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

Bug#375505: yaboot-installer: generated yaboot.conf is broken on IBM POWER hardware ...



clone 375505 -1 -2
reassign -1 yaboot
retitle -1 yaboot: ofpath fails on IBM JS21 blades
retitle -2 yaboot-installer: should write hvc serial console into append= line
retitle 375505 yaboot-installer: should only write magicboot= on NewWorld PowerMacs
thanks

On Mon, Jun 26, 2006 at 03:49:50PM +0200, Sven Luther wrote:
> When installing on IBM JS21 blades, as well as on the augsbourg quad-power5
> machines and p505 boxes, the install failed on yaboot-installer.
> 
> After investigation, three issues appeared :
> 
> 1) There was a device= line, with nothing behind it, which yaboot failed to
> parse at boot time, and thus rejected the config file or entries. The device =
> line is not really necessary, since yaboot is able to recognize the device
> from the OF chosen nodes. Removing this line by hand worked just fine.

I've cloned this to a yaboot bug, because apparently ofpath failed. If
you could give that cloned bug (I don't have the number yet, of course)
more details of the disks on this system and what the Open Firmware path
should be, then perhaps ofpath can be fixed.

That said, there's obviously no point in writing out an unparseable
configuration file, so I've changed yaboot-installer to avoid doing
this.

> 2) the magicboot=/usr/lib/yaboot/ofboot, and this is the one which caused
> yaboot-installer to fail. ybin doesn't support a magicboot line with a raw
> install, that is in a PReP boot partition, as is used on IBM POWER. Removing
> this line by hand eliminated that problem.

Fixed in r38733 (and will be in yaboot-installer 1.1.7), thanks.

> 3) These machines use a virtual serial console, namely /dev/hvc* (hvc0 on the
> blade center, but hvc1 on the augsbourg virtual partition). yaboot-installer
> needs to add the correct console=hvc* into the append line, in order to get a
> console after the reboot.

OK, seems sensible. I have a patch for this, which I'll commit once I
get the cloned bug number back from debbugs.

Thanks,

-- 
Colin Watson                                       [cjwatson@debian.org]



Reply to: