[2930] in linux-scsi channel archive
DPT 2044U and Quantum Fireball ST4.3
daemon@ATHENA.MIT.EDU (Zane Dodson)
Mon Dec 15 23:55:49 1997
From: Zane Dodson <dodson@mza.com>
To: linux-eata@trudi.zdv.Uni-Mainz.DE, linux-scsi@vger.rutgers.edu
Date: Mon, 15 Dec 1997 21:51:34 -0700 (MST)
Hello,
I'm having a problem with my DPT 2044U and a Quantum Fireball ST4.3. Below
is my hardware configuration:
Here's an excerpt from the system log for this machine running 2.0.30.
EATA (Extended Attachment) driver version: 2.59b
developed in co-operation with DPT
(c) 1993-96 Michael Neuffer, mike@i-Connect.Net
Registered HBAs:
HBA no. Boardtype Revis EATA Bus BaseIO IRQ DMA Ch ID Pr QS S/G IS
scsi0 : PM2044U v07K.Y 2.0c PCI 0x6110 11 BMST 1 7 N 64 252 Y
scsi0 : EATA (Extended Attachment) HBA driver
scsi : 1 host.
Vendor: QUANTUM Model: FIREBALL ST4.3S Rev: 0F0C
Type: Direct-Access ANSI SCSI revision: 02
Detected scsi disk sda at scsi0, channel 0, id 0, lun 0
scsi0: queue depth for target 0 on channel 0 set to 64
The Quantum drive is the only peripheral on the bus. It has its
terminator enabled jumper set. Also, I have turned ON termination in
the DPT PCI SCSI BIOS. So, my SCSI bus is properly terminated
at the adapter end and the disk end.
I have found that if I have tagged queuing enabled in the DPT PCI BIOS,
I can get severe filesystem corruption. However, disabling tagged queuing in
the DPT BIOS results in a stable system. I contacted both Quantum and
DPT. Quantum explained that their drive had no reported problems with
supporting tagged queuing, but the maximum depth was 8. DPT echoed similar
sentiments in that they had no reported problems with the combination of
hardware and tagged queuing.
Does anyone have any information about this particular hardware combination?
I have long been confused by the many driver choices I have. Is one
of the EATA drivers recommended above the others? (I'm using eata_dma at the
moment, but there's EATA-PIO and just EATA which I haven't yet tried.)
I have many machines with DPT controllers in them. I have a 2044U, 2144UW,
a 3334UW, and a 3334UDW with an SX4030UW/2 added on. Should I choose
the eata_dma driver for all these controllers? Also, I found in the
eata.c source file a listing of all the boot-time/command-line options for
the eata driver. However, I could not find a similar listing for the
eata_dma driver. Can I disable tagged queuing on a target-by-target basis?
Can I set the maximum queue depth on a target-by-target basis?
It is interesting that the queue depth of target 0 above is set to 64
(this is true regardless of the setting of the DPT BIOS option for tagged
queuing -- I suspect in the case where the DPT BIOS option for tagged
queuing is off, this parameter is ignored). What happens if the drive
only supports a queue depth of 8 and the driver attempts 64? Given the
filesystem errors that resulted, it would appear that the commands
are discarded and never processed.
Also, I found it interesting that in another machine with a 2144UW (a
Quantum Fireball ST4.3 on 0 and a Jaz drive on 2), the queue depth was
set to 32 for both targets. Is the total of 64 divided between the two
drives in this case? I haven't yet had a problem with this particular
hardware/software configuration, but it hasn't been used as heavily.
I'd be interested in a brief explanation of how the tagged queuing is
implemented and how the DPT negotiates this setting with each drive.
The theory of operation section in the DPT manual seems to indicate the
controller is capable of 64 outstanding commands and I'm assuming this
has to be divided among the target devices attached.
Any comments would be appreciated and more information can be provided.
Best regards,
Zane Dodson
dodson@mza.com
