#618 Mysterious firewire HD errors

[Anonymous]

I'm running Gentoo Linux with the 2.6.7 kernel
(Gentoo-sources, r3). I've run into baffling firewire
problems with this kernel as well as various versions
of kernels in the 2.6.x and 2.4.2x series, both
"vanilla" and Gentoo-optimized. I installed libraw
only today (version 0.9.0), recompiled the kernel
afterwards, but the symptoms stayed the same as
they have been for ages now. It should be noted
that today I tried two variations on the 2.6.7 kernel
sources: the ones emerged with Gentoo's Portage,
and another one with the same sources, except that
I replaced the entire IEEE1394 directory with the one
provided on the linux1394.org website.

The problem: I'm trying to use an external hard drive
via FireWire. The HD is a 120 GB Western Digital,
and I'm using it inside an external case for standard
IDE hard drives. The HD is functional, as far as I
have been able to determine by using it in another
machine.

My FW controller is based on Lucent's FW323
chipset, rev. 04. It as well as the external HD casing
have worked fine together when used in my Windows
XP box.

lspci output is as follows:

----------
lspci output:

0000:00:00.0 Host bridge: Intel Corp. 440BX/ZX/DX
- 82443BX/ZX/DX Host bridge (rev 03)
0000:00:01.0 PCI bridge: Intel Corp. 440BX/ZX/DX -
82443BX/ZX/DX AGP bridge (rev 03)
0000:00:04.0 ISA bridge: Intel Corp. 82371AB/EB/MB
PIIX4 ISA (rev 02)
0000:00:04.1 IDE interface: Intel Corp.
82371AB/EB/MB PIIX4 IDE (rev 01)
0000:00:04.2 USB Controller: Intel Corp.
82371AB/EB/MB PIIX4 USB (rev 01)
0000:00:04.3 Bridge: Intel Corp. 82371AB/EB/MB
PIIX4 ACPI (rev 02)
0000:00:09.0 Ethernet controller: 3Com Corporation
3c905B 100BaseTX [Cyclone] (rev 30)
0000:00:0a.0 FireWire (IEEE 1394): Lucent
Microelectronics FW323 (rev 04)
0000:01:00.0 VGA compatible controller: Matrox
Graphics, Inc. MGA G400 AGP (rev 03)
----------

Testing the setup today, the following happened.

I connected the HD case to my Firewire controller
and switched it on. dmesg | tail showed that it had
been detected without problems, and the HD became
/dev/sda. I used FDISK on it, created a partition
spanning the entire 120 gigs and formatted it as
ext3. Mounted it, dmesg output still normal, no
errors. Created some directories, changed their
permissions and owners, no problems.

Then I moved about 2 gigs of data from another disk
to the external HD. The operation seemed to take
significantly longer than it should have. The files
were there when I checked. I tried to move them via
network to my XP box using Samba, but the file
transfers stopped abruptly in the middle with an
error. At this point I re-checked dmesg. There were
errors such as these:

----------
Raw sense data:0x70 0x00 0x03 0x00 0x00 0x00
0x00 0x0a 0x00 0x00 0x00 0x00 0x08 0x03 0x00
0x00 0x00 0x00
end_request: I/O error, dev sda, sector 116133984
(the above error was repeated ad nauseam, with the
sector incremented by one in each, and finally
followed by:)
Buffer I/O error on device sda1, logical block
58066962
ieee1394: unsolicited response packet received - no
tlabel match
ieee1394: unsolicited response packet received - no
tlabel match
ieee1394: unsolicited response packet received - no
tlabel match
ieee1394: unsolicited response packet received - no
tlabel match
----------

The HD used to work just fine using a 2.6.x kernel
some time in the past, for a period of maybe two
weeks. One day the errors started appearing. I
moved on to a 2.4 kernel instead, and the disk
worked again. This time for a few days, and then
similar (if I remember correctly) symptoms as
described above appeared.

e2fsck -f shows no errors on the drive.

I have the following drivers compiled into the kernel:

IEEE 1394 (FireWire) Support
OHCI-394 Support
SBP-2 Support (Harddisks etc.) (but not the Phys
DMA option)
Raw IEEE1394 I/O Support

I have no idea how to proceed from here. Neither
Google or Gentoo Forums provided any relevant help
in solving this problem.

My e-mail is jarkko.laiho@nic.fi. I'd be extremely
grateful for a response.