rbd: only get snap context for write requests

The rbd_device->exists field can be updated asynchronously, changing from set to clear if a mapped snapshot disappears from the base image's snapshot context.

Currently, value of the "exists" flag is only read and modified under protection of the header semaphore, but that will change with the next patch. Making it atomic ensures this won't be a problem because the a the non-existence of device will be immediately known.

Signed-off-by Alex Elder <[email protected]>

---

  drivers/block/rbd.c |    9 +++++----
  1 file changed, 5 insertions(+), 4 deletions(-)

drivers/block/rbd.c +2 -2

index 1348825..a3b0d43 100644
@@ -229,7 +229,7 @@ struct rbd_device {
  	spinlock_t		lock;		/* queue lock */
 
  	struct rbd_image_header	header;
-	bool                    exists;
+	atomic_t		exists;
  	struct rbd_spec		*spec;
 
  	char			*header_name;

@@ -751,7 +751,7 @@ static int rbd_dev_set_mapping(struct rbd_device
 *rbd_dev)
  			goto done;
  		rbd_dev->mapping.read_only = true;
  	}
-	rbd_dev->exists = true;
+	atomic_set(&rbd_dev->exists, 1);
  done:
  	return ret;

  }

drivers/block/rbd.c +2 -1

@@ -1670,7 +1670,7 @@ static void rbd_rq_fn(struct request_queue *q)
  		/* Grab a reference to the snapshot context */
 
  		down_read(&rbd_dev->header_rwsem);
-		if (rbd_dev->exists) {
+		if (atomic_read(&rbd_dev->exists)) {
  			snapc = ceph_get_snap_context(rbd_dev->header.snapc);
  			rbd_assert(snapc != NULL);
  		}

@@ -2292,6 +2292,7 @@ struct rbd_device *rbd_dev_create(struct
 rbd_client *rbdc,
  		return NULL;
 
  	spin_lock_init(&rbd_dev->lock);
+	atomic_set(&rbd_dev->exists, 0);
  	INIT_LIST_HEAD(&rbd_dev->node);
  	INIT_LIST_HEAD(&rbd_dev->snaps);

  	init_rwsem(&rbd_dev->header_rwsem);

drivers/block/rbd.c +1 -1

@@ -2916,7 +2917,7 @@ static int rbd_dev_snaps_update(struct rbd_device
 *rbd_dev)
  			/* Existing snapshot not in the new snap context */
 
  			if (rbd_dev->spec->snap_id == snap->id)
-				rbd_dev->exists = false;
+				atomic_set(&rbd_dev->exists, 0);
  			rbd_remove_snap_dev(snap);
  			dout("%ssnap id %llu has been removed\n",

rbd_dev->spec->snap_id == snap->id ?

Right now we get the snapshot context for an rbd image (under protection of the header semaphore) for every request processed.

There's no need to get the snap context if we're doing a read, so avoid doing so in that case.

Note that we no longer need to hold the header semaphore to check the rbd_dev's existence flag.

Signed-off-by Alex Elder <[email protected]>

---

  drivers/block/rbd.c |   36 ++++++++++++++++++++----------------
  1 file changed, 20 insertions(+), 16 deletions(-)

drivers/block/rbd.c +20 -16

index a3b0d43..fd6a708 100644
@@ -1331,7 +1331,7 @@ static int rbd_do_op(struct request *rq,
  	} else {
  		opcode = CEPH_OSD_OP_READ;
  		flags = CEPH_OSD_FLAG_READ;
-		snapc = NULL;
+		rbd_assert(!snapc);
  		snapid = rbd_dev->spec->snap_id;
  		payload_len = 0;
  	}

@@ -1661,22 +1661,25 @@ static void rbd_rq_fn(struct request_queue *q)
  		}
  		spin_unlock_irq(q->queue_lock);
 
-		/* Stop writes to a read-only device */
-
-		result = -EROFS;
-		if (read_only && rq_data_dir(rq) == WRITE)
-			goto out_end_request;
-
-		/* Grab a reference to the snapshot context */
-
-		down_read(&rbd_dev->header_rwsem);
-		if (atomic_read(&rbd_dev->exists)) {
+		/* Write requests need a reference to the snapshot context */
+
+		if (rq_data_dir(rq) == WRITE) {
+			result = -EROFS;
+			if (read_only) /* Can't write to a read-only device */
+				goto out_end_request;
+
+			/*
+			 * Note that each osd request will take its
+			 * own reference to the snapshot context
+			 * supplied.  The reference we take here
+			 * just guarantees the one we provide stays
+			 * valid.
+			 */
+			down_read(&rbd_dev->header_rwsem);
  			snapc = ceph_get_snap_context(rbd_dev->header.snapc);
+			up_read(&rbd_dev->header_rwsem);
  			rbd_assert(snapc != NULL);
-		}
-		up_read(&rbd_dev->header_rwsem);
-
-		if (!snapc) {
+		} else if (!atomic_read(&rbd_dev->exists)) {
  			rbd_assert(rbd_dev->spec->snap_id != CEPH_NOSNAP);
  			dout("request for non-existent snapshot");
  			result = -ENXIO;

@@ -1687,7 +1690,8 @@ static void rbd_rq_fn(struct request_queue *q)
  				blk_rq_pos(rq) * SECTOR_SIZE,
  				blk_rq_bytes(rq), rq->bio);
  out_end_request:
-		ceph_put_snap_context(snapc);
+		if (snapc)
+			ceph_put_snap_context(snapc);
  		spin_lock_irq(q->queue_lock);
  		if (result < 0)
  			__blk_end_request_all(rq, result);