X-Git-Url: http://git.cascardo.info/?a=blobdiff_plain;f=block%2Fblk-mq.c;h=1fa9dd153fde22a976483ef3b5da3ec39f0f458e;hb=ac1d8590d3ae855d747b12c9e85dfb40bda709e4;hp=cee96234bf589393c464b491ea27a2495b1d889e;hpb=14ec77f352cb00ab8425ec2af03bd7e529eefe24;p=cascardo%2Flinux.git

diff --git a/block/blk-mq.c b/block/blk-mq.c
index cee96234bf58..1fa9dd153fde 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -194,27 +194,9 @@ static void blk_mq_rq_ctx_init(struct request_queue *q, struct blk_mq_ctx *ctx,
 }
 
 static struct request *__blk_mq_alloc_request(struct blk_mq_hw_ctx *hctx,
-					      gfp_t gfp, bool reserved,
-					      int rw)
+					      gfp_t gfp, bool reserved)
 {
-	struct request *req;
-	bool is_flush = false;
-	/*
-	 * flush need allocate a request, leave at least one request for
-	 * non-flush IO to avoid deadlock
-	 */
-	if ((rw & REQ_FLUSH) && !(rw & REQ_FLUSH_SEQ)) {
-		if (atomic_inc_return(&hctx->pending_flush) >=
-		    hctx->queue_depth - hctx->reserved_tags - 1) {
-			atomic_dec(&hctx->pending_flush);
-			return NULL;
-		}
-		is_flush = true;
-	}
-	req = blk_mq_alloc_rq(hctx, gfp, reserved);
-	if (!req && is_flush)
-		atomic_dec(&hctx->pending_flush);
-	return req;
+	return blk_mq_alloc_rq(hctx, gfp, reserved);
 }
 
 static struct request *blk_mq_alloc_request_pinned(struct request_queue *q,
@@ -227,7 +209,7 @@ static struct request *blk_mq_alloc_request_pinned(struct request_queue *q,
 		struct blk_mq_ctx *ctx = blk_mq_get_ctx(q);
 		struct blk_mq_hw_ctx *hctx = q->mq_ops->map_queue(q, ctx->cpu);
 
-		rq = __blk_mq_alloc_request(hctx, gfp & ~__GFP_WAIT, reserved, rw);
+		rq = __blk_mq_alloc_request(hctx, gfp & ~__GFP_WAIT, reserved);
 		if (rq) {
 			blk_mq_rq_ctx_init(q, ctx, rq, rw);
 			break;
@@ -244,15 +226,14 @@ 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, bool reserved)
+struct request *blk_mq_alloc_request(struct request_queue *q, int rw, gfp_t gfp)
 {
 	struct request *rq;
 
 	if (blk_mq_queue_enter(q))
 		return NULL;
 
-	rq = blk_mq_alloc_request_pinned(q, rw, gfp, reserved);
+	rq = blk_mq_alloc_request_pinned(q, rw, gfp, false);
 	if (rq)
 		blk_mq_put_ctx(rq->mq_ctx);
 	return rq;
@@ -276,7 +257,7 @@ EXPORT_SYMBOL(blk_mq_alloc_reserved_request);
 /*
  * Re-init and set pdu, if we have it
  */
-static void blk_mq_rq_init(struct blk_mq_hw_ctx *hctx, struct request *rq)
+void blk_mq_rq_init(struct blk_mq_hw_ctx *hctx, struct request *rq)
 {
 	blk_rq_init(hctx->queue, rq);
 
@@ -290,9 +271,6 @@ static void __blk_mq_free_request(struct blk_mq_hw_ctx *hctx,
 	const int tag = rq->tag;
 	struct request_queue *q = rq->q;
 
-	if ((rq->cmd_flags & REQ_FLUSH) && !(rq->cmd_flags & REQ_FLUSH_SEQ))
-		atomic_dec(&hctx->pending_flush);
-
 	blk_mq_rq_init(hctx, rq);
 	blk_mq_put_tag(hctx->tags, tag);
 
@@ -326,7 +304,7 @@ static void blk_mq_bio_endio(struct request *rq, struct bio *bio, int error)
 		bio_endio(bio, error);
 }
 
