Return to BSD News archive
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msuinfo!gmi!usenet.eel.ufl.edu!usenet.cis.ufl.edu!eng.ufl.edu!spool.mu.edu!news.clark.edu!netnews.nwnet.net!news.u.washington.edu!tzs
From: tzs@u.washington.edu (Tim Smith)
Newsgroups: comp.os.386bsd.development
Subject: Re: Where to buy NCR 810 PCI SCSI cards?
Date: 20 Sep 1994 05:03:58 GMT
Organization: University of Washington School of Law, Class of '95
Lines: 41
Message-ID: <35lqfu$1i1@news.u.washington.edu>
References: <199408172054.PAA07862@MindBender.HeadCandy.com> <35ahfqINN15u8@rs1.rrz.uni-koeln.de> <35ge0p$d9l@news.u.washington.edu> <35i7luINNg88@rs1.rrz.uni-koeln.de>
NNTP-Posting-Host: stein1.u.washington.edu
Stefan Esser <se@fileserv1.MI.Uni-Koeln.DE> wrote:
>|> Speaking of NCR host adaptors, someone should see if NCR is willing to
>|> release the interface to the SDMS BIOS. If they would, it would greatly
>|> simplify supporting all NCR host adaptors under {Free,Net}BSD.
>
>Does SMDS support host adapters with
>other NCR chips than the 53c810/53c825 ?
Yes. SDMS supports the 700 family and the 96. I don't recall if they
ever released SDMS for the 400.
>|> Anyway, if NCR would release the interface documents for this code, then
>|> one could write a single driver that would work with all present and
>|> future systems that contain the SDMS BIOS.
>
>Yes. But I wouldn't expect such a driver to
>offer the same performance as the current
>driver!
>
>Does SMDS support Tagged Command Queues ?
Yes, it supports tagged queues. The SDMS BIOS does not maintain the queues,
though. That is the job of the driver. Similarly, the SDSM BIOS does not
itself handle things like negotiating for synchronous mode--that is up to
the driver. The BIOS handles all the low level details (e.g., it will handle
sending and receiving the messages, but it is up to the driver to interpret
those messages. Of course, the BIOS does handle messages like DISCONNECT
and COMMAND COMPLETE).
Basically, you give it a pointer to a structure that describes the command
you wish to execute, and it executes the command. If the command is a
"normal" command (e.g., it will select the target, send some message bytes,
do a command phase, maybe do a data phase, and then do a status and send a
completion message (possibly disconnecting in there), then the SDMS BIOS
handles the command completely. If anything unusual happens (e.g., someone
tries to negotitate synchronos), the BIOS asks the driver for help. The
driver does *NOT* have to know anything about the SCSI chip. The BIOS
handles that. All the driver has to deal with when it is asked for help
is high level SCSI stuff.
--Tim Smith