https://bugzilla.novell.com/show_bug.cgi?id=333753
User teheo@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=333753#c41
--- Comment #41 from Tejun Heo 2009-01-21 17:42:32 MST ---
There are two problems here.
1. Certain dm raid layouts put the metadata at the end of the disk and the
"end" of the disk is determined with HPA turned on. On openSUSE, HPA is turned
off by default to remain compatible with the original IDE behavior and I think
several other distros do so too. Anyways, with HPA unlocked, the end of disk
is no longer the place where the BIOS wants the OS to think it is so the array
detection fails. This can be worked around by turning off HPA unlocking but
more graceful solution would be teaching dm that the device has two sizes - the
BIOS one and the native one and try the BIOS one first.
2. On reboot, probably the BIOS doesn't bother to re-enable or check HPA status
and fails miserably with the HPA unlocked. This is BIOS's fault as BIOS
shouldn't assume the system configuration to be in specific state on reboot but
it might be workaroundable by making libata restore HPA setting on detach, but
I don't know it's just messy. Earlier nv boards had a lot of hardware and BIOS
problems, so it's not too surprising tho.
Anyways, the dual size thing has been on my to do list for quite some time now.
I think it's about time I get on to it.
And, BTW, it's probably best to simply avoid those BIOS RAIDs. There simply is
no advantage over md raid other than supporting direct booting from the array
and it can easily be worked around by making a small boot partition on md
raid1.
--
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.
