home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.amiga.misc
- Path: sparky!uunet!news.smith.edu!sophia.smith.edu!jfieber
- From: jfieber@sophia.smith.edu (John Fieber)
- Subject: Re: Why doesn't Amiga core-dump? *sigh*
- Message-ID: <1992Nov19.142226.27864@sophia.smith.edu>
- Sender: root@sophia.smith.edu (Operator)
- Organization: Smith College
- References: <92323.003545BGT101@psuvm.psu.edu> <1992Nov18.133139.21744@ifi.unizh.ch> <92323.182614BGT101@psuvm.psu.edu>
- Date: Thu, 19 Nov 1992 14:22:26 GMT
- Lines: 26
-
- In article <92323.182614BGT101@psuvm.psu.edu> Blaise Tarr <BGT101@psuvm.psu.edu> writes:
- >on the road to memory protection. When a message is passed (or rather a
- >pointer to the message), the area with the message essentially belongs to
- >the process receiving the message. The MMU tables can be modified so the
- >receiving process has access to the message until it has sent a ReplyMsg().
-
- ...BUT, how is the OS going to know what memory constitues the
- message? If I use a message to pass a pointer to a structure
- full of pointers to other bits of data, there could be dozens of
- little pieces of message floating around. The only thing that
- can know for sure the extent of the message is the application
- sending it. The receiving application can usually find out but
- even it has to have some pre-programmed knowledge of what to look
- for.
-
- Sure, I can now use MEMF_PUBLIC when I allocate that memory but
- that still leaves about 99% of current programs that would fall
- flat on their face. Though I will never say never, I cannot at
- this moment see any way to (1) make protected memory happen (2)
- not break existing software and (3) not take a huge performance
- hit in the process.
-
- -john
- --
- === jfieber@sophia.smith.edu ================================================
- ======================================= Come up and be a kite! --K. Bush ===
-