[Clfs-support] ppc64 recovery, and yaboot ?

[Ken Moffat]

On Tue, Sep 24, 2013 at 04:22:55AM +0100, Ken Moffat wrote:
> 
> 2. Any ideas what I've done wrong in compiling/installing yaboot ?
> 
 I'm still not sure exactly what the problem was, but my money is
now on e2fsprogs.  We're both using binutils-2.23.2, gcc-4.8.1, and
e2fsprogs-1.42.8, I think.  I'm using glibc-2.18, you are using an
eglibc version of 2.18 so that might make a difference.

 The first thing I did was try gentoo's full yaboot mix - new
version of ybin/ofpath from 2010, their -p0 stubfuncs and then a
-p0 new-ofpath-devspec patch to alter their ofpath.  No idea if that
actually differs from what is in cross-lfs, but it didn't help.

 Then I tried making an ext2 partition for kernels (I'd screwed up
partitioning years ago, and have a spare 1000MB which was not in
use) - again, didn't help.

 My still-working 2010 system has yaboot-1.3.13 (carried over from
what we did a few years before that), and it seemed to be fine
despite gentoo's claims that the ofpath needed to be changed for
"recent kernels".  Of course that system is using a very old udev so
it has all the sysfs compatability settings turned on.  The big
difference with 1.3.13, and why I now suspect e2fsprogs, is that it
was before debian got the maintainer to remove the e2fsprogs code.
In other words, it includes a static lib taken from an old version
of e2fsprogs and doesn't need to link to a system version of
libext2fs.a when building.

 So, I built 1.3.13 by hand.  Seemed to work, so to double-check I
put it into my script, built it, and then built a new 1.3.13 kernel,
installed that, and ran ybin from my (excuse the swearing 8)
LFS-7.4-based system.  I'm now booted into that. :-)

 With luck, people following the book won't have any of this
aggravation.  For 1.3.13 I used the -fix-1 and -ofpath_sata-1
patches, with the current seds and "make PREFIX=/usr ; make
PREFIX=/usr install" - hopefully nobody will get that desperate, but
this information is offered in case it is useful.

 Also, I very much doubt that using parted instead of the old
mac-fdisk has any bearing on this - my previous build used an
earlier version of parted and omitted mac-fdisk without any issues.

ĸen
-- 
das eine Mal als Tragödie, dieses Mal als Farce