Date: Mon, 30 Mar 1998 14:10:24 +0300 (IDT)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Nate Eldredge <eldredge AT ap DOT net>
cc: Edward Cashin <edc AT arches DOT uga DOT edu>, djgpp AT delorie DOT com,
        Charles Sandmann <sandmann AT clio DOT rice DOT edu>
Subject: Re: djgpp malloc and freeing memory on exit
In-Reply-To: <199803292223.OAA18960@adit.ap.net>
Message-ID: <Pine.SUN.3.91.980330140958.19525I-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk


On Sun, 29 Mar 1998, Nate Eldredge wrote:

> It seemed likely to me that they needed to be explicitly freed, judging from
> the trouble DJGPP's exit code goes to to free them (the 16-bit exit code and
> so forth). Is that just paranoia, then?

No, not paranoia.  AFAIK, some DPMI hosts (Windows?) don't free the
selectors we are allocating for special needs, like the 16-bit sbrk
helper.  I think the code you see was added to work around those DPMI
hosts and prevent selector leeks.

But the selectors allocated by the 32-bit code, including those
allocated by calling the DPMI functions, are automatically cleared,
IIRC.  Charles, is this correct?