This is a test release of the 4Mb page support in CWSDPMI.  This should be
considered a WIP for CWSDPMI r6.  The primary reason for this set of changes
is to decrease the DOS memory requirements for large memory space usage and
to allow high performance access to large amounts of physical memory (up to
2Gb).  This package contains the binary build for CWSDPMI.EXE and the
files with source changes from r5.

The 4Mb page support in this test release is not complete, and the program
has had very minimal testing.  Known issues:
 1) 4Mb regions are treated as locked or non-pageable.  This means you 
    cannot use virtual memory effectively with 4Mb pages.
 2) There is no way in this image to prevent using 4Mb pages.
 3) DPMI function 0x503 to resize memory is not supported if it requires the
    memory zone to be moved in physical memory and it contains a 4Mb page.

I have tested the distributed binary using over 780Mb of memory using 
Alybadix and saw fast performance (compared to the old r5 image, which
slowed substantially after 500Mb of memory usage).  I haven't tried nesting
images or anything fancy.  If you don't have 256Mb of memory or more you
probably don't want to mess with this test image!

Long term issues:
 1) Should 4Mb pages be "paged" to disk like 4K pages?  This might take up to
    several seconds just for the I/O to page-out/page-in a 4Mb page.
 2) If we don't allow paging of 4M pages, should this only be available when 
    swapping is disabled?  Or in the ring 0 version?  
 3) Should some amount of memory be reserved for non-4M pages?  Worse case
    example - system with 8Mb of memory (one 4Mb aligned page available).
 4) DPMI 1.0 extensions which handle "page" attributes won't work on 4Mb
    pages; so trying to protect or decommit a page in the middle of a 4Mb
    page is silently failed at best.

If you have good luck or problems with this test image you can send me an
email at sandmann@clio.rice.edu and tell me what you found.  Good luck.
