From:	sandmann AT clio DOT rice DOT edu (Charles Sandmann)
Message-Id:	<10212220633.AA01541@clio.rice.edu>
Subject:	Re: Patches for module cleanup
To:	djgpp-workers AT delorie DOT com
Date:	Sun, 22 Dec 2002 00:33:54 -0600 (CST)
In-Reply-To:	<Pine.SUN.3.91.1021222075734.28392E-100000@is> from "Eli Zaretskii" at Dec 22, 2002 08:00:33 AM
X-Mailer:	ELM [version 2.5 PL2]
Mime-Version:	1.0
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

> We should probably look into existing uses of the alias descriptor, to 
> make sure we don't break any working code.  There are libraries out there 
> that use it, and even debuggers, including GDB, might look at it (I'm 
> talking from memory, so I might be gravely mistaken).

Did that.  They can still use it (even though it hasn't been documented,
which is now on my list).  Only issue is if it is a hardware interrupt
handler (self written, since the wrappers handle this), they reference 
it directly, it's under CWSDPMI and it's been paged out.  

> If there's some way of solving the modularity problem that won't even 
> potentially break existing code, I think we should prefer it.

We could lock the page containing the two bytes, but I'd like to see
some code potentially broken first.

- Raw text -