From owner-freebsd-hackers  Fri Dec 13 11:43:00 1996
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id LAA08650
          for hackers-outgoing; Fri, 13 Dec 1996 11:43:00 -0800 (PST)
Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.211])
          by freefall.freebsd.org (8.8.4/8.8.4) with SMTP id LAA08645
          for <hackers@freebsd.org>; Fri, 13 Dec 1996 11:42:52 -0800 (PST)
Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id MAA15610; Fri, 13 Dec 1996 12:19:53 -0700
From: Terry Lambert <terry@lambert.org>
Message-Id: <199612131919.MAA15610@phaeton.artisoft.com>
Subject: Re: Multiple Buffer allocation of Shared Memory
To: scrappy@hub.org (Marc G. Fournier)
Date: Fri, 13 Dec 1996 12:19:52 -0700 (MST)
Cc: erich@lodgenet.com, hackers@freebsd.org
In-Reply-To: <Pine.NEB.3.95.961213002725.14115D-100000@hub.org> from "Marc G. Fournier" at Dec 13, 96 00:47:09 am
X-Mailer: ELM [version 2.4 PL24]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

> parent:
> 
> 	open named file (O_RDONLY, O_RDWR, or it doesn't matter?)
> 	ftruncate named file
> 	mmap it using something similar to:
>   	   mmap(0, <size of file>, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
> 	close named file
> 
> child:
> 
> 	open named file (again, does the mode matter?)
> 	mmap it using something similar to:
>            mmap(0, <size of file>, PROT_READ, MAP_SHARED, fd, 0);
> 	close named file

I believe the mode does not matter.  Specifically, the mode is on the
fd, not on the vp.  All fd's point to the same vp.

One posible change in the future, however, is honoring the read/write
restrictions on the fd when the fd is mmap'ed.  This may be done so
that fd's inherited from suid/sgid modes and opened read-only *remain*
read-only independent of what you specify to mmap.

I'm not sure if fd's inherited this way can currently be mapped or not...
it's an interesting special case.

Either way, I'd suggest opening O_RDWR if you plan to later map the file
PROT_READ|PROT_WRITE.

> 	Is that the gist of it?  Then I can "read/write" the mmap'd area
> just like I would a similarly malloc()'d area, correct?

Yes.


					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.