Path: utzoo!attcan!uunet!mcvax!enea!kth!sics!uplog!thomas
From: thomas@uplog.se (Thomas Hameenaho)
Newsgroups: comp.arch
Subject: Re: NULL pointers (was: Software Distribution)
Keywords: software distribution, null pointers
Message-ID: <303@uplog.se>
Date: 3 Oct 88 07:29:05 GMT
References: <978@esunix.UUCP> <3112@pt.cs.cmu.edu>
Reply-To: thomas@uplog.UUCP (Thomas Hameenaho)
Organization: TeleLOGIC Uppsala AB
Lines: 22

In article <3112@pt.cs.cmu.edu> bsy@PLAY.MACH.CS.CMU.EDU (Bennet Yee) writes:
>Of course, if your machine core dumps on dereferencing NULL, just trap the
>fault (SIGSEGV), check the instruction that caused the trap, and patch up
>and return if it was due to dereferencing NULL; otherwise dump core.  If we
>presume that most pointers dereferenced are not NULL, then the run-time
>costs are small.  It's a very machine dependent solution, but certainly can
>be done on any Unix machine.

This problem should be easy to solve hadn't it been the case that many
Unix kernels are defunct when it comes to using the a.out header.
Our Pyramid (hint hint!) dumps core if the text segment doesn't start
on virtual address 0! The sysV 68020 machines we have can correctly
use the header. You can for instance have the entry point at address 4
and have a 0 at address 0 if you just want to be able to run the code as is.
Or if you prefer you can skip page 0 and trap any references to 0 and
once and for all fix the offending code.

-- 
Real life:	Thomas Hameenaho		Email:	thomas@uplog.{se,uucp}
Snail mail:	TeleLOGIC Uppsala AB		Phone:	+46 18 189406
		Box 1218			Fax:	+46 18 132039
		S - 751 42 Uppsala, Sweden