rbd: introduce image request state machine

Make it possible to schedule image requests on a workqueue.  This fixes
parent chain recursion added in the previous commit and lays the ground
for exclusive lock wait/wake improvements.

The "wait for pending subrequests and report first nonzero result" code
is generalized to be used by object request state machine.
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>
Reviewed-by: NDongsheng Yang <dongsheng.yang@easystack.cn>

drivers/block/rbd.c

@@ -203,6 +203,11 @@ struct rbd_client {
 	struct list_head	node;
 };
 
+struct pending_result {
+	int			result;		/* first nonzero result */
+	int			num_pending;
+};
+
 struct rbd_img_request;
 
 enum obj_request_type {
@@ -295,11 +300,18 @@ enum img_req_flags {
 	IMG_REQ_LAYERED,	/* ENOENT handling: normal = 0, layered = 1 */
 };
 
+enum rbd_img_state {
+	RBD_IMG_START = 1,
+	__RBD_IMG_OBJECT_REQUESTS,
+	RBD_IMG_OBJECT_REQUESTS,
+};
+
 struct rbd_img_request {
 	struct rbd_device	*rbd_dev;
 	enum obj_operation_type	op_type;
 	enum obj_request_type	data_type;
 	unsigned long		flags;
+	enum rbd_img_state	state;
 	union {
 		u64			snap_id;	/* for reads */
 		struct ceph_snap_context *snapc;	/* for writes */
@@ -308,12 +320,13 @@ struct rbd_img_request {
 		struct request		*rq;		/* block request */
 		struct rbd_obj_request	*obj_request;	/* obj req initiator */
 	};
-	spinlock_t		completion_lock;
-	int			result;	/* first nonzero obj_request result */
 
 	struct list_head	object_extents;	/* obj_req.ex structs */
-	u32			pending_count;
 
+	struct mutex		state_mutex;
+	struct pending_result	pending;
+	struct work_struct	work;
+	int			work_result;
 	struct kref		kref;
 };
 
@@ -592,6 +605,23 @@ static int _rbd_dev_v2_snap_features(struct rbd_device *rbd_dev, u64 snap_id,
 		u64 *snap_features);
 
 static void rbd_obj_handle_request(struct rbd_obj_request *obj_req, int result);
+static void rbd_img_handle_request(struct rbd_img_request *img_req, int result);
+
+/*
+ * Return true if nothing else is pending.
+ */
+static bool pending_result_dec(struct pending_result *pending, int *result)
+{
+	rbd_assert(pending->num_pending > 0);
+
+	if (*result && !pending->result)
+		pending->result = *result;
+	if (--pending->num_pending)
+		return false;
+
+	*result = pending->result;
+	return true;
+}
 
 static int rbd_open(struct block_device *bdev, fmode_t mode)
 {
@@ -1350,13 +1380,6 @@ static void rbd_obj_request_put(struct rbd_obj_request *obj_request)
 	kref_put(&obj_request->kref, rbd_obj_request_destroy);
 }
 
-static void rbd_img_request_get(struct rbd_img_request *img_request)
-{
-	dout("%s: img %p (was %d)\n", __func__, img_request,
-	     kref_read(&img_request->kref));
-	kref_get(&img_request->kref);
-}
-
 static void rbd_img_request_destroy(struct kref *kref);
 static void rbd_img_request_put(struct rbd_img_request *img_request)
 {
@@ -1373,7 +1396,6 @@ static inline void rbd_img_obj_request_add(struct rbd_img_request *img_request,
 
 	/* Image request now owns object's original reference */
 	obj_request->img_request = img_request;
-	img_request->pending_count++;
 	dout("%s: img %p obj %p\n", __func__, img_request, obj_request);
 }
 
@@ -1694,8 +1716,8 @@ static struct rbd_img_request *rbd_img_request_create(
 	if (rbd_dev_parent_get(rbd_dev))
 		img_request_layered_set(img_request);
 
