On Fri, Sep 22, 2017 at 03:14:05PM -0700, Bart Van Assche wrote: > Avoid that it can take 200 ms too long to wait for requests to > finish. Note: blk_mq_freeze_queue() uses a wait queue to wait > for ongoing requests to finish. > > Signed-off-by: Bart Van Assche <[email protected]> > Cc: Martin K. Petersen <[email protected]> > Cc: Ming Lei <[email protected]> > Cc: Christoph Hellwig <[email protected]> > Cc: Hannes Reinecke <[email protected]> > Cc: Johannes Thumshirn <[email protected]> > --- > drivers/scsi/scsi_lib.c | 16 +++++++++++++--- > 1 file changed, 13 insertions(+), 3 deletions(-) > > diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c > index ca84fd2d93ea..c5e66bb0c94d 100644 > --- a/drivers/scsi/scsi_lib.c > +++ b/drivers/scsi/scsi_lib.c > @@ -2939,10 +2939,20 @@ scsi_device_quiesce(struct scsi_device *sdev) > if (err) > return err; > > - scsi_run_queue(q); > - while (atomic_read(&sdev->device_busy)) { > - msleep_interruptible(200); > + if (q->mq_ops) { > + /* > + * Ensure write order of the PREEMPT_ONLY queue flag and > + * q_usage_counter. See also blk_queue_enter(). > + */ > + smp_wmb(); > + blk_mq_freeze_queue(q); > + blk_mq_unfreeze_queue(q); > + } else { > scsi_run_queue(q); > + while (atomic_read(&sdev->device_busy)) { > + msleep_interruptible(200); > + scsi_run_queue(q); > + }
Again, your patch doesn't fix scsi quiesce issue on block legacy, either on suspend path or revalidate path of transport spi. -- Ming
