Re: [PATCH 06/16] virtio_blk: drop config_enable

[Cornelia Huck <cornelia.huck@de.ibm.com>]

On Mon, 6 Oct 2014 15:31:10 +0300
"Michael S. Tsirkin" <mst@redhat.com> wrote:

> On Mon, Oct 06, 2014 at 02:20:38PM +0200, Cornelia Huck wrote:
> > On Mon, 6 Oct 2014 15:09:53 +0300
> > "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > 
> > > On Mon, Oct 06, 2014 at 01:42:29PM +0200, Cornelia Huck wrote:
> > > > On Sun, 5 Oct 2014 19:07:07 +0300
> > > > "Michael S. Tsirkin" <mst@redhat.com> wrote:
> > > > 
> > > > > Now that virtio core ensures config changes don't
> > > > > arrive during probing, drop config_enable flag
> > > > > in virtio blk.
> > > > > On removal, flush is now sufficient to guarantee that
> > > > > no change work is queued.
> > > > > 
> > > > > This help simplify the driver, and will allow
> > > > > setting DRIVER_OK earlier without losing config
> > > > > change notifications.
> > > > > 
> > > > > Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> > > > > ---
> > > > >  drivers/block/virtio_blk.c | 19 ++-----------------
> > > > >  1 file changed, 2 insertions(+), 17 deletions(-)
> > > > > 
> > > > > diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
> > > > > index 0a58140..c8cf6a1 100644
> > > > > --- a/drivers/block/virtio_blk.c
> > > > > +++ b/drivers/block/virtio_blk.c
> > > > 
> > > > > @@ -772,9 +766,7 @@ static void virtblk_remove(struct virtio_device *vdev)
> > > > >  	int refc;
> > > > > 
> > > > >  	/* Prevent config work handler from accessing the device. */
> > > > 
> > > > /* Common code ensures no further work will be queued. */
> > > > 
> > > > instead?
> > > 
> > > No, I think you missed the point:
> > > this comment now refers to the flush below: flush is required to
> > > ensure work handler is not running.
> > > 
> > > Agree?
> > 
> > I think we both mean the same thing.
> > 
> > Preventing the handler from access sounds to me more like "when the
> > handler starts running, it is prevented from accessing the
> > device" (like with setting config_enable, as the code did before). What
> > I meant was "common code has already ensured that our work-queueing
> > function will not be called, therefore flushing the workqueue is
> > enough."
> > 
> > (same for net)
> 
> OK so I'll rewrite this to
> 	/* Make sure no work handler is accessing the device. */
> ?
> I prefer not duplicating core guarantees in all devices.
> 

Sounds good to me.