Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net

Conflicts:
	net/bridge/br_mdb.c

Minor conflict in br_mdb.c, in 'net' we added a memset of the
on-stack 'ip' variable whereas in 'net-next' we assign a new
member 'vid'.

Signed-off-by: David S. Miller <davem@davemloft.net>

152 files changed, 5245 insertions, 2321 deletions

diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index 510040b04c96..b1dc51888048 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -540,8 +540,7 @@ static struct inode *v9fs_qid_iget(struct super_block *sb,
 	unlock_new_inode(inode);
 	return inode;
 error:
-	unlock_new_inode(inode);
-	iput(inode);
+	iget_failed(inode);
 	return ERR_PTR(retval);
 
 }
diff --git a/fs/9p/vfs_inode_dotl.c b/fs/9p/vfs_inode_dotl.c
index 09e4433717b8..e8aa57dc8d6d 100644
--- a/fs/9p/vfs_inode_dotl.c
+++ b/fs/9p/vfs_inode_dotl.c
@@ -149,8 +149,7 @@ static struct inode *v9fs_qid_iget_dotl(struct super_block *sb,
 	unlock_new_inode(inode);
 	return inode;
 error:
-	unlock_new_inode(inode);
-	iput(inode);
+	iget_failed(inode);
 	return ERR_PTR(retval);
 
 }
diff --git a/fs/affs/affs.h b/fs/affs/affs.h
index cffe8370fb44..c69a87eaf57d 100644
--- a/fs/affs/affs.h
+++ b/fs/affs/affs.h
@@ -64,7 +64,7 @@ struct affs_inode_info {
 /* short cut to get to the affs specific inode data */
 static inline struct affs_inode_info *AFFS_I(struct inode *inode)
 {
-	return list_entry(inode, struct affs_inode_info, vfs_inode);
+	return container_of(inode, struct affs_inode_info, vfs_inode);
 }
 
 /*
diff --git a/fs/autofs4/autofs_i.h b/fs/autofs4/autofs_i.h
index 5b700ef1e59d..c37149b929be 100644
--- a/fs/autofs4/autofs_i.h
+++ b/fs/autofs4/autofs_i.h
@@ -238,11 +238,6 @@ static inline u64 autofs4_get_ino(struct autofs_sb_info *sbi)
 	return d_inode(sbi->sb->s_root)->i_ino;
 }
 
-static inline int simple_positive(struct dentry *dentry)
-{
-	return d_really_is_positive(dentry) && !d_unhashed(dentry);
-}
-
 static inline void __autofs4_add_expiring(struct dentry *dentry)
 {
 	struct autofs_sb_info *sbi = autofs4_sbi(dentry->d_sb);
diff --git a/fs/befs/befs.h b/fs/befs/befs.h
index 1fead8d56a98..35d19e8731e3 100644
--- a/fs/befs/befs.h
+++ b/fs/befs/befs.h
@@ -112,7 +112,7 @@ BEFS_SB(const struct super_block *super)
 static inline struct befs_inode_info *
 BEFS_I(const struct inode *inode)
 {
-	return list_entry(inode, struct befs_inode_info, vfs_inode);
+	return container_of(inode, struct befs_inode_info, vfs_inode);
 }
 
 static inline befs_blocknr_t
diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
index cd46e4158830..6b659967898e 100644
--- a/fs/binfmt_elf.c
+++ b/fs/binfmt_elf.c
@@ -1530,7 +1530,7 @@ static int fill_files_note(struct memelfnote *note)
 		file = vma->vm_file;
 		if (!file)
 			continue;
-		filename = d_path(&file->f_path, name_curpos, remaining);
+		filename = file_path(file, name_curpos, remaining);
 		if (IS_ERR(filename)) {
 			if (PTR_ERR(filename) == -ENAMETOOLONG) {
 				vfree(data);
@@ -1540,7 +1540,7 @@ static int fill_files_note(struct memelfnote *note)
 			continue;
 		}
 
