Re: d-i daily build fails when building initrd (PowerMac7,3)
On Sun, Apr 02, 2006 at 04:26:27PM +0200, Michael Schmitz wrote:
> > > > That would be interesting, since the B50 is exactly what those corean guys
> > > > have all this trouble dealing with.
> > >
> > > Odd. We installed from a floppy using the woody netinstall CDs by just
> > > following the instructions found at
> > > http://www.kutilek.de/technik/debian-rs6k.en.php
> > You reqlise that sarge was released almost a year ago now, and that
> > debian-installer has almost nothing to do with the woody boot-floppies, right
> > ? :)
> I do in fact realize that. I had two students install these B50 boxes and
> that's what worked in the end. IIRC we tried to boot from sarge netinstall
> CD and that didn't even begin to boot.
netbooting is the correct way on those, and the only tested way.
> > There is no floppy support on either sarge or etch debian-installer for
> > powerpc prep or chrp boxes right now, only the unofficial powermac oldworld
> > miboot floppies.
> And the CDs are supposed to boot on the RS6000?
They should, but there seems to be some issue that need investigating. I fixed
something (missing icon) recently, so the daily builds and beta2 should be
better than sarge.
> > > I should be able to provide a system.map of the booting kernel so people
> > > can figure out what's missing in the stock kernels.
> > The problem was not the kernel, but it seems that yaboot-installer or
> With our install it was the kernel. Your mail did not specify detail
> beyond 'korean guys' which did not ring a bell.
Well, they have some name in a non-displayable (for me) charset, and a korean
mailing list address, i asked them to post on d-ppc this last month.
> > prep-installer, or plain dd was not able to create a booting system.
> yaboot works for booting these machines; a simple default config created
> using yabootconfig did the job. The kernel resides in / instead of /boot
> but that should not matter.
Yep, but since partman-prep and yaboot-installer is broken on those. Nobody
seems to be in a hurry to fix those though, as it is not over 3 month since
the problem was first observed.
> > Solution is to fix partman-prep, and see if yaboot-installer behaves then, or
> > to see if the excessive ramdisk size created by initramfs-tools caused the
> > compressed kernel + builtin initrd to cause trouble with the OFs installer.
> My bet would be on the latter. That should be fairly easy to test.
partman is one of the most ugly and spaguetti-like code to play with, i wish
you luck with it.
> > Or maybe simply that the addnote thingy is broken or something ? I don't
> > really know.
> > I am able to make the prep machines bootable with simply dding the kernel with
> > builtin ramdisk the prep partition
> So no yaboot used, and no size limits whatsoever?
Nope, the size limit seems to be in the kernel+builtin initrd put onto the