Merge tag 'v3.10-rc7' into for-3.11/drivers

Linux 3.10-rc7

Pull this in early to avoid doing it with the bcache merge,
since there are a number of changes to bcache between my old
base (3.10-rc1) and the new pull request.

8 files changed, 647 insertions, 442 deletions

diff --git a/drivers/block/brd.c b/drivers/block/brd.c
index f1a29f8e9d3..9bf4371755f 100644
--- a/drivers/block/brd.c
+++ b/drivers/block/brd.c
@@ -117,13 +117,13 @@ static struct page *brd_insert_page(struct brd_device *brd, sector_t sector)
 
 	spin_lock(&brd->brd_lock);
 	idx = sector >> PAGE_SECTORS_SHIFT;
+	page->index = idx;
 	if (radix_tree_insert(&brd->brd_pages, idx, page)) {
 		__free_page(page);
 		page = radix_tree_lookup(&brd->brd_pages, idx);
 		BUG_ON(!page);
 		BUG_ON(page->index != idx);
-	} else
-		page->index = idx;
+	}
 	spin_unlock(&brd->brd_lock);
 
 	radix_tree_preload_end();
diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c
index 6374dc10352..62b6c2cc80b 100644
--- a/drivers/block/cciss.c
+++ b/drivers/block/cciss.c
@@ -168,8 +168,6 @@ static irqreturn_t do_cciss_msix_intr(int irq, void *dev_id);
 static int cciss_open(struct block_device *bdev, fmode_t mode);
 static int cciss_unlocked_open(struct block_device *bdev, fmode_t mode);
 static void cciss_release(struct gendisk *disk, fmode_t mode);
-static int do_ioctl(struct block_device *bdev, fmode_t mode,
-		    unsigned int cmd, unsigned long arg);
 static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
 		       unsigned int cmd, unsigned long arg);
 static int cciss_getgeo(struct block_device *bdev, struct hd_geometry *geo);
@@ -235,7 +233,7 @@ static const struct block_device_operations cciss_fops = {
 	.owner = THIS_MODULE,
 	.open = cciss_unlocked_open,
 	.release = cciss_release,
-	.ioctl = do_ioctl,
+	.ioctl = cciss_ioctl,
 	.getgeo = cciss_getgeo,
 #ifdef CONFIG_COMPAT
 	.compat_ioctl = cciss_compat_ioctl,
@@ -1143,16 +1141,6 @@ static void cciss_release(struct gendisk *disk, fmode_t mode)
 	mutex_unlock(&cciss_mutex);
 }
 
-static int do_ioctl(struct block_device *bdev, fmode_t mode,
-		    unsigned cmd, unsigned long arg)
-{
-	int ret;
-	mutex_lock(&cciss_mutex);
-	ret = cciss_ioctl(bdev, mode, cmd, arg);
-	mutex_unlock(&cciss_mutex);
-	return ret;
-}
-
 #ifdef CONFIG_COMPAT
 
 static int cciss_ioctl32_passthru(struct block_device *bdev, fmode_t mode,
@@ -1179,7 +1167,7 @@ static int cciss_compat_ioctl(struct block_device *bdev, fmode_t mode,
 	case CCISS_REGNEWD:
 	case CCISS_RESCANDISK:
 	case CCISS_GETLUNINFO:
-		return do_ioctl(bdev, mode, cmd, arg);
+		return cciss_ioctl(bdev, mode, cmd, arg);
 
 	case CCISS_PASSTHRU32:
 		return cciss_ioctl32_passthru(bdev, mode, cmd, arg);
@@ -1219,7 +1207,7 @@ static int cciss_ioctl32_passthru(struct block_device *bdev, fmode_t mode,
 	if (err)
 		return -EFAULT;
 
-	err = do_ioctl(bdev, mode, CCISS_PASSTHRU, (unsigned long)p);
+	err = cciss_ioctl(bdev, mode, CCISS_PASSTHRU, (unsigned long)p);
 	if (err)
 		return err;
 	err |=
@@ -1261,7 +1249,7 @@ static int cciss_ioctl32_big_passthru(struct block_device *bdev, fmode_t mode,
 	if (err)
 		return -EFAULT;
 
-	err = do_ioctl(bdev, mode, CCISS_BIG_PASSTHRU, (unsigned long)p);
+	err = cciss_ioctl(bdev, mode, CCISS_BIG_PASSTHRU, (unsigned long)p);
 	if (err)
 		return err;
 	err |=
@@ -1311,11 +1299,14 @@ static int cciss_getpciinfo(ctlr_info_t *h, void __user *argp)
 static int cciss_getintinfo(ctlr_info_t *h, void __user *argp)
 {
 	cciss_coalint_struct intinfo;
+	unsigned long flags;
 
 	if (!argp)
 		return -EINVAL;
+	spin_lock_irqsave(&h->lock, flags);
 	intinfo.delay = readl(&h->cfgtable->HostWrite.CoalIntDelay);
 	intinfo.count = readl(&h->cfgtable->HostWrite.CoalIntCount);
+	spin_unlock_irqrestore(&h->lock, flags);
 	if (copy_to_user
 	    (argp, &intinfo, sizeof(cciss_coalint_struct)))
 		return -EFAULT;
@@ -1356,12 +1347,15 @@ static int cciss_setintinfo(ctlr_info_t *h, void __user *argp)
 static int cciss_getnodename(ctlr_info_t *h, void __user *argp)
 {
 	NodeName_type NodeName;
+	unsigned long flags;
 	int i;
 
