Debugging yaboot issues (was: Re: Call for testers: new yaboot package release - experimental installer)

[Benjamin Cama <benoar@free.fr>]

Hi Zsombor,

First, my apologies for being so slow to answer. And thank you for the
custom installer you built.

I am very thankfull for your enthusiasm, but I would like to point out
some thing that you're doing wrong. I am saying that not to deter anyone
to do anything, but to try to be more organized, as it has already been
pointed out by other people from Debian : debian-powerpc is often a mess
any nobody else will want to help us if we don't try to do things in a
more organized way. We already had no help from the d-i team about this
new yaboot release, let's hope that by making things the "right way"
we'll get more help.

Le dimanche 10 octobre 2010 à 12:42 +0200, Zsombor a écrit :
> * official 64-bit kernel replaced by a custom kernel [linux-image-2.6.32zsexp01]
>   with CONFIG_SCSI_PROC_FS=y   <- an idea to make it work on Mac

Don't do that. That /may/ work but it's not the right way to do that and
won't ever be accepted by the kernel team. This option has already been
deactivated years ago, and it's been years that yaboot should have
adapted to /sys. BTW, the error message is wrong in that it also
checks /sys for SCSI devices, but the /proc/scsi message has not been
updated regarding this change.

To help about this _specific_ issue, please report to #572869. Your SATA
controller should have been supported by yaboot, as I thought that
changes made to 1.3.16 fixed that, but it doesn't seem to be the case.
To help diagnose the problem, to anyone having this issue on a G5,
please also include the result of : ls /sys/bus/scsi/devices

> + added ehea nic module (for 64bit di-kernel only) <- to support Power6 netcard

A specific bug report should be opened to track that: Xavier, could you
do that ? The problem is that it happens on a custom image you made
Zsombor, and I'm not sure anyone will accept it as a valid BR.

That's where I am asking for help from the d-i team: how could we test
this new package from a d-i install disc ? Is there an easier and more
"officially-compliant" way, so that we can report bugs without doubts ?
Should we push the package to unstable directly ?

Regards,
benjamin