On Thu, Jan 3, 2019 at 5:54 PM Frank Scheiner <[email protected]> wrote: > > On 1/3/19 12:10, Mathieu Malaterre wrote: > > On Sun, Dec 30, 2018 at 9:54 PM Frank Scheiner <[email protected]> > > wrote: > >> So one actually cannot repair that issue from a G5 - or maybe not even > >> any issue with an HFS at all, because the same also happens when trying > >> to check a clean HFS with `fsck.hfs`. The result is that an existing > >> GRUB installation can no longer be upgraded on a G5 as soon as someone > >> hits that problem. Actually I'm also unsure if my HFS problems were > >> created by an unclean shutdown at all. > >> > >> One could solve this by recreating the HFS partition from scratch and > >> restoring the contents (incl. file blessing and file types (i.e. > >> `tbxi`)). Or maybe by rewriting a `dd`ed image from a clean HFS - if you > >> have one at hand. Trying to repair the partition with `fsck.hfs` from a > >> Mac mini G4 (via SATA to USB adapter) worked for me also. But all these > >> workarounds are unhandy to the least. > > > > The way you describe this bug makes me think of a 64bits vs 32bits > > issue. Next time this happen to you, use a foreign powerpc > > installation to run ppc32 binary on your G5. > > That's a good idea and I think we don't even have to wait for the next > time this issue hits me, as even for a clean HFS the `fsck.hfs` > segfaults. The significance of the result of such a test will not be the > same as for checking a borked HFS, but could show a tendency.
Right ! Good point. > Say, will a powerpc64 kernel from arch ppc64 work with a userland from > arch powerpc or do I need the powerpc64 kernel from arch powerpc? Or are > these kernels actually identical? If I understand your sentence, the answer is yes: they are identical. The only (main?) difference with amd64/i386 world is that you cannot run a ppc32 kernel on a ppc64 machine. > Because such a configuration I could setup easily by just changing the > root FS that the machine uses when network booting. > > Cheers, > Frank
