Essentially, it seems to die with a bunch of SCSI bus resets under heavy
disk I/O (as when mounting a disk). It'll stop accessing the drive after a
second or two, pause for 5-10 seconds, then the endless SCSI bus
resets start.
I'm using an IBM DGHS09Y 9GB SCA drive (also tried a Seagate ST32430WC,
and a Hitachi DK319H-18WC), with an SCA convertor (has both narrow and
wide connectors) and a multimode LVD/SE terminator (I've tried three
different ones so far) plugged into the last connector on the SCSI cable
(the SCA convertor does not have termination), which I've tried various
versions thus (none of these cables are more than 36 inches long):
1)The PVC 3-connector cable supplied with the Tekram card. 8" between term
connector and connector the drive is plugged into.
2)A Teflon 5-connector cable, with 4" between term connector and drive
connector.
3)A "twist 'n' flat" LVD-rated 5-connector cable, with 1/2" between drive
connector and term connector.
I've also tried slowing the SCSI bus down to 10MHz instead of 20MHz (also
tried disabling tagged commands and disconnection). If anything, that
should have fixed the problem if it were due to a cabling or termination
issue, but that didn't help much at all. Getting rid of the PVC cable
helped somewhat, but I suspect that the PVC cable was intermittent anyway.
This controller uses the Symbios Logic SYM53C875 Ultra-Wide chipset, and
I'm using the sym53c8xx driver with Linux (a Tekram-supplied driver is
available for this card, but I'm not sure how it differs from the
Linux-supplied driver so I haven't tried it).
The same drive, when plugged into a narrow-channel Adaptec 2940, works
flawlessly at 10MHz with a Seagate 2160N terminating the bus.
I'd also had this problem with the SYM53C875 controller found on the Intel
NL440BX "Nitelite" motherboard, with completly different hard drives
(Seagate) with active termination on the drive (no SCA convertors), but I
didn't pursue it then since there were other reasons for not using the
NL440BX motherboard (it only has 3 DIMM slots).
Below are some snips from the log (this error occured with both the stock
2.2.14 driver and the new patch found on
ftp://ftp.tux.org/pub/people/gerard-roudier):
May 14 13:27:58 test kernel: scsi : aborting command due to timeout : pid
1905,
scsi1, channel 0, id 4, lun 0 Read (6) 00 00 30 08 00
May 14 13:27:58 test kernel: sym53c8xx_abort: pid=1905 serial_number=1926
serial_number_at_timeout=1926
May 14 13:27:58 test kernel: SCSI host 1 abort (pid 1905) timed out -
resetting
May 14 13:27:58 test kernel: SCSI bus is being reset for host 1 channel 0.
May 14 13:27:58 test kernel: sym53c8xx_reset: pid=1905 reset_flags=2
serial_number=1926 serial_number_at_timeout=1926
May 14 13:27:58 test kernel: sym53c875E-0: restart (scsi reset).
May 14 13:27:58 test kernel: sym53c875E-0: Downloading SCSI SCRIPTS.
May 14 13:28:01 test kernel: sym53c875E-0-<4,*>: FAST-20 WIDE SCSI 40.0
MB/s (50 ns, offset 15)
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [EMAIL PROTECTED]
