From: Martin DOT Stromberg AT lu DOT erisoft DOT se (Martin Stromberg)
Newsgroups: comp.os.msdos.djgpp
Subject: Re: DMA under djgpp
Date: 13 Feb 1998 15:00:13 GMT
Organization: Ericsson Erisoft AB, Sweden
Lines: 19
Message-ID: <6c1n5t$4di$3@antares.lu.erisoft.se>
References: <199802130639 DOT WAA14453 AT adit DOT ap DOT net> <6c0u42$b92$1 AT pulp DOT ucs DOT ualberta DOT ca>
NNTP-Posting-Host: juno.lu.erisoft.se
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

B Hodge (bhodge AT gpu DOT srv DOT ualberta DOT ca) wrote:
: Nate Eldredge <eldredge AT ap DOT net> wrote:
: > Normally you allocate the DMA buffer in conventional memory. That way you
: > know its physical location, and that it's contiguous. Allocate 128K, then
: > find a 64K boundary within that.

: I always hear and see about that approach and get an ill feeling in my
: gut.  Wasting 64K like that seemse sorta pointless.  I see how it would
: probably require it if the buffer needed to be 64K, as it is next to
: impossible to get a perfectly aligned page 64K.

[ Klippa, klapp, kluppit excellent method to allocate DMA buffer. ]

It's impossible to selectively free parts if the 128K, I suppose?


Scratch-scratch-scratch,

							MartinS