-void blk_mq_complete_request(struct request *rq, int error)
+void blk_mq_end_io(struct request *rq, int error)
 {
 	struct bio *bio = rq->bio;
 	unsigned int bytes = 0;
@@ -351,48 +329,55 @@ void blk_mq_complete_request(struct request *rq, int error)
 	else
 		blk_mq_free_request(rq);
 }
+EXPORT_SYMBOL(blk_mq_end_io);
 
-void __blk_mq_end_io(struct request *rq, int error)
-{
-	if (!blk_mark_rq_complete(rq))
-		blk_mq_complete_request(rq, error);
-}
-
-static void blk_mq_end_io_remote(void *data)
+static void __blk_mq_complete_request_remote(void *data)
 {
 	struct request *rq = data;
 
-	__blk_mq_end_io(rq, rq->errors);
+	rq->q->softirq_done_fn(rq);
 }
 
-/*
- * End IO on this request on a multiqueue enabled driver. We'll either do
- * it directly inline, or punt to a local IPI handler on the matching
- * remote CPU.
- */
-void blk_mq_end_io(struct request *rq, int error)
+void __blk_mq_complete_request(struct request *rq)
 {
 	struct blk_mq_ctx *ctx = rq->mq_ctx;
 	int cpu;
 
-	if (!ctx->ipi_redirect)
-		return __blk_mq_end_io(rq, error);
+	if (!ctx->ipi_redirect) {
+		rq->q->softirq_done_fn(rq);
+		return;
+	}
 
 	cpu = get_cpu();
 	if (cpu != ctx->cpu && cpu_online(ctx->cpu)) {
-		rq->errors = error;
-		rq->csd.func = blk_mq_end_io_remote;
+		rq->csd.func = __blk_mq_complete_request_remote;
 		rq->csd.info = rq;
 		rq->csd.flags = 0;
 		__smp_call_function_single(ctx->cpu, &rq->csd, 0);
 	} else {
-		__blk_mq_end_io(rq, error);
+		rq->q->softirq_done_fn(rq);
 	}
 	put_cpu();
 }
-EXPORT_SYMBOL(blk_mq_end_io);
 
-static void blk_mq_start_request(struct request *rq)
+/**
+ * blk_mq_complete_request - end I/O on a request
+ * @rq:		the request being processed
+ *
+ * Description:
+ *	Ends all I/O on a request. It does not handle partial completions.
+ *	The actual completion happens out-of-order, through a IPI handler.
+ **/
+void blk_mq_complete_request(struct request *rq)
+{
+	if (unlikely(blk_should_fake_timeout(rq->q)))
+		return;
+	if (!blk_mark_rq_complete(rq))
+		__blk_mq_complete_request(rq);
+}
+EXPORT_SYMBOL(blk_mq_complete_request);
+
+static void blk_mq_start_request(struct request *rq, bool last)
 {
 	struct request_queue *q = rq->q;
 
@@ -405,6 +390,25 @@ static void blk_mq_start_request(struct request *rq)
 	 */
 	rq->deadline = jiffies + q->rq_timeout;
 	set_bit(REQ_ATOM_STARTED, &rq->atomic_flags);
+
+	if (q->dma_drain_size && blk_rq_bytes(rq)) {
+		/*
+		 * Make sure space for the drain appears.  We know we can do
+		 * this because max_hw_segments has been adjusted to be one
+		 * fewer than the device can handle.
+		 */
+		rq->nr_phys_segments++;
+	}
+
+	/*
+	 * Flag the last request in the series so that drivers know when IO
+	 * should be kicked off, if they don't do it on a per-request basis.
+	 *
+	 * Note: the flag isn't the only condition drivers should do kick off.
+	 * If drive is busy, the last request might not have the bit set.
+	 */
+	if (last)
+		rq->cmd_flags |= REQ_END;
 }
 
 static void blk_mq_requeue_request(struct request *rq)
@@ -413,6 +417,11 @@ static void blk_mq_requeue_request(struct request *rq)
 
 	trace_block_rq_requeue(q, rq);
 	clear_bit(REQ_ATOM_STARTED, &rq->atomic_flags);
+
+	rq->cmd_flags &= ~REQ_END;
+
+	if (q->dma_drain_size && blk_rq_bytes(rq))
+		rq->nr_phys_segments--;
 }
 
 struct blk_mq_timeout_data {
@@ -580,29 +589,8 @@ static void __blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx)
 
 		rq = list_first_entry(&rq_list, struct request, queuelist);
 		list_del_init(&rq->queuelist);
-		blk_mq_start_request(rq);
-
-		if (q->dma_drain_size && blk_rq_bytes(rq)) {
-			/*
-			 * make sure space for the drain appears we
-			 * know we can do this because max_hw_segments
-			 * has been adjusted to be one fewer than the
-			 * device can handle
-			 */
-			rq->nr_phys_segments++;
-		}
 
