Xref: news2.mv.net comp.os.msdos.djgpp:4003
From: Charles Sandmann <sandmann AT clio DOT rice DOT edu>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: DJGPP/Allegro/Memory Problem?
Date: Sat, 18 May 1996 21:35:29 CDT
Organization: Rice University, Houston, Texas
Lines: 15
Message-ID: <319e88f1.sandmann@clio.rice.edu>
References: <4nei97$gr5 AT madeline DOT INS DOT CWRU DOT Edu>
Reply-To: sandmann AT clio DOT rice DOT edu
NNTP-Posting-Host: clio.rice.edu
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

> Page fault cr2=10030028 at eip12e8; flags 3046
> eax=00000502 ebx=0002f278 ecx=00000000 edx=00030000 esi=00001003 edi=10030000
> ebp=00000000 esp=00001270 cs=a7 ds=af es=af fs=0 gs=0 ss=af error=0004

It appears none of the recommendations to fix this helped, so I looked at the
above error message in more detail and found the problem.  It's a bug in our
crt0.s stub.  In particular, we sbrk() memory for the stubinfo structure,
and if you are really unlucky (your program ends up too close to a 64K boundary
that it ends up being allocated in a different memory zone) it gets deallocated
before we get the final tidbit of info from it to exit.

A workaround would be to add some additional code/static memory to the 
program which would move the sbrk() problem.  Another workaround would be to
use unixy sbrk() if you aren't using near pointers/hw interrupts.  A more 
permanent fix will take some hacking on crt0 ...