Date: Wed, 8 Sep 1999 10:30:05 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender: eliz AT is
To: Rez <rividh DOT minusthispart AT earthlink DOT net>
cc: djgpp AT delorie DOT com
Subject: Re: WARNING: DOS is about to die
In-Reply-To: <37D5D74D.3BB@earthlink.net>
Message-ID: <Pine.SUN.3.91.990908102936.21899G@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp AT delorie DOT com
X-Mailing-List: djgpp AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk


On Tue, 7 Sep 1999, Rez wrote:

> Question, then: I beta-test for a coder who works with DJGPP under
> Win95. I test our app exclusively in real 	DOS. About half the time when
> there's a bug, the program runs just fine under Win95 in a DOS box (both
> for the coder and my testing partner) but segviols before it even gets
> fully loaded under real DOS with CWSDPMI. Sometimes this happens even
> with trivial bugs, like a wonked menu item. I realise this is vague, but
> anything in particular which causes this dichotomy (bug has no effect
> under Win95, dies spectacularly under DOS) ???  ISTM if we knew what
> sort of things cause this, we (ie. our coder) could avoid 'em up front.

It's all in the DJGPP FAQ: see section 9.1.

In a nutshell: CWSDPMI supports an extension of the DPMI spec that
allows to decommit the null page, and DJGPP uses that extension in the
startup code.  Thereafter, every NULL pointer dereference causes a
page fault when you run under CWSDPMI.