The 53c810 driver in the pre-2.0.31-2 and 2.1.47 kernels resets the
SCSI bus very often. 2.0.30 doesn't have that problem.
My setup is
ASUS SP3 board + intel 486 DX-66
16 MB RAM
NEC DSE 2100S 2.1 GB SCSI drive
Wangtek Streamer
The SCSI reset do not occur randomly. This time I have run my machine
for 2 days without problems. Then sometimes SCSI resets occured
nearly every minute, sometimes with 4 or 5 minutes inbetween, but
finally, after 30 minutes hanging th e machine completely. I have
just hit the reset button. Here are the last lines of the log (kernel
pre-2.0.31-2):
Aug 6 22:10:44 isnogud kernel: scsi : aborting command due to timeout : pid 354223, scsi0, channel 0, id 0, lun 0 Read (6) 04 ce 5a 08 00
Aug 6 22:10:44 isnogud kernel: ncr53c8xx_abort: pid=354223 serial_number=354276 serial_number_at_timeout=354276
Aug 6 22:10:48 isnogud kernel: scsi : aborting command due to timeout : pid 354224, scsi0, channel 0, id 0, lun 0 Read (6) 03 c9 ae 02 00
Aug 6 22:10:48 isnogud kernel: ncr53c8xx_abort: pid=354224 serial_number=354277 serial_number_at_timeout=354277
Aug 6 22:10:48 isnogud kernel: ncr53c810-0: abort ccb=00097820 (skip)
Aug 6 22:10:49 isnogud kernel: SCSI host 0 abort (pid 354224) timed out - resetting
Aug 6 22:10:49 isnogud kernel: SCSI bus is being reset for host 0 channel 0.
Aug 6 22:10:49 isnogud kernel: ncr53c8xx_reset: pid=354224 reset_flags=2 serial_number=354277 serial_number_at_timeout=354277
Aug 6 22:10:50 isnogud kernel: ncr53c810-0: restart (scsi reset).
Aug 6 22:10:52 isnogud kernel: ncr53c810-0-<0,0>: 5.0 MB/s (200 ns, offset 8)
Aug 6 22:11:50 isnogud kernel: scsi : aborting command due to timeout : pid 354857, scsi0, channel 0, id 0, lun 0 Write (6) 00 c0 84 02 00
Aug 6 22:11:50 isnogud kernel: ncr53c8xx_abort: pid=354857 serial_number=354914 serial_number_at_timeout=354914
Aug 6 22:11:50 isnogud kernel: ncr53c810-0: abort ccb=00097820 (skip)
Aug 6 22:11:52 isnogud kernel: SCSI host 0 abort (pid 354857) timed out - resetting
Aug 6 22:11:52 isnogud kernel: SCSI bus is being reset for host 0 channel 0.
Aug 6 22:11:53 isnogud kernel: ncr53c8xx_reset: pid=354857 reset_flags=2 serial_number=354914 serial_number_at_timeout=354914
Aug 6 22:11:53 isnogud kernel: ncr53c810-0: restart (scsi reset).
Aug 6 22:11:55 isnogud kernel: ncr53c810-0-<0,0>: 5.0 MB/s (200 ns, offset 8)
A few days ago I have seen this bug right after rebooting, when trying
to unpack a large tar file.
My impression is that these resets begin to occur, when large
transfers are done to or from the hard drive. Once started, the
resets occur almost continously or as a cause of even small transfers.
When I saw the resets right after rebooting a few days ago, it also
seemded that it makes a difference wether there is a tape in the
streamer. With no tape inserted, the resets seemed to occur less
often. But I may be wrong here.
Have others seen this bug too and is someone looking into it?
urs