From: sandmann AT clio DOT rice DOT edu (Charles Sandmann)
Message-Id: <9710071553.AA12865@clio.rice.edu>
Subject: CWSDPMI r4 beta
To: djgpp-workers AT delorie DOT com
Date: Tue, 7 Oct 1997 10:53:17 -0600 (CDT)
In-Reply-To:  
Content-Type: text
Precedence: bulk

In a fit which can only be described as temporary (?) insanity, I have 
put some bug fixes into CWSDPMI and fallen victim to the enhancement
syndrome.  I am looking for some victims^H^H^H^H^H^H beta testers to try 
out the new release before infecting the rest of the known universe.

Things I tried to do:
 Map HMA for non-VCPI clients
 Fix 0x800 mapping for 0x1... addresses
 Check 0x800 mapping for invalid addresses
 Use XMS extensions to support >64K XMS memory (NEED A TESTER!)
 More code size reduction tricks
 New license agreement (GPL + binaries, just leave me alone...)
 Dynamic allocation for swapfile usage bitmap; a new cwsparam field for 
    max swap usage.  So the user can decide how much virtual memory to
    allow in exchange for 32 bytes of real memory per Mb swap.  Thus
    the base footprint drops by 4Kb and runtime footprint increases 
    between 0 and 8Kb, depending on the amount of swap you want and how
    much disk you have free.  Now limited to 256Mb virtual.
 Unknown things I've forgotten

Things I still might do:
 Dynamic allocation for physical memory (to increase limit to 256Mb
    physical... this seemed silly a few years ago...)
 Hack the routines to put them in a form suitable for linking the pmode
    stub (standalone image)
 Make dynamic allocation limit 4Gb
 Something someone nags me enough about

Send me an email if you want to test the new version, have a strong 
opinion about some feature, or just want to irritate the hell out of me :-)