 	if (!argp)
 		return -EINVAL;
+	spin_lock_irqsave(&h->lock, flags);
 	for (i = 0; i < 16; i++)
 		NodeName[i] = readb(&h->cfgtable->ServerName[i]);
+	spin_unlock_irqrestore(&h->lock, flags);
 	if (copy_to_user(argp, NodeName, sizeof(NodeName_type)))
 		return -EFAULT;
 	return 0;
@@ -1398,10 +1392,13 @@ static int cciss_setnodename(ctlr_info_t *h, void __user *argp)
 static int cciss_getheartbeat(ctlr_info_t *h, void __user *argp)
 {
 	Heartbeat_type heartbeat;
+	unsigned long flags;
 
 	if (!argp)
 		return -EINVAL;
+	spin_lock_irqsave(&h->lock, flags);
 	heartbeat = readl(&h->cfgtable->HeartBeat);
+	spin_unlock_irqrestore(&h->lock, flags);
 	if (copy_to_user(argp, &heartbeat, sizeof(Heartbeat_type)))
 		return -EFAULT;
 	return 0;
@@ -1410,10 +1407,13 @@ static int cciss_getheartbeat(ctlr_info_t *h, void __user *argp)
 static int cciss_getbustypes(ctlr_info_t *h, void __user *argp)
 {
 	BusTypes_type BusTypes;
+	unsigned long flags;
 
 	if (!argp)
 		return -EINVAL;
+	spin_lock_irqsave(&h->lock, flags);
 	BusTypes = readl(&h->cfgtable->BusTypes);
+	spin_unlock_irqrestore(&h->lock, flags);
 	if (copy_to_user(argp, &BusTypes, sizeof(BusTypes_type)))
 		return -EFAULT;
 	return 0;
diff --git a/drivers/block/mtip32xx/mtip32xx.c b/drivers/block/mtip32xx/mtip32xx.c
index 847107ef0cc..20dd52a2f92 100644
--- a/drivers/block/mtip32xx/mtip32xx.c
+++ b/drivers/block/mtip32xx/mtip32xx.c
@@ -3002,7 +3002,8 @@ static int mtip_hw_debugfs_init(struct driver_data *dd)
 
 static void mtip_hw_debugfs_exit(struct driver_data *dd)
 {
-	debugfs_remove_recursive(dd->dfs_node);
+	if (dd->dfs_node)
+		debugfs_remove_recursive(dd->dfs_node);
 }
 
 
@@ -3863,7 +3864,7 @@ static void mtip_make_request(struct request_queue *queue, struct bio *bio)
 	struct driver_data *dd = queue->queuedata;
 	struct scatterlist *sg;
 	struct bio_vec *bvec;
-	int nents = 0;
+	int i, nents = 0;
 	int tag = 0, unaligned = 0;
 
 	if (unlikely(dd->dd_flag & MTIP_DDF_STOP_IO)) {
@@ -3921,11 +3922,12 @@ static void mtip_make_request(struct request_queue *queue, struct bio *bio)
 		}
 
 		/* Create the scatter list for this bio. */
-		bio_for_each_segment(bvec, bio, nents) {
+		bio_for_each_segment(bvec, bio, i) {
 			sg_set_page(&sg[nents],
 					bvec->bv_page,
 					bvec->bv_len,
 					bvec->bv_offset);
+			nents++;
 		}
 
 		/* Issue the read/write. */
diff --git a/drivers/block/nvme-core.c b/drivers/block/nvme-core.c
index 8efdfaa44a5..ce79a590b45 100644
--- a/drivers/block/nvme-core.c
+++ b/drivers/block/nvme-core.c
@@ -629,7 +629,7 @@ static int nvme_submit_bio_queue(struct nvme_queue *nvmeq, struct nvme_ns *ns,
 	struct nvme_command *cmnd;
 	struct nvme_iod *iod;
 	enum dma_data_direction dma_dir;
-	int cmdid, length, result = -ENOMEM;
+	int cmdid, length, result;
 	u16 control;
 	u32 dsmgmt;
 	int psegs = bio_phys_segments(ns->queue, bio);
@@ -640,6 +640,7 @@ static int nvme_submit_bio_queue(struct nvme_queue *nvmeq, struct nvme_ns *ns,
 			return result;
 	}
 
