From: sandmann AT clio DOT rice DOT edu (Charles Sandmann)
Message-Id: <10107281620.AA14010@clio.rice.edu>
Subject: Re: Make 3.791 on Windows 2000 test
To: eliz AT is DOT elta DOT co DOT il
Date: Sat, 28 Jul 2001 11:20:31 -0500 (CDT)
Cc: acottrel AT ihug DOT com DOT au, djgpp-workers AT delorie DOT com
In-Reply-To: <200107281612.MAA20431@delorie.com> from "Eli Zaretskii" at Jul 28, 2001 12:12:02 PM
X-Mailer: ELM [version 2.5 PL2]
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
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

> Is it true that Unixy sbrk is a much better default on NT and W2K?  

Probably so, but there isn't a lot of testing to confirm this.  If we
were to handle the wrap around the non-move sbrk() is probably just
as good.

> If so, it might be better to change the default automatically if we can
> do that without bloating the startup code.  To recognize NT, all we
> have to do is call a single DOS function (3306h) and compare the
> result with 0532h: if they compare equal, we are on NT/W2K.  This
> doesn't sound like a lot of code, does it?
> 
> (We should still let users override this by setting bits in
> __crt0_startup_flags, in case they have reasons to do so.)

I *HATE* futzing with the crt0 sbrk code.  Maybe I'll re-write it in C.
Then maybe it would get the attention it deserves.

I'd like to do more checking on the passing of signed/unsigned sbrk()
values.  I'd like to increase block sizes from 64K to larger dynamically.
I think we should check the selector limit sets to see if they stick,
and if not don't use the wrap around blocks.