Merge branch 'sched/urgent' into sched/core, to avoid conflicts

Signed-off-by: Ingo Molnar <mingo@kernel.org>

17 files changed, 1388 insertions, 481 deletions

diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c
index 18c76e84d540..68e3992e8838 100644
--- a/drivers/block/drbd/drbd_receiver.c
+++ b/drivers/block/drbd/drbd_receiver.c
@@ -469,24 +469,14 @@ static void drbd_wait_ee_list_empty(struct drbd_device *device,
 
 static int drbd_recv_short(struct socket *sock, void *buf, size_t size, int flags)
 {
-	mm_segment_t oldfs;
 	struct kvec iov = {
 		.iov_base = buf,
 		.iov_len = size,
 	};
 	struct msghdr msg = {
-		.msg_iovlen = 1,
-		.msg_iov = (struct iovec *)&iov,
 		.msg_flags = (flags ? flags : MSG_WAITALL | MSG_NOSIGNAL)
 	};
-	int rv;
-
-	oldfs = get_fs();
-	set_fs(KERNEL_DS);
-	rv = sock_recvmsg(sock, &msg, size, msg.msg_flags);
-	set_fs(oldfs);
-
-	return rv;
+	return kernel_recvmsg(sock, &msg, &iov, 1, size, msg.msg_flags);
 }
 
 static int drbd_recv(struct drbd_connection *connection, void *buf, size_t size)
diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index c8bf270b7890..6cb1beb47c25 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -237,7 +237,7 @@ static int __do_lo_send_write(struct file *file,
 	file_end_write(file);
 	if (likely(bw == len))
 		return 0;
-	printk(KERN_ERR "loop: Write error at byte offset %llu, length %i.\n",
+	printk_ratelimited(KERN_ERR "loop: Write error at byte offset %llu, length %i.\n",
 			(unsigned long long)pos, len);
 	if (bw >= 0)
 		bw = -EIO;
@@ -277,7 +277,7 @@ static int do_lo_send_write(struct loop_device *lo, struct bio_vec *bvec,
 		return __do_lo_send_write(lo->lo_backing_file,
 				page_address(page), bvec->bv_len,
 				pos);
-	printk(KERN_ERR "loop: Transfer error at byte offset %llu, "
+	printk_ratelimited(KERN_ERR "loop: Transfer error at byte offset %llu, "
 			"length %i.\n", (unsigned long long)pos, bvec->bv_len);
 	if (ret > 0)
 		ret = -EIO;
@@ -316,7 +316,7 @@ static int lo_send(struct loop_device *lo, struct bio *bio, loff_t pos)
 out:
 	return ret;
 fail:
-	printk(KERN_ERR "loop: Failed to allocate temporary page for write.\n");
+	printk_ratelimited(KERN_ERR "loop: Failed to allocate temporary page for write.\n");
 	ret = -ENOMEM;
 	goto out;
 }
@@ -345,7 +345,7 @@ lo_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
 		size = p->bsize;
 
 	if (lo_do_transfer(lo, READ, page, buf->offset, p->page, p->offset, size, IV)) {
-		printk(KERN_ERR "loop: transfer error block %ld\n",
+		printk_ratelimited(KERN_ERR "loop: transfer error block %ld\n",
 		       page->index);
 		size = -EINVAL;
 	}
diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c
index 2a1f26bd6640..56a027d6115e 100644
--- a/drivers/block/nbd.c
+++ b/drivers/block/nbd.c
@@ -630,37 +630,29 @@ static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *nbd,
 	}
  
 	case NBD_CLEAR_SOCK: {
-		struct file *file;
-
+		struct socket *sock = nbd->sock;
 		nbd->sock = NULL;
-		file = nbd->file;
-		nbd->file = NULL;
 		nbd_clear_que(nbd);
 		BUG_ON(!list_empty(&nbd->queue_head));
 		BUG_ON(!list_empty(&nbd->waiting_queue));
 		kill_bdev(bdev);
-		if (file)
-			fput(file);
+		if (sock)
+			sockfd_put(sock);
 		return 0;
 	}
 
 	case NBD_SET_SOCK: {
-		struct file *file;
-		if (nbd->file)
+		struct socket *sock;
+		int err;
+		if (nbd->sock)
 			return -EBUSY;
-		file = fget(arg);
-		if (file) {
-			struct inode *inode = file_inode(file);
-			if (S_ISSOCK(inode->i_mode)) {
-				nbd->file = file;
-				nbd->sock = SOCKET_I(inode);
-				if (max_part > 0)
-					bdev->bd_invalidated = 1;
-				nbd->disconnect = 0; /* we're connected now */
-				return 0;
-			} else {
-				fput(file);
-			}
+		sock = sockfd_lookup(arg, &err);
+		if (sock) {
+			nbd->sock = sock;
+			if (max_part > 0)
+				bdev->bd_invalidated = 1;
+			nbd->disconnect = 0; /* we're connected now */
+			return 0;
 		}
 		return -EINVAL;
 	}
@@ -697,12 +689,12 @@ static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *nbd,
 
 	case NBD_DO_IT: {
 		struct task_struct *thread;
-		struct file *file;
+		struct socket *sock;
 		int error;
 
 		if (nbd->pid)
 			return -EBUSY;
-		if (!nbd->file)
+		if (!nbd->sock)
 			return -EINVAL;
 
 		mutex_unlock(&nbd->tx_lock);
@@ -731,15 +723,15 @@ static int __nbd_ioctl(struct block_device *bdev, struct nbd_device *nbd,
 		if (error)
 			return error;
 		sock_shutdown(nbd, 0);
-		file = nbd->file;
-		nbd->file = NULL;
+		sock = nbd->sock;
+		nbd->sock = NULL;
 		nbd_clear_que(nbd);
 		dev_warn(disk_to_dev(nbd->disk), "queue cleared\n");
 		kill_bdev(bdev);
 		queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, nbd->disk->queue);
 		set_device_ro(bdev, false);
-		if (file)
-			fput(file);
+		if (sock)
+			sockfd_put(sock);
 		nbd->flags = 0;
 		nbd->bytesize = 0;
 		bdev->bd_inode->i_size = 0;
@@ -875,9 +867,7 @@ static int __init nbd_init(void)
 
 	for (i = 0; i < nbds_max; i++) {
 		struct gendisk *disk = nbd_dev[i].disk;
-		nbd_dev[i].file = NULL;
 		nbd_dev[i].magic = NBD_MAGIC;
-		nbd_dev[i].flags = 0;
 		INIT_LIST_HEAD(&nbd_dev[i].waiting_queue);
 		spin_lock_init(&nbd_dev[i].queue_lock);
 		INIT_LIST_HEAD(&nbd_dev[i].queue_head);
diff --git a/drivers/block/nvme-core.c b/drivers/block/nvme-core.c
index da085ff10d25..7c64fa756cce 100644
--- a/drivers/block/nvme-core.c
+++ b/drivers/block/nvme-core.c
@@ -1,6 +1,6 @@
 /*
  * NVM Express device driver
- * Copyright (c) 2011, Intel Corporation.
+ * Copyright (c) 2011-2014, Intel Corporation.
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms and conditions of the GNU General Public License,
@@ -20,10 +20,12 @@
 #include <linux/bio.h>
 #include <linux/bitops.h>
 #include <linux/blkdev.h>
+#include <linux/cpu.h>
 #include <linux/delay.h>
 #include <linux/errno.h>
 #include <linux/fs.h>
 #include <linux/genhd.h>
+#include <linux/hdreg.h>
 #include <linux/idr.h>
 #include <linux/init.h>
 #include <linux/interrupt.h>
@@ -35,6 +37,7 @@
 #include <linux/module.h>
 #include <linux/moduleparam.h>
 #include <linux/pci.h>
+#include <linux/percpu.h>
 #include <linux/poison.h>
 #include <linux/ptrace.h>
 #include <linux/sched.h>
@@ -47,6 +50,11 @@
 #define SQ_SIZE(depth)		(depth * sizeof(struct nvme_command))
 #define CQ_SIZE(depth)		(depth * sizeof(struct nvme_completion))
 #define ADMIN_TIMEOUT	(60 * HZ)
+#define IOD_TIMEOUT	(4 * NVME_IO_TIMEOUT)
+
+unsigned char io_timeout = 30;
+module_param(io_timeout, byte, 0644);
+MODULE_PARM_DESC(io_timeout, "timeout in seconds for I/O");
 
 static int nvme_major;
 module_param(nvme_major, int, 0);
@@ -58,6 +66,7 @@ static DEFINE_SPINLOCK(dev_list_lock);
 static LIST_HEAD(dev_list);
 static struct task_struct *nvme_thread;
 static struct workqueue_struct *nvme_workq;
+static wait_queue_head_t nvme_kthread_wait;
 
 static void nvme_reset_failed_dev(struct work_struct *ws);
 
@@ -74,6 +83,7 @@ struct async_cmd_info {
  * commands and one for I/O commands).
  */
 struct nvme_queue {
+	struct rcu_head r_head;
 	struct device *q_dmadev;
 	struct nvme_dev *dev;
 	char irqname[24];	/* nvme4294967295-65535\0 */
@@ -85,6 +95,7 @@ struct nvme_queue {
 	wait_queue_head_t sq_full;
 	wait_queue_t sq_cong_wait;
 	struct bio_list sq_cong;
+	struct list_head iod_bio;
 	u32 __iomem *q_db;
 	u16 q_depth;
 	u16 cq_vector;
@@ -95,6 +106,7 @@ struct nvme_queue {
 	u8 cq_phase;
 	u8 cqe_seen;
 	u8 q_suspended;
+	cpumask_var_t cpu_mask;
 	struct async_cmd_info cmdinfo;
 	unsigned long cmdid_data[];
 };
@@ -118,7 +130,7 @@ static inline void _nvme_check_size(void)
 	BUILD_BUG_ON(sizeof(struct nvme_smart_log) != 512);
 }
 
-typedef void (*nvme_completion_fn)(struct nvme_dev *, void *,
+typedef void (*nvme_completion_fn)(struct nvme_queue *, void *,
 						struct nvme_completion *);
 
 struct nvme_cmd_info {
@@ -190,7 +202,7 @@ static int alloc_cmdid_killable(struct nvme_queue *nvmeq, void *ctx,
 #define CMD_CTX_FLUSH		(0x318 + CMD_CTX_BASE)
 #define CMD_CTX_ABORT		(0x31C + CMD_CTX_BASE)
 
-static void special_completion(struct nvme_dev *dev, void *ctx,
+static void special_completion(struct nvme_queue *nvmeq, void *ctx,
 						struct nvme_completion *cqe)
 {
 	if (ctx == CMD_CTX_CANCELLED)
@@ -198,26 +210,26 @@ static void special_completion(struct nvme_dev *dev, void *ctx,
 	if (ctx == CMD_CTX_FLUSH)
 		return;
 	if (ctx == CMD_CTX_ABORT) {
-		++dev->abort_limit;
+		++nvmeq->dev->abort_limit;
 		return;
 	}
 	if (ctx == CMD_CTX_COMPLETED) {
-		dev_warn(&dev->pci_dev->dev,
+		dev_warn(nvmeq->q_dmadev,
 				"completed id %d twice on queue %d\n",
 				cqe->command_id, le16_to_cpup(&cqe->sq_id));
 		return;
 	}
 	if (ctx == CMD_CTX_INVALID) {
-		dev_warn(&dev->pci_dev->dev,
+		dev_warn(nvmeq->q_dmadev,
 				"invalid id %d completed on queue %d\n",
 				cqe->command_id, le16_to_cpup(&cqe->sq_id));
 		return;
 	}
 
-	dev_warn(&dev->pci_dev->dev, "Unknown special completion %p\n", ctx);
+	dev_warn(nvmeq->q_dmadev, "Unknown special completion %p\n", ctx);
 }
 
-static void async_completion(struct nvme_dev *dev, void *ctx,
+static void async_completion(struct nvme_queue *nvmeq, void *ctx,
 						struct nvme_completion *cqe)
 {
 	struct async_cmd_info *cmdinfo = ctx;
@@ -262,14 +274,34 @@ static void *cancel_cmdid(struct nvme_queue *nvmeq, int cmdid,
 	return ctx;
 }
 
-struct nvme_queue *get_nvmeq(struct nvme_dev *dev)
+static struct nvme_queue *raw_nvmeq(struct nvme_dev *dev, int qid)
+{
+	return rcu_dereference_raw(dev->queues[qid]);
+}
+
+static struct nvme_queue *get_nvmeq(struct nvme_dev *dev) __acquires(RCU)
+{
+	unsigned queue_id = get_cpu_var(*dev->io_queue);
+	rcu_read_lock();
+	return rcu_dereference(dev->queues[queue_id]);
+}
+
+static void put_nvmeq(struct nvme_queue *nvmeq) __releases(RCU)
 {
-	return dev->queues[get_cpu() + 1];
+	rcu_read_unlock();
+	put_cpu_var(nvmeq->dev->io_queue);
 }
 
-void put_nvmeq(struct nvme_queue *nvmeq)
+static struct nvme_queue *lock_nvmeq(struct nvme_dev *dev, int q_idx)
+							__acquires(RCU)
 {
-	put_cpu();
+	rcu_read_lock();
+	return rcu_dereference(dev->queues[q_idx]);
+}
+
+static void unlock_nvmeq(struct nvme_queue *nvmeq) __releases(RCU)
+{
+	rcu_read_unlock();
 }
 
 /**
@@ -284,6 +316,10 @@ static int nvme_submit_cmd(struct nvme_queue *nvmeq, struct nvme_command *cmd)
 	unsigned long flags;
 	u16 tail;
 	spin_lock_irqsave(&nvmeq->q_lock, flags);
+	if (nvmeq->q_suspended) {
+		spin_unlock_irqrestore(&nvmeq->q_lock, flags);
+		return -EBUSY;
+	}
 	tail = nvmeq->sq_tail;
 	memcpy(&nvmeq->sq_cmds[tail], cmd, sizeof(*cmd));
 	if (++tail == nvmeq->q_depth)
@@ -323,6 +359,7 @@ nvme_alloc_iod(unsigned nseg, unsigned nbytes, gfp_t gfp)
 		iod->npages = -1;
 		iod->length = nbytes;
 		iod->nents = 0;
+		iod->first_dma = 0ULL;
 		iod->start_time = jiffies;
 	}
 
@@ -371,19 +408,31 @@ static void nvme_end_io_acct(struct bio *bio, unsigned long start_time)
 	part_stat_unlock();
 }
 
-static void bio_completion(struct nvme_dev *dev, void *ctx,
+static void bio_completion(struct nvme_queue *nvmeq, void *ctx,
 						struct nvme_completion *cqe)
 {
 	struct nvme_iod *iod = ctx;
 	struct bio *bio = iod->private;
 	u16 status = le16_to_cpup(&cqe->status) >> 1;
 
+	if (unlikely(status)) {
+		if (!(status & NVME_SC_DNR ||
+				bio->bi_rw & REQ_FAILFAST_MASK) &&
+				(jiffies - iod->start_time) < IOD_TIMEOUT) {
+			if (!waitqueue_active(&nvmeq->sq_full))
+				add_wait_queue(&nvmeq->sq_full,
+							&nvmeq->sq_cong_wait);
+			list_add_tail(&iod->node, &nvmeq->iod_bio);
+			wake_up(&nvmeq->sq_full);
+			return;
+		}
+	}
 	if (iod->nents) {
-		dma_unmap_sg(&dev->pci_dev->dev, iod->sg, iod->nents,
+		dma_unmap_sg(nvmeq->q_dmadev, iod->sg, iod->nents,
 			bio_data_dir(bio) ? DMA_TO_DEVICE : DMA_FROM_DEVICE);
 		nvme_end_io_acct(bio, iod->start_time);
 	}
-	nvme_free_iod(dev, iod);
+	nvme_free_iod(nvmeq->dev, iod);
 	if (status)
 		bio_endio(bio, -EIO);
 	else
@@ -391,8 +440,8 @@ static void bio_completion(struct nvme_dev *dev, void *ctx,
 }
 