+	result = -ENOMEM;
 	iod = nvme_alloc_iod(psegs, bio->bi_size, GFP_ATOMIC);
 	if (!iod)
 		goto nomem;
@@ -977,6 +978,8 @@ static void nvme_cancel_ios(struct nvme_queue *nvmeq, bool timeout)
 
 		if (timeout && !time_after(now, info[cmdid].timeout))
 			continue;
+		if (info[cmdid].ctx == CMD_CTX_CANCELLED)
+			continue;
 		dev_warn(nvmeq->q_dmadev, "Cancelling I/O %d\n", cmdid);
 		ctx = cancel_cmdid(nvmeq, cmdid, &fn);
 		fn(nvmeq->dev, ctx, &cqe);
@@ -1206,7 +1209,7 @@ struct nvme_iod *nvme_map_user_pages(struct nvme_dev *dev, int write,
 
 	if (addr & 3)
 		return ERR_PTR(-EINVAL);
-	if (!length)
+	if (!length || length > INT_MAX - PAGE_SIZE)
 		return ERR_PTR(-EINVAL);
 
 	offset = offset_in_page(addr);
@@ -1227,7 +1230,8 @@ struct nvme_iod *nvme_map_user_pages(struct nvme_dev *dev, int write,
 	sg_init_table(sg, count);
 	for (i = 0; i < count; i++) {
 		sg_set_page(&sg[i], pages[i],
-				min_t(int, length, PAGE_SIZE - offset), offset);
+			    min_t(unsigned, length, PAGE_SIZE - offset),
+			    offset);
 		length -= (PAGE_SIZE - offset);
 		offset = 0;
 	}
@@ -1435,7 +1439,7 @@ static int nvme_user_admin_cmd(struct nvme_dev *dev,
 		nvme_free_iod(dev, iod);
 	}
 
-	if (!status && copy_to_user(&ucmd->result, &cmd.result,
+	if ((status >= 0) && copy_to_user(&ucmd->result, &cmd.result,
 							sizeof(cmd.result)))
 		status = -EFAULT;
 
@@ -1633,7 +1637,8 @@ static int set_queue_count(struct nvme_dev *dev, int count)
 
 static int nvme_setup_io_queues(struct nvme_dev *dev)
 {
-	int result, cpu, i, nr_io_queues, db_bar_size, q_depth;
+	struct pci_dev *pdev = dev->pci_dev;
+	int result, cpu, i, nr_io_queues, db_bar_size, q_depth, q_count;
 
 	nr_io_queues = num_online_cpus();
 	result = set_queue_count(dev, nr_io_queues);
@@ -1642,14 +1647,14 @@ static int nvme_setup_io_queues(struct nvme_dev *dev)
 	if (result < nr_io_queues)
 		nr_io_queues = result;
 
+	q_count = nr_io_queues;
 	/* Deregister the admin queue's interrupt */
 	free_irq(dev->entry[0].vector, dev->queues[0]);
 
 	db_bar_size = 4096 + ((nr_io_queues + 1) << (dev->db_stride + 3));
 	if (db_bar_size > 8192) {
 		iounmap(dev->bar);
-		dev->bar = ioremap(pci_resource_start(dev->pci_dev, 0),
-								db_bar_size);
+		dev->bar = ioremap(pci_resource_start(pdev, 0), db_bar_size);
 		dev->dbs = ((void __iomem *)dev->bar) + 4096;
 		dev->queues[0]->q_db = dev->dbs;
 	}
@@ -1657,19 +1662,36 @@ static int nvme_setup_io_queues(struct nvme_dev *dev)
 	for (i = 0; i < nr_io_queues; i++)
 		dev->entry[i].entry = i;
 	for (;;) {
-		result = pci_enable_msix(dev->pci_dev, dev->entry,
-								nr_io_queues);
+		result = pci_enable_msix(pdev, dev->entry, nr_io_queues);
 		if (result == 0) {
 			break;
 		} else if (result > 0) {
 			nr_io_queues = result;
 			continue;
 		} else {
-			nr_io_queues = 1;
+			nr_io_queues = 0;
 			break;
 		}
 	}
 
+	if (nr_io_queues == 0) {
+		nr_io_queues = q_count;
+		for (;;) {
+			result = pci_enable_msi_block(pdev, nr_io_queues);
+			if (result == 0) {
+				for (i = 0; i < nr_io_queues; i++)
+					dev->entry[i].vector = i + pdev->irq;
+				break;
+			} else if (result > 0) {
+				nr_io_queues = result;
+				continue;
+			} else {
+				nr_io_queues = 1;
+				break;
+			}
+		}
+	}
+
 	result = queue_request_irq(dev, dev->queues[0], "nvme admin");
 	/* XXX: handle failure here */
 
@@ -1850,7 +1872,10 @@ static void nvme_free_dev(struct kref *kref)
 {
 	struct nvme_dev *dev = container_of(kref, struct nvme_dev, kref);
 	nvme_dev_remove(dev);
-	pci_disable_msix(dev->pci_dev);
+	if (dev->pci_dev->msi_enabled)
+		pci_disable_msi(dev->pci_dev);
+	else if (dev->pci_dev->msix_enabled)
+		pci_disable_msix(dev->pci_dev);
 	iounmap(dev->bar);
 	nvme_release_instance(dev);
 	nvme_release_prp_pools(dev);
@@ -1923,8 +1948,14 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	INIT_LIST_HEAD(&dev->namespaces);
 	dev->pci_dev = pdev;
 	pci_set_drvdata(pdev, dev);
-	dma_set_mask(&pdev->dev, DMA_BIT_MASK(64));
-	dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(64));
+
+	if (!dma_set_mask(&pdev->dev, DMA_BIT_MASK(64)))
+		dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(64));
+	else if (!dma_set_mask(&pdev->dev, DMA_BIT_MASK(32)))
+		dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(32));
+	else
+		goto disable;
+
 	result = nvme_set_instance(dev);
 	if (result)
 		goto disable;
@@ -1977,7 +2008,10 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
  unmap:
 	iounmap(dev->bar);
  disable_msix:
-	pci_disable_msix(pdev);
+	if (dev->pci_dev->msi_enabled)
+		pci_disable_msi(dev->pci_dev);
+	else if (dev->pci_dev->msix_enabled)
+		pci_disable_msix(dev->pci_dev);
 	nvme_release_instance(dev);
 	nvme_release_prp_pools(dev);
  disable:
diff --git a/drivers/block/nvme-scsi.c b/drivers/block/nvme-scsi.c
index fed54b03989..102de2f52b5 100644
--- a/drivers/block/nvme-scsi.c
+++ b/drivers/block/nvme-scsi.c
@@ -44,7 +44,6 @@
 #include <linux/sched.h>
 #include <linux/slab.h>
 #include <linux/types.h>
-#include <linux/version.h>
 #include <scsi/sg.h>
 #include <scsi/scsi.h>
 
