Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!julius.cs.uiuc.edu!apple!agate!usenet
From: raymond@math.berkeley.edu (Raymond Chen)
Newsgroups: comp.sys.ibm.pc.misc
Subject: Re: extended memory
Keywords: 286 extended memory
Message-ID: <1990Dec6.222421.15840@agate.berkeley.edu>
Date: 6 Dec 90 22:24:21 GMT
References: <18373@rasp.eng.cam.ac.uk> <4467@altos86.Altos.COM>
Sender: usenet@agate.berkeley.edu (USENET Administrator)
Reply-To: raymond@math.berkeley.edu (Raymond Chen)
Organization: U.C. Berkeley
Lines: 14
In-Reply-To: rcollins@altos86.Altos.COM (Robert Collins)

In article <4467@altos86.Altos.COM>, rcollins@altos86 (Robert Collins) writes:
> [on using the memory from FFFF:0010 to FFFF:FFFF, known as the HMA]
>Leave A20 always enabled.  

This is dangerous.  Some software (typically, copy protection schemes)
rely on A20 wraparound to work properly.  The recommended solution is
to leave a stub in low memory.  The stub enables the A20, calls the
TSR, then disables the A20 when the TSR returns.

I strongly suggest using the XMS function calls (in HIMEM.SYS) to
negotiate access to the HMA.  That way, other programs will know that
you are using the HMA and won't stomp all over you.

Disclaimer:  I speak with no authority.  Proceed at your own risk.