Michael Schmitz writes:
Now, the fact that the package builds (except for the pmac-fdisk-cross
failure at the end, which is a consequence of ARCH being neither m68k now
powerpc) and the binaries actually run is not too surprising. I would be
amazed, however, if mac-fdisk does read and write correct Apple partition
tables on x86 (or any other little endian architecture, for that matter).
Did you successfully read a Mac-generated partition table, or generate on
that's properly recognized by a Mac?
Last I looked, the partition table entries would have to be converted to
host endianness before manipulating them in-core, and back-converted
before writing a changed table out to disk.
If you're interested in testing this, I can send you a copy of a partition
table in raw format which you can dump to a blank disk (or a loopback
device).
My use was on a Tivo2 series PVR with a usb-storage (to ide) adapter. I was
inspecting the drive for backup/restore options. This all started when I
used file(1) to see it uses a mac partition table. I have etch installed on
a dual booting iBook to. mac-fdisk appears to print the same partition
information regardless of what machine I ran it from (x86,ppc). Tigers'
disk utility sees these partitiions to (but not Linux, even with apple
partition support - odd). I used the offsets provided by mac-fdisk with
losetup -o offset to mount some of the ext2 file systems. Cool, tivo runs
Linux. How cool is that?.
Anyway, AFAIT the reading is just fine. If you want to torture test
anything, you don't need a real disk if you're just comparing how the x86
compiled binary does it, and the ppc one does. Just use a few MB file to
play with (it's just blocks after all...).
Thanks,
Scott.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
