*BSD News Article 34496


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msuinfo!agate!howland.reston.ans.net!math.ohio-state.edu!jussieu.fr!univ-lyon1.fr!swidir.switch.ch!scsing.switch.ch!news.dfn.de!rrz.uni-koeln.de!RRZ.Uni-Koeln.DE!RRZ.Uni-Koeln.DE!news
From: se@fileserv1.MI.Uni-Koeln.DE (Stefan Esser)
Newsgroups: comp.os.386bsd.questions
Subject: Re: NCR controllers
Date: 17 Aug 1994 22:05:04 GMT
Organization: Institute of Nuclear Physics, University of Cologne, Germany
Lines: 50
Distribution: world
Message-ID: <32u1igINN12k2@rs1.rrz.Uni-Koeln.DE>
References: <32smrq$si@agate.berkeley.edu>
NNTP-Posting-Host: fileserv1.mi.uni-koeln.de
Keywords: PCI,SCSI,FreeBSD,NetBSD,NCR
Bcc: wolf@dentaro.GUN.de

In article <32smrq$si@agate.berkeley.edu>, dim@soda.CSUA.Berkeley.EDU (D. Gerasimatos) writes:
|> 	I understand that the NCR 53c810 driver is now in the beta testing
|> 	stage (and hence should be in NetBSD 1.0?), but are there/will
|> 	there be any drivers for the FAST/WIDE NCR S1365 825 PCI controller?

Since I've been involved in the development of that driver, I'll 
comment on that question:

1) The driver has been available for FreeBSD > 1.1 for some
   time now. A few bugs and problems were fixed, most related
   to some device not being up to SCSI-2. There seem to be
   problems with some 486 motherboards with write back cache
   (ie. the ASUS SP3G), but it works if write through is 
   selected. This may be a chip set problem, since the driver
   seems to work with other systems with write back cache.

2) The driver can be found in the NetBSD-current source tree,
   so it might really make it into NetBSD-1.0 :).
   The Copyright note has been adapted to allow distribution 
   as part of the free *BSD.

3) FAST WIDE support is easy to add, but the correct operation 
   has to be verified. WIDE negotiation is similar in structure 
   to the negotiation of synchronous transfers. But there are 
   registers in the 825 where the last byte of a transfer may 
   be left in certain situations, and it must be written to host
   memory by the host CPU, then. This part of the driver has to 
   be checked under normal operation conditions and with forced
   errors, to verify correct operation. And that requires a 825
   controller board and a WIDE drive (at least for the duration
   of the tests, ie. 2 or three weeks).

4) If somebody makes a controller and drive available to us for 
   the purpose of testing WIDE transfers, then we'll start working 
   immediately. We can't afford to buy a WIDE drive now, and we 
   really don't need more disk capacity.

5) If somebody else wants to do the tests, we can make a driver 
   version with WIDE transfer negotiation available. But we don't
   expect it to be complete on first try (there have been quite a 
   few surprises working with the NCR chip :), so that person 
   ought to be able to understand the driver code (not written in 
   C, but in NCR SCRIPTS commands for the biggest part).

-- 
 Stefan Esser				Internet:	<se@MI.Uni-Koeln.DE>
 Mathematisches Institut		Tel:		+49 221 4706010
 Universitaet zu Koeln			FAX:		+49 221 4705160
 Weyertal 80
 50931 Koeln