Re: [Bug 12195] "dd" make kernel panic

[Mike Anderson <andmike@linux.vnet.ibm.com>]

bugme-daemon@bugzilla.kernel.org <bugme-daemon@bugzilla.kernel.org> wrote:
> http://bugzilla.kernel.org/show_bug.cgi?id=12195
> 
> 
> 
> 
> 
> ------- Comment #6 from ming.m.lin@intel.com  2008-12-11 18:27 -------
> 2.6.28-rc8 also panic

The blk_mark_rq_complete check should prevent completions from occurring on
already timed out requests unless the interaction previous mentioned between
mpt_fault_reset_work and the scsi eh thread requeue alows the REQ_ATOM_COMPLETE
bit to get cleared prior to the scsi_done being called from
mptscsih_flush_running_cmds. This did not look obvious to hit.

mpt_fault_reset_work
	mpt_HardResetHandler
		mpt_signal_reset
			mptsas_ioc_reset 
				mptscsih_flush_running_cmds
		mpt_do_ioc_recovery

When scsi_times_out is called there should not be a transportt->eh_timed_out,
or hostt->eh_timed_out set for mptsas which should lead to waking up the eh
thread. We will then call mptscsih_abort from the eh thread and it will return
success if the scsi command is not found leading to a possible requeue.

If you have time for another re-create it would be good to set some scsi
logging.
sysctl -w dev.scsi.logging_level=4100 # mlcomplete 1 and error 4
echo "1" > /proc/sys/kernel/sysrq # If needed
echo 9 > /proc/sysrq-trigger # Raise console log level

Also if you have more dmesg output prior to the error from the previous
failure runs that would be good to post also.

-andmike
--
Michael Anderson
andmike@linux.vnet.ibm.com