 /* length is in bytes.  gfp flags indicates whether we may sleep. */
-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 *dev, struct nvme_iod *iod, int total_len,
+								gfp_t gfp)
 {
 	struct dma_pool *pool;
 	int length = total_len;
@@ -405,7 +454,6 @@ int nvme_setup_prps(struct nvme_dev *dev, struct nvme_common_command *cmd,
 	dma_addr_t prp_dma;
 	int nprps, i;
 
-	cmd->prp1 = cpu_to_le64(dma_addr);
 	length -= (PAGE_SIZE - offset);
 	if (length <= 0)
 		return total_len;
@@ -420,7 +468,7 @@ int nvme_setup_prps(struct nvme_dev *dev, struct nvme_common_command *cmd,
 	}
 
 	if (length <= PAGE_SIZE) {
-		cmd->prp2 = cpu_to_le64(dma_addr);
+		iod->first_dma = dma_addr;
 		return total_len;
 	}
 
@@ -435,13 +483,12 @@ int nvme_setup_prps(struct nvme_dev *dev, struct nvme_common_command *cmd,
 
 	prp_list = dma_pool_alloc(pool, gfp, &prp_dma);
 	if (!prp_list) {
-		cmd->prp2 = cpu_to_le64(dma_addr);
+		iod->first_dma = dma_addr;
 		iod->npages = -1;
 		return (total_len - length) + PAGE_SIZE;
 	}
 	list[0] = prp_list;
 	iod->first_dma = prp_dma;
-	cmd->prp2 = cpu_to_le64(prp_dma);
 	i = 0;
 	for (;;) {
 		if (i == PAGE_SIZE / 8) {
@@ -480,10 +527,11 @@ static int nvme_split_and_submit(struct bio *bio, struct nvme_queue *nvmeq,
 
 	bio_chain(split, bio);
 
-	if (bio_list_empty(&nvmeq->sq_cong))
+	if (!waitqueue_active(&nvmeq->sq_full))
 		add_wait_queue(&nvmeq->sq_full, &nvmeq->sq_cong_wait);
 	bio_list_add(&nvmeq->sq_cong, split);
 	bio_list_add(&nvmeq->sq_cong, bio);
+	wake_up(&nvmeq->sq_full);
 
 	return 0;
 }
@@ -536,25 +584,13 @@ static int nvme_map_bio(struct nvme_queue *nvmeq, struct nvme_iod *iod,
 	return length;
 }
 
-/*
- * We reuse the small pool to allocate the 16-byte range here as it is not
- * worth having a special pool for these or additional cases to handle freeing
- * the iod.
- */
 static int nvme_submit_discard(struct nvme_queue *nvmeq, struct nvme_ns *ns,
 		struct bio *bio, struct nvme_iod *iod, int cmdid)
 {
-	struct nvme_dsm_range *range;
+	struct nvme_dsm_range *range =
+				(struct nvme_dsm_range *)iod_list(iod)[0];
 	struct nvme_command *cmnd = &nvmeq->sq_cmds[nvmeq->sq_tail];
 
-	range = dma_pool_alloc(nvmeq->dev->prp_small_pool, GFP_ATOMIC,
-							&iod->first_dma);
-	if (!range)
-		return -ENOMEM;
-
-	iod_list(iod)[0] = (__le64 *)range;
-	iod->npages = 0;
-
 	range->cattr = cpu_to_le32(0);
 	range->nlb = cpu_to_le32(bio->bi_iter.bi_size >> ns->lba_shift);
 	range->slba = cpu_to_le64(nvme_block_nr(ns, bio->bi_iter.bi_sector));
@@ -601,44 +637,22 @@ int nvme_submit_flush_data(struct nvme_queue *nvmeq, struct nvme_ns *ns)
 	return nvme_submit_flush(nvmeq, ns, cmdid);
 }
 
-/*
- * Called with local interrupts disabled and the q_lock held.  May not sleep.
- */
-static int nvme_submit_bio_queue(struct nvme_queue *nvmeq, struct nvme_ns *ns,
-								struct bio *bio)
+static int nvme_submit_iod(struct nvme_queue *nvmeq, struct nvme_iod *iod)
 {
+	struct bio *bio = iod->private;
+	struct nvme_ns *ns = bio->bi_bdev->bd_disk->private_data;
 	struct nvme_command *cmnd;
-	struct nvme_iod *iod;
-	enum dma_data_direction dma_dir;
-	int cmdid, length, result;
+	int cmdid;
 	u16 control;
 	u32 dsmgmt;
-	int psegs = bio_phys_segments(ns->queue, bio);
-
-	if ((bio->bi_rw & REQ_FLUSH) && psegs) {
-		result = nvme_submit_flush_data(nvmeq, ns);
-		if (result)
-			return result;
-	}
 
-	result = -ENOMEM;
-	iod = nvme_alloc_iod(psegs, bio->bi_iter.bi_size, GFP_ATOMIC);
-	if (!iod)
-		goto nomem;
-	iod->private = bio;
-
-	result = -EBUSY;
 	cmdid = alloc_cmdid(nvmeq, iod, bio_completion, NVME_IO_TIMEOUT);
 	if (unlikely(cmdid < 0))
-		goto free_iod;
+		return cmdid;
 
-	if (bio->bi_rw & REQ_DISCARD) {
-		result = nvme_submit_discard(nvmeq, ns, bio, iod, cmdid);
-		if (result)
-			goto free_cmdid;
-		return result;
-	}
-	if ((bio->bi_rw & REQ_FLUSH) && !psegs)
+	if (bio->bi_rw & REQ_DISCARD)
+		return nvme_submit_discard(nvmeq, ns, bio, iod, cmdid);
+	if ((bio->bi_rw & REQ_FLUSH) && !iod->nents)
 		return nvme_submit_flush(nvmeq, ns, cmdid);
 
 	control = 0;
@@ -652,42 +666,85 @@ static int nvme_submit_bio_queue(struct nvme_queue *nvmeq, struct nvme_ns *ns,
 		dsmgmt |= NVME_RW_DSM_FREQ_PREFETCH;
 
 	cmnd = &nvmeq->sq_cmds[nvmeq->sq_tail];
-
 	memset(cmnd, 0, sizeof(*cmnd));
-	if (bio_data_dir(bio)) {
-		cmnd->rw.opcode = nvme_cmd_write;
-		dma_dir = DMA_TO_DEVICE;
-	} else {
-		cmnd->rw.opcode = nvme_cmd_read;
-		dma_dir = DMA_FROM_DEVICE;
-	}
-
-	result = nvme_map_bio(nvmeq, iod, bio, dma_dir, psegs);
-	if (result <= 0)
-		goto free_cmdid;
-	length = result;
 
+	cmnd->rw.opcode = bio_data_dir(bio) ? nvme_cmd_write : nvme_cmd_read;
 	cmnd->rw.command_id = cmdid;
 	cmnd->rw.nsid = cpu_to_le32(ns->ns_id);
-	length = nvme_setup_prps(nvmeq->dev, &cmnd->common, iod, length,
-								GFP_ATOMIC);
+	cmnd->rw.prp1 = cpu_to_le64(sg_dma_address(iod->sg));
+	cmnd->rw.prp2 = cpu_to_le64(iod->first_dma);
 	cmnd->rw.slba = cpu_to_le64(nvme_block_nr(ns, bio->bi_iter.bi_sector));
-	cmnd->rw.length = cpu_to_le16((length >> ns->lba_shift) - 1);
+	cmnd->rw.length =
+		cpu_to_le16((bio->bi_iter.bi_size >> ns->lba_shift) - 1);
 	cmnd->rw.control = cpu_to_le16(control);
 	cmnd->rw.dsmgmt = cpu_to_le32(dsmgmt);
 
-	nvme_start_io_acct(bio);
 	if (++nvmeq->sq_tail == nvmeq->q_depth)
 		nvmeq->sq_tail = 0;
 	writel(nvmeq->sq_tail, nvmeq->q_db);
 
 	return 0;
+}
+
+/*
+ * Called with local interrupts disabled and the q_lock held.  May not sleep.
+ */
+static int nvme_submit_bio_queue(struct nvme_queue *nvmeq, struct nvme_ns *ns,
+								struct bio *bio)
+{
+	struct nvme_iod *iod;
+	int psegs = bio_phys_segments(ns->queue, bio);
+	int result;
+
+	if ((bio->bi_rw & REQ_FLUSH) && psegs) {
+		result = nvme_submit_flush_data(nvmeq, ns);
+		if (result)
+			return result;
+	}
+
+	iod = nvme_alloc_iod(psegs, bio->bi_iter.bi_size, GFP_ATOMIC);
+	if (!iod)
+		return -ENOMEM;
+
+	iod->private = bio;
+	if (bio->bi_rw & REQ_DISCARD) {
+		void *range;
+		/*
+		 * We reuse the small pool to allocate the 16-byte range here
+		 * as it is not worth having a special pool for these or
+		 * additional cases to handle freeing the iod.
+		 */
+		range = dma_pool_alloc(nvmeq->dev->prp_small_pool,
+						GFP_ATOMIC,
+						&iod->first_dma);
+		if (!range) {
+			result = -ENOMEM;
+			goto free_iod;
+		}
+		iod_list(iod)[0] = (__le64 *)range;
+		iod->npages = 0;
+	} else if (psegs) {
+		result = nvme_map_bio(nvmeq, iod, bio,
+			bio_data_dir(bio) ? DMA_TO_DEVICE : DMA_FROM_DEVICE,
+			psegs);
+		if (result <= 0)
+			goto free_iod;
+		if (nvme_setup_prps(nvmeq->dev, iod, result, GFP_ATOMIC) !=
+								result) {
+			result = -ENOMEM;
+			goto free_iod;
+		}
+		nvme_start_io_acct(bio);
+	}
+	if (unlikely(nvme_submit_iod(nvmeq, iod))) {
+		if (!waitqueue_active(&nvmeq->sq_full))
+			add_wait_queue(&nvmeq->sq_full, &nvmeq->sq_cong_wait);
+		list_add_tail(&iod->node, &nvmeq->iod_bio);
+	}
+	return 0;
 
- free_cmdid:
-	free_cmdid(nvmeq, cmdid, NULL);
  free_iod:
 	nvme_free_iod(nvmeq->dev, iod);
- nomem:
 	return result;
 }
 
@@ -711,7 +768,7 @@ static int nvme_process_cq(struct nvme_queue *nvmeq)
 		}
 
 		ctx = free_cmdid(nvmeq, cqe.command_id, &fn);
-		fn(nvmeq->dev, ctx, &cqe);
+		fn(nvmeq, ctx, &cqe);
 	}
 
 	/* If the controller ignores the cq head doorbell and continuously
@@ -747,7 +804,7 @@ static void nvme_make_request(struct request_queue *q, struct bio *bio)
 	if (!nvmeq->q_suspended && bio_list_empty(&nvmeq->sq_cong))
 		result = nvme_submit_bio_queue(nvmeq, ns, bio);
 	if (unlikely(result)) {
-		if (bio_list_empty(&nvmeq->sq_cong))
+		if (!waitqueue_active(&nvmeq->sq_full))
 			add_wait_queue(&nvmeq->sq_full, &nvmeq->sq_cong_wait);
 		bio_list_add(&nvmeq->sq_cong, bio);
 	}
@@ -791,7 +848,7 @@ struct sync_cmd_info {
 	int status;
 };
 
-static void sync_completion(struct nvme_dev *dev, void *ctx,
+static void sync_completion(struct nvme_queue *nvmeq, void *ctx,
 						struct nvme_completion *cqe)
 {
 	struct sync_cmd_info *cmdinfo = ctx;
@@ -804,27 +861,46 @@ static void sync_completion(struct nvme_dev *dev, void *ctx,
  * Returns 0 on success.  If the result is negative, it's a Linux error code;
  * if the result is positive, it's an NVM Express status code
  */
-int nvme_submit_sync_cmd(struct nvme_queue *nvmeq, struct nvme_command *cmd,
+static int nvme_submit_sync_cmd(struct nvme_dev *dev, int q_idx,
+						struct nvme_command *cmd,
 						u32 *result, unsigned timeout)
 {
-	int cmdid;
+	int cmdid, ret;
 	struct sync_cmd_info cmdinfo;
+	struct nvme_queue *nvmeq;
+
+	nvmeq = lock_nvmeq(dev, q_idx);
+	if (!nvmeq) {
+		unlock_nvmeq(nvmeq);
+		return -ENODEV;
+	}
 
 	cmdinfo.task = current;
 	cmdinfo.status = -EINTR;
 
-	cmdid = alloc_cmdid_killable(nvmeq, &cmdinfo, sync_completion,
-								timeout);
-	if (cmdid < 0)
+	cmdid = alloc_cmdid(nvmeq, &cmdinfo, sync_completion, timeout);
+	if (cmdid < 0) {
+		unlock_nvmeq(nvmeq);
 		return cmdid;
+	}
 	cmd->common.command_id = cmdid;
 
 	set_current_state(TASK_KILLABLE);
-	nvme_submit_cmd(nvmeq, cmd);
+	ret = nvme_submit_cmd(nvmeq, cmd);
+	if (ret) {
+		free_cmdid(nvmeq, cmdid, NULL);
+		unlock_nvmeq(nvmeq);
+		set_current_state(TASK_RUNNING);
+		return ret;
+	}
+	unlock_nvmeq(nvmeq);
 	schedule_timeout(timeout);
 
 	if (cmdinfo.status == -EINTR) {
-		nvme_abort_command(nvmeq, cmdid);
+		nvmeq = lock_nvmeq(dev, q_idx);
+		if (nvmeq)
+			nvme_abort_command(nvmeq, cmdid);
+		unlock_nvmeq(nvmeq);
 		return -EINTR;
 	}
 
@@ -845,20 +921,26 @@ static int nvme_submit_async_cmd(struct nvme_queue *nvmeq,
 		return cmdid;
 	cmdinfo->status = -EINTR;
 	cmd->common.command_id = cmdid;
-	nvme_submit_cmd(nvmeq, cmd);
-	return 0;
+	return nvme_submit_cmd(nvmeq, cmd);
 }
 
 int nvme_submit_admin_cmd(struct nvme_dev *dev, struct nvme_command *cmd,
 								u32 *result)
 {
-	return nvme_submit_sync_cmd(dev->queues[0], cmd, result, ADMIN_TIMEOUT);
+	return nvme_submit_sync_cmd(dev, 0, cmd, result, ADMIN_TIMEOUT);
+}
+
+int nvme_submit_io_cmd(struct nvme_dev *dev, struct nvme_command *cmd,
+								u32 *result)
+{
+	return nvme_submit_sync_cmd(dev, smp_processor_id() + 1, cmd, result,
+							NVME_IO_TIMEOUT);
 }
 
 static int nvme_submit_admin_cmd_async(struct nvme_dev *dev,
 		struct nvme_command *cmd, struct async_cmd_info *cmdinfo)
 {
-	return nvme_submit_async_cmd(dev->queues[0], cmd, cmdinfo,
+	return nvme_submit_async_cmd(raw_nvmeq(dev, 0), cmd, cmdinfo,
 								ADMIN_TIMEOUT);
 }
 
@@ -985,6 +1067,7 @@ static void nvme_abort_cmd(int cmdid, struct nvme_queue *nvmeq)
 	struct nvme_command cmd;
 	struct nvme_dev *dev = nvmeq->dev;
 	struct nvme_cmd_info *info = nvme_cmd_info(nvmeq);
+	struct nvme_queue *adminq;
 
 	if (!nvmeq->qid || info[cmdid].aborted) {
 		if (work_busy(&dev->reset_work))
@@ -1001,7 +1084,8 @@ static void nvme_abort_cmd(int cmdid, struct nvme_queue *nvmeq)
 	if (!dev->abort_limit)
 		return;
 
