; line 213 execute.c...
I am not sure it is all position independent, but I could be wrong.

christos