Return to BSD News archive
Path: sserve!newshost.anu.edu.au!munnari.oz.au!bruce.cs.monash.edu.au!harbinger.cc.monash.edu.au!yeshua.marcam.com!MathWorks.Com!europa.eng.gtefsd.com!howland.reston.ans.net!agate!ames!elroy.jpl.nasa.gov!ncar!noao!CS.Arizona.EDU!organpipe.uug.arizona.edu!afthree.as.arizona.edu!tom
From: tom@afthree.as.arizona.edu (Thomas J. Trebisky)
Newsgroups: comp.os.386bsd.questions
Subject: Re: DRQ for wd.c
Date: 22 Feb 1994 00:48:37 GMT
Organization: Steward Observatory, Univ. of Arizona, Tucson
Lines: 26
Message-ID: <2kbkp5$6dj@organpipe.uug.arizona.edu>
References: <2kb77b$so6@explorer.clark.net>
NNTP-Posting-Host: afthree.as.arizona.edu
Keywords: DRQ DMA IDE wd.c
In article <2kb77b$so6@explorer.clark.net> ack@clark.net (Eric S. Hvozda) writes:
>I was under the impression that IDE disks are capable of DMA.
so was I.
>Is this true?
No.
Amazingly, at one time in the history of PC's this decision was made and
it lingers to this day. All the more amazing is that the Floppy disk
does get a DMA line.
Actually, this is not as incredibly dumb as it sounds (well, maybe it is),
since a REP STOSB sequence can move data as fast as a DMA chip, in old
clunky PC's. This refers to the 8237 chip, which could never be clocked
terribly fast. With new LSI chipsets, dma can get respectable, but the
new faster PC's can really move along too. The rub is for a multitasking
OS like unix where it would be nice to schedule a transfer, let DMA
handle it, and get an interrupt when it was all done. Look at the wd.c
driver and see what actually gets done.
--
Tom Trebisky ttrebisky@as.arizona.edu
I learned very early the difference between knowing the name of
something and knowing something. --R. P. Feynman