Subject: Re: TK50Z/non-Sock
To: None <port-vax@NetBSD.ORG>
From: John Wilson <wilson@dbit.dbit.com>
List: port-vax
Date: 03/02/1998 11:13:20
>From: allisonp@world.std.com (Allison J Parent)

><	- read/write to the DMA buffer
><	- NCR doing DMA to/from the buffer
>
>NCR using what DMA, the MV-II dma?

(Uninformed conjecture:)

Since the 5380 has no built-in DMA controller of its own, that may explain
things -- a quick-and-dirty external DMA controller could be built for it
using just counters if the designers totally blew off any kind of arbitration
logic vs. the other on-board devices, in that case it would be important to
leave the 5380 alone while it's doing DMA.  I don't get that item about 9224
and 5380 register accesses being mutually exclusive, since of course they
couldn't be otherwise.  But if the 5380's DRQ and DACK lines are tied together
(or something similar) then it definitely needs to be left alone, that might
also explain why they used a private DMA buffer instead of (some portion of)
main memory, that avoids arbitration to main memory too.

John Wilson
D Bit