Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!ucsd!ames!uhccux!munnari.oz.au!bruce!alanf
From: alanf@bruce.cs.monash.OZ.AU (Alan Grant Finlay)
Newsgroups: comp.lang.prolog
Subject: Re: GC triggering and stack limit checking by MMU hardware
Summary: I agree
Keywords: GC, stack, heap, MMU
Message-ID: <2730@bruce.cs.monash.OZ.AU>
Date: 24 Jul 90 04:50:38 GMT
References: <1990Jul19.151524.22544@diku.dk> <3260@swi.swi.psy.uva.nl> <11079@alice.UUCP>
Organization: Monash Uni. Computer Science, Australia
Lines: 10


I agree with those who point out that using MMU segment violations to trap
stack overflow is not a good idea.  Apart from the reasons which have already
been given, where hardware is concerned it should be used as intended.  What
we really need is for hardware designers to recognise the need for multiple
stack overflow detection and provide a uniform mechanism to support it.
I suggest a simple "notify addr" instruction which saves a limited number of
addresses in hardware registers and generates a trap when these addresses are
accessed.  The point is that if you want to open a can then you should use
a can opener, not a hammer.