Merge branch 'for-linus' of git://git.kernel.dk/linux-block

Pull block fixes from Jens Axboe:
 "Unfortunately a few issues that warrant sending another pull request,
  even if I had hoped to avoid it. This contains:

   - A fix for multiqueue xen-blkback, on tear down / disconnect.

   - A few fixups for NVMe, including a wrong bit definition, fix for
     host memory buffers, and an nvme rdma page size fix"

* 'for-linus' of git://git.kernel.dk/linux-block:
  nvme: fix the definition of the doorbell buffer config support bit
  nvme-pci: use dma memory for the host memory buffer descriptors
  nvme-rdma: default MR page size to 4k
  xen-blkback: stop blkback thread of every queue in xen_blkif_disconnect

drivers/block/xen-blkback/xenbus.c

@@ -244,6 +244,7 @@ static int xen_blkif_disconnect(struct xen_blkif *blkif)
 {
 	struct pending_req *req, *n;
 	unsigned int j, r;
+	bool busy = false;
 
 	for (r = 0; r < blkif->nr_rings; r++) {
 		struct xen_blkif_ring *ring = &blkif->rings[r];
@@ -261,8 +262,10 @@ static int xen_blkif_disconnect(struct xen_blkif *blkif)
 		 * don't have any discard_io or other_io requests. So, checking
 		 * for inflight IO is enough.
 		 */
-		if (atomic_read(&ring->inflight) > 0)
-			return -EBUSY;
+		if (atomic_read(&ring->inflight) > 0) {
+			busy = true;
+			continue;
+		}
 
 		if (ring->irq) {
 			unbind_from_irqhandler(ring->irq, ring);
@@ -300,6 +303,9 @@ static int xen_blkif_disconnect(struct xen_blkif *blkif)
 		WARN_ON(i != (XEN_BLKIF_REQS_PER_PAGE * blkif->nr_ring_pages));
 		ring->active = false;
 	}
+	if (busy)
+		return -EBUSY;
+
 	blkif->nr_ring_pages = 0;
 	/*
 	 * blkif->rings was allocated in connect_ring, so we should free it in

drivers/nvme/host/pci.c

@@ -109,6 +109,7 @@ struct nvme_dev {
 	/* host memory buffer support: */
 	u64 host_mem_size;
 	u32 nr_host_mem_descs;
+	dma_addr_t host_mem_descs_dma;
 	struct nvme_host_mem_buf_desc *host_mem_descs;
 	void **host_mem_desc_bufs;
 };
@@ -1565,16 +1566,10 @@ static inline void nvme_release_cmb(struct nvme_dev *dev)
 
 static int nvme_set_host_mem(struct nvme_dev *dev, u32 bits)
 {
-	size_t len = dev->nr_host_mem_descs * sizeof(*dev->host_mem_descs);
+	u64 dma_addr = dev->host_mem_descs_dma;
 	struct nvme_command c;
-	u64 dma_addr;
 	int ret;
 
-	dma_addr = dma_map_single(dev->dev, dev->host_mem_descs, len,
-			DMA_TO_DEVICE);
-	if (dma_mapping_error(dev->dev, dma_addr))
-		return -ENOMEM;
-
 	memset(&c, 0, sizeof(c));
 	c.features.opcode	= nvme_admin_set_features;
 	c.features.fid		= cpu_to_le32(NVME_FEAT_HOST_MEM_BUF);
@@ -1591,7 +1586,6 @@ static int nvme_set_host_mem(struct nvme_dev *dev, u32 bits)
 			 "failed to set host mem (err %d, flags %#x).\n",
 			 ret, bits);
 	}
-	dma_unmap_single(dev->dev, dma_addr, len, DMA_TO_DEVICE);
 	return ret;
 }
 