-	spin_lock_init(&img_request->completion_lock);
 	INIT_LIST_HEAD(&img_request->object_extents);
+	mutex_init(&img_request->state_mutex);
 	kref_init(&img_request->kref);
 
 	dout("%s: rbd_dev %p %s -> img %p\n", __func__, rbd_dev,
@@ -2061,7 +2083,6 @@ static int __rbd_img_fill_request(struct rbd_img_request *img_req)
 		if (ret < 0)
 			return ret;
 		if (ret > 0) {
-			img_req->pending_count--;
 			rbd_img_obj_request_del(img_req, obj_req);
 			continue;
 		}
@@ -2071,6 +2092,7 @@ static int __rbd_img_fill_request(struct rbd_img_request *img_req)
 			return ret;
 	}
 
+	img_req->state = RBD_IMG_START;
 	return 0;
 }
 
@@ -2359,17 +2381,19 @@ static int rbd_img_fill_from_bvecs(struct rbd_img_request *img_req,
 					 &it);
 }
 
-static void rbd_img_request_submit(struct rbd_img_request *img_request)
+static void rbd_img_handle_request_work(struct work_struct *work)
 {
-	struct rbd_obj_request *obj_request;
+	struct rbd_img_request *img_req =
+	    container_of(work, struct rbd_img_request, work);
 
-	dout("%s: img %p\n", __func__, img_request);
-
-	rbd_img_request_get(img_request);
-	for_each_obj_request(img_request, obj_request)
-		rbd_obj_handle_request(obj_request, 0);
+	rbd_img_handle_request(img_req, img_req->work_result);
+}
 
-	rbd_img_request_put(img_request);
+static void rbd_img_schedule(struct rbd_img_request *img_req, int result)
+{
+	INIT_WORK(&img_req->work, rbd_img_handle_request_work);
+	img_req->work_result = result;
+	queue_work(rbd_wq, &img_req->work);
 }
 
 static int rbd_obj_read_object(struct rbd_obj_request *obj_req)
@@ -2421,7 +2445,8 @@ static int rbd_obj_read_from_parent(struct rbd_obj_request *obj_req)
 		return ret;
 	}
 
-	rbd_img_request_submit(child_img_req);
+	/* avoid parent chain recursion */
+	rbd_img_schedule(child_img_req, 0);
 	return 0;
 }
 
@@ -2756,6 +2781,7 @@ static bool __rbd_obj_handle_request(struct rbd_obj_request *obj_req,
 				     int *result)
 {
 	struct rbd_img_request *img_req = obj_req->img_request;
+	struct rbd_device *rbd_dev = img_req->rbd_dev;
 	bool done;
 
 	mutex_lock(&obj_req->state_mutex);
@@ -2765,59 +2791,113 @@ static bool __rbd_obj_handle_request(struct rbd_obj_request *obj_req,
 		done = rbd_obj_advance_write(obj_req, result);
 	mutex_unlock(&obj_req->state_mutex);
 
+	if (done && *result) {
+		rbd_assert(*result < 0);
+		rbd_warn(rbd_dev, "%s at objno %llu %llu~%llu result %d",
+			 obj_op_name(img_req->op_type), obj_req->ex.oe_objno,
+			 obj_req->ex.oe_off, obj_req->ex.oe_len, *result);
+	}
 	return done;
 }
 
-static void rbd_obj_end_request(struct rbd_obj_request *obj_req, int result)
+/*
+ * This is open-coded in rbd_img_handle_request() to avoid parent chain
+ * recursion.
+ */
+static void rbd_obj_handle_request(struct rbd_obj_request *obj_req, int result)
+{
+	if (__rbd_obj_handle_request(obj_req, &result))
+		rbd_img_handle_request(obj_req->img_request, result);
+}
+
+static void rbd_img_object_requests(struct rbd_img_request *img_req)
 {
-	struct rbd_img_request *img_req = obj_req->img_request;
+	struct rbd_obj_request *obj_req;
 
-	rbd_assert(result <= 0);
-	if (!result)
-		return;
+	rbd_assert(!img_req->pending.result && !img_req->pending.num_pending);
+
+	for_each_obj_request(img_req, obj_req) {
+		int result = 0;
 