-		/* d_path() fills at the end, move name down */
+		/* file_path() fills at the end, move name down */
 		/* n = strlen(filename) + 1: */
 		n = (name_curpos + remaining) - filename;
 		remaining = filename - name_curpos;
diff --git a/fs/block_dev.c b/fs/block_dev.c
index 4fe10f93db8a..198243717da5 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -152,6 +152,9 @@ blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, loff_t offset)
 	struct file *file = iocb->ki_filp;
 	struct inode *inode = file->f_mapping->host;
 
+	if (IS_DAX(inode))
+		return dax_do_io(iocb, inode, iter, offset, blkdev_get_block,
+				NULL, DIO_SKIP_DIO_COUNT);
 	return __blockdev_direct_IO(iocb, inode, I_BDEV(inode), iter, offset,
 				    blkdev_get_block, NULL, NULL,
 				    DIO_SKIP_DIO_COUNT);
@@ -443,6 +446,12 @@ long bdev_direct_access(struct block_device *bdev, sector_t sector,
 	long avail;
 	const struct block_device_operations *ops = bdev->bd_disk->fops;
 
+	/*
+	 * The device driver is allowed to sleep, in order to make the
+	 * memory directly accessible.
+	 */
+	might_sleep();
+
 	if (size < 0)
 		return size;
 	if (!ops->direct_access)
@@ -1170,6 +1179,7 @@ static int __blkdev_get(struct block_device *bdev, fmode_t mode, int for_part)
 		bdev->bd_disk = disk;
 		bdev->bd_queue = disk->queue;
 		bdev->bd_contains = bdev;
+		bdev->bd_inode->i_flags = disk->fops->direct_access ? S_DAX : 0;
 		if (!partno) {
 			ret = -ENXIO;
 			bdev->bd_part = disk_get_part(disk, partno);
diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index 0ef5cc13fae2..81220b2203c6 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -44,6 +44,8 @@
 #define BTRFS_INODE_IN_DELALLOC_LIST		9
 #define BTRFS_INODE_READDIO_NEED_LOCK		10
 #define BTRFS_INODE_HAS_PROPS		        11
+/* DIO is ready to submit */
+#define BTRFS_INODE_DIO_READY		        12
 /*
  * The following 3 bits are meant only for the btree inode.
  * When any of them is set, it means an error happened while writing an
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 80a9aefb0c46..aac314e14188 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -1778,6 +1778,7 @@ struct btrfs_fs_info {
 	spinlock_t unused_bgs_lock;
 	struct list_head unused_bgs;
 	struct mutex unused_bg_unpin_mutex;
+	struct mutex delete_unused_bgs_mutex;
 
 	/* For btrfs to record security options */
 	struct security_mnt_opts security_opts;
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 3f43bfea3684..a9aadb2ad525 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1751,6 +1751,7 @@ static int cleaner_kthread(void *arg)
 {
 	struct btrfs_root *root = arg;
 	int again;
+	struct btrfs_trans_handle *trans;
 
 	do {
 		again = 0;
@@ -1772,7 +1773,6 @@ static int cleaner_kthread(void *arg)
 		}
 
 		btrfs_run_delayed_iputs(root);
-		btrfs_delete_unused_bgs(root->fs_info);
 		again = btrfs_clean_one_deleted_snapshot(root);
 		mutex_unlock(&root->fs_info->cleaner_mutex);
 
