https://bugzilla.novell.com/show_bug.cgi?id=328388
User teheo@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=328388#c44
--- Comment #44 from Tejun Heo 2008-01-11 19:39:38 MST ---
Maybe the module is not loaded from initrd? Tiago, please check the sysfs
file.
Ah.. we're between rock and hard place here. IDE has always unlocked HPAs
which was a good decision when it was made because HPA was often used to work
around BIOS limitations. As time passes, that usage has faded out but by then
HPA unlocking was widely used and so it couldn't be turned off.
At first, libata didn't have HPA unlocking support but later when it was added,
it had to be turned on to keep compatibility for people migrating from IDE.
HPA unlocking frees up areas at the end of the disk so unlocking doesn't
usually harm anything. Of course someone came up with the brilliant idea of
putting metadata at the end of disk - which can be clipped in several different
ways. :-(
<rant>This whole HPA on SATA thing is non-sense from the beginning. BIOS
programs HPA during boot up and restores it during resume using ACPI _GTF.
Unfortunately, hardreset also clears HPA setting and unless BIOS uses the new
ATA ACPI layout, it has no way to reprogram HPA after hardresets and there are
lots of things which can trigger hardreset in SATA and when that happens device
size changes on the fly and revalidation fails and all hell breaks loose. I
wish vendors learn to let go of their attachment to this little hidden area.
You know, users are not out there to get you.</rant>
Rant aside, I think what can be done here is to make block devices dual-sized
such that code which tries to look for metadata at the end of a disk can first
try the native max address and then what the BIOS likes people to think how
large the drive is. Which formats put metadata at the end?
Thanks.
--
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.