-	rbd_warn(img_req->rbd_dev, "%s at objno %llu %llu~%llu result %d",
-		 obj_op_name(img_req->op_type), obj_req->ex.oe_objno,
-		 obj_req->ex.oe_off, obj_req->ex.oe_len, result);
-	if (!img_req->result)
-		img_req->result = result;
+		if (__rbd_obj_handle_request(obj_req, &result)) {
+			if (result) {
+				img_req->pending.result = result;
+				return;
+			}
+		} else {
+			img_req->pending.num_pending++;
+		}
+	}
 }
 
-static void rbd_img_end_request(struct rbd_img_request *img_req)
+static bool rbd_img_advance(struct rbd_img_request *img_req, int *result)
 {
-	rbd_assert(!test_bit(IMG_REQ_CHILD, &img_req->flags));
+again:
+	switch (img_req->state) {
+	case RBD_IMG_START:
+		rbd_assert(!*result);
 
-	blk_mq_end_request(img_req->rq,
-			   errno_to_blk_status(img_req->result));
-	rbd_img_request_put(img_req);
+		rbd_img_object_requests(img_req);
+		if (!img_req->pending.num_pending) {
+			*result = img_req->pending.result;
+			img_req->state = RBD_IMG_OBJECT_REQUESTS;
+			goto again;
+		}
+		img_req->state = __RBD_IMG_OBJECT_REQUESTS;
+		return false;
+	case __RBD_IMG_OBJECT_REQUESTS:
+		if (!pending_result_dec(&img_req->pending, result))
+			return false;
+		/* fall through */
+	case RBD_IMG_OBJECT_REQUESTS:
+		return true;
+	default:
+		BUG();
+	}
 }
 
-static void rbd_obj_handle_request(struct rbd_obj_request *obj_req, int result)
+/*
+ * Return true if @img_req is completed.
+ */
+static bool __rbd_img_handle_request(struct rbd_img_request *img_req,
+				     int *result)
 {
-	struct rbd_img_request *img_req;
+	struct rbd_device *rbd_dev = img_req->rbd_dev;
+	bool done;
 
-again:
-	if (!__rbd_obj_handle_request(obj_req, &result))
-		return;
+	mutex_lock(&img_req->state_mutex);
+	done = rbd_img_advance(img_req, result);
+	mutex_unlock(&img_req->state_mutex);
 
-	img_req = obj_req->img_request;
-	spin_lock(&img_req->completion_lock);
-	rbd_obj_end_request(obj_req, result);
-	rbd_assert(img_req->pending_count);
-	if (--img_req->pending_count) {
-		spin_unlock(&img_req->completion_lock);
-		return;
+	if (done && *result) {
+		rbd_assert(*result < 0);
+		rbd_warn(rbd_dev, "%s%s result %d",
+		      test_bit(IMG_REQ_CHILD, &img_req->flags) ? "child " : "",
+		      obj_op_name(img_req->op_type), *result);
 	}
+	return done;
+}
+
+static void rbd_img_handle_request(struct rbd_img_request *img_req, int result)
+{
+again:
+	if (!__rbd_img_handle_request(img_req, &result))
+		return;
 
-	spin_unlock(&img_req->completion_lock);
-	rbd_assert(img_req->result <= 0);
 	if (test_bit(IMG_REQ_CHILD, &img_req->flags)) {
-		obj_req = img_req->obj_request;
-		result = img_req->result;
+		struct rbd_obj_request *obj_req = img_req->obj_request;
+
 		rbd_img_request_put(img_req);
-		goto again;
+		if (__rbd_obj_handle_request(obj_req, &result)) {
+			img_req = obj_req->img_request;
+			goto again;
+		}
+	} else {
+		struct request *rq = img_req->rq;
+
+		rbd_img_request_put(img_req);
+		blk_mq_end_request(rq, errno_to_blk_status(result));
 	}
-	rbd_img_end_request(img_req);
 }
 
 static const struct rbd_client_id rbd_empty_cid;
@@ -3933,10 +4013,10 @@ static void rbd_queue_workfn(struct work_struct *work)
 	else
 		result = rbd_img_fill_from_bio(img_request, offset, length,
 					       rq->bio);
-	if (result || !img_request->pending_count)
+	if (result)
 		goto err_img_request;
 
-	rbd_img_request_submit(img_request);
+	rbd_img_handle_request(img_request, 0);
 	if (must_be_locked)
 		up_read(&rbd_dev->lock_rwsem);
 	return;