[Debian-ppc64-devel] Installation Report

[Sven Luther]

Please don't drop debian-boot from the CC list.

On Mon, Feb 13, 2006 at 05:05:11PM +0100, "Nölle, Christian" wrote:
> Sven Luther wrote:
> 
> >> Comments/Problems:
> >> 
> >> We had problems with creating filesystems. The installer fails to
> >> create the prep-Partition properly, so we had to exit to shell,
> >> chroot 
> >> into the target and create in manually with cfdisk.
> > 
> > Can you give :
> > 
> >   1) give me the exact scenario of this "failed to create a prep
> >   partition" thingy.
> 
> Ok, here comes the thingy: The installer fires up the partitioning dialog
> and suggest a certain layout with a partition #1, which should become prep, 
> but states it as:
> 
> #1 primary    8.2 MB B KÂ Â

This is the prep partition, or should be, i also don't know why the name is
such garbled.

> When selecting it, it states a prep partition with the boot flag on. You continue 
> and partman only lists an ext3 for / and a swap. We then returned to partman dialog and
> selected it again, same as above. Interesting on that is, if you select the type in the
> detailed view, it gives you the list of possible types with prep on top stating:
> 
> 10 partman/method_long/prep doesn't exist

This is a partman bug, i trully hate partman's hermetic design, let's hope
someone shows up and help fix this.

> If you choose that entry, the installers ends up in an endless loop of a senseless dialog
> with the very meaningful text ???- ???] and the two options "Go Back" and "Continue", which
> both lead you back to the little meaningful dialog with two options....and so forth.. :)

This is a bug indeed.

> After applying changes to disk and trying to install yaboot via the menu it states:
> 
> No hard disks were found which have an "Apple_Bootstrap" partition.
> You must create an 819200-byte partition with type "Apple_Bootstrap".

Sure, this is a yaboot-installer bug, you don't need ab apple partition, but
yaboot-installer is for some reason not noticing this.

> This lead us to the cfdisk solution in the chroot, which worked.

Kind of.

> >   2) give me the exact /proc/cpuinfo output ?
> 
> Sorry for that, my definition in the first mail was wrong. Here is the output:
> 
> processor       : 0
> cpu             : POWER5 (gr)
> clock           : 1504.352000MHz
> revision        : 2.2
> 
> processor       : 1
> cpu             : POWER5 (gr)
> clock           : 1504.352000MHz
> revision        : 2.2
> 
> processor       : 2
> cpu             : POWER5 (gr)
> clock           : 1504.352000MHz
> revision        : 2.2
> 
> processor       : 3
> cpu             : POWER5 (gr)
> clock           : 1504.352000MHz
> revision        : 2.2
> 
> processor       : 4
> cpu             : POWER5 (gr)
> clock           : 1504.352000MHz
> revision        : 2.2
> 
> processor       : 5
> cpu             : POWER5 (gr)
> clock           : 1504.352000MHz
> revision        : 2.2
> 
> processor       : 6
> cpu             : POWER5 (gr)
> clock           : 1504.352000MHz
> revision        : 2.2
> 
> processor       : 7
> cpu             : POWER5 (gr)
> clock           : 1504.352000MHz
> revision        : 2.2
> 
> timebase        : 188044000
> machine         : CHRP IBM,9124-720

Neqt; so 8-way CHRP IBM,9124-720, this should be detected as chrp_ibm
correctly, and partman-prep should work, and yaboot-installer should know not
to expect an apple partition.

> >> As we have no video-card in that machine, we used the ssh-Feature of the
> >> installer. As we finished the 1st stage, we were prompted with an
> >> error message that network-console-config is not avaliable and we
> >> have to continue without ssh in the 2nd stage *sigh*.
> > 
> > Yeah, not nice.
> 
> You name it.
> 
> 
> >> fgconsole: VT_GETSTATE Invalid Argument
> >> 
> >> No input is possible, could not continue with the installation, as we
> >> only have the serial console and not the ssh feature avaliable (see
> >> above). Any hints!?
> > 
> > Not entirely sure about this one, maybe waldi can comment ?
> 
> Ok.
> 
> > Well, i can help you solve the first issue, and then once this is
> > done, we can tackle the second, how does this sound ? 
> 
> Good for me :)

So, let's follow in this track.

Friendly,

Sven Luther