Merge git://git.infradead.org/users/willy/linux-nvme

Pull NVMe driver updates from Matthew Wilcox:
 "Various updates to the NVMe driver.  The most user-visible change is
  that drive hotplugging now works and CPU hotplug while an NVMe drive
  is installed should also work better"

* git://git.infradead.org/users/willy/linux-nvme:
  NVMe: Retry failed commands with non-fatal errors
  NVMe: Add getgeo to block ops
  NVMe: Start-stop nvme_thread during device add-remove.
  NVMe: Make I/O timeout a module parameter
  NVMe: CPU hot plug notification
  NVMe: per-cpu io queues
  NVMe: Replace DEFINE_PCI_DEVICE_TABLE
  NVMe: Fix divide-by-zero in nvme_trans_io_get_num_cmds
  NVMe: IOCTL path RCU protect queue access
  NVMe: RCU protected access to io queues
  NVMe: Initialize device reference count earlier
  NVMe: Add CONFIG_PM_SLEEP to suspend/resume functions

drivers/block/nvme-scsi.c

@@ -1562,13 +1562,14 @@ static int nvme_trans_send_fw_cmd(struct nvme_ns *ns, struct sg_io_hdr *hdr,
 			res = PTR_ERR(iod);
 			goto out;
 		}
-		length = nvme_setup_prps(dev, &c.common, iod, tot_len,
-								GFP_KERNEL);
+		length = nvme_setup_prps(dev, iod, tot_len, GFP_KERNEL);
 		if (length != tot_len) {
 			res = -ENOMEM;
 			goto out_unmap;
 		}
 
+		c.dlfw.prp1 = cpu_to_le64(sg_dma_address(iod->sg));
+		c.dlfw.prp2 = cpu_to_le64(iod->first_dma);
 		c.dlfw.numd = cpu_to_le32((tot_len/BYTES_TO_DWORDS) - 1);
 		c.dlfw.offset = cpu_to_le32(offset/BYTES_TO_DWORDS);
 	} else if (opcode == nvme_admin_activate_fw) {
@@ -2033,7 +2034,6 @@ static int nvme_trans_do_nvme_io(struct nvme_ns *ns, struct sg_io_hdr *hdr,
 	int res = SNTI_TRANSLATION_SUCCESS;
 	int nvme_sc;
 	struct nvme_dev *dev = ns->dev;
-	struct nvme_queue *nvmeq;
 	u32 num_cmds;
 	struct nvme_iod *iod;
 	u64 unit_len;
@@ -2045,7 +2045,7 @@ static int nvme_trans_do_nvme_io(struct nvme_ns *ns, struct sg_io_hdr *hdr,
 	struct nvme_command c;
 	u8 opcode = (is_write ? nvme_cmd_write : nvme_cmd_read);
 	u16 control;
-	u32 max_blocks = nvme_block_nr(ns, dev->max_hw_sectors);
+	u32 max_blocks = queue_max_hw_sectors(ns->queue);
 
 	num_cmds = nvme_trans_io_get_num_cmds(hdr, cdb_info, max_blocks);
 
@@ -2093,8 +2093,7 @@ static int nvme_trans_do_nvme_io(struct nvme_ns *ns, struct sg_io_hdr *hdr,
 			res = PTR_ERR(iod);
 			goto out;
 		}
-		retcode = nvme_setup_prps(dev, &c.common, iod, unit_len,
-							GFP_KERNEL);
+		retcode = nvme_setup_prps(dev, iod, unit_len, GFP_KERNEL);
 		if (retcode != unit_len) {
 			nvme_unmap_user_pages(dev,
 				(is_write) ? DMA_TO_DEVICE : DMA_FROM_DEVICE,
@@ -2103,21 +2102,12 @@ static int nvme_trans_do_nvme_io(struct nvme_ns *ns, struct sg_io_hdr *hdr,
 			res = -ENOMEM;
 			goto out;
 		}
+		c.rw.prp1 = cpu_to_le64(sg_dma_address(iod->sg));
+		c.rw.prp2 = cpu_to_le64(iod->first_dma);
 
 		nvme_offset += unit_num_blocks;
 
-		nvmeq = get_nvmeq(dev);
-		/*
-		 * Since nvme_submit_sync_cmd sleeps, we can't keep
-		 * preemption disabled.  We may be preempted at any
-		 * point, and be rescheduled to a different CPU.  That
-		 * will cause cacheline bouncing, but no additional
-		 * races since q_lock already protects against other
-		 * CPUs.
-		 */
-		put_nvmeq(nvmeq);
-		nvme_sc = nvme_submit_sync_cmd(nvmeq, &c, NULL,
-						NVME_IO_TIMEOUT);
+		nvme_sc = nvme_submit_io_cmd(dev, &c, NULL);
 		if (nvme_sc != NVME_SC_SUCCESS) {
 			nvme_unmap_user_pages(dev,
 				(is_write) ? DMA_TO_DEVICE : DMA_FROM_DEVICE,
@@ -2644,7 +2634,6 @@ static int nvme_trans_start_stop(struct nvme_ns *ns, struct sg_io_hdr *hdr,
 {
 	int res = SNTI_TRANSLATION_SUCCESS;
 	int nvme_sc;
-	struct nvme_queue *nvmeq;
 	struct nvme_command c;
 	u8 immed, pcmod, pc, no_flush, start;
 
@@ -2671,10 +2660,7 @@ static int nvme_trans_start_stop(struct nvme_ns *ns, struct sg_io_hdr *hdr,
 			c.common.opcode = nvme_cmd_flush;
 			c.common.nsid = cpu_to_le32(ns->ns_id);
 
