Re: kern/42154 (boot installcd fails)

[List Mail User <track%Plectere.com@localhost>]

The following reply was made to PR kern/42154; it has been noted by GNATS.

From: List Mail User <track%Plectere.com@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: kern/42154 (boot installcd fails)
Date: Mon, 19 Oct 2009 00:31:12 -0700 (PDT)

 >...
 >From: David Holland <dholland-bugs%netbsd.org@localhost>
 >To: gnats-bugs%NetBSD.org@localhost
 >Cc: 
 >Subject: Re: kern/42154 (boot installcd fails)
 >Date: Mon, 19 Oct 2009 03:58:36 +0000
 >
 > On Mon, Oct 19, 2009 at 02:45:02AM +0000, List Mail User wrote:
 >  >   I can reproduce this on _some_ machines using an external USB DVD.
 > 
 > ah, USB... :(
 > 
 > Which kind of USB controller is it? Although since in the bootloader
 > it's going through the BIOS and not through our USB stack it
 > theoretically shouldn't matter... or so I'd think...
 > 
 > -- 
 > David A. Holland
 > dholland%netbsd.org@localhost
 > 
 >
        All EHCI, but any underlying OHCI case works - all my failures
 are on Intel ICH7-9s (don't have an ICH10 and everything before ICH7
 seems to always work);  Those which still work include various ATI, Via
 and Sis chipsets, but also some of the same Intel ones (e.g. more recent
 ICH7s from 945GC based boards fail after 5.99.13 but boot upto that
 point and one "HP" ICH9 notebook can boot, but another "Compaq" model,
 which is slightly newer, cannot) - So it isn't simply the chipset (i.e.
 different 945xx boards, from different manufacturers act differently
 and different ICH9 boards from the same company also act differently)
 nor is it just USB, since the laptop that fails reports its DVD is IDE
 (unfortunately, that one is used by my spouse).
 
        Besides, at this point (i.e. during bootxx_cd9660) the USB stack
 isn't up yet and the machines are running BIOS emulation code to boot
 from an external device (which rules out testing on the old ICH[2-5] boxes
 I have - most won't boot over USB and those that can are "production" hosts
 I can't easily bring down, though I might be able to add and IDE CD or DVD
 to some of the old gear which has been shelved and is lying around).
 
        I have quite a few small (and recent) Atom hosts affected by this
 (ICH7 in a 945GC chipset w/ Atom 330 CPUs) and can easily test on one host
 which is a "hot spare" (now running 5.99.20) or on a "cold spare" (which I
 think has 5.0 already on it, but is just sitting on a shelf).  Most other
 machines aren't generally available for testing (though a recent spate of
 upgrades showed this problem on most of my "new" machines, using the same
 USB DVD and disk(s) to upgrade them all - i386 or amd64).