Hi,

John Paul Adrian Glaubitz wrote:
> Please use the latest image I created yesterday:
> https://cdimage.debian.org/cdimage/ports/snapshots/2020-10-13/debian-10.0.0-powerpc-NETINST-1.iso

This contains 1 mountable partition in MBR and 3 partitions in an
Apple Partition Map (APM), of which partition 2 contains a HFS+ filesystem
with bocksize 2048. This blocksize is not supported by Linux, AFAIK.


To avoid any ambiguities i would mount /dev/sdb with -t iso9660.

Wayward idea: Would a CD or DVD with a copy of the ISO work in /dev/sr0 ?


-------------------------------------------------------------------
Theoretical considerations:

Lennart Sorensen wrote:
> Does using dd work for a powermac like it does on PCs?

I am not aware of other ways to present the ISO to the machine.
The boot equipment should work on HDD and CD/DVD.

If i understood the original problem correctly, then the boot process
began to work and failed only when Linux was already up. So it can hardly
be a total failure of the ISO to tell the firmware what's wanted.

The computer's hardware under firmware control thus works with the USB
stick. Linux sees the CHRP partition. So it, too, was able to read from
the stick. Its reason to refuse mounting /dev/sdb1 must be in a more
subtle shortcomming.

Since Linux was able to read the partition table, it might be able to
mount /dev/sdb.


> Looking at the iso it appears to have a DOS partition table, [...]
> and an apple partition map

It has a CHRP partition in MBR and an Apple Partition Map partition
pointing to a HFS+ with creator and type
  chrp tbxi
attributed to
  /System/Library/CoreServices/BootX
and blessing
  ppc_bootdir
to
  /System/Library/CoreServices
It is supposed to represent the same data directories and files as
the ISO. But as said, Linux will take offense from the block size.

(Actually that size is a necessity only if GPT is present. Maybe one
should make experiments with xorrisofs option -hfsplus-block-size 512.
But that would be a completely different adventure.)


> Certainly the size in /proc/partitions appears to agree with
> the overall size and the dos partition table which should be the entire
> iso file system, so mounting as iso9660 should have worked.

Yes. It is not the APM partition 1 which has only 915 * 2048 bytes.
The data partition is number 2 and has 158181 * 2048 bytes = 316362 KiB.
This size does not match the display from /proc/partitions either.

The MBR partition begins at LBA 0. So the block addresses in the ISO 9660
should work with /dev/sdb and /dev/sdb1 alike. Well, /dev/sdb1 doesn't.

-----------------------------------------------------------------------

Have a nice day :)

Thomas

Reply via email to