NetBSD-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Hitachi HDD need LBA48-QUIRK



Hi NetBSD-users,

i am trying to use RAIDframe mode 5 on 500GB Hitachi
"Deskstar" series disks. I use the satalink driver[1].
While writing initial paririty on the disks i
discovered that these discs have also problems with
LBA48-addressing like the Seagates. I got "error
reading fsbn 268435424 of 268435424 - 268435455". So i
changed my LBA48_TRESHHOLD to 0xfffffe0 instead the
default 268435455 in my dev/ata/wd.c and built a new
kernel. Everything seemed fine, the raid5 got
correctly initialized, i made a LFS on it and then i
put about some 10s of GB of data on it.

Because the manpage for raidctl says that it is
"highly recommended that before using the RAID driver
for real file systems that the system administrator(s)
become quite familiar with the use of raidctl, and
that they understand how the component reconstruction
process works" i manually faild one of my disks and
started reconstruction. After about one hour the
reconstruction progress stopped with the same error
message "error reading fsbn 268435424".

Finally i set my LBA48_TRESHHOLD to 0xfffffd0 and
reconstruction finished.

Conclusion: Hitachi drives have problems with
LBA-border.
Fix: Someone should add HD* disks like my HDP725050 to
the wd_quirk_table in ata/wd.clike the following:

This should work for E7K/P7K/T7K [2] "Deskstars": 
{   "HD[P-T][0-9][0-9][0-9][0-9][2-9][0-9]*",
          WD_QUIRK_FORCE_LBA48 }

It would be nice to see an option for specifying
LBA48_TRESHHOLD in kernel konfiguration file too.

Hope this helps someone,
Martin Seibold, Wels, Austria

[1] I first had Promise TX4 controller - this one is
totally broaken. First of all, it came without a BIOS.
The DOS BIOS flashing utility needs to be on floppy,
it can't be installed from a DOS boot CD. Windows
flashing util needed Promise driver first. Anyway, it
made "missing interrupts" with older and newer
controller-BIOS versions and none of these versions
had let me entered TX4-BIOS. After trying every option
in PC BIOS that could make sense, manually configuring
IRQs, disabling every component on my mainboard and
making 5 different kernels with various wd0 flags i
gave up after i hat invested approx. 30hrs of work
with this crappy controller. Promise will never get
any cent from me. pdcsata.c says "busmaster registers
.. spread all over the controller's register space :(.
They are also 4 bytes sized, with some specific
extentions in the extra bits. It also seems that the
IDEDMA_CTL register isn't available." Finally i
exchanged it for dawicontrol DC-154 with Sil3114 chip
which works fine without any troubles with satalink. I
got a few euros back from the dealer, too ;-)

[2] I don't know exactly which Hitachi series are
affected by the LBA-border problems. At least the
500GB P-version is. The code should work with all
P/T/E-Deskstars 200GB and up.


      Lesen Sie Ihre E-Mails jetzt einfach von unterwegs.
www.yahoo.de/go


Home | Main Index | Thread Index | Old Index