@@ -1781,6 +1781,16 @@ static int cleaner_kthread(void *arg)
 		 * needn't do anything special here.
 		 */
 		btrfs_run_defrag_inodes(root->fs_info);
+
+		/*
+		 * Acquires fs_info->delete_unused_bgs_mutex to avoid racing
+		 * with relocation (btrfs_relocate_chunk) and relocation
+		 * acquires fs_info->cleaner_mutex (btrfs_relocate_block_group)
+		 * after acquiring fs_info->delete_unused_bgs_mutex. So we
+		 * can't hold, nor need to, fs_info->cleaner_mutex when deleting
+		 * unused block groups.
+		 */
+		btrfs_delete_unused_bgs(root->fs_info);
 sleep:
 		if (!try_to_freeze() && !again) {
 			set_current_state(TASK_INTERRUPTIBLE);
@@ -1789,6 +1799,34 @@ sleep:
 			__set_current_state(TASK_RUNNING);
 		}
 	} while (!kthread_should_stop());
+
+	/*
+	 * Transaction kthread is stopped before us and wakes us up.
+	 * However we might have started a new transaction and COWed some
+	 * tree blocks when deleting unused block groups for example. So
+	 * make sure we commit the transaction we started to have a clean
+	 * shutdown when evicting the btree inode - if it has dirty pages
+	 * when we do the final iput() on it, eviction will trigger a
+	 * writeback for it which will fail with null pointer dereferences
+	 * since work queues and other resources were already released and
+	 * destroyed by the time the iput/eviction/writeback is made.
+	 */
+	trans = btrfs_attach_transaction(root);
+	if (IS_ERR(trans)) {
+		if (PTR_ERR(trans) != -ENOENT)
+			btrfs_err(root->fs_info,
+				  "cleaner transaction attach returned %ld",
+				  PTR_ERR(trans));
+	} else {
+		int ret;
+
+		ret = btrfs_commit_transaction(trans, root);
+		if (ret)
+			btrfs_err(root->fs_info,
+				  "cleaner open transaction commit returned %d",
+				  ret);
+	}
+
 	return 0;
 }
 
@@ -2492,6 +2530,7 @@ int open_ctree(struct super_block *sb,
 	spin_lock_init(&fs_info->unused_bgs_lock);
 	rwlock_init(&fs_info->tree_mod_log_lock);
 	mutex_init(&fs_info->unused_bg_unpin_mutex);
+	mutex_init(&fs_info->delete_unused_bgs_mutex);
 	mutex_init(&fs_info->reloc_mutex);
 	mutex_init(&fs_info->delalloc_root_mutex);
 	seqlock_init(&fs_info->profiles_lock);
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 38b76cc02f48..1c2bd1723e40 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -9889,6 +9889,8 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		}
 		spin_unlock(&fs_info->unused_bgs_lock);
 
+		mutex_lock(&root->fs_info->delete_unused_bgs_mutex);
+
 		/* Don't want to race with allocators so take the groups_sem */
 		down_write(&space_info->groups_sem);
 		spin_lock(&block_group->lock);
@@ -9983,6 +9985,7 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 end_trans:
 		btrfs_end_transaction(trans, root);
 next:
+		mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		btrfs_put_block_group(block_group);
 		spin_lock(&fs_info->unused_bgs_lock);
 	}
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 795d754327a7..b823fac91c92 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1748,7 +1748,7 @@ static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
 	}
 
 	current->backing_dev_info = inode_to_bdi(inode);
-	err = file_remove_suid(file);
+	err = file_remove_privs(file);
 	if (err) {
 		mutex_unlock(&inode->i_mutex);
 		goto out;
diff --git a/fs/btrfs/inode-map.c b/fs/btrfs/inode-map.c
index f6a596d5a637..d4a582ac3f73 100644
--- a/fs/btrfs/inode-map.c
+++ b/fs/btrfs/inode-map.c
@@ -246,6 +246,7 @@ void btrfs_unpin_free_ino(struct btrfs_root *root)
 {
 	struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
 	struct rb_root *rbroot = &root->free_ino_pinned->free_space_offset;
+	spinlock_t *rbroot_lock = &root->free_ino_pinned->tree_lock;
 	struct btrfs_free_space *info;
 	struct rb_node *n;
 	u64 count;
@@ -254,24 +255,30 @@ void btrfs_unpin_free_ino(struct btrfs_root *root)
 		return;
 
 	while (1) {
+		bool add_to_ctl = true;
+
+		spin_lock(rbroot_lock);
 		n = rb_first(rbroot);
-		if (!n)
+		if (!n) {
+			spin_unlock(rbroot_lock);
 			break;
+		}
 
 		info = rb_entry(n, struct btrfs_free_space, offset_index);
 		BUG_ON(info->bitmap); /* Logic error */
 
 		if (info->offset > root->ino_cache_progress)
-			goto free;
+			add_to_ctl = false;
 		else if (info->offset + info->bytes > root->ino_cache_progress)
 			count = root->ino_cache_progress - info->offset + 1;
 		else
 			count = info->bytes;
 
