From:	drupp AT cs DOT washington DOT edu (Douglas Rupp)
Message-Id:	<199603271541.HAA26418@june.cs.washington.edu>
Subject:	Re: Itimer/Signal handler bug?
To:	eliz AT is DOT elta DOT co DOT il (Eli Zaretskii)
Date:	Wed, 27 Mar 1996 07:41:18 -0800 (PST)
Cc:	drupp AT cs DOT washington DOT edu, sandmann AT clio DOT rice DOT edu, djgpp-v2 AT delorie DOT com
In-Reply-To:	<Pine.SUN.3.91.960313174212.10443D-100000@is> from "Eli Zaretskii" at Mar 13, 96 05:48:26 pm
MIME-Version:	1.0

> 
> If you are talking about the Swappable Data Area (SDA), then beware: it 
> would probably work under native DOS, but I'd expect it to crash and burn 
> under Win 3.11 and Win95 when 32-bit file access is used (to say nothing 
> about WinNT and Linux).  All these work by catching DOS calls and 
> redirecting them into totally different code and data structures.  Now 
> guess: what are the chances that MS designed those replacements for DOS 
> to be reentrant?
> 
> The code to get the address of the SDA can be found on stat.c in the DJGPP
> library sources.  There is no DOS call to swap it in or out AFAIK;  you
> will have to do it yourself.  There are more details about this technique
> in the chapter on TSRs in ``Undocumented DOS'', 2nd ed. 
> 

I gave up on time-slicing, since it's not required by the Ada language.  W/o
it the threads library works great and Ada tasking is now fully implemented
in the DOS version of GNAT, and will be in the next release.

The next thing to do is get stack checking working, for which I need a
sigaltstack() function.  Any thoughts on this?

- Raw text -