Mailing-List: contact dev-help@apr.apache.org; run by ezmlm
Precedence: bulk
Message-ID: <4679119E.30509@apache.org>
Date: Wed, 20 Jun 2007 07:38:06 -0400
From: "William A. Rowe, Jr." <wrowe@apache.org>
User-Agent: Thunderbird 1.5.0.12 (X11/20070530)
MIME-Version: 1.0
To: Peter Samuelson <peter@p12n.org>
CC: Manfred Rebentisch <mrebentisch@comparat.de>,  dev@apr.apache.org
Subject: Re: SHM / RMM pointer values
References: <200706150923.58591.mrebentisch@comparat.de>
 <20070620042048.GA31225@p12n.org>
In-Reply-To: <20070620042048.GA31225@p12n.org>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit

Exactly.  Good call Peter.

The bottom line - shim pointers need to be apr_size_t indexes into
the defined SHM segment.

Peter Samuelson wrote:
> [Manfred Rebentisch]
>> struct mystr {
>> 	struct mystr* next;
>> 	char* data;
>> 	size_t dsize;
>> };
> 
> I'm pretty sure you can't portably implement shared memory mapped to a
> fixed base address in all processes.  Thus pointers simply _cannot_ be
> shared as pointers, you need to make them size_t values relative to the
> shm segment base address, or something.
> 
> On many systems you can _request_ that a shm segment be mapped to a
> particular address, but the OS may or may not be able (and willing) to
> honor that.  And on other systems you can't even do that.  So apr
> doesn't support that feature.