Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 26 Mar 1997 01:28:20 -0800 (PST)
From:      asami@vader.cs.berkeley.edu (Satoshi Asami)
To:        phk@freebsd.org
Cc:        gibbs@freebsd.org, current@freebsd.org, se@freebsd.org
Subject:   Re: RELENG_2_2, ahc and Atlas II, YES!!
Message-ID:  <199703260928.BAA01900@silvia.HIP.Berkeley.EDU>
In-Reply-To: <8145.859321269@critter> (message from Poul-Henning Kamp on Tue, 25 Mar 1997 21:21:09 %2B0100)

next in thread | previous in thread | raw e-mail | index | archive | help
 * I upgraded the box that's been giving me trouble to RELENG_2_2 as of
 * this morning, danish time and it went straight through the operation
 * that would until now 100% probability hang it.
 * 
 * Thanks!

What's your firmware revision?  Wide or narrow?  My "QUANTUM XP34550W
LXQ1" used to corrupt files with 2940UW (and ftp.qntm.com doesn't have 
firmware upgrades for wide AII drives....).

 * Now for the question:
 * 
 * 	Assume a machine with a 3940 in it.  The bios will take bus A
 * 	first, then bus B.  FreeBSD does the same thing.
 * 
 * 	Now add a 2940 to it.  The bios goes 3940:A, 3940:B, 2940,
 * 	FreeBSD goes 2940, 3940:A, 3940:B
 * 
 * why ?
 * 
 * I can understand that the order in the PCI slots matter and all that,
 * but shouldn't we be able to do it in the same order the BIOS does it ?
 * It's a semi-bad nuisance otherwise...

Wow.  I have never seen a BIOS that wants to do it that way.  AFAIK
FreeBSD always probes all devices on bus N before looking into bus N+1
(and since the 3940 has two SCSI chips on a secondary PCI bus, it will
always be probed after the 2940).  Your BIOS seems to be doing it in
depth first order (or postorder, although I surely hope that's not the
case ;).

We (Stefan and I) fixed the bridge handling code in sys/pci/pci.c rev
1.51 so that the kernel uses the PCI bus numbers assigned by the BIOS.
It was necessary to boot on some machines when there were more than
one 3940 in there.

But for what you say -- I don't even know if it's possible to fix.
Does the BIOS leave enough information around to determine if it's
done the disk probes in depth-first order?

Satoshi



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199703260928.BAA01900>