scsi timeout with 2.4

hi,

i don't know if it is a kernel issue at all, it could be a hardware issue too, but i can't tell so please have a look.

i am using linux 2.4.20 on a PowerPC (PReP) with a "LSI Logic / Symbios Logic (formerly NCR) 53c825 (rev 13)" scsi controller (onboard). i am using the symbios driver (CONFIG_SCSI_SYM53C8XX) because the other alternatives (NCR53C8XX, timeouts during boot, booting stops. SYM53C8XX_2, timeouts during use, unuseable) are not working.

i have these timouts when there is heavy traffic on the controller, it's reproduceable, it happens only *sometimes* during "normal" use. when these timeouts occur, load goes up to 10-16, the system is unuseable, then, after a few seconds, the system status seems to be ok again, load is going down to normal.

i know, the controller and the machine is a bit old, but it serves my needs. on the controller there are 3 scsi disks (2x18 GB IBM xServer, 1x36GB eServer -- yes High-End scsi disks on a low-end controller, but the disks were cheap, somehow.)

here are the timeouts:

scsi : aborting command due to timeout : pid 3463623, scsi0, channel 0,\ id 1, lun 0 Read (10) 00 01 d0 1e 9a 00 01 00 00
sym53c8xx_abort: pid=3463623 serial_number=3463623 serial_number_at_timeout=3463623
SCSI host 0 abort (pid 3463623) timed out - resetting
SCSI bus is being reset for host 0 channel 0.
sym53c8xx_reset: pid=3463623 reset_flags=2 serial_number=3463623\ serial_number_at_timeout=3463623 sym53c825a-0-<0,*>: FAST-10 WIDE SCSI 20.0 MB/s (100.0 ns, offset 8)
sym53c825a-0-<1,*>: FAST-10 WIDE SCSI 20.0 MB/s (100.0 ns, offset 8)
sym53c825a-0-<2,*>: FAST-10 WIDE SCSI 20.0 MB/s (100.0 ns, offset 8)

the first line ("scsi: abort...") is sometimes repeated, 2 or up to 7 times. the last 3 lines are always repeated 3 times, as you can see above. and the "id" is not always "1" but all three id's are affected. while i don't believe that the disks are bad, i suspect the scsi controller perhaps, or is it a driver issue and can i tune something?
the driver is statically compiled into the kernel with default settings.
(DEFAULT_TAGS=8, MAX_TAGS=32, SYNC=20).

what is your opinion on that? the controller, the disks (perhaps too new for the controller?) or the driver?

Thank you for your time,
Christian.

root@sheep:~# cat /proc/scsi/sym53c8xx/0
General information:
  Chip sym53c825a, device id 0x3, revision id 0x13
  On PCI bus 0, device 12, function 0, IRQ 15
  Synchronous period factor 25, max commands per lun 32
root@sheep:~#
root@sheep:~# cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
  Vendor: IBM-ESXS Model: ST318305LW    !# Rev: B245
  Type:   Direct-Access                    ANSI SCSI revision: 03
Host: scsi0 Channel: 00 Id: 01 Lun: 00
  Vendor: IBM-ESXS Model: ST318305LW    !# Rev: B244
  Type:   Direct-Access                    ANSI SCSI revision: 03
Host: scsi0 Channel: 00 Id: 02 Lun: 00
  Vendor: IBM-ESXS Model: ST336607LW    FN Rev: B258
  Type:   Direct-Access                    ANSI SCSI revision: 03
root@sheep:~#
root@sheep:~# cat /proc/cpuinfo
cpu		: 604r
clock		: ???
revision	: 49.2 (pvr 0009 3102)
bogomips	: 299.00
machine		: PReP Utah (Powerstack II Pro4000)
L2		: 512Kb, parity disabled SRAM:synchronous,pipelined,no parity
root@sheep:~#

BOFH excuse #274:

It was OK before you touched it.