@@ -1654,7 +1653,7 @@ static void nvme_trans_modesel_save_bd(struct nvme_ns *ns, u8 *parm_list,
 	}
 }
 
-static u16 nvme_trans_modesel_get_mp(struct nvme_ns *ns, struct sg_io_hdr *hdr,
+static int nvme_trans_modesel_get_mp(struct nvme_ns *ns, struct sg_io_hdr *hdr,
 					u8 *mode_page, u8 page_code)
 {
 	int res = SNTI_TRANSLATION_SUCCESS;
diff --git a/drivers/block/pktcdvd.c b/drivers/block/pktcdvd.c
index 3c08983e600..f5d0ea11d9f 100644
--- a/drivers/block/pktcdvd.c
+++ b/drivers/block/pktcdvd.c
@@ -83,7 +83,8 @@
 
 #define MAX_SPEED 0xffff
 
-#define ZONE(sector, pd) (((sector) + (pd)->offset) & ~((pd)->settings.size - 1))
+#define ZONE(sector, pd) (((sector) + (pd)->offset) & \
+			~(sector_t)((pd)->settings.size - 1))
 
 static DEFINE_MUTEX(pktcdvd_mutex);
 static struct pktcdvd_device *pkt_devs[MAX_WRITERS];
diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c
index ca63104136e..49394e3f31b 100644
--- a/drivers/block/rbd.c
+++ b/drivers/block/rbd.c
@@ -55,6 +55,39 @@
 #define	SECTOR_SHIFT	9
 #define	SECTOR_SIZE	(1ULL << SECTOR_SHIFT)
 
+/*
+ * Increment the given counter and return its updated value.
+ * If the counter is already 0 it will not be incremented.
+ * If the counter is already at its maximum value returns
+ * -EINVAL without updating it.
+ */
+static int atomic_inc_return_safe(atomic_t *v)
+{
+	unsigned int counter;
+
+	counter = (unsigned int)__atomic_add_unless(v, 1, 0);
+	if (counter <= (unsigned int)INT_MAX)
+		return (int)counter;
+
+	atomic_dec(v);
+
+	return -EINVAL;
+}
+
+/* Decrement the counter.  Return the resulting value, or -EINVAL */
+static int atomic_dec_return_safe(atomic_t *v)
+{
+	int counter;
+
+	counter = atomic_dec_return(v);
+	if (counter >= 0)
+		return counter;
+
+	atomic_inc(v);
+
+	return -EINVAL;
+}
+
 #define RBD_DRV_NAME "rbd"
 #define RBD_DRV_NAME_LONG "rbd (rados block device)"
 
@@ -100,21 +133,20 @@
  * block device image metadata (in-memory version)
  */
 struct rbd_image_header {
-	/* These four fields never change for a given rbd image */
+	/* These six fields never change for a given rbd image */
 	char *object_prefix;
-	u64 features;
 	__u8 obj_order;
 	__u8 crypt_type;
 	__u8 comp_type;
+	u64 stripe_unit;
+	u64 stripe_count;
+	u64 features;		/* Might be changeable someday? */
 
 	/* The remaining fields need to be updated occasionally */
 	u64 image_size;
 	struct ceph_snap_context *snapc;
-	char *snap_names;
-	u64 *snap_sizes;
-
-	u64 stripe_unit;
-	u64 stripe_count;
+	char *snap_names;	/* format 1 only */
+	u64 *snap_sizes;	/* format 1 only */
 };
 
 /*
@@ -225,6 +257,7 @@ struct rbd_obj_request {
 		};
 	};
 	struct page		**copyup_pages;
+	u32			copyup_page_count;
 
 	struct ceph_osd_request	*osd_req;
 
@@ -257,6 +290,7 @@ struct rbd_img_request {
 		struct rbd_obj_request	*obj_request;	/* obj req initiator */
 	};
 	struct page		**copyup_pages;
+	u32			copyup_page_count;
 	spinlock_t		completion_lock;/* protects next_completion */
 	u32			next_completion;
 	rbd_img_callback_t	callback;
@@ -311,6 +345,7 @@ struct rbd_device {
 
 	struct rbd_spec		*parent_spec;
 	u64			parent_overlap;
+	atomic_t		parent_ref;
 	struct rbd_device	*parent;
 
 	/* protects updating the header */
@@ -359,7 +394,8 @@ static ssize_t rbd_add(struct bus_type *bus, const char *buf,
 		       size_t count);
 static ssize_t rbd_remove(struct bus_type *bus, const char *buf,
 			  size_t count);
-static int rbd_dev_image_probe(struct rbd_device *rbd_dev);
+static int rbd_dev_image_probe(struct rbd_device *rbd_dev, bool mapping);
+static void rbd_spec_put(struct rbd_spec *spec);
 
 static struct bus_attribute rbd_bus_attrs[] = {
 	__ATTR(add, S_IWUSR, NULL, rbd_add),
@@ -426,7 +462,8 @@ static void rbd_img_parent_read(struct rbd_obj_request *obj_request);
 static void rbd_dev_remove_parent(struct rbd_device *rbd_dev);
 
 static int rbd_dev_refresh(struct rbd_device *rbd_dev);
-static int rbd_dev_v2_refresh(struct rbd_device *rbd_dev);
+static int rbd_dev_v2_header_onetime(struct rbd_device *rbd_dev);
+static int rbd_dev_v2_header_info(struct rbd_device *rbd_dev);
 static const char *rbd_dev_v2_snap_name(struct rbd_device *rbd_dev,
 					u64 snap_id);
 static int _rbd_dev_v2_snap_size(struct rbd_device *rbd_dev, u64 snap_id,
@@ -482,8 +519,8 @@ static const struct block_device_operations rbd_bd_ops = {
 };
 
 /*
- * Initialize an rbd client instance.
- * We own *ceph_opts.
+ * Initialize an rbd client instance.  Success or not, this function
+ * consumes ceph_opts.
  */
 static struct rbd_client *rbd_client_create(struct ceph_options *ceph_opts)
 {
@@ -638,7 +675,8 @@ static int parse_rbd_opts_token(char *c, void *private)
 
 /*
  * Get a ceph client with specific addr and configuration, if one does
- * not exist create it.
+ * not exist create it.  Either way, ceph_opts is consumed by this
+ * function.
  */
 static struct rbd_client *rbd_get_client(struct ceph_options *ceph_opts)
 {
@@ -726,88 +764,123 @@ static bool rbd_dev_ondisk_valid(struct rbd_image_header_ondisk *ondisk)
 }
 
 /*
- * Create a new header structure, translate header format from the on-disk
- * header.
+ * Fill an rbd image header with information from the given format 1
+ * on-disk header.
  */
-static int rbd_header_from_disk(struct rbd_image_header *header,
+static int rbd_header_from_disk(struct rbd_device *rbd_dev,
 				 struct rbd_image_header_ondisk *ondisk)
 {
+	struct rbd_image_header *header = &rbd_dev->header;
+	bool first_time = header->object_prefix == NULL;
+	struct ceph_snap_context *snapc;
+	char *object_prefix = NULL;
+	char *snap_names = NULL;
+	u64 *snap_sizes = NULL;
 	u32 snap_count;
-	size_t len;
 	size_t size;
+	int ret = -ENOMEM;
 	u32 i;
 
-	memset(header, 0, sizeof (*header));
+	/* Allocate this now to avoid having to handle failure below */
 
-	snap_count = le32_to_cpu(ondisk->snap_count);
+	if (first_time) {
+		size_t len;
 
-	len = strnlen(ondisk->object_prefix, sizeof (ondisk->object_prefix));
-	header->object_prefix = kmalloc(len + 1, GFP_KERNEL);
-	if (!header->object_prefix)
-		return -ENOMEM;
-	memcpy(header->object_prefix, ondisk->object_prefix, len);
-	header->object_prefix[len] = '\0';
+		len = strnlen(ondisk->object_prefix,
+				sizeof (ondisk->object_prefix));
+		object_prefix = kmalloc(len + 1, GFP_KERNEL);
+		if (!object_prefix)
+			return -ENOMEM;
+		memcpy(object_prefix, ondisk->object_prefix, len);
+		object_prefix[len] = '\0';
+	}
+
+	/* Allocate the snapshot context and fill it in */
 
+	snap_count = le32_to_cpu(ondisk->snap_count);
+	snapc = ceph_create_snap_context(snap_count, GFP_KERNEL);
+	if (!snapc)
+		goto out_err;
+	snapc->seq = le64_to_cpu(ondisk->snap_seq);
 	if (snap_count) {
+		struct rbd_image_snap_ondisk *snaps;
 		u64 snap_names_len = le64_to_cpu(ondisk->snap_names_len);
 
-		/* Save a copy of the snapshot names */
+		/* We'll keep a copy of the snapshot names... */
 
-		if (snap_names_len > (u64) SIZE_MAX)
-			return -EIO;
-		header->snap_names = kmalloc(snap_names_len, GFP_KERNEL);
-		if (!header->snap_names)
+		if (snap_names_len > (u64)SIZE_MAX)
+			goto out_2big;
+		snap_names = kmalloc(snap_names_len, GFP_KERNEL);
+		if (!snap_names)
+			goto out_err;
+
+		/* ...as well as the array of their sizes. */
+
+		size = snap_count * sizeof (*header->snap_sizes);
+		snap_sizes = kmalloc(size, GFP_KERNEL);
+		if (!snap_sizes)
 			goto out_err;
+
 		/*
-		 * Note that rbd_dev_v1_header_read() guarantees
-		 * the ondisk buffer we're working with has
+		 * Copy the names, and fill in each snapshot's id
+		 * and size.
+		 *
+		 * Note that rbd_dev_v1_header_info() guarantees the
+		 * ondisk buffer we're working with has
 		 * snap_names_len bytes beyond the end of the
 		 * snapshot id array, this memcpy() is safe.
 		 */
-		memcpy(header->snap_names, &ondisk->snaps[snap_count],
-			snap_names_len);
+		memcpy(snap_names, &ondisk->snaps[snap_count], snap_names_len);
+		snaps = ondisk->snaps;
+		for (i = 0; i < snap_count; i++) {
+			snapc->snaps[i] = le64_to_cpu(snaps[i].id);
+			snap_sizes[i] = le64_to_cpu(snaps[i].image_size);
+		}
+	}
 
-		/* Record each snapshot's size */
+	/* We won't fail any more, fill in the header */
 
-		size = snap_count * sizeof (*header->snap_sizes);
-		header->snap_sizes = kmalloc(size, GFP_KERNEL);
-		if (!header->snap_sizes)
-			goto out_err;
-		for (i = 0; i < snap_count; i++)
-			header->snap_sizes[i] =
-				le64_to_cpu(ondisk->snaps[i].image_size);
+	down_write(&rbd_dev->header_rwsem);
+	if (first_time) {
+		header->object_prefix = object_prefix;
+		header->obj_order = ondisk->options.order;
+		header->crypt_type = ondisk->options.crypt_type;
+		header->comp_type = ondisk->options.comp_type;
+		/* The rest aren't used for format 1 images */
+		header->stripe_unit = 0;
+		header->stripe_count = 0;
+		header->features = 0;
 	} else {
-		header->snap_names = NULL;
-		header->snap_sizes = NULL;
+		ceph_put_snap_context(header->snapc);
+		kfree(header->snap_names);
+		kfree(header->snap_sizes);
 	}
 
-	header->features = 0;	/* No features support in v1 images */
-	header->obj_order = ondisk->options.order;
-	header->crypt_type = ondisk->options.crypt_type;
-	header->comp_type = ondisk->options.comp_type;
-
-	/* Allocate and fill in the snapshot context */
+	/* The remaining fields always get updated (when we refresh) */
 
 	header->image_size = le64_to_cpu(ondisk->image_size);
+	header->snapc = snapc;
+	header->snap_names = snap_names;
+	header->snap_sizes = snap_sizes;
 
-	header->snapc = ceph_create_snap_context(snap_count, GFP_KERNEL);
-	if (!header->snapc)
-		goto out_err;
-	header->snapc->seq = le64_to_cpu(ondisk->snap_seq);
-	for (i = 0; i < snap_count; i++)
-		header->snapc->snaps[i] = le64_to_cpu(ondisk->snaps[i].id);
+	/* Make sure mapping size is consistent with header info */
 
-	return 0;
+	if (rbd_dev->spec->snap_id == CEPH_NOSNAP || first_time)
+		if (rbd_dev->mapping.size != header->image_size)
+			rbd_dev->mapping.size = header->image_size;
 
+	up_write(&rbd_dev->header_rwsem);
+
+	return 0;
+out_2big:
+	ret = -EIO;
 out_err:
-	kfree(header->snap_sizes);
-	header->snap_sizes = NULL;
-	kfree(header->snap_names);
-	header->snap_names = NULL;
-	kfree(header->object_prefix);
-	header->object_prefix = NULL;
+	kfree(snap_sizes);
+	kfree(snap_names);
+	ceph_put_snap_context(snapc);
+	kfree(object_prefix);
 
-	return -ENOMEM;
+	return ret;
 }
 
 static const char *_rbd_dev_v1_snap_name(struct rbd_device *rbd_dev, u32 which)
@@ -934,20 +1007,11 @@ static int rbd_snap_features(struct rbd_device *rbd_dev, u64 snap_id,
 
 static int rbd_dev_mapping_set(struct rbd_device *rbd_dev)
 {
-	const char *snap_name = rbd_dev->spec->snap_name;
-	u64 snap_id;
+	u64 snap_id = rbd_dev->spec->snap_id;
 	u64 size = 0;
 	u64 features = 0;
 	int ret;
 
-	if (strcmp(snap_name, RBD_SNAP_HEAD_NAME)) {
-		snap_id = rbd_snap_id_by_name(rbd_dev, snap_name);
-		if (snap_id == CEPH_NOSNAP)
-			return -ENOENT;
-	} else {
-		snap_id = CEPH_NOSNAP;
-	}
-
 	ret = rbd_snap_size(rbd_dev, snap_id, &size);
 	if (ret)
 		return ret;
@@ -958,11 +1022,6 @@ static int rbd_dev_mapping_set(struct rbd_device *rbd_dev)
 	rbd_dev->mapping.size = size;
 	rbd_dev->mapping.features = features;
 
-	/* If we are mapping a snapshot it must be marked read-only */
-
-	if (snap_id != CEPH_NOSNAP)
-		rbd_dev->mapping.read_only = true;
-
 	return 0;
 }
 
@@ -970,14 +1029,6 @@ static void rbd_dev_mapping_clear(struct rbd_device *rbd_dev)
 {
 	rbd_dev->mapping.size = 0;
 	rbd_dev->mapping.features = 0;
-	rbd_dev->mapping.read_only = true;
-}
-
-static void rbd_dev_clear_mapping(struct rbd_device *rbd_dev)
-{
-	rbd_dev->mapping.size = 0;
-	rbd_dev->mapping.features = 0;
-	rbd_dev->mapping.read_only = true;
 }
 
 static const char *rbd_segment_name(struct rbd_device *rbd_dev, u64 offset)
@@ -985,12 +1036,16 @@ static const char *rbd_segment_name(struct rbd_device *rbd_dev, u64 offset)
 	char *name;
 	u64 segment;
 	int ret;
+	char *name_format;
 
 	name = kmem_cache_alloc(rbd_segment_name_cache, GFP_NOIO);
 	if (!name)
 		return NULL;
 	segment = offset >> rbd_dev->header.obj_order;
-	ret = snprintf(name, MAX_OBJ_NAME_SIZE + 1, "%s.%012llx",
+	name_format = "%s.%012llx";
+	if (rbd_dev->image_format == 2)
+		name_format = "%s.%016llx";
+	ret = snprintf(name, MAX_OBJ_NAME_SIZE + 1, name_format,
 			rbd_dev->header.object_prefix, segment);
 	if (ret < 0 || ret > MAX_OBJ_NAME_SIZE) {
 		pr_err("error formatting segment name for #%llu (%d)\n",
@@ -1342,20 +1397,18 @@ static void rbd_obj_request_put(struct rbd_obj_request *obj_request)
 	kref_put(&obj_request->kref, rbd_obj_request_destroy);
 }
 
-static void rbd_img_request_get(struct rbd_img_request *img_request)
-{
-	dout("%s: img %p (was %d)\n", __func__, img_request,
-		atomic_read(&img_request->kref.refcount));
-	kref_get(&img_request->kref);
-}
-
+static bool img_request_child_test(struct rbd_img_request *img_request);
+static void rbd_parent_request_destroy(struct kref *kref);
 static void rbd_img_request_destroy(struct kref *kref);
 static void rbd_img_request_put(struct rbd_img_request *img_request)
 {
 	rbd_assert(img_request != NULL);
 	dout("%s: img %p (was %d)\n", __func__, img_request,
 		atomic_read(&img_request->kref.refcount));
-	kref_put(&img_request->kref, rbd_img_request_destroy);
+	if (img_request_child_test(img_request))
+		kref_put(&img_request->kref, rbd_parent_request_destroy);
+	else
+		kref_put(&img_request->kref, rbd_img_request_destroy);
 }
 
 static inline void rbd_img_obj_request_add(struct rbd_img_request *img_request,
@@ -1472,6 +1525,12 @@ static void img_request_child_set(struct rbd_img_request *img_request)
 	smp_mb();
 }
 
+static void img_request_child_clear(struct rbd_img_request *img_request)
+{
+	clear_bit(IMG_REQ_CHILD, &img_request->flags);
+	smp_mb();
+}
+
 static bool img_request_child_test(struct rbd_img_request *img_request)
 {
 	smp_mb();
@@ -1484,6 +1543,12 @@ static void img_request_layered_set(struct rbd_img_request *img_request)
 	smp_mb();
 }
 
+static void img_request_layered_clear(struct rbd_img_request *img_request)
+{
+	clear_bit(IMG_REQ_LAYERED, &img_request->flags);
+	smp_mb();
+}
+
 static bool img_request_layered_test(struct rbd_img_request *img_request)
 {
 	smp_mb();
@@ -1827,6 +1892,74 @@ static void rbd_obj_request_destroy(struct kref *kref)
 	kmem_cache_free(rbd_obj_request_cache, obj_request);
 }
 
+/* It's OK to call this for a device with no parent */
+
+static void rbd_spec_put(struct rbd_spec *spec);
+static void rbd_dev_unparent(struct rbd_device *rbd_dev)
+{
+	rbd_dev_remove_parent(rbd_dev);
+	rbd_spec_put(rbd_dev->parent_spec);
+	rbd_dev->parent_spec = NULL;
+	rbd_dev->parent_overlap = 0;
+}
+
+/*
+ * Parent image reference counting is used to determine when an
+ * image's parent fields can be safely torn down--after there are no
+ * more in-flight requests to the parent image.  When the last
+ * reference is dropped, cleaning them up is safe.
+ */
+static void rbd_dev_parent_put(struct rbd_device *rbd_dev)
+{
+	int counter;
+
+	if (!rbd_dev->parent_spec)
+		return;
+
+	counter = atomic_dec_return_safe(&rbd_dev->parent_ref);
+	if (counter > 0)
+		return;
+
+	/* Last reference; clean up parent data structures */
+
+	if (!counter)
+		rbd_dev_unparent(rbd_dev);
+	else
+		rbd_warn(rbd_dev, "parent reference underflow\n");
+}
+
+/*
+ * If an image has a non-zero parent overlap, get a reference to its
+ * parent.
+ *
+ * We must get the reference before checking for the overlap to
+ * coordinate properly with zeroing the parent overlap in
+ * rbd_dev_v2_parent_info() when an image gets flattened.  We
+ * drop it again if there is no overlap.
+ *
+ * Returns true if the rbd device has a parent with a non-zero
+ * overlap and a reference for it was successfully taken, or
+ * false otherwise.
+ */
+static bool rbd_dev_parent_get(struct rbd_device *rbd_dev)
+{
+	int counter;
+
+	if (!rbd_dev->parent_spec)
+		return false;
+
+	counter = atomic_inc_return_safe(&rbd_dev->parent_ref);
+	if (counter > 0 && rbd_dev->parent_overlap)
+		return true;
+
+	/* Image was flattened, but parent is not yet torn down */
+
+	if (counter < 0)
+		rbd_warn(rbd_dev, "parent reference overflow\n");
+
+	return false;
+}
+
 /*
  * Caller is responsible for filling in the list of object requests
  * that comprises the image request, and the Linux request pointer
@@ -1835,8 +1968,7 @@ static void rbd_obj_request_destroy(struct kref *kref)
 static struct rbd_img_request *rbd_img_request_create(
 					struct rbd_device *rbd_dev,
 					u64 offset, u64 length,
-					bool write_request,
-					bool child_request)
+					bool write_request)
 {
 	struct rbd_img_request *img_request;
 
@@ -1861,9 +1993,7 @@ static struct rbd_img_request *rbd_img_request_create(
 	} else {
 		img_request->snap_id = rbd_dev->spec->snap_id;
 	}
-	if (child_request)
-		img_request_child_set(img_request);
-	if (rbd_dev->parent_spec)
+	if (rbd_dev_parent_get(rbd_dev))
 		img_request_layered_set(img_request);
 	spin_lock_init(&img_request->completion_lock);
 	img_request->next_completion = 0;
@@ -1873,9 +2003,6 @@ static struct rbd_img_request *rbd_img_request_create(
 	INIT_LIST_HEAD(&img_request->obj_requests);
 	kref_init(&img_request->kref);
 
-	rbd_img_request_get(img_request);	/* Avoid a warning */
-	rbd_img_request_put(img_request);	/* TEMPORARY */
-
 	dout("%s: rbd_dev %p %s %llu/%llu -> img %p\n", __func__, rbd_dev,
 		write_request ? "write" : "read", offset, length,
 		img_request);
@@ -1897,15 +2024,54 @@ static void rbd_img_request_destroy(struct kref *kref)
 		rbd_img_obj_request_del(img_request, obj_request);
 	rbd_assert(img_request->obj_request_count == 0);
 
+	if (img_request_layered_test(img_request)) {
+		img_request_layered_clear(img_request);
+		rbd_dev_parent_put(img_request->rbd_dev);
+	}
+
 	if (img_request_write_test(img_request))
 		ceph_put_snap_context(img_request->snapc);
 
-	if (img_request_child_test(img_request))
-		rbd_obj_request_put(img_request->obj_request);
-
 	kmem_cache_free(rbd_img_request_cache, img_request);
 }
 
+static struct rbd_img_request *rbd_parent_request_create(
+					struct rbd_obj_request *obj_request,
+					u64 img_offset, u64 length)
+{
+	struct rbd_img_request *parent_request;
+	struct rbd_device *rbd_dev;
+
+	rbd_assert(obj_request->img_request);
+	rbd_dev = obj_request->img_request->rbd_dev;
+
+	parent_request = rbd_img_request_create(rbd_dev->parent,
+						img_offset, length, false);
+	if (!parent_request)
+		return NULL;
+
+	img_request_child_set(parent_request);
+	rbd_obj_request_get(obj_request);
+	parent_request->obj_request = obj_request;
+
+	return parent_request;
+}
+
+static void rbd_parent_request_destroy(struct kref *kref)
+{
+	struct rbd_img_request *parent_request;
+	struct rbd_obj_request *orig_request;
+
+	parent_request = container_of(kref, struct rbd_img_request, kref);
+	orig_request = parent_request->obj_request;
+
+	parent_request->obj_request = NULL;
+	rbd_obj_request_put(orig_request);
+	img_request_child_clear(parent_request);
+
+	rbd_img_request_destroy(kref);
+}
+
 static bool rbd_img_obj_end_request(struct rbd_obj_request *obj_request)
 {
 	struct rbd_img_request *img_request;
@@ -2114,7 +2280,7 @@ rbd_img_obj_copyup_callback(struct rbd_obj_request *obj_request)
 {
 	struct rbd_img_request *img_request;
 	struct rbd_device *rbd_dev;
-	u64 length;
+	struct page **pages;
 	u32 page_count;
 
 	rbd_assert(obj_request->type == OBJ_REQUEST_BIO);
@@ -2124,12 +2290,14 @@ rbd_img_obj_copyup_callback(struct rbd_obj_request *obj_request)
 
 	rbd_dev = img_request->rbd_dev;
 	rbd_assert(rbd_dev);
-	length = (u64)1 << rbd_dev->header.obj_order;
-	page_count = (u32)calc_pages_for(0, length);
 
-	rbd_assert(obj_request->copyup_pages);