@@ -1609,7 +1603,9 @@ static void nvme_free_host_mem(struct nvme_dev *dev)
 
 	kfree(dev->host_mem_desc_bufs);
 	dev->host_mem_desc_bufs = NULL;
-	kfree(dev->host_mem_descs);
+	dma_free_coherent(dev->dev,
+			dev->nr_host_mem_descs * sizeof(*dev->host_mem_descs),
+			dev->host_mem_descs, dev->host_mem_descs_dma);
 	dev->host_mem_descs = NULL;
 }
 
@@ -1617,6 +1613,7 @@ static int nvme_alloc_host_mem(struct nvme_dev *dev, u64 min, u64 preferred)
 {
 	struct nvme_host_mem_buf_desc *descs;
 	u32 chunk_size, max_entries, len;
+	dma_addr_t descs_dma;
 	int i = 0;
 	void **bufs;
 	u64 size = 0, tmp;
@@ -1627,7 +1624,8 @@ static int nvme_alloc_host_mem(struct nvme_dev *dev, u64 min, u64 preferred)
 	tmp = (preferred + chunk_size - 1);
 	do_div(tmp, chunk_size);
 	max_entries = tmp;
-	descs = kcalloc(max_entries, sizeof(*descs), GFP_KERNEL);
+	descs = dma_zalloc_coherent(dev->dev, max_entries * sizeof(*descs),
+			&descs_dma, GFP_KERNEL);
 	if (!descs)
 		goto out;
 
@@ -1661,6 +1659,7 @@ static int nvme_alloc_host_mem(struct nvme_dev *dev, u64 min, u64 preferred)
 	dev->nr_host_mem_descs = i;
 	dev->host_mem_size = size;
 	dev->host_mem_descs = descs;
+	dev->host_mem_descs_dma = descs_dma;
 	dev->host_mem_desc_bufs = bufs;
 	return 0;
 
@@ -1674,7 +1673,8 @@ static int nvme_alloc_host_mem(struct nvme_dev *dev, u64 min, u64 preferred)
 
 	kfree(bufs);
 out_free_descs:
-	kfree(descs);
+	dma_free_coherent(dev->dev, max_entries * sizeof(*descs), descs,
+			descs_dma);
 out:
 	/* try a smaller chunk size if we failed early */
 	if (chunk_size >= PAGE_SIZE * 2 && (i == 0 || size < min)) {

drivers/nvme/host/rdma.c

@@ -920,7 +920,11 @@ static int nvme_rdma_map_sg_fr(struct nvme_rdma_queue *queue,
 	struct nvme_keyed_sgl_desc *sg = &c->common.dptr.ksgl;
 	int nr;
 
-	nr = ib_map_mr_sg(req->mr, req->sg_table.sgl, count, NULL, PAGE_SIZE);
+	/*
+	 * Align the MR to a 4K page size to match the ctrl page size and
+	 * the block virtual boundary.
+	 */
+	nr = ib_map_mr_sg(req->mr, req->sg_table.sgl, count, NULL, SZ_4K);
 	if (nr < count) {
 		if (nr < 0)
 			return nr;
@@ -1583,7 +1587,7 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl)
 		goto out_cleanup_queue;
 
 	ctrl->ctrl.max_hw_sectors =
-		(ctrl->max_fr_pages - 1) << (PAGE_SHIFT - 9);
+		(ctrl->max_fr_pages - 1) << (ilog2(SZ_4K) - 9);
 
 	error = nvme_init_identify(&ctrl->ctrl);
 	if (error)

include/linux/nvme.h

@@ -254,7 +254,7 @@ enum {
 	NVME_CTRL_VWC_PRESENT			= 1 << 0,
 	NVME_CTRL_OACS_SEC_SUPP                 = 1 << 0,
 	NVME_CTRL_OACS_DIRECTIVES		= 1 << 5,
-	NVME_CTRL_OACS_DBBUF_SUPP		= 1 << 7,
+	NVME_CTRL_OACS_DBBUF_SUPP		= 1 << 8,
 };
 
 struct nvme_lbaf {