blk-mq: fix for flush deadlock

The flush state machine takes in a struct request, which then is
submitted multiple times to the underling driver.  The old block code
requeses the same request for each of those, so it does not have an
issue with tapping into the request pool.  The new one on the other hand
allocates a new request for each of the actualy steps of the flush
sequence. If have already allocated all of the tags for IO, we will
fail allocating the flush request.

Set aside a reserved request just for flushes.

Signed-off-by: Jens Axboe <axboe@kernel.dk>

4 files changed, 15 insertions, 5 deletions

diff --git a/block/blk-core.c b/block/blk-core.c
index 3bb9e9f7f87e..9677c6525ed8 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1102,7 +1102,7 @@ static struct request *blk_old_get_request(struct request_queue *q, int rw,
 struct request *blk_get_request(struct request_queue *q, int rw, gfp_t gfp_mask)
 {
 	if (q->mq_ops)
-		return blk_mq_alloc_request(q, rw, gfp_mask);
+		return blk_mq_alloc_request(q, rw, gfp_mask, false);
 	else
 		return blk_old_get_request(q, rw, gfp_mask);
 }
diff --git a/block/blk-flush.c b/block/blk-flush.c
index 3e4cc9c7890a..331e627301ea 100644
--- a/block/blk-flush.c
+++ b/block/blk-flush.c
@@ -286,7 +286,7 @@ static void mq_flush_work(struct work_struct *work)
 
 	/* We don't need set REQ_FLUSH_SEQ, it's for consistency */
 	rq = blk_mq_alloc_request(q, WRITE_FLUSH|REQ_FLUSH_SEQ,
-		__GFP_WAIT|GFP_ATOMIC);
+		__GFP_WAIT|GFP_ATOMIC, true);
 	rq->cmd_type = REQ_TYPE_FS;
 	rq->end_io = flush_end_io;
 
diff --git a/block/blk-mq.c b/block/blk-mq.c
index ac804c635040..2dc8de86d0d2 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -210,14 +210,15 @@ static struct request *blk_mq_alloc_request_pinned(struct request_queue *q,
 	return rq;
 }
 
-struct request *blk_mq_alloc_request(struct request_queue *q, int rw, gfp_t gfp)
+struct request *blk_mq_alloc_request(struct request_queue *q, int rw,
+		gfp_t gfp, bool reserved)
 {
 	struct request *rq;
 
 	if (blk_mq_queue_enter(q))
 		return NULL;
 
-	rq = blk_mq_alloc_request_pinned(q, rw, gfp, false);
+	rq = blk_mq_alloc_request_pinned(q, rw, gfp, reserved);
 	blk_mq_put_ctx(rq->mq_ctx);
 	return rq;
 }
@@ -1327,6 +1328,15 @@ struct request_queue *blk_mq_init_queue(struct blk_mq_reg *reg,
 		reg->queue_depth = BLK_MQ_MAX_DEPTH;
 	}
 
+	/*
+	 * Set aside a tag for flush requests.  It will only be used while
+	 * another flush request is in progress but outside the driver.
+	 *
+	 * TODO: only allocate if flushes are supported
+	 */
+	reg->queue_depth++;
+	reg->reserved_tags++;
+
 	if (reg->queue_depth < (reg->reserved_tags + BLK_MQ_TAG_MIN))
 		return ERR_PTR(-EINVAL);
 
diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h
index 3368b97bee73..ab0e9b2025b3 100644
--- a/include/linux/blk-mq.h
+++ b/include/linux/blk-mq.h
@@ -124,7 +124,7 @@ void blk_mq_insert_request(struct request_queue *, struct request *, bool);
 void blk_mq_run_queues(struct request_queue *q, bool async);
 void blk_mq_free_request(struct request *rq);
 bool blk_mq_can_queue(struct blk_mq_hw_ctx *);
-struct request *blk_mq_alloc_request(struct request_queue *q, int rw, gfp_t gfp);
+struct request *blk_mq_alloc_request(struct request_queue *q, int rw, gfp_t gfp, bool reserved);
 struct request *blk_mq_alloc_reserved_request(struct request_queue *q, int rw, gfp_t gfp);
 struct request *blk_mq_rq_from_tag(struct request_queue *q, unsigned int tag);