block: recursive merge requests

In a workload, thread 1 accesses a, a+2, ..., thread 2 accesses a+1, a+3,....
When the requests are flushed to queue, a and a+1 are merged to (a, a+1), a+2
and a+3 too to (a+2, a+3), but (a, a+1) and (a+2, a+3) aren't merged.

If we do recursive merge for such interleave access, some workloads throughput
get improvement. A recent worload I'm checking on is swap, below change
boostes the throughput around 5% ~ 10%.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

block/elevator.c

@@ -458,6 +458,7 @@ static bool elv_attempt_insert_merge(struct request_queue *q,
 				     struct request *rq)
 {
 	struct request *__rq;
+	bool ret;
 
 	if (blk_queue_nomerges(q))
 		return false;
@@ -471,14 +472,21 @@ static bool elv_attempt_insert_merge(struct request_queue *q,
 	if (blk_queue_noxmerges(q))
 		return false;
 
+	ret = false;
 	/*
 	 * See if our hash lookup can find a potential backmerge.
 	 */
-	__rq = elv_rqhash_find(q, blk_rq_pos(rq));
-	if (__rq && blk_attempt_req_merge(q, __rq, rq))
-		return true;
+	while (1) {
+		__rq = elv_rqhash_find(q, blk_rq_pos(rq));
+		if (!__rq || !blk_attempt_req_merge(q, __rq, rq))
+			break;
 
-	return false;
+		/* The merged request could be merged with others, try again */
+		ret = true;
+		rq = __rq;
+	}
+
+	return ret;
 }
 
 void elv_merged_request(struct request_queue *q, struct request *rq, int type)