-	a_cmdid = alloc_cmdid(dev->queues[0], CMD_CTX_ABORT, special_completion,
+	adminq = rcu_dereference(dev->queues[0]);
+	a_cmdid = alloc_cmdid(adminq, CMD_CTX_ABORT, special_completion,
 								ADMIN_TIMEOUT);
 	if (a_cmdid < 0)
 		return;
@@ -1018,7 +1102,7 @@ static void nvme_abort_cmd(int cmdid, struct nvme_queue *nvmeq)
 
 	dev_warn(nvmeq->q_dmadev, "Aborting I/O %d QID %d\n", cmdid,
 							nvmeq->qid);
-	nvme_submit_cmd(dev->queues[0], &cmd);
+	nvme_submit_cmd(adminq, &cmd);
 }
 
 /**
@@ -1051,23 +1135,38 @@ static void nvme_cancel_ios(struct nvme_queue *nvmeq, bool timeout)
 		dev_warn(nvmeq->q_dmadev, "Cancelling I/O %d QID %d\n", cmdid,
 								nvmeq->qid);
 		ctx = cancel_cmdid(nvmeq, cmdid, &fn);
-		fn(nvmeq->dev, ctx, &cqe);
+		fn(nvmeq, ctx, &cqe);
 	}
 }
 
-static void nvme_free_queue(struct nvme_queue *nvmeq)
+static void nvme_free_queue(struct rcu_head *r)
 {
+	struct nvme_queue *nvmeq = container_of(r, struct nvme_queue, r_head);
+
 	spin_lock_irq(&nvmeq->q_lock);
 	while (bio_list_peek(&nvmeq->sq_cong)) {
 		struct bio *bio = bio_list_pop(&nvmeq->sq_cong);
 		bio_endio(bio, -EIO);
 	}
+	while (!list_empty(&nvmeq->iod_bio)) {
+		static struct nvme_completion cqe = {
+			.status = cpu_to_le16(
+				(NVME_SC_ABORT_REQ | NVME_SC_DNR) << 1),
+		};
+		struct nvme_iod *iod = list_first_entry(&nvmeq->iod_bio,
+							struct nvme_iod,
+							node);
+		list_del(&iod->node);
+		bio_completion(nvmeq, iod, &cqe);
+	}
 	spin_unlock_irq(&nvmeq->q_lock);
 
 	dma_free_coherent(nvmeq->q_dmadev, CQ_SIZE(nvmeq->q_depth),
 				(void *)nvmeq->cqes, nvmeq->cq_dma_addr);
 	dma_free_coherent(nvmeq->q_dmadev, SQ_SIZE(nvmeq->q_depth),
 					nvmeq->sq_cmds, nvmeq->sq_dma_addr);
+	if (nvmeq->qid)
+		free_cpumask_var(nvmeq->cpu_mask);
 	kfree(nvmeq);
 }
 
@@ -1076,9 +1175,10 @@ static void nvme_free_queues(struct nvme_dev *dev, int lowest)
 	int i;
 
 	for (i = dev->queue_count - 1; i >= lowest; i--) {
-		nvme_free_queue(dev->queues[i]);
+		struct nvme_queue *nvmeq = raw_nvmeq(dev, i);
+		rcu_assign_pointer(dev->queues[i], NULL);
+		call_rcu(&nvmeq->r_head, nvme_free_queue);
 		dev->queue_count--;
-		dev->queues[i] = NULL;
 	}
 }
 
@@ -1098,6 +1198,7 @@ static int nvme_suspend_queue(struct nvme_queue *nvmeq)
 		return 1;
 	}
 	nvmeq->q_suspended = 1;
+	nvmeq->dev->online_queues--;
 	spin_unlock_irq(&nvmeq->q_lock);
 
 	irq_set_affinity_hint(vector, NULL);
@@ -1116,7 +1217,7 @@ static void nvme_clear_queue(struct nvme_queue *nvmeq)
 
 static void nvme_disable_queue(struct nvme_dev *dev, int qid)
 {
-	struct nvme_queue *nvmeq = dev->queues[qid];
+	struct nvme_queue *nvmeq = raw_nvmeq(dev, qid);
 
 	if (!nvmeq)
 		return;
@@ -1152,6 +1253,9 @@ static struct nvme_queue *nvme_alloc_queue(struct nvme_dev *dev, int qid,
 	if (!nvmeq->sq_cmds)
 		goto free_cqdma;
 
+	if (qid && !zalloc_cpumask_var(&nvmeq->cpu_mask, GFP_KERNEL))
+		goto free_sqdma;
+
 	nvmeq->q_dmadev = dmadev;
 	nvmeq->dev = dev;
 	snprintf(nvmeq->irqname, sizeof(nvmeq->irqname), "nvme%dq%d",
@@ -1162,15 +1266,20 @@ static struct nvme_queue *nvme_alloc_queue(struct nvme_dev *dev, int qid,
 	init_waitqueue_head(&nvmeq->sq_full);
 	init_waitqueue_entry(&nvmeq->sq_cong_wait, nvme_thread);
 	bio_list_init(&nvmeq->sq_cong);
+	INIT_LIST_HEAD(&nvmeq->iod_bio);
 	nvmeq->q_db = &dev->dbs[qid * 2 * dev->db_stride];
 	nvmeq->q_depth = depth;
 	nvmeq->cq_vector = vector;
 	nvmeq->qid = qid;
 	nvmeq->q_suspended = 1;
 	dev->queue_count++;
+	rcu_assign_pointer(dev->queues[qid], nvmeq);
 
 	return nvmeq;
 
+ free_sqdma:
+	dma_free_coherent(dmadev, SQ_SIZE(depth), (void *)nvmeq->sq_cmds,
+							nvmeq->sq_dma_addr);
  free_cqdma:
 	dma_free_coherent(dmadev, CQ_SIZE(depth), (void *)nvmeq->cqes,
 							nvmeq->cq_dma_addr);
@@ -1203,6 +1312,7 @@ static void nvme_init_queue(struct nvme_queue *nvmeq, u16 qid)
 	memset((void *)nvmeq->cqes, 0, CQ_SIZE(nvmeq->q_depth));
 	nvme_cancel_ios(nvmeq, false);
 	nvmeq->q_suspended = 0;
+	dev->online_queues++;
 }
 
 static int nvme_create_queue(struct nvme_queue *nvmeq, int qid)
@@ -1311,12 +1421,11 @@ static int nvme_configure_admin_queue(struct nvme_dev *dev)
 	if (result < 0)
 		return result;
 
-	nvmeq = dev->queues[0];
+	nvmeq = raw_nvmeq(dev, 0);
 	if (!nvmeq) {
 		nvmeq = nvme_alloc_queue(dev, 0, 64, 0);
 		if (!nvmeq)
 			return -ENOMEM;
-		dev->queues[0] = nvmeq;
 	}
 
 	aqa = nvmeq->q_depth - 1;
@@ -1418,7 +1527,6 @@ void nvme_unmap_user_pages(struct nvme_dev *dev, int write,
 static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio)
 {
 	struct nvme_dev *dev = ns->dev;
-	struct nvme_queue *nvmeq;
 	struct nvme_user_io io;
 	struct nvme_command c;
 	unsigned length, meta_len;
@@ -1492,22 +1600,14 @@ static int nvme_submit_io(struct nvme_ns *ns, struct nvme_user_io __user *uio)
 		c.rw.metadata = cpu_to_le64(meta_dma_addr);
 	}
 
-	length = nvme_setup_prps(dev, &c.common, iod, length, GFP_KERNEL);
+	length = nvme_setup_prps(dev, iod, length, GFP_KERNEL);
+	c.rw.prp1 = cpu_to_le64(sg_dma_address(iod->sg));
+	c.rw.prp2 = cpu_to_le64(iod->first_dma);
 
-	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);
 	if (length != (io.nblocks + 1) << ns->lba_shift)
 		status = -ENOMEM;
-	else if (!nvmeq || nvmeq->q_suspended)
-		status = -EBUSY;
 	else
-		status = nvme_submit_sync_cmd(nvmeq, &c, NULL, NVME_IO_TIMEOUT);
+		status = nvme_submit_io_cmd(dev, &c, NULL);
 
 	if (meta_len) {
 		if (status == NVME_SC_SUCCESS && !(io.opcode & 1)) {
@@ -1572,8 +1672,9 @@ static int nvme_user_admin_cmd(struct nvme_dev *dev,
 								length);
 		if (IS_ERR(iod))
 			return PTR_ERR(iod);
-		length = nvme_setup_prps(dev, &c.common, iod, length,
-								GFP_KERNEL);
+		length = nvme_setup_prps(dev, iod, length, GFP_KERNEL);
+		c.common.prp1 = cpu_to_le64(sg_dma_address(iod->sg));
+		c.common.prp2 = cpu_to_le64(iod->first_dma);
 	}
 
 	timeout = cmd.timeout_ms ? msecs_to_jiffies(cmd.timeout_ms) :
@@ -1581,8 +1682,7 @@ static int nvme_user_admin_cmd(struct nvme_dev *dev,
 	if (length != cmd.data_len)
 		status = -ENOMEM;
 	else
-		status = nvme_submit_sync_cmd(dev->queues[0], &c, &cmd.result,
-								timeout);
+		status = nvme_submit_sync_cmd(dev, 0, &c, &cmd.result, timeout);
 
 	if (cmd.data_len) {
 		nvme_unmap_user_pages(dev, cmd.opcode & 1, iod);
@@ -1653,25 +1753,51 @@ static void nvme_release(struct gendisk *disk, fmode_t mode)
 	kref_put(&dev->kref, nvme_free_dev);
 }
 
+static int nvme_getgeo(struct block_device *bd, struct hd_geometry *geo)
+{
+	/* some standard values */
+	geo->heads = 1 << 6;
+	geo->sectors = 1 << 5;
+	geo->cylinders = get_capacity(bd->bd_disk) >> 11;
+	return 0;
+}
+
 static const struct block_device_operations nvme_fops = {
 	.owner		= THIS_MODULE,
 	.ioctl		= nvme_ioctl,
 	.compat_ioctl	= nvme_compat_ioctl,
 	.open		= nvme_open,
 	.release	= nvme_release,
+	.getgeo		= nvme_getgeo,
 };
 
+static void nvme_resubmit_iods(struct nvme_queue *nvmeq)
+{
+	struct nvme_iod *iod, *next;
+
+	list_for_each_entry_safe(iod, next, &nvmeq->iod_bio, node) {
+		if (unlikely(nvme_submit_iod(nvmeq, iod)))
+			break;
+		list_del(&iod->node);
+		if (bio_list_empty(&nvmeq->sq_cong) &&
+						list_empty(&nvmeq->iod_bio))
+			remove_wait_queue(&nvmeq->sq_full,
+						&nvmeq->sq_cong_wait);
+	}
+}
+
 static void nvme_resubmit_bios(struct nvme_queue *nvmeq)
 {
 	while (bio_list_peek(&nvmeq->sq_cong)) {
 		struct bio *bio = bio_list_pop(&nvmeq->sq_cong);
 		struct nvme_ns *ns = bio->bi_bdev->bd_disk->private_data;
 
-		if (bio_list_empty(&nvmeq->sq_cong))
+		if (bio_list_empty(&nvmeq->sq_cong) &&
+						list_empty(&nvmeq->iod_bio))
 			remove_wait_queue(&nvmeq->sq_full,
 							&nvmeq->sq_cong_wait);
 		if (nvme_submit_bio_queue(nvmeq, ns, bio)) {
-			if (bio_list_empty(&nvmeq->sq_cong))
+			if (!waitqueue_active(&nvmeq->sq_full))
 				add_wait_queue(&nvmeq->sq_full,
 							&nvmeq->sq_cong_wait);
 			bio_list_add_head(&nvmeq->sq_cong, bio);
@@ -1700,8 +1826,10 @@ static int nvme_kthread(void *data)
 				queue_work(nvme_workq, &dev->reset_work);
 				continue;
 			}
+			rcu_read_lock();
 			for (i = 0; i < dev->queue_count; i++) {
-				struct nvme_queue *nvmeq = dev->queues[i];
+				struct nvme_queue *nvmeq =
+						rcu_dereference(dev->queues[i]);
 				if (!nvmeq)
 					continue;
 				spin_lock_irq(&nvmeq->q_lock);
@@ -1710,9 +1838,11 @@ static int nvme_kthread(void *data)
 				nvme_process_cq(nvmeq);
 				nvme_cancel_ios(nvmeq, true);
 				nvme_resubmit_bios(nvmeq);
+				nvme_resubmit_iods(nvmeq);
  unlock:
 				spin_unlock_irq(&nvmeq->q_lock);
 			}
+			rcu_read_unlock();
 		}
 		spin_unlock(&dev_list_lock);
 		schedule_timeout(round_jiffies_relative(HZ));
@@ -1787,6 +1917,143 @@ static struct nvme_ns *nvme_alloc_ns(struct nvme_dev *dev, unsigned nsid,
 	return NULL;
 }
 
+static int nvme_find_closest_node(int node)
+{
+	int n, val, min_val = INT_MAX, best_node = node;
+
+	for_each_online_node(n) {
+		if (n == node)
+			continue;
+		val = node_distance(node, n);
+		if (val < min_val) {
+			min_val = val;
+			best_node = n;
+		}
+	}
+	return best_node;
+}
+
+static void nvme_set_queue_cpus(cpumask_t *qmask, struct nvme_queue *nvmeq,
+								int count)
+{
+	int cpu;
+	for_each_cpu(cpu, qmask) {
+		if (cpumask_weight(nvmeq->cpu_mask) >= count)
+			break;
+		if (!cpumask_test_and_set_cpu(cpu, nvmeq->cpu_mask))
+			*per_cpu_ptr(nvmeq->dev->io_queue, cpu) = nvmeq->qid;
+	}
+}
+
+static void nvme_add_cpus(cpumask_t *mask, const cpumask_t *unassigned_cpus,
+	const cpumask_t *new_mask, struct nvme_queue *nvmeq, int cpus_per_queue)
+{
+	int next_cpu;
+	for_each_cpu(next_cpu, new_mask) {
+		cpumask_or(mask, mask, get_cpu_mask(next_cpu));
+		cpumask_or(mask, mask, topology_thread_cpumask(next_cpu));
+		cpumask_and(mask, mask, unassigned_cpus);
+		nvme_set_queue_cpus(mask, nvmeq, cpus_per_queue);
+	}
+}
+
+static void nvme_create_io_queues(struct nvme_dev *dev)
+{
+	unsigned i, max;
+
+	max = min(dev->max_qid, num_online_cpus());
+	for (i = dev->queue_count; i <= max; i++)
+		if (!nvme_alloc_queue(dev, i, dev->q_depth, i - 1))
+			break;
+
+	max = min(dev->queue_count - 1, num_online_cpus());
+	for (i = dev->online_queues; i <= max; i++)
+		if (nvme_create_queue(raw_nvmeq(dev, i), i))
+			break;
+}
+
+/*
+ * If there are fewer queues than online cpus, this will try to optimally
+ * assign a queue to multiple cpus by grouping cpus that are "close" together:
+ * thread siblings, core, socket, closest node, then whatever else is
+ * available.
+ */
+static void nvme_assign_io_queues(struct nvme_dev *dev)
+{
+	unsigned cpu, cpus_per_queue, queues, remainder, i;
+	cpumask_var_t unassigned_cpus;
+
+	nvme_create_io_queues(dev);
+
+	queues = min(dev->online_queues - 1, num_online_cpus());
+	if (!queues)
+		return;
+
+	cpus_per_queue = num_online_cpus() / queues;
+	remainder = queues - (num_online_cpus() - queues * cpus_per_queue);
+
+	if (!alloc_cpumask_var(&unassigned_cpus, GFP_KERNEL))
+		return;
+
+	cpumask_copy(unassigned_cpus, cpu_online_mask);
+	cpu = cpumask_first(unassigned_cpus);
+	for (i = 1; i <= queues; i++) {
+		struct nvme_queue *nvmeq = lock_nvmeq(dev, i);
+		cpumask_t mask;
+
+		cpumask_clear(nvmeq->cpu_mask);
+		if (!cpumask_weight(unassigned_cpus)) {
+			unlock_nvmeq(nvmeq);
+			break;
+		}
+
+		mask = *get_cpu_mask(cpu);
+		nvme_set_queue_cpus(&mask, nvmeq, cpus_per_queue);
+		if (cpus_weight(mask) < cpus_per_queue)
+			nvme_add_cpus(&mask, unassigned_cpus,
+				topology_thread_cpumask(cpu),
+				nvmeq, cpus_per_queue);
+		if (cpus_weight(mask) < cpus_per_queue)
+			nvme_add_cpus(&mask, unassigned_cpus,
+				topology_core_cpumask(cpu),
+				nvmeq, cpus_per_queue);
+		if (cpus_weight(mask) < cpus_per_queue)
+			nvme_add_cpus(&mask, unassigned_cpus,
+				cpumask_of_node(cpu_to_node(cpu)),
+				nvmeq, cpus_per_queue);
+		if (cpus_weight(mask) < cpus_per_queue)
+			nvme_add_cpus(&mask, unassigned_cpus,
+				cpumask_of_node(
+					nvme_find_closest_node(
+						cpu_to_node(cpu))),
+				nvmeq, cpus_per_queue);
+		if (cpus_weight(mask) < cpus_per_queue)
+			nvme_add_cpus(&mask, unassigned_cpus,
+				unassigned_cpus,
+				nvmeq, cpus_per_queue);
+
+		WARN(cpumask_weight(nvmeq->cpu_mask) != cpus_per_queue,
+			"nvme%d qid:%d mis-matched queue-to-cpu assignment\n",
+			dev->instance, i);
+
+		irq_set_affinity_hint(dev->entry[nvmeq->cq_vector].vector,
+							nvmeq->cpu_mask);
+		cpumask_andnot(unassigned_cpus, unassigned_cpus,
+						nvmeq->cpu_mask);
+		cpu = cpumask_next(cpu, unassigned_cpus);
+		if (remainder && !--remainder)
+			cpus_per_queue++;
+		unlock_nvmeq(nvmeq);
+	}
+	WARN(cpumask_weight(unassigned_cpus), "nvme%d unassigned online cpus\n",
+								dev->instance);
+	i = 0;
+	cpumask_andnot(unassigned_cpus, cpu_possible_mask, cpu_online_mask);
+	for_each_cpu(cpu, unassigned_cpus)
+		*per_cpu_ptr(dev->io_queue, cpu) = (i++ % queues) + 1;
+	free_cpumask_var(unassigned_cpus);
+}
+
 static int set_queue_count(struct nvme_dev *dev, int count)
 {
 	int status;
@@ -1805,13 +2072,26 @@ static size_t db_bar_size(struct nvme_dev *dev, unsigned nr_io_queues)
 	return 4096 + ((nr_io_queues + 1) * 8 * dev->db_stride);
 }
 