-		__btrfs_add_free_space(ctl, info->offset, count);
-free:
 		rb_erase(&info->offset_index, rbroot);
-		kfree(info);
+		spin_unlock(rbroot_lock);
+		if (add_to_ctl)
+			__btrfs_add_free_space(ctl, info->offset, count);
+		kmem_cache_free(btrfs_free_space_cachep, info);
 	}
 }
 
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 855935f6671a..b33c0cf02668 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -4989,8 +4989,9 @@ static void evict_inode_truncate_pages(struct inode *inode)
 	/*
 	 * Keep looping until we have no more ranges in the io tree.
 	 * We can have ongoing bios started by readpages (called from readahead)
-	 * that didn't get their end io callbacks called yet or they are still
-	 * in progress ((extent_io.c:end_bio_extent_readpage()). This means some
+	 * that have their endio callback (extent_io.c:end_bio_extent_readpage)
+	 * still in progress (unlocked the pages in the bio but did not yet
+	 * unlocked the ranges in the io tree). Therefore this means some
 	 * ranges can still be locked and eviction started because before
 	 * submitting those bios, which are executed by a separate task (work
 	 * queue kthread), inode references (inode->i_count) were not taken
@@ -7546,6 +7547,7 @@ unlock:
 
 		current->journal_info = outstanding_extents;
 		btrfs_free_reserved_data_space(inode, len);
+		set_bit(BTRFS_INODE_DIO_READY, &BTRFS_I(inode)->runtime_flags);
 	}
 
 	/*
@@ -7871,8 +7873,6 @@ static void btrfs_endio_direct_write(struct bio *bio, int err)
 	struct bio *dio_bio;
 	int ret;
 
-	if (err)
-		goto out_done;
 again:
 	ret = btrfs_dec_test_first_ordered_pending(inode, &ordered,
 						   &ordered_offset,
@@ -7895,7 +7895,6 @@ out_test:
 		ordered = NULL;
 		goto again;
 	}
-out_done:
 	dio_bio = dip->dio_bio;
 
 	kfree(dip);
@@ -8163,9 +8162,8 @@ out_err:
 static void btrfs_submit_direct(int rw, struct bio *dio_bio,
 				struct inode *inode, loff_t file_offset)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct btrfs_dio_private *dip;
-	struct bio *io_bio;
+	struct btrfs_dio_private *dip = NULL;
+	struct bio *io_bio = NULL;
 	struct btrfs_io_bio *btrfs_bio;
 	int skip_sum;
 	int write = rw & REQ_WRITE;
@@ -8182,7 +8180,7 @@ static void btrfs_submit_direct(int rw, struct bio *dio_bio,
 	dip = kzalloc(sizeof(*dip), GFP_NOFS);
 	if (!dip) {
 		ret = -ENOMEM;
-		goto free_io_bio;
+		goto free_ordered;
 	}
 
 	dip->private = dio_bio->bi_private;
@@ -8210,25 +8208,55 @@ static void btrfs_submit_direct(int rw, struct bio *dio_bio,
 
 	if (btrfs_bio->end_io)
 		btrfs_bio->end_io(btrfs_bio, ret);
-free_io_bio:
-	bio_put(io_bio);
 
 free_ordered:
 	/*
-	 * If this is a write, we need to clean up the reserved space and kill
-	 * the ordered extent.
+	 * If we arrived here it means either we failed to submit the dip
+	 * or we either failed to clone the dio_bio or failed to allocate the
+	 * dip. If we cloned the dio_bio and allocated the dip, we can just
+	 * call bio_endio against our io_bio so that we get proper resource
+	 * cleanup if we fail to submit the dip, otherwise, we must do the
+	 * same as btrfs_endio_direct_[write|read] because we can't call these
+	 * callbacks - they require an allocated dip and a clone of dio_bio.
 	 */
-	if (write) {
-		struct btrfs_ordered_extent *ordered;
-		ordered = btrfs_lookup_ordered_extent(inode, file_offset);
-		if (!test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags) &&
-		    !test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags))
-			btrfs_free_reserved_extent(root, ordered->start,
-						   ordered->disk_len, 1);
-		btrfs_put_ordered_extent(ordered);
-		btrfs_put_ordered_extent(ordered);
+	if (io_bio && dip) {
+		bio_endio(io_bio, ret);
+		/*
+		 * The end io callbacks free our dip, do the final put on io_bio
+		 * and all the cleanup and final put for dio_bio (through
+		 * dio_end_io()).
+		 */
+		dip = NULL;
+		io_bio = NULL;
+	} else {
+		if (write) {
+			struct btrfs_ordered_extent *ordered;
+
+			ordered = btrfs_lookup_ordered_extent(inode,
+							      file_offset);
+			set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
+			/*
+			 * Decrements our ref on the ordered extent and removes
+			 * the ordered extent from the inode's ordered tree,
+			 * doing all the proper resource cleanup such as for the
+			 * reserved space and waking up any waiters for this
+			 * ordered extent (through btrfs_remove_ordered_extent).
+			 */
+			btrfs_finish_ordered_io(ordered);
+		} else {
+			unlock_extent(&BTRFS_I(inode)->io_tree, file_offset,
+			      file_offset + dio_bio->bi_iter.bi_size - 1);
+		}
+		clear_bit(BIO_UPTODATE, &dio_bio->bi_flags);
+		/*
+		 * Releases and cleans up our dio_bio, no need to bio_put()
+		 * nor bio_endio()/bio_io_error() against dio_bio.
+		 */
+		dio_end_io(dio_bio, ret);
 	}
-	bio_endio(dio_bio, ret);
+	if (io_bio)
+		bio_put(io_bio);
+	kfree(dip);
 }
 
 static ssize_t check_direct_IO(struct btrfs_root *root, struct kiocb *iocb,
@@ -8330,9 +8358,18 @@ static ssize_t btrfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
 				   btrfs_submit_direct, flags);
 	if (iov_iter_rw(iter) == WRITE) {
 		current->journal_info = NULL;
-		if (ret < 0 && ret != -EIOCBQUEUED)
-			btrfs_delalloc_release_space(inode, count);
-		else if (ret >= 0 && (size_t)ret < count)
+		if (ret < 0 && ret != -EIOCBQUEUED) {
+			/*
+			 * If the error comes from submitting stage,
+			 * btrfs_get_blocsk_direct() has free'd data space,
+			 * and metadata space will be handled by
+			 * finish_ordered_fn, don't do that again to make
+			 * sure bytes_may_use is correct.
+			 */
+			if (!test_and_clear_bit(BTRFS_INODE_DIO_READY,
+				     &BTRFS_I(inode)->runtime_flags))
+				btrfs_delalloc_release_space(inode, count);
+		} else if (ret >= 0 && (size_t)ret < count)
 			btrfs_delalloc_release_space(inode,
 						     count - (size_t)ret);
 	}
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index c86b835da7a8..5d91776e12a2 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -87,7 +87,8 @@ struct btrfs_ioctl_received_subvol_args_32 {
 
 
 static int btrfs_clone(struct inode *src, struct inode *inode,
-		       u64 off, u64 olen, u64 olen_aligned, u64 destoff);
+		       u64 off, u64 olen, u64 olen_aligned, u64 destoff,
+		       int no_time_update);
 
 /* Mask out flags that are inappropriate for the given type of inode. */
 static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags)
@@ -2765,14 +2766,11 @@ out:
 	return ret;
 }
 
-static struct page *extent_same_get_page(struct inode *inode, u64 off)
+static struct page *extent_same_get_page(struct inode *inode, pgoff_t index)
 {
 	struct page *page;
-	pgoff_t index;
 	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
 
-	index = off >> PAGE_CACHE_SHIFT;
-
 	page = grab_cache_page(inode->i_mapping, index);
 	if (!page)
 		return NULL;
@@ -2793,6 +2791,20 @@ static struct page *extent_same_get_page(struct inode *inode, u64 off)
 	return page;
 }
 
+static int gather_extent_pages(struct inode *inode, struct page **pages,
+			       int num_pages, u64 off)
+{
+	int i;
+	pgoff_t index = off >> PAGE_CACHE_SHIFT;
+
+	for (i = 0; i < num_pages; i++) {
+		pages[i] = extent_same_get_page(inode, index + i);
+		if (!pages[i])
+			return -ENOMEM;
+	}
+	return 0;
+}
+
 static inline void lock_extent_range(struct inode *inode, u64 off, u64 len)
 {
 	/* do any pending delalloc/csum calc on src, one way or
@@ -2818,52 +2830,120 @@ static inline void lock_extent_range(struct inode *inode, u64 off, u64 len)
 	}
 }
 
-static void btrfs_double_unlock(struct inode *inode1, u64 loff1,
-				struct inode *inode2, u64 loff2, u64 len)
+static void btrfs_double_inode_unlock(struct inode *inode1, struct inode *inode2)
 {
-	unlock_extent(&BTRFS_I(inode1)->io_tree, loff1, loff1 + len - 1);
-	unlock_extent(&BTRFS_I(inode2)->io_tree, loff2, loff2 + len - 1);
-
 	mutex_unlock(&inode1->i_mutex);
 	mutex_unlock(&inode2->i_mutex);
 }
 
-static void btrfs_double_lock(struct inode *inode1, u64 loff1,
-			      struct inode *inode2, u64 loff2, u64 len)
+static void btrfs_double_inode_lock(struct inode *inode1, struct inode *inode2)
+{
+	if (inode1 < inode2)
+		swap(inode1, inode2);
+
+	mutex_lock_nested(&inode1->i_mutex, I_MUTEX_PARENT);
+	if (inode1 != inode2)
+		mutex_lock_nested(&inode2->i_mutex, I_MUTEX_CHILD);
+}
+
+static void btrfs_double_extent_unlock(struct inode *inode1, u64 loff1,
+				      struct inode *inode2, u64 loff2, u64 len)
+{
+	unlock_extent(&BTRFS_I(inode1)->io_tree, loff1, loff1 + len - 1);
+	unlock_extent(&BTRFS_I(inode2)->io_tree, loff2, loff2 + len - 1);
+}
+
+static void btrfs_double_extent_lock(struct inode *inode1, u64 loff1,
+				     struct inode *inode2, u64 loff2, u64 len)
 {
 	if (inode1 < inode2) {
 		swap(inode1, inode2);
 		swap(loff1, loff2);
 	}
-
-	mutex_lock_nested(&inode1->i_mutex, I_MUTEX_PARENT);
 	lock_extent_range(inode1, loff1, len);
-	if (inode1 != inode2) {
-		mutex_lock_nested(&inode2->i_mutex, I_MUTEX_CHILD);
+	if (inode1 != inode2)
 		lock_extent_range(inode2, loff2, len);
+}
+
+struct cmp_pages {
+	int		num_pages;
+	struct page	**src_pages;
+	struct page	**dst_pages;
+};
+
+static void btrfs_cmp_data_free(struct cmp_pages *cmp)
+{
+	int i;
+	struct page *pg;
+
+	for (i = 0; i < cmp->num_pages; i++) {
+		pg = cmp->src_pages[i];
+		if (pg)
+			page_cache_release(pg);
+		pg = cmp->dst_pages[i];
+		if (pg)
+			page_cache_release(pg);
+	}
+	kfree(cmp->src_pages);
+	kfree(cmp->dst_pages);
+}
+
+static int btrfs_cmp_data_prepare(struct inode *src, u64 loff,
+				  struct inode *dst, u64 dst_loff,
+				  u64 len, struct cmp_pages *cmp)
+{
+	int ret;
+	int num_pages = PAGE_CACHE_ALIGN(len) >> PAGE_CACHE_SHIFT;
+	struct page **src_pgarr, **dst_pgarr;
+
+	/*
+	 * We must gather up all the pages before we initiate our
+	 * extent locking. We use an array for the page pointers. Size
+	 * of the array is bounded by len, which is in turn bounded by
+	 * BTRFS_MAX_DEDUPE_LEN.
+	 */
+	src_pgarr = kzalloc(num_pages * sizeof(struct page *), GFP_NOFS);
+	dst_pgarr = kzalloc(num_pages * sizeof(struct page *), GFP_NOFS);
+	if (!src_pgarr || !dst_pgarr) {
+		kfree(src_pgarr);
+		kfree(dst_pgarr);
+		return -ENOMEM;
 	}
+	cmp->num_pages = num_pages;
+	cmp->src_pages = src_pgarr;
+	cmp->dst_pages = dst_pgarr;
+
+	ret = gather_extent_pages(src, cmp->src_pages, cmp->num_pages, loff);
+	if (ret)
+		goto out;
+
+	ret = gather_extent_pages(dst, cmp->dst_pages, cmp->num_pages, dst_loff);
+
+out:
+	if (ret)
+		btrfs_cmp_data_free(cmp);
+	return 0;
 }
 
 static int btrfs_cmp_data(struct inode *src, u64 loff, struct inode *dst,
-			  u64 dst_loff, u64 len)
+			  u64 dst_loff, u64 len, struct cmp_pages *cmp)
 {
 	int ret = 0;
+	int i;
 	struct page *src_page, *dst_page;
 	unsigned int cmp_len = PAGE_CACHE_SIZE;
 	void *addr, *dst_addr;
 
+	i = 0;
 	while (len) {
 		if (len < PAGE_CACHE_SIZE)
 			cmp_len = len;
 
-		src_page = extent_same_get_page(src, loff);
-		if (!src_page)
-			return -EINVAL;
-		dst_page = extent_same_get_page(dst, dst_loff);
-		if (!dst_page) {
-			page_cache_release(src_page);
-			return -EINVAL;
-		}
+		BUG_ON(i >= cmp->num_pages);
+
+		src_page = cmp->src_pages[i];
+		dst_page = cmp->dst_pages[i];
+
 		addr = kmap_atomic(src_page);
 		dst_addr = kmap_atomic(dst_page);
 
@@ -2875,15 +2955,12 @@ static int btrfs_cmp_data(struct inode *src, u64 loff, struct inode *dst,
 
 		kunmap_atomic(addr);
 		kunmap_atomic(dst_addr);
-		page_cache_release(src_page);
-		page_cache_release(dst_page);
 
 		if (ret)
 			break;
 
-		loff += cmp_len;
-		dst_loff += cmp_len;
 		len -= cmp_len;
+		i++;
 	}
 
 	return ret;
@@ -2914,27 +2991,62 @@ static int btrfs_extent_same(struct inode *src, u64 loff, u64 olen,
 {
 	int ret;
 	u64 len = olen;
+	struct cmp_pages cmp;
+	int same_inode = 0;
+	u64 same_lock_start = 0;
+	u64 same_lock_len = 0;
 
-	/*
-	 * btrfs_clone() can't handle extents in the same file
-	 * yet. Once that works, we can drop this check and replace it
-	 * with a check for the same inode, but overlapping extents.
-	 */
 	if (src == dst)
-		return -EINVAL;
+		same_inode = 1;
 
 	if (len == 0)
 		return 0;
 
-	btrfs_double_lock(src, loff, dst, dst_loff, len);
+	if (same_inode) {
+		mutex_lock(&src->i_mutex);
 
-	ret = extent_same_check_offsets(src, loff, &len, olen);
-	if (ret)
-		goto out_unlock;
+		ret = extent_same_check_offsets(src, loff, &len, olen);
+		if (ret)
+			goto out_unlock;
 
