*BSD News Article 14585


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!spool.mu.edu!uunet!pipex!sunic!isgate!veda.is!adam
From: adam@veda.is (Adam David)
Newsgroups: comp.os.386bsd.questions
Subject: Re: Cache testing software.
Message-ID: <C5L051.D5t@veda.is>
Date: 16 Apr 93 14:55:47 GMT
References: <1993Apr15.154147.10556@cm.cf.ac.uk> <1993Apr16.065919.103320@eratu.rz.uni-konstanz.de>
Organization: Veda Systems, Iceland
Lines: 25

zh@news.uni-konstanz.de (Z. Horvat) writes:

>The told me that they gave my "old" board to someone who is now
>successfully running OS2 on it.
>This sounds strange to me, as i would have expected that one would
>not even be able to run DOS software properly on such a board.

The DOS and OS/2 probably *expect* the motherboard to be broken in
this way (cache and DMA hardware subsystems unaware of each other),
and therefore compensate for it in software. 386bsd 0.1 on the other
hand expects the motherboard to have integrated cache and DMA, and is
less forgiving in this respect.

If someone points me to the relevant code, I'll gladly put some time
into fixing this (I am assuming that locating the danger zone in the
source would take me far longer than making it safe). I do not expect
to be making any 100% fixes, because this will (hopefully) be completely
overhauled in 0.2 anyway. I do however have a pressing incentive to get
this gaping hole covered (there seems to be no easy way to tell whether
the motherboard is broken without installing 386bsd and trying to use
it for serious work). Also I would welcome some advice about how to
program the 386/486 to mark certain areas as uncacheable.

--
Adam D. (adam@veda.is)