+static int nvme_cpu_notify(struct notifier_block *self,
+				unsigned long action, void *hcpu)
+{
+	struct nvme_dev *dev = container_of(self, struct nvme_dev, nb);
+	switch (action) {
+	case CPU_ONLINE:
+	case CPU_DEAD:
+		nvme_assign_io_queues(dev);
+		break;
+	}
+	return NOTIFY_OK;
+}
+
 static int nvme_setup_io_queues(struct nvme_dev *dev)
 {
-	struct nvme_queue *adminq = dev->queues[0];
+	struct nvme_queue *adminq = raw_nvmeq(dev, 0);
 	struct pci_dev *pdev = dev->pci_dev;
-	int result, cpu, i, vecs, nr_io_queues, size, q_depth;
+	int result, i, vecs, nr_io_queues, size;
 
-	nr_io_queues = num_online_cpus();
+	nr_io_queues = num_possible_cpus();
 	result = set_queue_count(dev, nr_io_queues);
 	if (result < 0)
 		return result;
@@ -1830,7 +2110,7 @@ static int nvme_setup_io_queues(struct nvme_dev *dev)
 			size = db_bar_size(dev, nr_io_queues);
 		} while (1);
 		dev->dbs = ((void __iomem *)dev->bar) + 4096;
-		dev->queues[0]->q_db = dev->dbs;
+		adminq->q_db = dev->dbs;
 	}
 
 	/* Deregister the admin queue's interrupt */
@@ -1856,6 +2136,7 @@ static int nvme_setup_io_queues(struct nvme_dev *dev)
 	 * number of interrupts.
 	 */
 	nr_io_queues = vecs;
+	dev->max_qid = nr_io_queues;
 
 	result = queue_request_irq(dev, adminq, adminq->irqname);
 	if (result) {
@@ -1864,49 +2145,13 @@ static int nvme_setup_io_queues(struct nvme_dev *dev)
 	}
 
 	/* Free previously allocated queues that are no longer usable */
-	spin_lock(&dev_list_lock);
-	for (i = dev->queue_count - 1; i > nr_io_queues; i--) {
-		struct nvme_queue *nvmeq = dev->queues[i];
-
-		spin_lock_irq(&nvmeq->q_lock);
-		nvme_cancel_ios(nvmeq, false);
-		spin_unlock_irq(&nvmeq->q_lock);
-
-		nvme_free_queue(nvmeq);
-		dev->queue_count--;
-		dev->queues[i] = NULL;
-	}
-	spin_unlock(&dev_list_lock);
-
-	cpu = cpumask_first(cpu_online_mask);
-	for (i = 0; i < nr_io_queues; i++) {
-		irq_set_affinity_hint(dev->entry[i].vector, get_cpu_mask(cpu));
-		cpu = cpumask_next(cpu, cpu_online_mask);
-	}
-
-	q_depth = min_t(int, NVME_CAP_MQES(readq(&dev->bar->cap)) + 1,
-								NVME_Q_DEPTH);
-	for (i = dev->queue_count - 1; i < nr_io_queues; i++) {
-		dev->queues[i + 1] = nvme_alloc_queue(dev, i + 1, q_depth, i);
-		if (!dev->queues[i + 1]) {
-			result = -ENOMEM;
-			goto free_queues;
-		}
-	}
-
-	for (; i < num_possible_cpus(); i++) {
-		int target = i % rounddown_pow_of_two(dev->queue_count - 1);
-		dev->queues[i + 1] = dev->queues[target + 1];
-	}
+	nvme_free_queues(dev, nr_io_queues + 1);
+	nvme_assign_io_queues(dev);
 
-	for (i = 1; i < dev->queue_count; i++) {
-		result = nvme_create_queue(dev->queues[i], i);
-		if (result) {
-			for (--i; i > 0; i--)
-				nvme_disable_queue(dev, i);
-			goto free_queues;
-		}
-	}
+	dev->nb.notifier_call = &nvme_cpu_notify;
+	result = register_hotcpu_notifier(&dev->nb);
+	if (result)
+		goto free_queues;
 
 	return 0;
 
@@ -1985,6 +2230,7 @@ static int nvme_dev_add(struct nvme_dev *dev)
 
 static int nvme_dev_map(struct nvme_dev *dev)
 {
+	u64 cap;
 	int bars, result = -ENOMEM;
 	struct pci_dev *pdev = dev->pci_dev;
 
@@ -2008,7 +2254,9 @@ static int nvme_dev_map(struct nvme_dev *dev)
 		result = -ENODEV;
 		goto unmap;
 	}
-	dev->db_stride = 1 << NVME_CAP_STRIDE(readq(&dev->bar->cap));
+	cap = readq(&dev->bar->cap);
+	dev->q_depth = min_t(int, NVME_CAP_MQES(cap) + 1, NVME_Q_DEPTH);
+	dev->db_stride = 1 << NVME_CAP_STRIDE(cap);
 	dev->dbs = ((void __iomem *)dev->bar) + 4096;
 
 	return 0;
@@ -2164,7 +2412,7 @@ static void nvme_disable_io_queues(struct nvme_dev *dev)
 	atomic_set(&dq.refcount, 0);
 	dq.worker = &worker;
 	for (i = dev->queue_count - 1; i > 0; i--) {
-		struct nvme_queue *nvmeq = dev->queues[i];
+		struct nvme_queue *nvmeq = raw_nvmeq(dev, i);
 
 		if (nvme_suspend_queue(nvmeq))
 			continue;
@@ -2177,19 +2425,38 @@ static void nvme_disable_io_queues(struct nvme_dev *dev)
 	kthread_stop(kworker_task);
 }
 
+/*
+* Remove the node from the device list and check
+* for whether or not we need to stop the nvme_thread.
+*/
+static void nvme_dev_list_remove(struct nvme_dev *dev)
+{
+	struct task_struct *tmp = NULL;
+
+	spin_lock(&dev_list_lock);
+	list_del_init(&dev->node);
+	if (list_empty(&dev_list) && !IS_ERR_OR_NULL(nvme_thread)) {
+		tmp = nvme_thread;
+		nvme_thread = NULL;
+	}
+	spin_unlock(&dev_list_lock);
+
+	if (tmp)
+		kthread_stop(tmp);
+}
+
 static void nvme_dev_shutdown(struct nvme_dev *dev)
 {
 	int i;
 
 	dev->initialized = 0;
+	unregister_hotcpu_notifier(&dev->nb);
 
-	spin_lock(&dev_list_lock);
-	list_del_init(&dev->node);
-	spin_unlock(&dev_list_lock);
+	nvme_dev_list_remove(dev);
 
 	if (!dev->bar || (dev->bar && readl(&dev->bar->csts) == -1)) {
 		for (i = dev->queue_count - 1; i >= 0; i--) {
-			struct nvme_queue *nvmeq = dev->queues[i];
+			struct nvme_queue *nvmeq = raw_nvmeq(dev, i);
 			nvme_suspend_queue(nvmeq);
 			nvme_clear_queue(nvmeq);
 		}
@@ -2282,6 +2549,7 @@ static void nvme_free_dev(struct kref *kref)
 	struct nvme_dev *dev = container_of(kref, struct nvme_dev, kref);
 
 	nvme_free_namespaces(dev);
+	free_percpu(dev->io_queue);
 	kfree(dev->queues);
 	kfree(dev->entry);
 	kfree(dev);
@@ -2325,6 +2593,7 @@ static const struct file_operations nvme_dev_fops = {
 static int nvme_dev_start(struct nvme_dev *dev)
 {
 	int result;
+	bool start_thread = false;
 
 	result = nvme_dev_map(dev);
 	if (result)
@@ -2335,9 +2604,24 @@ static int nvme_dev_start(struct nvme_dev *dev)
 		goto unmap;
 
 	spin_lock(&dev_list_lock);
+	if (list_empty(&dev_list) && IS_ERR_OR_NULL(nvme_thread)) {
+		start_thread = true;
+		nvme_thread = NULL;
+	}
 	list_add(&dev->node, &dev_list);
 	spin_unlock(&dev_list_lock);
 
+	if (start_thread) {
+		nvme_thread = kthread_run(nvme_kthread, NULL, "nvme");
+		wake_up(&nvme_kthread_wait);
+	} else
+		wait_event_killable(nvme_kthread_wait, nvme_thread);
+
+	if (IS_ERR_OR_NULL(nvme_thread)) {
+		result = nvme_thread ? PTR_ERR(nvme_thread) : -EINTR;
+		goto disable;
+	}
+
 	result = nvme_setup_io_queues(dev);
 	if (result && result != -EBUSY)
 		goto disable;
@@ -2346,9 +2630,7 @@ static int nvme_dev_start(struct nvme_dev *dev)
 
  disable:
 	nvme_disable_queue(dev, 0);
-	spin_lock(&dev_list_lock);
-	list_del_init(&dev->node);
-	spin_unlock(&dev_list_lock);
+	nvme_dev_list_remove(dev);
  unmap:
 	nvme_dev_unmap(dev);
 	return result;
@@ -2367,18 +2649,10 @@ static int nvme_remove_dead_ctrl(void *arg)
 
 static void nvme_remove_disks(struct work_struct *ws)
 {
-	int i;
 	struct nvme_dev *dev = container_of(ws, struct nvme_dev, reset_work);
 
 	nvme_dev_remove(dev);
-	spin_lock(&dev_list_lock);
-	for (i = dev->queue_count - 1; i > 0; i--) {
-		BUG_ON(!dev->queues[i] || !dev->queues[i]->q_suspended);
-		nvme_free_queue(dev->queues[i]);
-		dev->queue_count--;
-		dev->queues[i] = NULL;
-	}
-	spin_unlock(&dev_list_lock);
+	nvme_free_queues(dev, 1);
 }
 
 static int nvme_dev_resume(struct nvme_dev *dev)
@@ -2441,6 +2715,9 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 								GFP_KERNEL);
 	if (!dev->queues)
 		goto free;
+	dev->io_queue = alloc_percpu(unsigned short);
+	if (!dev->io_queue)
+		goto free;
 
 	INIT_LIST_HEAD(&dev->namespaces);
 	dev->reset_workfn = nvme_reset_failed_dev;
@@ -2455,6 +2732,7 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	if (result)
 		goto release;
 
+	kref_init(&dev->kref);
 	result = nvme_dev_start(dev);
 	if (result) {
 		if (result == -EBUSY)
@@ -2462,7 +2740,6 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 		goto release_pools;
 	}
 
-	kref_init(&dev->kref);
 	result = nvme_dev_add(dev);
 	if (result)
 		goto shutdown;
@@ -2491,6 +2768,7 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
  release:
 	nvme_release_instance(dev);
  free:
+	free_percpu(dev->io_queue);
 	kfree(dev->queues);
 	kfree(dev->entry);
 	kfree(dev);
@@ -2517,6 +2795,7 @@ static void nvme_remove(struct pci_dev *pdev)
 	nvme_dev_remove(dev);
 	nvme_dev_shutdown(dev);
 	nvme_free_queues(dev, 0);
+	rcu_barrier();
 	nvme_release_instance(dev);
 	nvme_release_prp_pools(dev);
 	kref_put(&dev->kref, nvme_free_dev);
@@ -2529,6 +2808,7 @@ static void nvme_remove(struct pci_dev *pdev)
 #define nvme_slot_reset NULL
 #define nvme_error_resume NULL
 
+#ifdef CONFIG_PM_SLEEP
 static int nvme_suspend(struct device *dev)
 {
 	struct pci_dev *pdev = to_pci_dev(dev);
@@ -2549,6 +2829,7 @@ static int nvme_resume(struct device *dev)
 	}
 	return 0;
 }
+#endif
 
 static SIMPLE_DEV_PM_OPS(nvme_dev_pm_ops, nvme_suspend, nvme_resume);
 
@@ -2563,7 +2844,7 @@ static const struct pci_error_handlers nvme_err_handler = {
 /* Move to pci_ids.h later */
 #define PCI_CLASS_STORAGE_EXPRESS	0x010802
 
-static DEFINE_PCI_DEVICE_TABLE(nvme_id_table) = {
+static const struct pci_device_id nvme_id_table[] = {
 	{ PCI_DEVICE_CLASS(PCI_CLASS_STORAGE_EXPRESS, 0xffffff) },
 	{ 0, }
 };
@@ -2585,14 +2866,11 @@ static int __init nvme_init(void)
 {
 	int result;
 
-	nvme_thread = kthread_run(nvme_kthread, NULL, "nvme");
-	if (IS_ERR(nvme_thread))
-		return PTR_ERR(nvme_thread);
+	init_waitqueue_head(&nvme_kthread_wait);
 
-	result = -ENOMEM;
 	nvme_workq = create_singlethread_workqueue("nvme");
 	if (!nvme_workq)
-		goto kill_kthread;
+		return -ENOMEM;
 
 	result = register_blkdev(nvme_major, "nvme");
 	if (result < 0)
@@ -2609,8 +2887,6 @@ static int __init nvme_init(void)
 	unregister_blkdev(nvme_major, "nvme");
  kill_workq:
 	destroy_workqueue(nvme_workq);
- kill_kthread:
-	kthread_stop(nvme_thread);
 	return result;
 }
 
@@ -2619,11 +2895,11 @@ static void __exit nvme_exit(void)
 	pci_unregister_driver(&nvme_driver);
 	unregister_blkdev(nvme_major, "nvme");
 	destroy_workqueue(nvme_workq);
-	kthread_stop(nvme_thread);
+	BUG_ON(nvme_thread && !IS_ERR(nvme_thread));
 }
 
