From: sandmann AT clio DOT rice DOT edu (Charles Sandmann)
Message-Id: <10212202123.AA19192@clio.rice.edu>
Subject: Re: Patches for module cleanup
To: djgpp-workers AT delorie DOT com
Date: Fri, 20 Dec 2002 15:23:41 -0600 (CST)
In-Reply-To: <3E036F4F.6E9C8D1D@phekda.freeserve.co.uk> from "Richard Dawe" at Dec 20, 2002 07:28:15 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

> > Proposed patches for modularity.  Smaller, more modular fix for ds_alias
> > included (just copy the 2 bytes to locked area for locked only usage).
> 
> Could you elaborate on what the last sentence means please? Other than
> clean-up, does this change allow us to support something that we didn't
> before?

We have a locked area of memory in the exception handler area which is
touched by the interrupt handers.  This is where we used to store 
ds_alias.  If I moved it to crt0 I would have needed to lock those 2 
bytes (an additional lock call code).  Instead __djgpp_ds_alias stays
public and is in non-locked memory, and a copy __excep_ds_alias is 
private to the exception handler stuff in it's locked block.

The wrappers also copy __djgpp_ds_alias to their locked block already.

So no new functionality, just making the locked copy private.