home *** CD-ROM | disk | FTP | other *** search
- Path: w351zrz.zrz.TU-Berlin.DE!rawneiha
- From: rawneiha@w351zrz.zrz.TU-Berlin.DE (Philipp Boerker)
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: Amiga Memory Map Needed! / ExecBase
- Date: 9 Feb 1996 12:09:43 GMT
- Organization: Technical University of Berlin, Germany
- Message-ID: <4ffdi7$rgc@news.cs.tu-berlin.de>
- References: <4f9itc$ld8@rock101.genie.net> <4facl8$ga0@brachio.zrz.TU-Berlin.DE> <wulfraedDMFxn6.L3y@netcom.com> <4fcnev$b3j@brachio.zrz.TU-Berlin.DE> <4fd3fj$548@serpens.rhein.de>
- NNTP-Posting-Host: w351zrz.zrz.tu-berlin.de
- Mime-Version: 1.0
- Content-Type: text/plain; charset=iso-8859-1
- Content-Transfer-Encoding: 8bit
-
- mlelstv@serpens.rhein.de (Michael van Elst) writes:
-
-
- >>It will have it from an fixed address of course, but this address could
- >>vary from computer to computer, from OS version to OS version. No user code
- >>should rely on fixed addresses or other hardware. Supervisor code has to.
-
- [rude corrections cut here ;) ]
-
- >The only method that avoids any absolute address would be to pass a pointer
- >to every new task, which wouldn't be consistent (interrupts still have to
- >fetch ExecBase from a fixed address) and requires extra magic (because the
- >first task has to magically "know" the pointer).
-
- This fixed address could be a constant that differs from compilation to
- compilation of the OS.
-
-
- > And all this just for aesthetics with not a single advantage in
- > functionality.
-
- It wouldn't be to much effort to pass ExecBase at the beginning. The user
- code would have to store it (move.l a6,ExecBase) and go on reading it from
- there (move.l ExecBase,a6). Where is the problem?
-
- Regards,
- Phil.
-
-