 MODULE_AUTHOR("Matthew Wilcox <willy@linux.intel.com>");
 MODULE_LICENSE("GPL");
-MODULE_VERSION("0.8");
+MODULE_VERSION("0.9");
 module_init(nvme_init);
 module_exit(nvme_exit);
diff --git a/drivers/block/nvme-scsi.c b/drivers/block/nvme-scsi.c
index 4a0ceb64e269..2c3f5be06da1 100644
--- a/drivers/block/nvme-scsi.c
+++ b/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),
diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c
index 34898d53395b..4c95b503b09e 100644
--- a/drivers/block/rbd.c
+++ b/drivers/block/rbd.c
@@ -1654,7 +1654,7 @@ static void rbd_osd_req_callback(struct ceph_osd_request *osd_req,
 	if (osd_req->r_result < 0)
 		obj_request->result = osd_req->r_result;
 
-	BUG_ON(osd_req->r_num_ops > 2);
+	rbd_assert(osd_req->r_num_ops <= CEPH_OSD_MAX_OP);
 
 	/*
 	 * We support a 64-bit length, but ultimately it has to be
@@ -1662,11 +1662,15 @@ static void rbd_osd_req_callback(struct ceph_osd_request *osd_req,
 	 */
 	obj_request->xferred = osd_req->r_reply_op_len[0];
 	rbd_assert(obj_request->xferred < (u64)UINT_MAX);
+
 	opcode = osd_req->r_ops[0].op;
 	switch (opcode) {
 	case CEPH_OSD_OP_READ:
 		rbd_osd_read_callback(obj_request);
 		break;
+	case CEPH_OSD_OP_SETALLOCHINT:
+		rbd_assert(osd_req->r_ops[1].op == CEPH_OSD_OP_WRITE);
+		/* fall through */
 	case CEPH_OSD_OP_WRITE:
 		rbd_osd_write_callback(obj_request);
 		break;
@@ -1715,9 +1719,16 @@ static void rbd_osd_req_format_write(struct rbd_obj_request *obj_request)
 			snapc, CEPH_NOSNAP, &mtime);
 }
 
+/*
+ * Create an osd request.  A read request has one osd op (read).
+ * A write request has either one (watch) or two (hint+write) osd ops.
+ * (All rbd data writes are prefixed with an allocation hint op, but
+ * technically osd watch is a write request, hence this distinction.)
+ */
 static struct ceph_osd_request *rbd_osd_req_create(
 					struct rbd_device *rbd_dev,
 					bool write_request,
+					unsigned int num_ops,
 					struct rbd_obj_request *obj_request)
 {
 	struct ceph_snap_context *snapc = NULL;
@@ -1733,10 +1744,13 @@ static struct ceph_osd_request *rbd_osd_req_create(
 			snapc = img_request->snapc;
 	}
 
-	/* Allocate and initialize the request, for the single op */
+	rbd_assert(num_ops == 1 || (write_request && num_ops == 2));
+
+	/* Allocate and initialize the request, for the num_ops ops */
 
 	osdc = &rbd_dev->rbd_client->client->osdc;
-	osd_req = ceph_osdc_alloc_request(osdc, snapc, 1, false, GFP_ATOMIC);
+	osd_req = ceph_osdc_alloc_request(osdc, snapc, num_ops, false,
+					  GFP_ATOMIC);
 	if (!osd_req)
 		return NULL;	/* ENOMEM */
 
@@ -1756,8 +1770,8 @@ static struct ceph_osd_request *rbd_osd_req_create(
 
 /*
  * Create a copyup osd request based on the information in the
- * object request supplied.  A copyup request has two osd ops,
- * a copyup method call, and a "normal" write request.
+ * object request supplied.  A copyup request has three osd ops,
+ * a copyup method call, a hint op, and a write op.
  */
 static struct ceph_osd_request *
 rbd_osd_req_create_copyup(struct rbd_obj_request *obj_request)
@@ -1773,12 +1787,12 @@ rbd_osd_req_create_copyup(struct rbd_obj_request *obj_request)
 	rbd_assert(img_request);
 	rbd_assert(img_request_write_test(img_request));
 
-	/* Allocate and initialize the request, for the two ops */
+	/* Allocate and initialize the request, for the three ops */
 
 	snapc = img_request->snapc;
 	rbd_dev = img_request->rbd_dev;
 	osdc = &rbd_dev->rbd_client->client->osdc;
-	osd_req = ceph_osdc_alloc_request(osdc, snapc, 2, false, GFP_ATOMIC);
+	osd_req = ceph_osdc_alloc_request(osdc, snapc, 3, false, GFP_ATOMIC);
 	if (!osd_req)
 		return NULL;	/* ENOMEM */
 
@@ -2178,6 +2192,7 @@ static int rbd_img_request_fill(struct rbd_img_request *img_request,
 		const char *object_name;
 		u64 offset;
 		u64 length;
+		unsigned int which = 0;
 
 		object_name = rbd_segment_name(rbd_dev, img_offset);
 		if (!object_name)
@@ -2190,6 +2205,7 @@ static int rbd_img_request_fill(struct rbd_img_request *img_request,
 		rbd_segment_name_free(object_name);
 		if (!obj_request)
 			goto out_unwind;
+
 		/*
 		 * set obj_request->img_request before creating the
 		 * osd_request so that it gets the right snapc
@@ -2207,7 +2223,7 @@ static int rbd_img_request_fill(struct rbd_img_request *img_request,
 								clone_size,
 								GFP_ATOMIC);
 			if (!obj_request->bio_list)
-				goto out_partial;
+				goto out_unwind;
 		} else {
 			unsigned int page_count;
 
@@ -2220,19 +2236,27 @@ static int rbd_img_request_fill(struct rbd_img_request *img_request,
 		}
 
 		osd_req = rbd_osd_req_create(rbd_dev, write_request,
-						obj_request);
+					     (write_request ? 2 : 1),
+					     obj_request);
 		if (!osd_req)
-			goto out_partial;
+			goto out_unwind;
 		obj_request->osd_req = osd_req;
 		obj_request->callback = rbd_img_obj_callback;
 
-		osd_req_op_extent_init(osd_req, 0, opcode, offset, length,
-						0, 0);
+		if (write_request) {
+			osd_req_op_alloc_hint_init(osd_req, which,
+					     rbd_obj_bytes(&rbd_dev->header),
+					     rbd_obj_bytes(&rbd_dev->header));
+			which++;
+		}
+
+		osd_req_op_extent_init(osd_req, which, opcode, offset, length,
+				       0, 0);
 		if (type == OBJ_REQUEST_BIO)
-			osd_req_op_extent_osd_data_bio(osd_req, 0,
+			osd_req_op_extent_osd_data_bio(osd_req, which,
 					obj_request->bio_list, length);
 		else
-			osd_req_op_extent_osd_data_pages(osd_req, 0,
+			osd_req_op_extent_osd_data_pages(osd_req, which,
 					obj_request->pages, length,
 					offset & ~PAGE_MASK, false, false);
 
@@ -2249,11 +2273,9 @@ static int rbd_img_request_fill(struct rbd_img_request *img_request,
 
 	return 0;
 
-out_partial:
-	rbd_obj_request_put(obj_request);
 out_unwind:
 	for_each_obj_request_safe(img_request, obj_request, next_obj_request)
-		rbd_obj_request_put(obj_request);
+		rbd_img_obj_request_del(img_request, obj_request);
 
 	return -ENOMEM;
 }
@@ -2353,7 +2375,7 @@ rbd_img_obj_parent_read_full_callback(struct rbd_img_request *img_request)
 
 	/*
 	 * The original osd request is of no use to use any more.
-	 * We need a new one that can hold the two ops in a copyup
+	 * We need a new one that can hold the three ops in a copyup
 	 * request.  Allocate the new copyup osd request for the
 	 * original request, and release the old one.
 	 */
@@ -2372,17 +2394,22 @@ rbd_img_obj_parent_read_full_callback(struct rbd_img_request *img_request)
 	osd_req_op_cls_request_data_pages(osd_req, 0, pages, parent_length, 0,
 						false, false);
 
-	/* Then the original write request op */
+	/* Then the hint op */
+
+	osd_req_op_alloc_hint_init(osd_req, 1, rbd_obj_bytes(&rbd_dev->header),
+				   rbd_obj_bytes(&rbd_dev->header));
+
+	/* And the original write request op */
 
 	offset = orig_request->offset;
 	length = orig_request->length;
-	osd_req_op_extent_init(osd_req, 1, CEPH_OSD_OP_WRITE,
+	osd_req_op_extent_init(osd_req, 2, CEPH_OSD_OP_WRITE,
 					offset, length, 0, 0);
 	if (orig_request->type == OBJ_REQUEST_BIO)
-		osd_req_op_extent_osd_data_bio(osd_req, 1,
+		osd_req_op_extent_osd_data_bio(osd_req, 2,
 					orig_request->bio_list, length);
 	else
-		osd_req_op_extent_osd_data_pages(osd_req, 1,
+		osd_req_op_extent_osd_data_pages(osd_req, 2,
 					orig_request->pages, length,
 					offset & ~PAGE_MASK, false, false);
 
@@ -2603,8 +2630,8 @@ static int rbd_img_obj_exists_submit(struct rbd_obj_request *obj_request)
 
 	rbd_assert(obj_request->img_request);
 	rbd_dev = obj_request->img_request->rbd_dev;
-	stat_request->osd_req = rbd_osd_req_create(rbd_dev, false,
-						stat_request);
+	stat_request->osd_req = rbd_osd_req_create(rbd_dev, false, 1,
+						   stat_request);
 	if (!stat_request->osd_req)
 		goto out;
 	stat_request->callback = rbd_img_obj_exists_callback;
@@ -2807,7 +2834,8 @@ static int rbd_obj_notify_ack_sync(struct rbd_device *rbd_dev, u64 notify_id)
 		return -ENOMEM;
 
 	ret = -ENOMEM;
-	obj_request->osd_req = rbd_osd_req_create(rbd_dev, false, obj_request);
+	obj_request->osd_req = rbd_osd_req_create(rbd_dev, false, 1,
+						  obj_request);
 	if (!obj_request->osd_req)
 		goto out;
 
@@ -2870,7 +2898,8 @@ static int __rbd_dev_header_watch_sync(struct rbd_device *rbd_dev, bool start)
 	if (!obj_request)
 		goto out_cancel;
 
-	obj_request->osd_req = rbd_osd_req_create(rbd_dev, true, obj_request);
+	obj_request->osd_req = rbd_osd_req_create(rbd_dev, true, 1,
+						  obj_request);
 	if (!obj_request->osd_req)
 		goto out_cancel;
 
@@ -2978,7 +3007,8 @@ static int rbd_obj_method_sync(struct rbd_device *rbd_dev,
 	obj_request->pages = pages;
 	obj_request->page_count = page_count;
 
-	obj_request->osd_req = rbd_osd_req_create(rbd_dev, false, obj_request);
+	obj_request->osd_req = rbd_osd_req_create(rbd_dev, false, 1,
+						  obj_request);
 	if (!obj_request->osd_req)
 		goto out;
 
@@ -3211,7 +3241,8 @@ static int rbd_obj_read_sync(struct rbd_device *rbd_dev,
 	obj_request->pages = pages;
 	obj_request->page_count = page_count;
 
-	obj_request->osd_req = rbd_osd_req_create(rbd_dev, false, obj_request);
+	obj_request->osd_req = rbd_osd_req_create(rbd_dev, false, 1,
+						  obj_request);
 	if (!obj_request->osd_req)
 		goto out;
 
diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index 0eace43cea11..6d8a87f252de 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@@ -158,6 +158,7 @@ static int virtio_queue_rq(struct blk_mq_hw_ctx *hctx, struct request *req)
 	unsigned long flags;
 	unsigned int num;
 	const bool last = (req->cmd_flags & REQ_END) != 0;
+	int err;
 
 	BUG_ON(req->nr_phys_segments + 2 > vblk->sg_elems);
 
@@ -198,11 +199,16 @@ static int virtio_queue_rq(struct blk_mq_hw_ctx *hctx, struct request *req)
 	}
 
 	spin_lock_irqsave(&vblk->vq_lock, flags);
-	if (__virtblk_add_req(vblk->vq, vbr, vbr->sg, num) < 0) {
+	err = __virtblk_add_req(vblk->vq, vbr, vbr->sg, num);
+	if (err) {
 		virtqueue_kick(vblk->vq);
 		spin_unlock_irqrestore(&vblk->vq_lock, flags);
 		blk_mq_stop_hw_queue(hctx);
-		return BLK_MQ_RQ_QUEUE_BUSY;
+		/* Out of mem doesn't actually happen, since we fall back
+		 * to direct descriptors */
+		if (err == -ENOMEM || err == -ENOSPC)
+			return BLK_MQ_RQ_QUEUE_BUSY;
+		return BLK_MQ_RQ_QUEUE_ERROR;
 	}
 
 	if (last)
@@ -485,10 +491,11 @@ static struct blk_mq_ops virtio_mq_ops = {
 static struct blk_mq_reg virtio_mq_reg = {
 	.ops		= &virtio_mq_ops,
 	.nr_hw_queues	= 1,
-	.queue_depth	= 64,
+	.queue_depth	= 0, /* Set in virtblk_probe */
 	.numa_node	= NUMA_NO_NODE,
 	.flags		= BLK_MQ_F_SHOULD_MERGE,
 };
+module_param_named(queue_depth, virtio_mq_reg.queue_depth, uint, 0444);
 
 static int virtblk_init_vbr(void *data, struct blk_mq_hw_ctx *hctx,
 			     struct request *rq, unsigned int nr)
@@ -553,6 +560,13 @@ static int virtblk_probe(struct virtio_device *vdev)
 		goto out_free_vq;
 	}
 
+	/* Default queue sizing is to fill the ring. */
+	if (!virtio_mq_reg.queue_depth) {
+		virtio_mq_reg.queue_depth = vblk->vq->num_free;
+		/* ... but without indirect descs, we use 2 descs per req */
+		if (!virtio_has_feature(vdev, VIRTIO_RING_F_INDIRECT_DESC))
+			virtio_mq_reg.queue_depth /= 2;
+	}
 	virtio_mq_reg.cmd_size =
 		sizeof(struct virtblk_req) +
 		sizeof(struct scatterlist) * sg_elems;
diff --git a/drivers/block/zram/Kconfig b/drivers/block/zram/Kconfig
index 3450be850399..6489c0fd0ea6 100644
--- a/drivers/block/zram/Kconfig
+++ b/drivers/block/zram/Kconfig
@@ -15,6 +15,16 @@ config ZRAM
 
 	  See zram.txt for more information.
 
