NetBSD Problem Report #4184

Received: (qmail 6470 invoked from network); 29 Sep 1997 04:42:41 -0000
Message-Id: <199709290444.VAA24626@nooksack.ldc.cs.wwu.edu>
Date: Sun, 28 Sep 1997 21:44:43 -0700 (PDT)
From: lennox@cs.columbia.edu
To: gnats-bugs@gnats.netbsd.org
Subject: esp driver crashes in boot with parity-violating CD-ROM drive turned on
X-Send-Pr-Version: www-1.0

>Number:         4184
>Category:       port-mac68k
>Synopsis:       esp driver crashes in boot with parity-violating CD-ROM drive turned on
>Confidential:   no
>Severity:       serious
>Priority:       low
>Responsible:    briggs
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Sep 28 21:50:02 +0000 1997
>Closed-Date:    Sun May 30 21:49:46 +0000 1999
>Last-Modified:  Sun May 30 21:50:57 +0000 1999
>Originator:     Jonathan Lennox
>Release:        netBSD-current 1.2G -- Wed. Aug 13
>Organization:
Columbia University
>Environment:
(Machine will not boot -- Macintosh Quadra 950)

>Description:
When I boot MacBSD on my Quadra 950, if my <NEC, CD-ROM DRIVE:210, 1.0>
external CD-ROM drive is powered on, the kernel boot freezes during the
esp driver's probe of scsibus1, after approximately 17 messages each of
"esp1: SCSI bus parity error" and "data: interput, but no count left"
(intermingled).

If the driver won't support the bad hardware (which would be reasonable,
even if somewhat unfortunate) it should at least ignore it.  The drive
works under MacOS.

>How-To-Repeat:
Boot the kernel with the CD-ROM drive powered on.


>Fix:
Booting without the drive or with it powered off is a workaround.

>Release-Note:
>Audit-Trail:
State-Changed-From-To: open->suspended 
State-Changed-By: briggs 
State-Changed-When: Thu Oct 2 06:32:14 PDT 1997 
State-Changed-Why:  
I think that this is not caused by the esp driver not handling parity properly, 
but rather by the esp driver not supporting the second channel in the Q900/950 
systems properly. 
Responsible-Changed-From-To: gnats-admin->briggs 
Responsible-Changed-By: scottr 
Responsible-Changed-When: Mon Nov 17 14:40:47 PST 1997 
Responsible-Changed-Why:  
State-Changed-From-To: suspended->feedback 
State-Changed-By: scottr 
State-Changed-When: Tue Sep 29 22:25:33 PDT 1998 
State-Changed-Why:  
Why did the state change? (Ctrl-D to end) 
A race condition in the SCSI (now SCSI+ATAPI) code was discovered in August 
by a NetBSD/atari user.  There was also a long-standing issue where a 
structure was being referenced after it was `freed' by the driver.  Both 
of these have been fixed, and I'm curious to know if the issue highlighted 
by this PR is still around... 
State-Changed-From-To: feedback->closed 
State-Changed-By: briggs 
State-Changed-When: Sun May 30 14:49:46 PDT 1999 
State-Changed-Why:  
esp driver on Q950 does seem to be operating correctly for me. 
>Unformatted:

Home	PR Database Search