bug#102: [kernel] bug#102: second SCSI on C3000 doesn't work


X-PA-RISC Linux-PR-Message: report 102
X-PA-RISC Linux-PR-Package: kernel
X-Loop: daniel_frazier@hp.com
Received: via spool by 102-bugs@bugs.parisc-linux.org id=B102.98714708610735
          (code B ref 102); Fri, 13 Apr 2001 07:33:01 GMT
Message-Id: <200104130724.BAA31057@puffin.external.hp.com>
To: Paul Bame <bame@fc.hp.com>, 102@bugs.parisc-linux.org
In-Reply-To: Your message of "Thu, 12 Apr 2001 16:57:06 MDT."
Date: Fri, 13 Apr 2001 01:24:56 -0600
From: Grant Grundler <grundler@puffin.external.hp.com>

Paul Bame wrote:
> The second (actually the first) SCSI controller on C3000 doesn't work.
> This one's hooked to the narrow SCSI connector on the back, which had
> a disk and proper termination at the time of the test, which the boot
> rom saw just fine.  This boot log is with terminator but no disk, and
> I think it's the same as when the disk is attached:

I'm surprised we hadn't filed a bug on this before. Thanks for doing it!

> sym53c896-0: suspicious SCSI data while resetting the BUS.
> sym53c896-0: dp1,d15-8,dp0,d7-0,rst,req,ack,bsy,sel,atn,msg,c/d,i/o
>	 = 0x4000100, expecting 0x100
>   etc.!
> sym53c896-0: giving up ...

I suspect the problem is that the SCSI interface chip is capable of
running "wide" (16-bit data path) and the sym53c8xx assumes that's how
it's wired. The upper bits *probably* are not grounded since the interface
out the back is narrow.
Can anyone confirm this? (ie has 896 or c3k motherboard knowledge?)

[ And I just saw Richard's mail saying parity is wrong for upper bits ]

HPUX queries PDC to determine the SCSI ID (+ other attributes)
and programs the 896 chip accordingly *before* the SCSI bus probe.
I assume bus width is one of the attributes but haven't checked.


Grant Grundler
parisc-linux {PCI|IOMMU|SMP} hacker