+config ZRAM_LZ4_COMPRESS
+	bool "Enable LZ4 algorithm support"
+	depends on ZRAM
+	select LZ4_COMPRESS
+	select LZ4_DECOMPRESS
+	default n
+	help
+	  This option enables LZ4 compression algorithm support. Compression
+	  algorithm can be changed using `comp_algorithm' device attribute.
+
 config ZRAM_DEBUG
 	bool "Compressed RAM block device debug support"
 	depends on ZRAM
diff --git a/drivers/block/zram/Makefile b/drivers/block/zram/Makefile
index cb0f9ced6a93..be0763ff57a2 100644
--- a/drivers/block/zram/Makefile
+++ b/drivers/block/zram/Makefile
@@ -1,3 +1,5 @@
-zram-y	:=	zram_drv.o
+zram-y	:=	zcomp_lzo.o zcomp.o zram_drv.o
+
+zram-$(CONFIG_ZRAM_LZ4_COMPRESS) += zcomp_lz4.o
 
 obj-$(CONFIG_ZRAM)	+=	zram.o
diff --git a/drivers/block/zram/zcomp.c b/drivers/block/zram/zcomp.c
new file mode 100644
index 000000000000..f1ff39a3d1c1
--- /dev/null
+++ b/drivers/block/zram/zcomp.c
@@ -0,0 +1,353 @@
+/*
+ * Copyright (C) 2014 Sergey Senozhatsky.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/err.h>
+#include <linux/slab.h>
+#include <linux/wait.h>
+#include <linux/sched.h>
+
+#include "zcomp.h"
+#include "zcomp_lzo.h"
+#ifdef CONFIG_ZRAM_LZ4_COMPRESS
+#include "zcomp_lz4.h"
+#endif
+
+/*
+ * single zcomp_strm backend
+ */
+struct zcomp_strm_single {
+	struct mutex strm_lock;
+	struct zcomp_strm *zstrm;
+};
+
+/*
+ * multi zcomp_strm backend
+ */
+struct zcomp_strm_multi {
+	/* protect strm list */
+	spinlock_t strm_lock;
+	/* max possible number of zstrm streams */
+	int max_strm;
+	/* number of available zstrm streams */
+	int avail_strm;
+	/* list of available strms */
+	struct list_head idle_strm;
+	wait_queue_head_t strm_wait;
+};
+
+static struct zcomp_backend *backends[] = {
+	&zcomp_lzo,
+#ifdef CONFIG_ZRAM_LZ4_COMPRESS
+	&zcomp_lz4,
+#endif
+	NULL
+};
+
+static struct zcomp_backend *find_backend(const char *compress)
+{
+	int i = 0;
+	while (backends[i]) {
+		if (sysfs_streq(compress, backends[i]->name))
+			break;
+		i++;
+	}
+	return backends[i];
+}
+
+static void zcomp_strm_free(struct zcomp *comp, struct zcomp_strm *zstrm)
+{
+	if (zstrm->private)
+		comp->backend->destroy(zstrm->private);
+	free_pages((unsigned long)zstrm->buffer, 1);
+	kfree(zstrm);
+}
+
+/*
+ * allocate new zcomp_strm structure with ->private initialized by
+ * backend, return NULL on error
+ */
+static struct zcomp_strm *zcomp_strm_alloc(struct zcomp *comp)
+{
+	struct zcomp_strm *zstrm = kmalloc(sizeof(*zstrm), GFP_KERNEL);
+	if (!zstrm)
+		return NULL;
+
+	zstrm->private = comp->backend->create();
+	/*
+	 * allocate 2 pages. 1 for compressed data, plus 1 extra for the
+	 * case when compressed size is larger than the original one
+	 */
+	zstrm->buffer = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO, 1);
+	if (!zstrm->private || !zstrm->buffer) {
+		zcomp_strm_free(comp, zstrm);
+		zstrm = NULL;
+	}
+	return zstrm;
+}
+
+/*
+ * get idle zcomp_strm or wait until other process release
+ * (zcomp_strm_release()) one for us
+ */
+static struct zcomp_strm *zcomp_strm_multi_find(struct zcomp *comp)
+{
+	struct zcomp_strm_multi *zs = comp->stream;
+	struct zcomp_strm *zstrm;
+
+	while (1) {
+		spin_lock(&zs->strm_lock);
+		if (!list_empty(&zs->idle_strm)) {
+			zstrm = list_entry(zs->idle_strm.next,
+					struct zcomp_strm, list);
+			list_del(&zstrm->list);
+			spin_unlock(&zs->strm_lock);
+			return zstrm;
+		}
+		/* zstrm streams limit reached, wait for idle stream */
+		if (zs->avail_strm >= zs->max_strm) {
+			spin_unlock(&zs->strm_lock);
+			wait_event(zs->strm_wait, !list_empty(&zs->idle_strm));
+			continue;
+		}
+		/* allocate new zstrm stream */
+		zs->avail_strm++;
+		spin_unlock(&zs->strm_lock);
+
+		zstrm = zcomp_strm_alloc(comp);
+		if (!zstrm) {
+			spin_lock(&zs->strm_lock);
+			zs->avail_strm--;
+			spin_unlock(&zs->strm_lock);
+			wait_event(zs->strm_wait, !list_empty(&zs->idle_strm));
+			continue;
+		}
+		break;
+	}
+	return zstrm;
+}
+
+/* add stream back to idle list and wake up waiter or free the stream */
+static void zcomp_strm_multi_release(struct zcomp *comp, struct zcomp_strm *zstrm)
+{
+	struct zcomp_strm_multi *zs = comp->stream;
+
+	spin_lock(&zs->strm_lock);
+	if (zs->avail_strm <= zs->max_strm) {
+		list_add(&zstrm->list, &zs->idle_strm);
+		spin_unlock(&zs->strm_lock);
+		wake_up(&zs->strm_wait);
+		return;
+	}
+
+	zs->avail_strm--;
+	spin_unlock(&zs->strm_lock);
+	zcomp_strm_free(comp, zstrm);
+}
+
+/* change max_strm limit */
+static bool zcomp_strm_multi_set_max_streams(struct zcomp *comp, int num_strm)
+{
+	struct zcomp_strm_multi *zs = comp->stream;
+	struct zcomp_strm *zstrm;
+
+	spin_lock(&zs->strm_lock);
+	zs->max_strm = num_strm;
+	/*
+	 * if user has lowered the limit and there are idle streams,
+	 * immediately free as much streams (and memory) as we can.
+	 */
+	while (zs->avail_strm > num_strm && !list_empty(&zs->idle_strm)) {
+		zstrm = list_entry(zs->idle_strm.next,
+				struct zcomp_strm, list);
+		list_del(&zstrm->list);
+		zcomp_strm_free(comp, zstrm);
+		zs->avail_strm--;
+	}
+	spin_unlock(&zs->strm_lock);
+	return true;
+}
+
+static void zcomp_strm_multi_destroy(struct zcomp *comp)
+{
+	struct zcomp_strm_multi *zs = comp->stream;
+	struct zcomp_strm *zstrm;
+
+	while (!list_empty(&zs->idle_strm)) {
+		zstrm = list_entry(zs->idle_strm.next,
+				struct zcomp_strm, list);
+		list_del(&zstrm->list);
+		zcomp_strm_free(comp, zstrm);
+	}
+	kfree(zs);
+}
+
+static int zcomp_strm_multi_create(struct zcomp *comp, int max_strm)
+{
+	struct zcomp_strm *zstrm;
+	struct zcomp_strm_multi *zs;
+
+	comp->destroy = zcomp_strm_multi_destroy;
+	comp->strm_find = zcomp_strm_multi_find;
+	comp->strm_release = zcomp_strm_multi_release;
+	comp->set_max_streams = zcomp_strm_multi_set_max_streams;
+	zs = kmalloc(sizeof(struct zcomp_strm_multi), GFP_KERNEL);
+	if (!zs)
+		return -ENOMEM;
+
+	comp->stream = zs;
+	spin_lock_init(&zs->strm_lock);
+	INIT_LIST_HEAD(&zs->idle_strm);
+	init_waitqueue_head(&zs->strm_wait);
+	zs->max_strm = max_strm;
+	zs->avail_strm = 1;
+
+	zstrm = zcomp_strm_alloc(comp);
+	if (!zstrm) {
+		kfree(zs);
+		return -ENOMEM;
+	}
+	list_add(&zstrm->list, &zs->idle_strm);
+	return 0;
+}
+
+static struct zcomp_strm *zcomp_strm_single_find(struct zcomp *comp)
+{
+	struct zcomp_strm_single *zs = comp->stream;
+	mutex_lock(&zs->strm_lock);
+	return zs->zstrm;
+}
+
+static void zcomp_strm_single_release(struct zcomp *comp,
+		struct zcomp_strm *zstrm)
+{
+	struct zcomp_strm_single *zs = comp->stream;
+	mutex_unlock(&zs->strm_lock);
+}
+
+static bool zcomp_strm_single_set_max_streams(struct zcomp *comp, int num_strm)
+{
+	/* zcomp_strm_single support only max_comp_streams == 1 */
+	return false;
+}
+
+static void zcomp_strm_single_destroy(struct zcomp *comp)
+{
+	struct zcomp_strm_single *zs = comp->stream;
+	zcomp_strm_free(comp, zs->zstrm);
+	kfree(zs);
+}
+
+static int zcomp_strm_single_create(struct zcomp *comp)
+{
+	struct zcomp_strm_single *zs;
+
+	comp->destroy = zcomp_strm_single_destroy;
+	comp->strm_find = zcomp_strm_single_find;
+	comp->strm_release = zcomp_strm_single_release;
+	comp->set_max_streams = zcomp_strm_single_set_max_streams;
+	zs = kmalloc(sizeof(struct zcomp_strm_single), GFP_KERNEL);
+	if (!zs)
+		return -ENOMEM;
+
+	comp->stream = zs;
+	mutex_init(&zs->strm_lock);
+	zs->zstrm = zcomp_strm_alloc(comp);
+	if (!zs->zstrm) {
+		kfree(zs);
+		return -ENOMEM;
+	}
+	return 0;
+}
+
+/* show available compressors */
+ssize_t zcomp_available_show(const char *comp, char *buf)
+{
+	ssize_t sz = 0;
+	int i = 0;
+
+	while (backends[i]) {
+		if (sysfs_streq(comp, backends[i]->name))
+			sz += scnprintf(buf + sz, PAGE_SIZE - sz - 2,
+					"[%s] ", backends[i]->name);
+		else
+			sz += scnprintf(buf + sz, PAGE_SIZE - sz - 2,
+					"%s ", backends[i]->name);
+		i++;
+	}
+	sz += scnprintf(buf + sz, PAGE_SIZE - sz, "\n");
+	return sz;
+}
+
+bool zcomp_set_max_streams(struct zcomp *comp, int num_strm)
+{
+	return comp->set_max_streams(comp, num_strm);
+}
+
+struct zcomp_strm *zcomp_strm_find(struct zcomp *comp)
+{
+	return comp->strm_find(comp);
+}
+
+void zcomp_strm_release(struct zcomp *comp, struct zcomp_strm *zstrm)
+{
+	comp->strm_release(comp, zstrm);
+}
+
+int zcomp_compress(struct zcomp *comp, struct zcomp_strm *zstrm,
+		const unsigned char *src, size_t *dst_len)
+{
+	return comp->backend->compress(src, zstrm->buffer, dst_len,
+			zstrm->private);
+}
+
+int zcomp_decompress(struct zcomp *comp, const unsigned char *src,
+		size_t src_len, unsigned char *dst)
+{
+	return comp->backend->decompress(src, src_len, dst);
+}
+
+void zcomp_destroy(struct zcomp *comp)
+{
+	comp->destroy(comp);
+	kfree(comp);
+}
+
+/*
+ * search available compressors for requested algorithm.
+ * allocate new zcomp and initialize it. return compressing
+ * backend pointer or ERR_PTR if things went bad. ERR_PTR(-EINVAL)
+ * if requested algorithm is not supported, ERR_PTR(-ENOMEM) in
+ * case of allocation error.
+ */
+struct zcomp *zcomp_create(const char *compress, int max_strm)
+{
+	struct zcomp *comp;
+	struct zcomp_backend *backend;
+
+	backend = find_backend(compress);
+	if (!backend)
+		return ERR_PTR(-EINVAL);
+
+	comp = kzalloc(sizeof(struct zcomp), GFP_KERNEL);
+	if (!comp)
+		return ERR_PTR(-ENOMEM);
+
+	comp->backend = backend;
+	if (max_strm > 1)
+		zcomp_strm_multi_create(comp, max_strm);
+	else
+		zcomp_strm_single_create(comp);
+	if (!comp->stream) {
+		kfree(comp);
+		return ERR_PTR(-ENOMEM);
+	}
+	return comp;
+}
diff --git a/drivers/block/zram/zcomp.h b/drivers/block/zram/zcomp.h
new file mode 100644
index 000000000000..c59d1fca72c0
--- /dev/null
+++ b/drivers/block/zram/zcomp.h
@@ -0,0 +1,68 @@
+/*
+ * Copyright (C) 2014 Sergey Senozhatsky.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#ifndef _ZCOMP_H_
+#define _ZCOMP_H_
+
+#include <linux/mutex.h>
+
+struct zcomp_strm {
+	/* compression/decompression buffer */
+	void *buffer;
+	/*
+	 * The private data of the compression stream, only compression
+	 * stream backend can touch this (e.g. compression algorithm
+	 * working memory)
+	 */
+	void *private;
+	/* used in multi stream backend, protected by backend strm_lock */
+	struct list_head list;
+};
+
+/* static compression backend */
+struct zcomp_backend {
+	int (*compress)(const unsigned char *src, unsigned char *dst,
+			size_t *dst_len, void *private);
+
+	int (*decompress)(const unsigned char *src, size_t src_len,
+			unsigned char *dst);
+
+	void *(*create)(void);
+	void (*destroy)(void *private);
+
+	const char *name;
+};
+
+/* dynamic per-device compression frontend */
+struct zcomp {
+	void *stream;
+	struct zcomp_backend *backend;
+
+	struct zcomp_strm *(*strm_find)(struct zcomp *comp);
+	void (*strm_release)(struct zcomp *comp, struct zcomp_strm *zstrm);
+	bool (*set_max_streams)(struct zcomp *comp, int num_strm);
+	void (*destroy)(struct zcomp *comp);
+};
+
+ssize_t zcomp_available_show(const char *comp, char *buf);
+
+struct zcomp *zcomp_create(const char *comp, int max_strm);
+void zcomp_destroy(struct zcomp *comp);
+
+struct zcomp_strm *zcomp_strm_find(struct zcomp *comp);
+void zcomp_strm_release(struct zcomp *comp, struct zcomp_strm *zstrm);
+
+int zcomp_compress(struct zcomp *comp, struct zcomp_strm *zstrm,
+		const unsigned char *src, size_t *dst_len);
+
+int zcomp_decompress(struct zcomp *comp, const unsigned char *src,
+		size_t src_len, unsigned char *dst);
+
+bool zcomp_set_max_streams(struct zcomp *comp, int num_strm);
+#endif /* _ZCOMP_H_ */
diff --git a/drivers/block/zram/zcomp_lz4.c b/drivers/block/zram/zcomp_lz4.c
new file mode 100644
index 000000000000..f2afb7e988c3
--- /dev/null
+++ b/drivers/block/zram/zcomp_lz4.c
@@ -0,0 +1,47 @@
+/*
+ * Copyright (C) 2014 Sergey Senozhatsky.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/lz4.h>
+
+#include "zcomp_lz4.h"
+
+static void *zcomp_lz4_create(void)
+{
+	return kzalloc(LZ4_MEM_COMPRESS, GFP_KERNEL);
+}
+
+static void zcomp_lz4_destroy(void *private)
+{
+	kfree(private);
+}
+
+static int zcomp_lz4_compress(const unsigned char *src, unsigned char *dst,
+		size_t *dst_len, void *private)
+{
+	/* return  : Success if return 0 */
+	return lz4_compress(src, PAGE_SIZE, dst, dst_len, private);
+}
+
+static int zcomp_lz4_decompress(const unsigned char *src, size_t src_len,
+		unsigned char *dst)
+{
+	size_t dst_len = PAGE_SIZE;
+	/* return  : Success if return 0 */
+	return lz4_decompress_unknownoutputsize(src, src_len, dst, &dst_len);
+}
+
+struct zcomp_backend zcomp_lz4 = {
+	.compress = zcomp_lz4_compress,
+	.decompress = zcomp_lz4_decompress,
+	.create = zcomp_lz4_create,
+	.destroy = zcomp_lz4_destroy,
+	.name = "lz4",
+};
diff --git a/drivers/block/zram/zcomp_lz4.h b/drivers/block/zram/zcomp_lz4.h
new file mode 100644
index 000000000000..60613fb29dd8
--- /dev/null
+++ b/drivers/block/zram/zcomp_lz4.h
@@ -0,0 +1,17 @@
+/*
+ * Copyright (C) 2014 Sergey Senozhatsky.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#ifndef _ZCOMP_LZ4_H_
+#define _ZCOMP_LZ4_H_
+
+#include "zcomp.h"
+
+extern struct zcomp_backend zcomp_lz4;
+
+#endif /* _ZCOMP_LZ4_H_ */
diff --git a/drivers/block/zram/zcomp_lzo.c b/drivers/block/zram/zcomp_lzo.c
new file mode 100644
index 000000000000..da1bc47d588e
--- /dev/null
+++ b/drivers/block/zram/zcomp_lzo.c
@@ -0,0 +1,47 @@
+/*
+ * Copyright (C) 2014 Sergey Senozhatsky.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/lzo.h>
+
+#include "zcomp_lzo.h"
+
+static void *lzo_create(void)
+{
+	return kzalloc(LZO1X_MEM_COMPRESS, GFP_KERNEL);
+}
+
+static void lzo_destroy(void *private)
+{
+	kfree(private);
+}
+
+static int lzo_compress(const unsigned char *src, unsigned char *dst,
+		size_t *dst_len, void *private)
+{
+	int ret = lzo1x_1_compress(src, PAGE_SIZE, dst, dst_len, private);
+	return ret == LZO_E_OK ? 0 : ret;
+}
+
+static int lzo_decompress(const unsigned char *src, size_t src_len,
+		unsigned char *dst)
+{
+	size_t dst_len = PAGE_SIZE;
+	int ret = lzo1x_decompress_safe(src, src_len, dst, &dst_len);
+	return ret == LZO_E_OK ? 0 : ret;
+}
+
+struct zcomp_backend zcomp_lzo = {
+	.compress = lzo_compress,
+	.decompress = lzo_decompress,
+	.create = lzo_create,
+	.destroy = lzo_destroy,
+	.name = "lzo",
+};
diff --git a/drivers/block/zram/zcomp_lzo.h b/drivers/block/zram/zcomp_lzo.h
new file mode 100644
index 000000000000..128c5807fa14
--- /dev/null
+++ b/drivers/block/zram/zcomp_lzo.h
@@ -0,0 +1,17 @@
+/*
+ * Copyright (C) 2014 Sergey Senozhatsky.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#ifndef _ZCOMP_LZO_H_
+#define _ZCOMP_LZO_H_
+
+#include "zcomp.h"
+
+extern struct zcomp_backend zcomp_lzo;
+
+#endif /* _ZCOMP_LZO_H_ */
diff --git a/drivers/block/zram/zram_drv.c b/drivers/block/zram/zram_drv.c
index 51c557cfd92b..9849b5233bf4 100644
--- a/drivers/block/zram/zram_drv.c
+++ b/drivers/block/zram/zram_drv.c
@@ -29,19 +29,36 @@
 #include <linux/genhd.h>
 #include <linux/highmem.h>
 #include <linux/slab.h>
-#include <linux/lzo.h>
 #include <linux/string.h>
 #include <linux/vmalloc.h>
+#include <linux/err.h>
 
