Although qla2xxx driver is calling schedule_{,delayed}_work() from 10 locations, I assume that flush_scheduled_work() from qlt_stop_phase1() needs to flush only works scheduled by qlt_sched_sess_work(), for this loop continues while "struct qla_tgt"->sess_works_list is not empty. Signed-off-by: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx> --- This patch is only compile tested. Changes in v4: Repost as a new thread. Changes in v3: Use flush_work(). Changes in v2: Use per "struct qla_tgt" workqueue. drivers/scsi/qla2xxx/qla_target.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/scsi/qla2xxx/qla_target.c b/drivers/scsi/qla2xxx/qla_target.c index 9013c162d4aa..b0cb463cf032 100644 --- a/drivers/scsi/qla2xxx/qla_target.c +++ b/drivers/scsi/qla2xxx/qla_target.c @@ -1559,7 +1559,7 @@ int qlt_stop_phase1(struct qla_tgt *tgt) spin_lock_irqsave(&tgt->sess_work_lock, flags); do { spin_unlock_irqrestore(&tgt->sess_work_lock, flags); - flush_scheduled_work(); + flush_work(&tgt->sess_work); spin_lock_irqsave(&tgt->sess_work_lock, flags); } while (!list_empty(&tgt->sess_works_list)); spin_unlock_irqrestore(&tgt->sess_work_lock, flags); -- 2.18.4