[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#228654: installation-reports: ide-detect lockup on Alpha PWS500a



On Tue, Jan 20, 2004 at 04:33:44AM -0600, John Lightsey wrote:

> > Thanks for the report.  We have 2.4.24 kernel images in the archive for
> > d-i, but I hadn't yet switched to using them for the daily builds.  This
> > is done now, and tomorrow's ISOs should be 2.4.24-based.  Would you be
> > willing to give them a try?

> I'd be happy to.  Is there any way of skipping the IDE detection or 
> telling it not to try different chipsets?

<cough>  The best method I've found so far is to switch to console #2 as
soon as the language question comes up, and remove the troublesome
modules from /lib/modules/<version>/kernel/.  You can also boot the
installer with DEBCONF_PRIORITY=low, in which case it will still try to
probe all the modules in order, but at least it will let you specify
options to insmod that will break the attempt to load it (pain...).

Our real goal, of course, is to get the autodetection right so it
doesn't *try* IDE chipset modules that will cause problems.

> > Have you filed a bug on the kernel-image packages about the tulip/de4x5
> > issue, by chance?

> No, I'm 99% certain that the problem isn't really in the ethernet drivers.  
> It's the machine's buggy PCI/IDE chipset causing the drivers to fail.  If I 
> put another network card in one of the PCI slots it will also fail.  From my 
> understanding, the problems with these particular machines have been known 
> for a while.

> http://lists.debian.org/debian-alpha/2003/debian-alpha-200308/msg00082.html

> The odd thing is that the 2.4.2x Debian kernels do boot and appear to 
> function properly (outside of bringing up a network interface.)  I 
> would expect it to work in the installer if you can still skip the 
> network configuration.

Hrm, very strange.  And on your system, which module does d-i try for
your ethernet card -- the one that hangs, or the one that just doesn't
work?

Anyway, the images should be posted by this point -- feel free to give
them a try.

Thanks again,
-- 
Steve Langasek
postmodern programmer

Attachment: pgpkzAtLezOqK.pgp
Description: PGP signature


Reply to: