foonvme-passthru-wip

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

4 files changed, 33 insertions, 24 deletions

diff --git a/block/blk-exec.c b/block/blk-exec.c
index 5fa3d735b9c8..44c4a2ffa708 100644
--- a/block/blk-exec.c
+++ b/block/blk-exec.c
@@ -66,7 +66,7 @@ EXPORT_SYMBOL_GPL(__blk_execute_rq_nowait);
 void blk_execute_rq_nowait(struct gendisk *bd_disk, struct request *rq,
 			   int at_head, rq_end_io_fn *done)
 {
-	__blk_execute_rq_nowait(bd_disk, rq, false, true, done);
+	__blk_execute_rq_nowait(bd_disk, rq, at_head, true, done);
 }
 EXPORT_SYMBOL_GPL(blk_execute_rq_nowait);
 
diff --git a/block/blk-mq.c b/block/blk-mq.c
index ce94adc5118f..9ef6f7bb3ca3 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1840,6 +1840,20 @@ void __blk_mq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
 	blk_mq_hctx_mark_pending(hctx, ctx);
 }
 
+static void blk_add_rq_to_plug(struct blk_plug *plug, struct request *rq)
+{
+	list_add_tail(&rq->queuelist, &plug->mq_list);
+	plug->rq_count++;
+	if (!plug->multiple_queues && !list_is_singular(&plug->mq_list)) {
+		struct request *tmp;
+
+		tmp = list_first_entry(&plug->mq_list, struct request,
+						queuelist);
+		if (tmp->q != rq->q)
+			plug->multiple_queues = true;
+	}
+}
+
 /**
  * blk_mq_request_bypass_insert - Insert a request at dispatch list.
  * @rq: Pointer to request to be inserted.
@@ -1853,16 +1867,21 @@ void blk_mq_request_bypass_insert(struct request *rq, bool at_head,
 				  bool run_queue)
 {
 	struct blk_mq_hw_ctx *hctx = rq->mq_hctx;
+	struct blk_plug *plug = current->plug;
 
-	spin_lock(&hctx->lock);
-	if (at_head)
-		list_add(&rq->queuelist, &hctx->dispatch);
-	else
-		list_add_tail(&rq->queuelist, &hctx->dispatch);
-	spin_unlock(&hctx->lock);
+	if (plug) {
+		blk_add_rq_to_plug(plug, rq);
+	} else {
+		spin_lock(&hctx->lock);
+		if (at_head)
+			list_add(&rq->queuelist, &hctx->dispatch);
+		else
+			list_add_tail(&rq->queuelist, &hctx->dispatch);
+		spin_unlock(&hctx->lock);
 
-	if (run_queue)
-		blk_mq_run_hw_queue(hctx, false);
+		if (run_queue)
+			blk_mq_run_hw_queue(hctx, false);
+	}
 }
 
 void blk_mq_insert_requests(struct blk_mq_hw_ctx *hctx, struct blk_mq_ctx *ctx,
@@ -2121,20 +2140,6 @@ void blk_mq_try_issue_list_directly(struct blk_mq_hw_ctx *hctx,
 		hctx->queue->mq_ops->commit_rqs(hctx);
 }
 
-static void blk_add_rq_to_plug(struct blk_plug *plug, struct request *rq)
-{
-	list_add_tail(&rq->queuelist, &plug->mq_list);
-	plug->rq_count++;
-	if (!plug->multiple_queues && !list_is_singular(&plug->mq_list)) {
-		struct request *tmp;
-
-		tmp = list_first_entry(&plug->mq_list, struct request,
-						queuelist);
-		if (tmp->q != rq->q)
-			plug->multiple_queues = true;
-	}
-}
-
 /*
  * Allow 4x BLK_MAX_REQUEST_COUNT requests on plug queue for multiple
  * queues. This is important for md arrays to benefit from merging
@@ -2278,6 +2283,7 @@ blk_qc_t blk_mq_submit_bio(struct bio *bio)
 		}
 	} else if ((q->nr_hw_queues > 1 && is_sync) ||
 			!data.hctx->dispatch_busy) {
+
 		/*
 		 * There is no scheduler and we can try to send directly
 		 * to the hardware.
diff --git a/drivers/nvme/host/ioctl.c b/drivers/nvme/host/ioctl.c
index 0d93a79900e4..7040090a745a 100644
--- a/drivers/nvme/host/ioctl.c
+++ b/drivers/nvme/host/ioctl.c
@@ -171,13 +171,14 @@ int nvme_rq_map_user_fixedb(struct request_queue *q, struct request *rq,
 		return ret;
 	iter_count = iov_iter_count(&iter);
 	nr_segs = iter.nr_segs;
+	printk("segs here %d\n", (int)nr_segs);
 
 	if (!iter_count || (iter_count >> 9) > queue_max_hw_sectors(q))
 		return -EINVAL;
 	if (nr_segs > queue_max_segments(q))
 		return -EINVAL;
 	/* no iovecs to alloc, as we already have a BVEC iterator */
-	bio = bio_kmalloc(gfp_mask, 0);
+	bio = bio_alloc(gfp_mask, 0);
 	if (!bio)
 		return -ENOMEM;
 
diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index b82492cd7503..69af7555ff30 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -857,6 +857,8 @@ static blk_status_t nvme_map_data(struct nvme_dev *dev, struct request *req,
 		}
 	}
 
+	if (!strncmp(current->comm, "fio", 3))
+		printk("segs %d\n", blk_rq_nr_phys_segments(req));
 	iod->dma_len = 0;
 	iod->sg = mempool_alloc(dev->iod_mempool, GFP_ATOMIC);
 	if (!iod->sg)