 #include "zram_drv.h"
 
 /* Globals */
 static int zram_major;
 static struct zram *zram_devices;
+static const char *default_compressor = "lzo";
 
 /* Module params (documentation at end) */
 static unsigned int num_devices = 1;
 
+#define ZRAM_ATTR_RO(name)						\
+static ssize_t zram_attr_##name##_show(struct device *d,		\
+				struct device_attribute *attr, char *b)	\
+{									\
+	struct zram *zram = dev_to_zram(d);				\
+	return scnprintf(b, PAGE_SIZE, "%llu\n",			\
+		(u64)atomic64_read(&zram->stats.name));			\
+}									\
+static struct device_attribute dev_attr_##name =			\
+	__ATTR(name, S_IRUGO, zram_attr_##name##_show, NULL);
+
+static inline int init_done(struct zram *zram)
+{
+	return zram->meta != NULL;
+}
+
 static inline struct zram *dev_to_zram(struct device *dev)
 {
 	return (struct zram *)dev_to_disk(dev)->private_data;
@@ -52,92 +69,114 @@ static ssize_t disksize_show(struct device *dev,
 {
 	struct zram *zram = dev_to_zram(dev);
 
-	return sprintf(buf, "%llu\n", zram->disksize);
+	return scnprintf(buf, PAGE_SIZE, "%llu\n", zram->disksize);
 }
 
 static ssize_t initstate_show(struct device *dev,
 		struct device_attribute *attr, char *buf)
 {
+	u32 val;
 	struct zram *zram = dev_to_zram(dev);
 
-	return sprintf(buf, "%u\n", zram->init_done);
-}
-
-static ssize_t num_reads_show(struct device *dev,
-		struct device_attribute *attr, char *buf)
-{
-	struct zram *zram = dev_to_zram(dev);
+	down_read(&zram->init_lock);
+	val = init_done(zram);
+	up_read(&zram->init_lock);
 
-	return sprintf(buf, "%llu\n",
-			(u64)atomic64_read(&zram->stats.num_reads));
+	return scnprintf(buf, PAGE_SIZE, "%u\n", val);
 }
 
-static ssize_t num_writes_show(struct device *dev,
+static ssize_t orig_data_size_show(struct device *dev,
 		struct device_attribute *attr, char *buf)
 {
 	struct zram *zram = dev_to_zram(dev);
 
-	return sprintf(buf, "%llu\n",
-			(u64)atomic64_read(&zram->stats.num_writes));
+	return scnprintf(buf, PAGE_SIZE, "%llu\n",
+		(u64)(atomic64_read(&zram->stats.pages_stored)) << PAGE_SHIFT);
 }
 
-static ssize_t invalid_io_show(struct device *dev,
+static ssize_t mem_used_total_show(struct device *dev,
 		struct device_attribute *attr, char *buf)
 {
+	u64 val = 0;
 	struct zram *zram = dev_to_zram(dev);
+	struct zram_meta *meta = zram->meta;
 
-	return sprintf(buf, "%llu\n",
-			(u64)atomic64_read(&zram->stats.invalid_io));
-}
-
-static ssize_t notify_free_show(struct device *dev,
-		struct device_attribute *attr, char *buf)
-{
-	struct zram *zram = dev_to_zram(dev);
+	down_read(&zram->init_lock);
+	if (init_done(zram))
+		val = zs_get_total_size_bytes(meta->mem_pool);
+	up_read(&zram->init_lock);
 
-	return sprintf(buf, "%llu\n",
-			(u64)atomic64_read(&zram->stats.notify_free));
+	return scnprintf(buf, PAGE_SIZE, "%llu\n", val);
 }
 
-static ssize_t zero_pages_show(struct device *dev,
+static ssize_t max_comp_streams_show(struct device *dev,
 		struct device_attribute *attr, char *buf)
 {
+	int val;
 	struct zram *zram = dev_to_zram(dev);
 
-	return sprintf(buf, "%u\n", atomic_read(&zram->stats.pages_zero));
+	down_read(&zram->init_lock);
+	val = zram->max_comp_streams;
+	up_read(&zram->init_lock);
+
+	return scnprintf(buf, PAGE_SIZE, "%d\n", val);
 }
 
-static ssize_t orig_data_size_show(struct device *dev,
-		struct device_attribute *attr, char *buf)
+static ssize_t max_comp_streams_store(struct device *dev,
+		struct device_attribute *attr, const char *buf, size_t len)
 {
+	int num;
 	struct zram *zram = dev_to_zram(dev);
+	int ret;
 
-	return sprintf(buf, "%llu\n",
-		(u64)(atomic_read(&zram->stats.pages_stored)) << PAGE_SHIFT);
-}
+	ret = kstrtoint(buf, 0, &num);
+	if (ret < 0)
+		return ret;
+	if (num < 1)
+		return -EINVAL;
 
-static ssize_t compr_data_size_show(struct device *dev,
-		struct device_attribute *attr, char *buf)
-{
-	struct zram *zram = dev_to_zram(dev);
+	down_write(&zram->init_lock);
+	if (init_done(zram)) {
+		if (!zcomp_set_max_streams(zram->comp, num)) {
+			pr_info("Cannot change max compression streams\n");
+			ret = -EINVAL;
+			goto out;
+		}
+	}
 
-	return sprintf(buf, "%llu\n",
-			(u64)atomic64_read(&zram->stats.compr_size));
+	zram->max_comp_streams = num;
+	ret = len;
+out:
+	up_write(&zram->init_lock);
+	return ret;
 }
 
-static ssize_t mem_used_total_show(struct device *dev,
+static ssize_t comp_algorithm_show(struct device *dev,
 		struct device_attribute *attr, char *buf)
 {
-	u64 val = 0;
+	size_t sz;
 	struct zram *zram = dev_to_zram(dev);
-	struct zram_meta *meta = zram->meta;
 
 	down_read(&zram->init_lock);
-	if (zram->init_done)
-		val = zs_get_total_size_bytes(meta->mem_pool);
+	sz = zcomp_available_show(zram->compressor, buf);
 	up_read(&zram->init_lock);
 
-	return sprintf(buf, "%llu\n", val);
+	return sz;
+}
+
+static ssize_t comp_algorithm_store(struct device *dev,
+		struct device_attribute *attr, const char *buf, size_t len)
+{
+	struct zram *zram = dev_to_zram(dev);
+	down_write(&zram->init_lock);
+	if (init_done(zram)) {
+		up_write(&zram->init_lock);
+		pr_info("Can't change algorithm for initialized device\n");
+		return -EBUSY;
+	}
+	strlcpy(zram->compressor, buf, sizeof(zram->compressor));
+	up_write(&zram->init_lock);
+	return len;
 }
 
 /* flag operations needs meta->tb_lock */
@@ -192,8 +231,6 @@ static inline int valid_io_request(struct zram *zram, struct bio *bio)
 static void zram_meta_free(struct zram_meta *meta)
 {
 	zs_destroy_pool(meta->mem_pool);
-	kfree(meta->compress_workmem);
-	free_pages((unsigned long)meta->compress_buffer, 1);
 	vfree(meta->table);
 	kfree(meta);
 }
@@ -205,22 +242,11 @@ static struct zram_meta *zram_meta_alloc(u64 disksize)
 	if (!meta)
 		goto out;
 
-	meta->compress_workmem = kzalloc(LZO1X_MEM_COMPRESS, GFP_KERNEL);
-	if (!meta->compress_workmem)
-		goto free_meta;
-
-	meta->compress_buffer =
-		(void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO, 1);
-	if (!meta->compress_buffer) {
-		pr_err("Error allocating compressor buffer space\n");
-		goto free_workmem;
-	}
-
 	num_pages = disksize >> PAGE_SHIFT;
 	meta->table = vzalloc(num_pages * sizeof(*meta->table));
 	if (!meta->table) {
 		pr_err("Error allocating zram address table\n");
-		goto free_buffer;
+		goto free_meta;
 	}
 
 	meta->mem_pool = zs_create_pool(GFP_NOIO | __GFP_HIGHMEM);
@@ -230,15 +256,10 @@ static struct zram_meta *zram_meta_alloc(u64 disksize)
 	}
 
 	rwlock_init(&meta->tb_lock);
-	mutex_init(&meta->buffer_lock);
 	return meta;
 
 free_table:
 	vfree(meta->table);
-free_buffer:
-	free_pages((unsigned long)meta->compress_buffer, 1);
-free_workmem:
-	kfree(meta->compress_workmem);
 free_meta:
 	kfree(meta);
 	meta = NULL;
@@ -288,7 +309,6 @@ static void zram_free_page(struct zram *zram, size_t index)
 {
 	struct zram_meta *meta = zram->meta;
 	unsigned long handle = meta->table[index].handle;
-	u16 size = meta->table[index].size;
 
 	if (unlikely(!handle)) {
 		/*
@@ -297,21 +317,15 @@ static void zram_free_page(struct zram *zram, size_t index)
 		 */
 		if (zram_test_flag(meta, index, ZRAM_ZERO)) {
 			zram_clear_flag(meta, index, ZRAM_ZERO);
-			atomic_dec(&zram->stats.pages_zero);
+			atomic64_dec(&zram->stats.zero_pages);
 		}
 		return;
 	}
 
-	if (unlikely(size > max_zpage_size))
-		atomic_dec(&zram->stats.bad_compress);
-
 	zs_free(meta->mem_pool, handle);
 
-	if (size <= PAGE_SIZE / 2)
-		atomic_dec(&zram->stats.good_compress);
-
-	atomic64_sub(meta->table[index].size, &zram->stats.compr_size);
-	atomic_dec(&zram->stats.pages_stored);
+	atomic64_sub(meta->table[index].size, &zram->stats.compr_data_size);
+	atomic64_dec(&zram->stats.pages_stored);
 
 	meta->table[index].handle = 0;
 	meta->table[index].size = 0;
@@ -319,8 +333,7 @@ static void zram_free_page(struct zram *zram, size_t index)
 
 static int zram_decompress_page(struct zram *zram, char *mem, u32 index)
 {
-	int ret = LZO_E_OK;
-	size_t clen = PAGE_SIZE;
+	int ret = 0;
 	unsigned char *cmem;
 	struct zram_meta *meta = zram->meta;
 	unsigned long handle;
@@ -340,12 +353,12 @@ static int zram_decompress_page(struct zram *zram, char *mem, u32 index)
 	if (size == PAGE_SIZE)
 		copy_page(mem, cmem);
 	else
-		ret = lzo1x_decompress_safe(cmem, size,	mem, &clen);
+		ret = zcomp_decompress(zram->comp, cmem, size, mem);
 	zs_unmap_object(meta->mem_pool, handle);
 	read_unlock(&meta->tb_lock);
 
 	/* Should NEVER happen. Return bio error if it does. */
-	if (unlikely(ret != LZO_E_OK)) {
+	if (unlikely(ret)) {
 		pr_err("Decompression failed! err=%d, page=%u\n", ret, index);
 		atomic64_inc(&zram->stats.failed_reads);
 		return ret;
@@ -388,7 +401,7 @@ static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec,
 
 	ret = zram_decompress_page(zram, uncmem, index);
 	/* Should NEVER happen. Return bio error if it does. */
-	if (unlikely(ret != LZO_E_OK))
+	if (unlikely(ret))
 		goto out_cleanup;
 
 	if (is_partial_io(bvec))
@@ -413,11 +426,10 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
 	struct page *page;
 	unsigned char *user_mem, *cmem, *src, *uncmem = NULL;
 	struct zram_meta *meta = zram->meta;
+	struct zcomp_strm *zstrm;
 	bool locked = false;
 
 	page = bvec->bv_page;
-	src = meta->compress_buffer;
-
 	if (is_partial_io(bvec)) {
 		/*
 		 * This is a partial IO. We need to read the full page
@@ -433,7 +445,7 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
 			goto out;
 	}
 
-	mutex_lock(&meta->buffer_lock);
+	zstrm = zcomp_strm_find(zram->comp);
 	locked = true;
 	user_mem = kmap_atomic(page);
 
@@ -454,28 +466,25 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
 		zram_set_flag(meta, index, ZRAM_ZERO);
 		write_unlock(&zram->meta->tb_lock);
 
-		atomic_inc(&zram->stats.pages_zero);
+		atomic64_inc(&zram->stats.zero_pages);
 		ret = 0;
 		goto out;
 	}
 
-	ret = lzo1x_1_compress(uncmem, PAGE_SIZE, src, &clen,
-			       meta->compress_workmem);
+	ret = zcomp_compress(zram->comp, zstrm, uncmem, &clen);
 	if (!is_partial_io(bvec)) {
 		kunmap_atomic(user_mem);
 		user_mem = NULL;
 		uncmem = NULL;
 	}
 
-	if (unlikely(ret != LZO_E_OK)) {
+	if (unlikely(ret)) {
 		pr_err("Compression failed! err=%d\n", ret);
 		goto out;
 	}
-
+	src = zstrm->buffer;
 	if (unlikely(clen > max_zpage_size)) {
-		atomic_inc(&zram->stats.bad_compress);
 		clen = PAGE_SIZE;
-		src = NULL;
 		if (is_partial_io(bvec))
 			src = uncmem;
 	}
@@ -497,6 +506,8 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
 		memcpy(cmem, src, clen);
 	}
 
+	zcomp_strm_release(zram->comp, zstrm);
+	locked = false;
 	zs_unmap_object(meta->mem_pool, handle);
 
 	/*
@@ -511,49 +522,88 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
 	write_unlock(&zram->meta->tb_lock);
 