-			nvmeq = get_nvmeq(ns->dev);
-			put_nvmeq(nvmeq);
-			nvme_sc = nvme_submit_sync_cmd(nvmeq, &c, NULL, NVME_IO_TIMEOUT);
-
+			nvme_sc = nvme_submit_io_cmd(ns->dev, &c, NULL);
 			res = nvme_trans_status_code(hdr, nvme_sc);
 			if (res)
 				goto out;
@@ -2697,15 +2683,12 @@ static int nvme_trans_synchronize_cache(struct nvme_ns *ns,
 	int res = SNTI_TRANSLATION_SUCCESS;
 	int nvme_sc;
 	struct nvme_command c;
-	struct nvme_queue *nvmeq;
 
 	memset(&c, 0, sizeof(c));
 	c.common.opcode = nvme_cmd_flush;
 	c.common.nsid = cpu_to_le32(ns->ns_id);
 
-	nvmeq = get_nvmeq(ns->dev);
-	put_nvmeq(nvmeq);
-	nvme_sc = nvme_submit_sync_cmd(nvmeq, &c, NULL, NVME_IO_TIMEOUT);
+	nvme_sc = nvme_submit_io_cmd(ns->dev, &c, NULL);
 
 	res = nvme_trans_status_code(hdr, nvme_sc);
 	if (res)
@@ -2872,7 +2855,6 @@ static int nvme_trans_unmap(struct nvme_ns *ns, struct sg_io_hdr *hdr,
 	struct nvme_dev *dev = ns->dev;
 	struct scsi_unmap_parm_list *plist;
 	struct nvme_dsm_range *range;
-	struct nvme_queue *nvmeq;
 	struct nvme_command c;
 	int i, nvme_sc, res = -ENOMEM;
 	u16 ndesc, list_len;
@@ -2914,10 +2896,7 @@ static int nvme_trans_unmap(struct nvme_ns *ns, struct sg_io_hdr *hdr,
 	c.dsm.nr = cpu_to_le32(ndesc - 1);
 	c.dsm.attributes = cpu_to_le32(NVME_DSMGMT_AD);
 
-	nvmeq = get_nvmeq(dev);
-	put_nvmeq(nvmeq);
-
-	nvme_sc = nvme_submit_sync_cmd(nvmeq, &c, NULL, NVME_IO_TIMEOUT);
+	nvme_sc = nvme_submit_io_cmd(dev, &c, NULL);
 	res = nvme_trans_status_code(hdr, nvme_sc);
 
 	dma_free_coherent(&dev->pci_dev->dev, ndesc * sizeof(*range),

include/linux/nvme.h

@@ -66,20 +66,25 @@ enum {
 
 #define NVME_VS(major, minor)	(major << 16 | minor)
 
-#define NVME_IO_TIMEOUT	(5 * HZ)
+extern unsigned char io_timeout;
+#define NVME_IO_TIMEOUT	(io_timeout * HZ)
 
 /*
  * Represents an NVM Express device.  Each nvme_dev is a PCI function.
  */
 struct nvme_dev {
 	struct list_head node;
-	struct nvme_queue **queues;
+	struct nvme_queue __rcu **queues;
+	unsigned short __percpu *io_queue;
 	u32 __iomem *dbs;
 	struct pci_dev *pci_dev;
 	struct dma_pool *prp_page_pool;
 	struct dma_pool *prp_small_pool;
 	int instance;
-	int queue_count;
+	unsigned queue_count;
+	unsigned online_queues;
+	unsigned max_qid;
+	int q_depth;
 	u32 db_stride;
 	u32 ctrl_config;
 	struct msix_entry *entry;
@@ -89,6 +94,7 @@ struct nvme_dev {
 	struct miscdevice miscdev;
 	work_func_t reset_workfn;
 	struct work_struct reset_work;
+	struct notifier_block nb;
 	char name[12];
 	char serial[20];
 	char model[40];
@@ -131,6 +137,7 @@ struct nvme_iod {
 	int length;		/* Of data, in bytes */
 	unsigned long start_time;
 	dma_addr_t first_dma;
+	struct list_head node;
 	struct scatterlist sg[0];
 };
 
@@ -146,16 +153,12 @@ static inline u64 nvme_block_nr(struct nvme_ns *ns, sector_t sector)
  */
 void nvme_free_iod(struct nvme_dev *dev, struct nvme_iod *iod);
 
-int nvme_setup_prps(struct nvme_dev *dev, struct nvme_common_command *cmd,
-			struct nvme_iod *iod, int total_len, gfp_t gfp);
+int nvme_setup_prps(struct nvme_dev *, struct nvme_iod *, int , gfp_t);
 struct nvme_iod *nvme_map_user_pages(struct nvme_dev *dev, int write,
 				unsigned long addr, unsigned length);
 void nvme_unmap_user_pages(struct nvme_dev *dev, int write,
 			struct nvme_iod *iod);
-struct nvme_queue *get_nvmeq(struct nvme_dev *dev);
-void put_nvmeq(struct nvme_queue *nvmeq);
-int nvme_submit_sync_cmd(struct nvme_queue *nvmeq, struct nvme_command *cmd,
-						u32 *result, unsigned timeout);
+int nvme_submit_io_cmd(struct nvme_dev *, struct nvme_command *, u32 *);
 int nvme_submit_flush_data(struct nvme_queue *nvmeq, struct nvme_ns *ns);
 int nvme_submit_admin_cmd(struct nvme_dev *, struct nvme_command *,
 							u32 *result);

include/uapi/linux/nvme.h

@@ -434,6 +434,7 @@ enum {
 	NVME_SC_REFTAG_CHECK		= 0x284,
 	NVME_SC_COMPARE_FAILED		= 0x285,
 	NVME_SC_ACCESS_DENIED		= 0x286,
+	NVME_SC_DNR			= 0x4000,
 };
 
 struct nvme_completion {