-		/*
-		 * Last request in the series. Flag it as such, this
-		 * enables drivers to know when IO should be kicked off,
-		 * if they don't do it on a per-request basis.
-		 *
-		 * Note: the flag isn't the only condition drivers
-		 * should do kick off. If drive is busy, the last
-		 * request might not have the bit set.
-		 */
-		if (list_empty(&rq_list))
-			rq->cmd_flags |= REQ_END;
+		blk_mq_start_request(rq, list_empty(&rq_list));
 
 		ret = q->mq_ops->queue_rq(hctx, rq);
 		switch (ret) {
@@ -620,8 +608,8 @@ static void __blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx)
 			break;
 		default:
 			pr_err("blk-mq: bad return on queue: %d\n", ret);
-			rq->errors = -EIO;
 		case BLK_MQ_RQ_QUEUE_ERROR:
+			rq->errors = -EIO;
 			blk_mq_end_io(rq, rq->errors);
 			break;
 		}
@@ -939,14 +927,14 @@ static void blk_mq_make_request(struct request_queue *q, struct bio *bio)
 	hctx = q->mq_ops->map_queue(q, ctx->cpu);
 
 	trace_block_getrq(q, bio, rw);
-	rq = __blk_mq_alloc_request(hctx, GFP_ATOMIC, false, bio->bi_rw);
+	rq = __blk_mq_alloc_request(hctx, GFP_ATOMIC, false);
 	if (likely(rq))
-		blk_mq_rq_ctx_init(q, ctx, rq, bio->bi_rw);
+		blk_mq_rq_ctx_init(q, ctx, rq, rw);
 	else {
 		blk_mq_put_ctx(ctx);
 		trace_block_sleeprq(q, bio, rw);
-		rq = blk_mq_alloc_request_pinned(q, bio->bi_rw,
-				__GFP_WAIT|GFP_ATOMIC, false);
+		rq = blk_mq_alloc_request_pinned(q, rw, __GFP_WAIT|GFP_ATOMIC,
+							false);
 		ctx = rq->mq_ctx;
 		hctx = q->mq_ops->map_queue(q, ctx->cpu);
 	}
@@ -1223,9 +1211,7 @@ static int blk_mq_init_hw_queues(struct request_queue *q,
 		hctx->queue_num = i;
 		hctx->flags = reg->flags;
 		hctx->queue_depth = reg->queue_depth;
-		hctx->reserved_tags = reg->reserved_tags;
 		hctx->cmd_size = reg->cmd_size;
-		atomic_set(&hctx->pending_flush, 0);
 
 		blk_mq_init_cpu_notifier(&hctx->cpu_notifier,
 						blk_mq_hctx_notify, hctx);
@@ -1399,12 +1385,20 @@ struct request_queue *blk_mq_init_queue(struct blk_mq_reg *reg,
 	if (reg->timeout)
 		blk_queue_rq_timeout(q, reg->timeout);
 
+	if (reg->ops->complete)
+		blk_queue_softirq_done(q, reg->ops->complete);
+
 	blk_mq_init_flush(q);
 	blk_mq_init_cpu_queues(q, reg->nr_hw_queues);
 
-	if (blk_mq_init_hw_queues(q, reg, driver_data))
+	q->flush_rq = kzalloc(round_up(sizeof(struct request) + reg->cmd_size,
+				cache_line_size()), GFP_KERNEL);
+	if (!q->flush_rq)
 		goto err_hw;
 
+	if (blk_mq_init_hw_queues(q, reg, driver_data))
+		goto err_flush_rq;
+
 	blk_mq_map_swqueue(q);
 
 	mutex_lock(&all_q_mutex);
@@ -1412,6 +1406,9 @@ struct request_queue *blk_mq_init_queue(struct blk_mq_reg *reg,
 	mutex_unlock(&all_q_mutex);
 
 	return q;
+
+err_flush_rq:
+	kfree(q->flush_rq);
 err_hw:
 	kfree(q->mq_map);
 err_map: