https://bugzilla.novell.com/show_bug.cgi?id=841392
https://bugzilla.novell.com/show_bug.cgi?id=841392#c12
Marcus Schaefer changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|ASSIGNED |NEEDINFO
InfoProvider| |snwint@suse.com
--- Comment #12 from Marcus Schaefer 2013-10-08 19:27:20 UTC ---
that's right, so I was able to reproduce the problem when building a 32bit iso
without efi support. I have no idea why we build the 32bit iso without
efi support ?
anyway without efi isohybrid is called as follows:
EXEC [/usr/bin/isohybrid --offset 0 --id 0x472ef954 --type 0x83
/tmp/mytest/iso-vmxFlavour/LimeJeOS-openSUSE-12.3.i686-1.12.3.iso 2>&1]
The offset set to 0 caused the broken partition table. But I remember Robert
had a good reason to set the offset to 0 in order to allow to boot the iso as
iso on different bios implementations. When I set the offset to 64 I got a
correct partition table output in parted
next I tested the iso as disk with persistent writing as reported in this bug.
I got one step further but I found that the iso header is only available on the
entire disk device /dev/sda. Thus when /dev/sda is mounted to access the data
from the iso all "partitions" are set as busy by the kernel. Because of this
the mkfs tools refused to run on the persistent write partition because the
kernel report it as busy. I know we workaround that by a losetup on /dev/sda
but that's nasty and also not required if isohybrid is called in --uefi mode
So I still think what isohybrid writes here is broken, I would expect that the
iso header is available on the partition device
if we would build the 32bit iso also with firmware="efi" we would solve the
problem by another workaround
Thoughts ?
--
Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.