From owner-freebsd-hackers  Mon Feb 19 06:15:05 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id GAA25792
          for hackers-outgoing; Mon, 19 Feb 1996 06:15:05 -0800 (PST)
Received: from specgw.spec.co.jp (specgw.spec.co.jp [202.32.13.1])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id GAA25777
          for <hackers@FreeBSD.org>; Mon, 19 Feb 1996 06:14:59 -0800 (PST)
Received: from localhost (uucp@localhost) by specgw.spec.co.jp (8.6.5/3.3Wb-SPEC) with UUCP id WAA16606; Mon, 19 Feb 1996 22:49:17 +0900
Received: by tama.spec.co.jp (8.7.3/6.4J.5)
	id WAA01167; Mon, 19 Feb 1996 22:56:25 +0900 (JST)
From: Atsushi Murai <amurai@spec.co.jp>
Message-Id: <199602191356.WAA01167@tama.spec.co.jp>
Subject: Re: BT946C problems. Can you help?
To: gibbs@freefall.freebsd.org (Justin T. Gibbs)
Date: Mon, 19 Feb 1996 22:56:24 +0900 (JST)
Cc: adamp@ovid.com, jkh@cdrom.com, hackers@FreeBSD.org, nirva@zynet.com
In-Reply-To: <199602182048.MAA02944@freefall.freebsd.org> from "Justin T. Gibbs" at Feb 18, 96 12:48:41 pm
Reply-To: amurai@spec.co.jp
X-Mailer: ELM [version 2.4 PL23]
Content-Type: text
Sender: owner-hackers@FreeBSD.org
Precedence: bulk

> And you ensured that the kernel config for the bt driver matched these
> settings every time?  In 2.1R, the bt driver used an ISA probe to
> find the BT cards, so it cannot change its irq or dma channel on the fly
> like new pci probe can even if the driver does properly report what the
> card is set too.

I don't think so. His Bt is 747C for EISA bus, right? Even *ISA
probe*, Just only I/O port is correct, bt driver can get a correct IRQ
information from cards and using it. (See device bt lines in GENERIC
conf. PCI bus may remap a these resources without notic to card as you
say.)

> I'm sure that it is mismatched configuration settings.  Boot with -c and
> set them up correctly to match your hardware and I bet the card will work.
> Its most likely the irq since the driver polls during the initial probe
> and doesn't require the irq to be set correctly to work during that
> phase of the boot process.

Same reason as above.

As far as I can tell you as follows,

 1. If you enabel write back cache, try to turn off(write through) in BIOS(?) )
 2. Do Auto SCSI(tm) "DMA Test" both write back/thourgh cache.
 3. Check SCSI bus termination, Each end of cable should has a terminater.
 4. I have experience VLB Video Card do memory corruption with bt747s on
    AMI Enterprise IV. -> Change a Video Card.

Atsushi.
-- 
Atsushi Murai                                       Internet: amurai@spec.co.jp
System Planning and Engineering Co,.Ltd.            Voice   : +81-33833-5341