NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/comp/lang/cplus/19886

Internet Message Format | 1993-01-25 | 2.0 KB

Path: sparky!uunet!vnet.ibm.com
From: [e-mail] (Jim Meyers)
Message-ID: <[e-mail]>
Date: Mon, 25 Jan 93 18:15:06 EST
Newsgroups: comp.lang.c++
Subject: Re: Shared Memory Objects
Lines: 36

In <[e-mail]> [e-mail] (Charles Clark) writes:
>
>Straying from the orignal subject a little, how can the location of the
>shared memory segments be coordinated?  The vtable location is fixed at
>link edit time, thus one would know then whether there is a conflict.
>Since the creation and attachment of the shared memory segment is at run-
>time, how can one ensure that the vm address of each of the cooperating
>processes will be available?
>
>       --Charles

This is an implementation-dependent problem. An easy way to do this
(in AIX) is to place all shared objects (and their vtables) in a single shared
library. AIX ensures that all processes accessing the shared library will
load the shared library code at the same virtual address. Thus, the location
of the vtables is consistent. To ensure that the virtual addresses of
shared objects is consistent, the process which instantiates the objects can
send the address to the clients. The clients can control the virtual address
of the shared memory segment via shmat(), and control the virtual address
of their "local" the shared memory objects via the new operator's placement
syntax.

Again, this is for AIX. This solution can also be implemented in OS/2 by
placing shared memory code (and vtables) in a DLL. Jordan describes Cfront-
dependent solutions for both SunOS and System V.

--Jim

-------------------------------------------------------------------------
Jim Meyers                           | Voice:    919-543-5999
IBM Networking Systems               | Fax:      [phone]
LAN Systems Workstation Technology   |
Research Triangle Park, NC 27709     | Internet: [e-mail]
-------------------------------------------------------------------------