> I have an Intraserver SCSI card with a couple SCSI disks, and a phasewriter
> rewriteable optical drive I use for backups. Under 2.2.14 this arrangement
> works fine, but with the 2.2.16 the dumps to the PD drive lock up. Below are
> some pertinent lines from the system logs.
>
> These are the driver discovering the devices:
>
> Jun 28 10:46:02 icarus kernel: sym53c875-0-<0,*>: WIDE SCSI (16 bit) enabled.
> Jun 28 10:46:03 icarus kernel: sym53c875-0-<0,*>: FAST-10 WIDE SCSI 20.0 MB/s (100 ns, offset 15)
> Jun 28 10:46:03 icarus kernel: SCSI device sda: hdwr sector= 512 bytes. Sectors= 4406960 [2151 MB] [2.2 GB]
> Jun 28 10:46:03 icarus kernel: sym53c875-0-<1,*>: WIDE SCSI (16 bit) enabled.
> Jun 28 10:46:03 icarus kernel: sym53c875-0-<1,*>: FAST-20 WIDE SCSI 40.0 MB/s (50 ns, offset 16)
> Jun 28 10:46:03 icarus kernel: SCSI device sdb: hdwr sector= 512 bytes. Sectors= 17774160 [8678 MB] [8.7 GB]
> Jun 28 10:46:03 icarus kernel: sym53c875-0-<2,*>: FAST-5 SCSI 5.0 MB/s (200 ns, offset 16)
> Jun 28 10:46:03 icarus kernel: SCSI device sdc: hdwr sector= 512 bytes. Sectors= 1298496 [634 MB] [0.6 GB]
> Jun 28 10:46:03 icarus kernel: sdc: Write Protect is off
>
>
> I find these in the log at the time when the dump process hangs:
>
> Jun 29 03:10:00 icarus kernel: sym53c875-0-<0,0>: QUEUE FULL! 8 busy, 7 disconnected CCBs
> Jun 29 03:10:00 icarus kernel: sym53c875-0-<0,0>: tagged command queue depth set to 7
> Jun 29 03:10:00 icarus kernel: sym53c875-0-<0,0>: QUEUE FULL! 7 busy, 6 disconnected CCBs
> Jun 29 03:10:00 icarus kernel: sym53c875-0-<0,0>: tagged command queue depth set to 6
> Jun 29 03:10:00 icarus kernel: sym53c875-0-<0,0>: QUEUE FULL! 6 busy, 5 disconnected CCBs
> Jun 29 03:10:00 icarus kernel: sym53c875-0-<0,0>: tagged command queue depth set to 5
> Jun 29 03:10:00 icarus kernel: sym53c875-0-<0,0>: QUEUE FULL! 5 busy, 4 disconnected CCBs
> Jun 29 03:10:00 icarus kernel: sym53c875-0-<0,0>: tagged command queue depth set to 4
> Jun 29 03:10:00 icarus kernel: sym53c875-0-<0,0>: QUEUE FULL! 4 busy, 3 disconnected CCBs
> Jun 29 03:10:00 icarus kernel: sym53c875-0-<0,0>: tagged command queue depth set to 3
> Jun 29 03:10:08 icarus kernel: sym53c875-0-<0,0>: QUEUE FULL! 3 busy, 2 disconnected CCBs
> Jun 29 03:10:18 icarus kernel: sym53c875-0-<0,0>: tagged command queue depth set to 2
> Jun 29 03:10:20 icarus kernel: sym53c875-0-<0,0>: QUEUE FULL! 2 busy, 1 disconnected CCBs
> Jun 29 03:10:27 icarus kernel: sym53c875-0-<0,0>: QUEUE FULL! 2 busy, 1 disconnected CCBs
> Jun 29 03:10:27 icarus kernel: sym53c875-0-<0,0>: tagged command queueing disabled
The device returned QUEUE FULL with only 1 command currently queued to it.
Then the driver disabled tagged command queuing as soon as this has been
possible (no command currently queued to the device).
This looks like the expected behaviour, but the situation is unusual.
Note that it has been observed with Atlas I drives firmare L912 with write
caching enabled flooded with bunches of small IOs.
If, as it seems, the drive is an Atlas I / L912, you may try a firmware
upgrade. Such an upgrade exists, but I donnot remember the URL.
I would also recommend you, even if the drive may not be the cause of the
hang-up, to disable write-caching for this drive. This reduces a lot the
pressure on the firmware and decreases the probability to trigger
firmware/software problems.
Gérard.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/