Date: Sun, 1 Mar 1998 09:06:26 +0100 (MET)
From: Beastly Wolf <beast@lintilla2.df.lth.se>
To: J Lothian <jlothian@holyrood.ed.ac.uk>
cc: pups@minnie.cs.adfa.oz.au
Subject: Re: RL02 meets BSD
In-Reply-To: <199802281618.QAA20434@holyrood.ed.ac.uk>
Message-ID: <Pine.SUN.3.91.980301090322.20308B-100000@lintilla2.df.lth.se>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-pups@minnie.cs.adfa.oz.au
Precedence: bulk

Both systems have "pirate" drive controllers and they have cards
that do pass grant signals.
If I do not remember wrongly, I think that only RQDX-1 had the
"feature" of not passing the grant chain.
But we placed all RQDX controller at the bottom anyhow even
though they worked further up.
THis is of academical interest only since I do not have holes
in the grant chain and do not have an RQDX controllers AND
I have devices below the drive controller in the first case
that do work!
/Lars

On Sat, 28 Feb 1998, J Lothian wrote:

> I've had an RL02 working on an 11/750 running
> BSD4.3. The RL11 controller only has 16 words of
> buffer memory, so you've got to make sure that other
> devices don't hog the bus too much. In particular,
> UDAs should have their DMA burst set to something like 1.
> Setting it much higher causes the RL11 to get data lates
> as its silo overflows before it gets access to the bus
> to do DMA. However, the RLV12 seems to have a much bigger
> silo (256 bytes?), so this should't be a problem for it. 
> 
> The only other things I can think of are bus grant problems
> &c. If you're using an RQDX3 in the same machine, bear in mind
> that it doesn't pass the grants, and so should be the last 
> device on the bus. 
> 
> James
> 
>