From owner-freebsd-bugs  Sun Apr 13 18:20:05 1997
Return-Path: <owner-bugs>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id SAA07018
          for bugs-outgoing; Sun, 13 Apr 1997 18:20:05 -0700 (PDT)
Received: (from gnats@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id SAA06991;
          Sun, 13 Apr 1997 18:20:02 -0700 (PDT)
Date: Sun, 13 Apr 1997 18:20:02 -0700 (PDT)
Message-Id: <199704140120.SAA06991@freefall.freebsd.org>
To: freebsd-bugs
Cc: 
From: "Robert S. Miles" <rsm@spyder.ssw.com>
Subject: Re: kern/3260: FreeBSD 2.2.1 (and 2.2) npx0 boot problem on IBM Aptiva M71 
Reply-To: "Robert S. Miles" <rsm@spyder.ssw.com>
Sender: owner-bugs@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

The following reply was made to PR kern/3260; it has been noted by GNATS.

From: "Robert S. Miles" <rsm@spyder.ssw.com>
To: "Jordan K. Hubbard" <jkh@time.cdrom.com>
Cc: rsm@spyder.ssw.com, freebsd-gnats-submit@freebsd.org
Subject: Re: kern/3260: FreeBSD 2.2.1 (and 2.2) npx0 boot problem on IBM Aptiva M71 
Date: Sun, 13 Apr 1997 21:17:24 -0400

 > > This problem does not occur with the 2.1.7 boot floppy, i.e. that does
 > > properly run "sysinstall" after autoconfig.
 > > 
 > > Using NetBSD I was able to build custom FreeBSD kernels with which to
 > > diagnose the problem.  Only when I excluded the npx0 device would the
 > > kernel stop hanging the machine.  Subsequently, I discovered that the
 > > boot floppy's kernel can be made to work, either by disabling the npx0
 > > device in CLI mode, or setting npx0 flags to 7 in CLI mode.
 > 
 > This sounds like the same broken behavior we've seen on certain
 > laptops.  Apparently, the new optimized-for-Pentium bcopy code works
 > great on 99.9% of the machines out there, and then there are a few on
 > which it just does not.  Could you try setting the npx0 flags to
 > 1 and see if it still boots?  That would at least point the finger
 > directly at the bcopy() code (7 turns off a number of optimized
 > features, not just bcopy()).
 > 
 > If you are interested in debugging this, Bruce Evans is your man
 > <bde@freebsd.org> since he added the optimized code in the first
 > place.  Thanks!
 
 Setting npx0 flags to 1 does the trick too, so it is the bcopy() code
 (or my hardware!) that's at fault.  I will follow up with Bruce Evans
 to see if he is interested in diagnosing this further.  Thanks for the
 quick response!
 
 				-Bob