Date: Wed, 24 Jan 2001 10:11:04 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender: eliz AT is
To: Martin Str|mberg <ams AT ludd DOT luth DOT se>
cc: djgpp-workers AT delorie DOT com
Subject: Re: Debugging on 386
In-Reply-To: <200101231805.TAA21501@father.ludd.luth.se>
Message-ID: <Pine.SUN.3.91.1010124101047.25819F-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk


On Tue, 23 Jan 2001, Martin Str|mberg wrote:

> According to Eli Zaretskii:
> > What happens if you say "c" instead of "n" at that point?  Does the
> > program run normally then?
> 
> No, I get Exiting due to signal SIGFPE, etc...
> 
> However if I do not put any breakpoint at main, the program runs
> properly.

Yes, that matches my experience back when I tried running GDB with
387=n in the environment.

Could you try doing the same in FSDB?  I'd like to establish if this
has anything to do with GDB's own code.

Note that FSDB uses hardware-assisted breakpoints for the first 4
breakpoints, so for a fair comparison either use hbr in GDB or put 4
watchpoints on some random data in FSDB, so that the debug registers
are taken, before you put your breakpoint.  (Actually, both variations
are probably worth trying, in case using hbr in GDB somehow solves the
problem, or at least produces a different effect that could give some
insight.)