 	/* Update stats */
-	atomic64_add(clen, &zram->stats.compr_size);
-	atomic_inc(&zram->stats.pages_stored);
-	if (clen <= PAGE_SIZE / 2)
-		atomic_inc(&zram->stats.good_compress);
-
+	atomic64_add(clen, &zram->stats.compr_data_size);
+	atomic64_inc(&zram->stats.pages_stored);
 out:
 	if (locked)
-		mutex_unlock(&meta->buffer_lock);
+		zcomp_strm_release(zram->comp, zstrm);
 	if (is_partial_io(bvec))
 		kfree(uncmem);
-
 	if (ret)
 		atomic64_inc(&zram->stats.failed_writes);
 	return ret;
 }
 
 static int zram_bvec_rw(struct zram *zram, struct bio_vec *bvec, u32 index,
-			int offset, struct bio *bio, int rw)
+			int offset, struct bio *bio)
 {
 	int ret;
+	int rw = bio_data_dir(bio);
 
-	if (rw == READ)
+	if (rw == READ) {
+		atomic64_inc(&zram->stats.num_reads);
 		ret = zram_bvec_read(zram, bvec, index, offset, bio);
-	else
+	} else {
+		atomic64_inc(&zram->stats.num_writes);
 		ret = zram_bvec_write(zram, bvec, index, offset);
+	}
 
 	return ret;
 }
 
+/*
+ * zram_bio_discard - handler on discard request
+ * @index: physical block index in PAGE_SIZE units
+ * @offset: byte offset within physical block
+ */
+static void zram_bio_discard(struct zram *zram, u32 index,
+			     int offset, struct bio *bio)
+{
+	size_t n = bio->bi_iter.bi_size;
+
+	/*
+	 * zram manages data in physical block size units. Because logical block
+	 * size isn't identical with physical block size on some arch, we
+	 * could get a discard request pointing to a specific offset within a
+	 * certain physical block.  Although we can handle this request by
+	 * reading that physiclal block and decompressing and partially zeroing
+	 * and re-compressing and then re-storing it, this isn't reasonable
+	 * because our intent with a discard request is to save memory.  So
+	 * skipping this logical block is appropriate here.
+	 */
+	if (offset) {
+		if (n < offset)
+			return;
+
+		n -= offset;
+		index++;
+	}
+
+	while (n >= PAGE_SIZE) {
+		/*
+		 * Discard request can be large so the lock hold times could be
+		 * lengthy.  So take the lock once per page.
+		 */
+		write_lock(&zram->meta->tb_lock);
+		zram_free_page(zram, index);
+		write_unlock(&zram->meta->tb_lock);
+		index++;
+		n -= PAGE_SIZE;
+	}
+}
+
 static void zram_reset_device(struct zram *zram, bool reset_capacity)
 {
 	size_t index;
 	struct zram_meta *meta;
 
 	down_write(&zram->init_lock);
-	if (!zram->init_done) {
+	if (!init_done(zram)) {
 		up_write(&zram->init_lock);
 		return;
 	}
 
 	meta = zram->meta;
-	zram->init_done = 0;
-
 	/* Free all pages that are still in this zram device */
 	for (index = 0; index < zram->disksize >> PAGE_SHIFT; index++) {
 		unsigned long handle = meta->table[index].handle;
@@ -563,6 +613,9 @@ static void zram_reset_device(struct zram *zram, bool reset_capacity)
 		zs_free(meta->mem_pool, handle);
 	}
 
+	zcomp_destroy(zram->comp);
+	zram->max_comp_streams = 1;
+
 	zram_meta_free(zram->meta);
 	zram->meta = NULL;
 	/* Reset stats */
@@ -574,37 +627,14 @@ static void zram_reset_device(struct zram *zram, bool reset_capacity)
 	up_write(&zram->init_lock);
 }
 
-static void zram_init_device(struct zram *zram, struct zram_meta *meta)
-{
-	if (zram->disksize > 2 * (totalram_pages << PAGE_SHIFT)) {
-		pr_info(
-		"There is little point creating a zram of greater than "
-		"twice the size of memory since we expect a 2:1 compression "
-		"ratio. Note that zram uses about 0.1%% of the size of "
-		"the disk when not in use so a huge zram is "
-		"wasteful.\n"
-		"\tMemory Size: %lu kB\n"
-		"\tSize you selected: %llu kB\n"
-		"Continuing anyway ...\n",
-		(totalram_pages << PAGE_SHIFT) >> 10, zram->disksize >> 10
-		);
-	}
-
-	/* zram devices sort of resembles non-rotational disks */
-	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, zram->disk->queue);
-
-	zram->meta = meta;
-	zram->init_done = 1;
-
-	pr_debug("Initialization done!\n");
-}
-
 static ssize_t disksize_store(struct device *dev,
 		struct device_attribute *attr, const char *buf, size_t len)
 {
 	u64 disksize;
+	struct zcomp *comp;
 	struct zram_meta *meta;
 	struct zram *zram = dev_to_zram(dev);
+	int err;
 
 	disksize = memparse(buf, NULL);
 	if (!disksize)
@@ -614,20 +644,35 @@ static ssize_t disksize_store(struct device *dev,
 	meta = zram_meta_alloc(disksize);
 	if (!meta)
 		return -ENOMEM;
+
+	comp = zcomp_create(zram->compressor, zram->max_comp_streams);
+	if (IS_ERR(comp)) {
+		pr_info("Cannot initialise %s compressing backend\n",
+				zram->compressor);
+		err = PTR_ERR(comp);
+		goto out_free_meta;
+	}
+
 	down_write(&zram->init_lock);
-	if (zram->init_done) {
-		up_write(&zram->init_lock);
-		zram_meta_free(meta);
+	if (init_done(zram)) {
 		pr_info("Cannot change disksize for initialized device\n");
-		return -EBUSY;
+		err = -EBUSY;
+		goto out_destroy_comp;
 	}
 
+	zram->meta = meta;
+	zram->comp = comp;
 	zram->disksize = disksize;
 	set_capacity(zram->disk, zram->disksize >> SECTOR_SHIFT);
-	zram_init_device(zram, meta);
 	up_write(&zram->init_lock);
-
 	return len;
+
+out_destroy_comp:
+	up_write(&zram->init_lock);
+	zcomp_destroy(comp);
+out_free_meta:
+	zram_meta_free(meta);
+	return err;
 }
 
 static ssize_t reset_store(struct device *dev,
@@ -671,26 +716,23 @@ out:
 	return ret;
 }
 
-static void __zram_make_request(struct zram *zram, struct bio *bio, int rw)
+static void __zram_make_request(struct zram *zram, struct bio *bio)
 {
 	int offset;
 	u32 index;
 	struct bio_vec bvec;
 	struct bvec_iter iter;
 
-	switch (rw) {
-	case READ:
-		atomic64_inc(&zram->stats.num_reads);
-		break;
-	case WRITE:
-		atomic64_inc(&zram->stats.num_writes);
-		break;
-	}
-
 	index = bio->bi_iter.bi_sector >> SECTORS_PER_PAGE_SHIFT;
 	offset = (bio->bi_iter.bi_sector &
 		  (SECTORS_PER_PAGE - 1)) << SECTOR_SHIFT;
 
+	if (unlikely(bio->bi_rw & REQ_DISCARD)) {
+		zram_bio_discard(zram, index, offset, bio);
+		bio_endio(bio, 0);
+		return;
+	}
+
 	bio_for_each_segment(bvec, bio, iter) {
 		int max_transfer_size = PAGE_SIZE - offset;
 
@@ -705,16 +747,15 @@ static void __zram_make_request(struct zram *zram, struct bio *bio, int rw)
 			bv.bv_len = max_transfer_size;
 			bv.bv_offset = bvec.bv_offset;
 
-			if (zram_bvec_rw(zram, &bv, index, offset, bio, rw) < 0)
+			if (zram_bvec_rw(zram, &bv, index, offset, bio) < 0)
 				goto out;
 
 			bv.bv_len = bvec.bv_len - max_transfer_size;
 			bv.bv_offset += max_transfer_size;
-			if (zram_bvec_rw(zram, &bv, index+1, 0, bio, rw) < 0)
+			if (zram_bvec_rw(zram, &bv, index + 1, 0, bio) < 0)
 				goto out;
 		} else
-			if (zram_bvec_rw(zram, &bvec, index, offset, bio, rw)
-			    < 0)
+			if (zram_bvec_rw(zram, &bvec, index, offset, bio) < 0)
 				goto out;
 
 		update_position(&index, &offset, &bvec);
@@ -736,7 +777,7 @@ static void zram_make_request(struct request_queue *queue, struct bio *bio)
 	struct zram *zram = queue->queuedata;
 
 	down_read(&zram->init_lock);
-	if (unlikely(!zram->init_done))
+	if (unlikely(!init_done(zram)))
 		goto error;
 
 	if (!valid_io_request(zram, bio)) {
@@ -744,7 +785,7 @@ static void zram_make_request(struct request_queue *queue, struct bio *bio)
 		goto error;
 	}
 
-	__zram_make_request(zram, bio, bio_data_dir(bio));
+	__zram_make_request(zram, bio);
 	up_read(&zram->init_lock);
 
 	return;
@@ -778,14 +819,21 @@ static DEVICE_ATTR(disksize, S_IRUGO | S_IWUSR,
 		disksize_show, disksize_store);
 static DEVICE_ATTR(initstate, S_IRUGO, initstate_show, NULL);
 static DEVICE_ATTR(reset, S_IWUSR, NULL, reset_store);
-static DEVICE_ATTR(num_reads, S_IRUGO, num_reads_show, NULL);
-static DEVICE_ATTR(num_writes, S_IRUGO, num_writes_show, NULL);
-static DEVICE_ATTR(invalid_io, S_IRUGO, invalid_io_show, NULL);
-static DEVICE_ATTR(notify_free, S_IRUGO, notify_free_show, NULL);
-static DEVICE_ATTR(zero_pages, S_IRUGO, zero_pages_show, NULL);
 static DEVICE_ATTR(orig_data_size, S_IRUGO, orig_data_size_show, NULL);
-static DEVICE_ATTR(compr_data_size, S_IRUGO, compr_data_size_show, NULL);
 static DEVICE_ATTR(mem_used_total, S_IRUGO, mem_used_total_show, NULL);
+static DEVICE_ATTR(max_comp_streams, S_IRUGO | S_IWUSR,
+		max_comp_streams_show, max_comp_streams_store);
+static DEVICE_ATTR(comp_algorithm, S_IRUGO | S_IWUSR,
+		comp_algorithm_show, comp_algorithm_store);
+
+ZRAM_ATTR_RO(num_reads);
+ZRAM_ATTR_RO(num_writes);
+ZRAM_ATTR_RO(failed_reads);
+ZRAM_ATTR_RO(failed_writes);
+ZRAM_ATTR_RO(invalid_io);
+ZRAM_ATTR_RO(notify_free);
+ZRAM_ATTR_RO(zero_pages);
+ZRAM_ATTR_RO(compr_data_size);
 
 static struct attribute *zram_disk_attrs[] = {
 	&dev_attr_disksize.attr,
@@ -793,12 +841,16 @@ static struct attribute *zram_disk_attrs[] = {
 	&dev_attr_reset.attr,
 	&dev_attr_num_reads.attr,
 	&dev_attr_num_writes.attr,
+	&dev_attr_failed_reads.attr,
+	&dev_attr_failed_writes.attr,
 	&dev_attr_invalid_io.attr,
 	&dev_attr_notify_free.attr,
 	&dev_attr_zero_pages.attr,
 	&dev_attr_orig_data_size.attr,
 	&dev_attr_compr_data_size.attr,
 	&dev_attr_mem_used_total.attr,
+	&dev_attr_max_comp_streams.attr,
+	&dev_attr_comp_algorithm.attr,
 	NULL,
 };
 
@@ -839,7 +891,8 @@ static int create_device(struct zram *zram, int device_id)
 
 	/* Actual capacity set using syfs (/sys/block/zram<id>/disksize */
 	set_capacity(zram->disk, 0);
-
+	/* zram devices sort of resembles non-rotational disks */
+	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, zram->disk->queue);
 	/*
 	 * To ensure that we always get PAGE_SIZE aligned
 	 * and n*PAGE_SIZED sized I/O requests.
@@ -849,6 +902,21 @@ static int create_device(struct zram *zram, int device_id)
 					ZRAM_LOGICAL_BLOCK_SIZE);
 	blk_queue_io_min(zram->disk->queue, PAGE_SIZE);
 	blk_queue_io_opt(zram->disk->queue, PAGE_SIZE);
+	zram->disk->queue->limits.discard_granularity = PAGE_SIZE;
+	zram->disk->queue->limits.max_discard_sectors = UINT_MAX;
+	/*
+	 * zram_bio_discard() will clear all logical blocks if logical block
+	 * size is identical with physical block size(PAGE_SIZE). But if it is
+	 * different, we will skip discarding some parts of logical blocks in
+	 * the part of the request range which isn't aligned to physical block
+	 * size.  So we can't ensure that all discarded logical blocks are
+	 * zeroed.
+	 */
+	if (ZRAM_LOGICAL_BLOCK_SIZE == PAGE_SIZE)
+		zram->disk->queue->limits.discard_zeroes_data = 1;
+	else
+		zram->disk->queue->limits.discard_zeroes_data = 0;
+	queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, zram->disk->queue);
 
 	add_disk(zram->disk);
 
@@ -858,8 +926,9 @@ static int create_device(struct zram *zram, int device_id)
 		pr_warn("Error creating sysfs group");
 		goto out_free_disk;
 	}
-
-	zram->init_done = 0;
+	strlcpy(zram->compressor, default_compressor, sizeof(zram->compressor));
+	zram->meta = NULL;
+	zram->max_comp_streams = 1;
 	return 0;
 
 out_free_disk:
diff --git a/drivers/block/zram/zram_drv.h b/drivers/block/zram/zram_drv.h
index ad8aa35bae00..7f21c145e317 100644
--- a/drivers/block/zram/zram_drv.h
+++ b/drivers/block/zram/zram_drv.h
@@ -16,9 +16,10 @@
 #define _ZRAM_DRV_H_
 
 #include <linux/spinlock.h>
-#include <linux/mutex.h>
 #include <linux/zsmalloc.h>
 
+#include "zcomp.h"
+
 /*
  * Some arbitrary value. This is just to catch
  * invalid value for num_devices module parameter.
@@ -64,38 +65,33 @@ enum zram_pageflags {
 struct table {
 	unsigned long handle;
 	u16 size;	/* object size (excluding header) */
-	u8 count;	/* object ref count (not yet used) */
 	u8 flags;
 } __aligned(4);
 
 struct zram_stats {
-	atomic64_t compr_size;	/* compressed size of pages stored */
+	atomic64_t compr_data_size;	/* compressed size of pages stored */
 	atomic64_t num_reads;	/* failed + successful */
 	atomic64_t num_writes;	/* --do-- */
 	atomic64_t failed_reads;	/* should NEVER! happen */
 	atomic64_t failed_writes;	/* can happen when memory is too low */
 	atomic64_t invalid_io;	/* non-page-aligned I/O requests */
 	atomic64_t notify_free;	/* no. of swap slot free notifications */
-	atomic_t pages_zero;		/* no. of zero filled pages */
-	atomic_t pages_stored;	/* no. of pages currently stored */
-	atomic_t good_compress;	/* % of pages with compression ratio<=50% */
-	atomic_t bad_compress;	/* % of pages with compression ratio>=75% */
+	atomic64_t zero_pages;		/* no. of zero filled pages */
+	atomic64_t pages_stored;	/* no. of pages currently stored */
 };
 
 struct zram_meta {
 	rwlock_t tb_lock;	/* protect table */
-	void *compress_workmem;
-	void *compress_buffer;
 	struct table *table;
 	struct zs_pool *mem_pool;
-	struct mutex buffer_lock; /* protect compress buffers */
 };
 
 struct zram {
 	struct zram_meta *meta;
 	struct request_queue *queue;
 	struct gendisk *disk;
-	int init_done;
+	struct zcomp *comp;
+
 	/* Prevent concurrent execution of device init, reset and R/W request */
 	struct rw_semaphore init_lock;
 	/*
@@ -103,7 +99,8 @@ struct zram {
 	 * we can store in a disk.
 	 */
 	u64 disksize;	/* bytes */
-
+	int max_comp_streams;
 	struct zram_stats stats;
+	char compressor[10];
 };
 #endif