On Mon, 2008-10-27 at 10:51 +0900, Tejun Heo wrote:
OK, so if we take the first trace as definitive, since it shows the same
command going around twice (being freed and reallocated from the slab in
between), it does tend to imply the one on eh_cmd_q is bogus.
Could you print out all scsi commands going into ata_qc_issue so we can
see if there's a clue in where this one is coming from?
Well, it was worth a shot.