Path: utzoo!attcan!uunet!zaphod.mps.ohio-state.edu!swrinde!ucsd!network.ucsd.edu!euclid.ucsd.edu!athing
From: athing@euclid.ucsd.edu (Bill Athing)
Newsgroups: comp.windows.ms
Subject: Memory Allocation/GlobalAlloc
Message-ID: <3935@network.ucsd.edu>
Date: 25 Oct 90 07:32:12 GMT
Sender: news@network.ucsd.edu
Reply-To: athing@euclid.ucsd.edu (Bill Athing)
Organization: none
Lines: 17
Nntp-Posting-Host: euclid.ucsd.edu


I am trying to allocate several buffers under MS windows that must be
aligned on 64k page boundries for purposes of DMA transfer. I have a
kludge solution right now, where if I want a 64k buffer, I request a 
128k buffer and use only half of it. This is, needless to say, wasteful,
and obnoxious in terms of proper behaviour in a cooperative multitasking
environment. 

Is there a better way to achieve the same results? Can I be guaranteed
that if I take a large chunk from global alloc, find the start address,
free up that block, allocate enough so that the next GlobalAlloc call
will give me the alignment I need, then free up the block I used to
force the third call to be DMA page aligned? There's got to be a better
way.

Thanks for your help,
Bill Athing