X-Authentication-Warning: delorie.com: mail set sender to djgpp-bounces using -f
From: RadSurfer AT yahoo DOT com (Radical NetSurfer)
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Achieving 486-Compatibility (2)
Date: 7 Feb 2004 12:09:06 -0800
Organization: http://groups.google.com
Lines: 46
Message-ID: <c9d8c82b.0402071209.6bd72830@posting.google.com>
References: <c9d8c82b DOT 0402061627 DOT 49f37f53 AT posting DOT google DOT com> <3405-Sat07Feb2004162009+0200-eliz AT elta DOT co DOT il>
NNTP-Posting-Host: 65.120.100.70
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit
X-Trace: posting.google.com 1076184546 11017 127.0.0.1 (7 Feb 2004 20:09:06 GMT)
X-Complaints-To: groups-abuse AT google DOT com
NNTP-Posting-Date: Sat, 7 Feb 2004 20:09:06 +0000 (UTC)
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

"Eli Zaretskii" <eliz AT elta DOT co DOT il> wrote in message news:<3405-Sat07Feb2004162009+0200-eliz AT elta DOT co DOT il>...
> > From: RadSurfer AT yahoo DOT com (Radical NetSurfer)
> > Newsgroups: comp.os.msdos.djgpp
> > Date: 6 Feb 2004 16:27:18 -0800
> > 
> > Exiting due to signal SIGSEGV
> > Page fault at eip=00014def, error=0004
> > 
> > What is that error = 0004 ?????????
> 
> For Page Fault, the error code of 4 means that your program attempted
> to read from a bad address.  (This is explained in section 12.2 of the
> FAQ, btw.)
> 
> It would be a good idea to post a complete crash traceback, complete
> with the full register dump.  Running SYMIFY (or, better yet,
> bfdsymify) on it would be even better.  Given that information,
> perhaps someone around here could get some useful ideas about possible
> causes for the crash.
> 
> > the exact same identical program does seem to run
> > properly on my P4 with no problems at all....
> > 
> > Any ideas?
> 
> It's a bug.  Page Faults cannot be caused by a bad opcode (which would
> be the case if the program couldn't run on a 486): those trigger
> SIGILL, not SIGSEGV.  I suspect that the reason for the different
> behavior is something else, like memory amount/configuration, for
> example.  Windows is quite lenient towards bad pointers.

Bullseye! On the P4's, it is through a DOS-shell we run my application under
 Win9x, and on those other machines it was "pure" ms-dos (no windoze)!
 The program does use malloc() but I check for proper return values;
 and the program does do quite a bit of String_manipulation.
 Time for a print-out and some good old-fashion analysis!

 Here's a question:
 Say I wanted to _deliberately_ attempt to get the P4 to actually
 behave the same way as these other machines;  
 Attempt to boot to ms-dos only?
 Is there some compile-option that would force using CWSDPMI (?),
 and not the Windoze DMPI,  etc.? Could prove useful maybe.
 
> As DJ suggested: debugging the program is the way to find out what
> went wrong.