X-Authentication-Warning: delorie.com: mail set sender to djgpp-bounces using -f
From: "Charles Sandmann" <sandmann AT clio DOT rice DOT edu>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Large memory usage CWSDPMI test available
Date: Mon,  3 Dec 2001 13:40:41
Organization: Aspen Technology, Inc.
Lines: 18
Message-ID: <3c0b80d9.sandmann@clio.rice.edu>
References: <3c0aaaa6 DOT sandmann AT clio DOT rice DOT edu> <9ug2gf$pq8$1 AT antares DOT lu DOT erisoft DOT se>
NNTP-Posting-Host: dcloan.hou.aspentech.com
X-Trace: selma.aspentech.com 1007408916 28962 10.32.115.107 (3 Dec 2001 19:48:36 GMT)
X-Complaints-To: postmaster AT aspentech DOT com
NNTP-Posting-Date: 3 Dec 2001 19:48:36 GMT
X-NewsEditor: ED-1.5.8
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

> : This also gives an early glimpse of the the way the 4Mb paging stuff might
> : work (interesting for those who do DMA or mess with page tables).  In 
> : particular, I haven't decided how this should work (if at all) with 
> : virtual memory.  Does anyone want to bother with virtual memory if they
> : have more than 500Mb of physical memory and are running under DOS?
> 
> So when this is enabled there's no swap?

Currently there is swap, but only for 4Kb pages.  Since the 4Mb pages are 
locked (and can quickly eat up all the memory) the paging behavior might not
be what you want.  It's the little details like this that make the new 
features a pain to implement.

I could page 4Mb pages, but even one fault would be a very long event on most 
hardware.  I could limit the total number of 4Mb pages... or ...

I suspect I'll end up with a new parameter or heuristic.  But making it 
available early may generate some good suggestions :-)