On Tue, 2017-10-17 at 08:43 +0800, Ming Lei wrote:
> On Mon, Oct 16, 2017 at 04:29:04PM -0700, Bart Van Assche wrote:
> > [ ... ]
> > int
> > scsi_device_quiesce(struct scsi_device *sdev)
> > {
> > + struct request_queue *q = sdev->request_queue;
> > int err;
> >
> > + /*
> > + * It is allowed to call scsi_device_quiesce() multiple times from
> > + * the same context but concurrent scsi_device_quiesce() calls are
> > + * not allowed.
> > + */
> > + WARN_ON_ONCE(sdev->quiesced_by && sdev->quiesced_by != current);
>
> If the above line and comment is removed, everything is still fine,
> either nested calling or concurrent calling, isn't it?
If scsi_device_quiesce() is called concurrently from two different threads
then the first scsi_device_resume() call will resume I/O for *both* contexts.
That's not what the callers expect. If scsi_device_quiesce() and
scsi_device_resume() are called concurrently that would be even worse. I think
we *really* should know whether callers serialize scsi_device_quiesce() and
scsi_device_resume() calls properly. Hence the WARN_ON_ONCE() statement.
Bart.