-	ret = extent_same_check_offsets(dst, dst_loff, &len, olen);
-	if (ret)
-		goto out_unlock;
+		/*
+		 * Single inode case wants the same checks, except we
+		 * don't want our length pushed out past i_size as
+		 * comparing that data range makes no sense.
+		 *
+		 * extent_same_check_offsets() will do this for an
+		 * unaligned length at i_size, so catch it here and
+		 * reject the request.
+		 *
+		 * This effectively means we require aligned extents
+		 * for the single-inode case, whereas the other cases
+		 * allow an unaligned length so long as it ends at
+		 * i_size.
+		 */
+		if (len != olen) {
+			ret = -EINVAL;
+			goto out_unlock;
+		}
+
+		/* Check for overlapping ranges */
+		if (dst_loff + len > loff && dst_loff < loff + len) {
+			ret = -EINVAL;
+			goto out_unlock;
+		}
+
+		same_lock_start = min_t(u64, loff, dst_loff);
+		same_lock_len = max_t(u64, loff, dst_loff) + len - same_lock_start;
+	} else {
+		btrfs_double_inode_lock(src, dst);
+
+		ret = extent_same_check_offsets(src, loff, &len, olen);
+		if (ret)
+			goto out_unlock;
+
+		ret = extent_same_check_offsets(dst, dst_loff, &len, olen);
+		if (ret)
+			goto out_unlock;
+	}
 
 	/* don't make the dst file partly checksummed */
 	if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
@@ -2943,12 +3055,32 @@ static int btrfs_extent_same(struct inode *src, u64 loff, u64 olen,
 		goto out_unlock;
 	}
 
-	ret = btrfs_cmp_data(src, loff, dst, dst_loff, len);
+	ret = btrfs_cmp_data_prepare(src, loff, dst, dst_loff, olen, &cmp);
+	if (ret)
+		goto out_unlock;
+
+	if (same_inode)
+		lock_extent_range(src, same_lock_start, same_lock_len);
+	else
+		btrfs_double_extent_lock(src, loff, dst, dst_loff, len);
+
+	/* pass original length for comparison so we stay within i_size */
+	ret = btrfs_cmp_data(src, loff, dst, dst_loff, olen, &cmp);
 	if (ret == 0)
-		ret = btrfs_clone(src, dst, loff, olen, len, dst_loff);
+		ret = btrfs_clone(src, dst, loff, olen, len, dst_loff, 1);
+
+	if (same_inode)
+		unlock_extent(&BTRFS_I(src)->io_tree, same_lock_start,
+			      same_lock_start + same_lock_len - 1);
+	else
+		btrfs_double_extent_unlock(src, loff, dst, dst_loff, len);
 
+	btrfs_cmp_data_free(&cmp);
 out_unlock:
-	btrfs_double_unlock(src, loff, dst, dst_loff, len);
+	if (same_inode)
+		mutex_unlock(&src->i_mutex);
+	else
+		btrfs_double_inode_unlock(src, dst);
 
 	return ret;
 }
@@ -3100,13 +3232,15 @@ static int clone_finish_inode_update(struct btrfs_trans_handle *trans,
 				     struct inode *inode,
 				     u64 endoff,
 				     const u64 destoff,
-				     const u64 olen)
+				     const u64 olen,
+				     int no_time_update)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	int ret;
 
 	inode_inc_iversion(inode);
-	inode->i_mtime = inode->i_ctime = CURRENT_TIME;
+	if (!no_time_update)
+		inode->i_mtime = inode->i_ctime = CURRENT_TIME;
 	/*
 	 * We round up to the block size at eof when determining which
 	 * extents to clone above, but shouldn't round up the file size.
@@ -3191,13 +3325,13 @@ static void clone_update_extent_map(struct inode *inode,
  * @inode: Inode to clone to
  * @off: Offset within source to start clone from
  * @olen: Original length, passed by user, of range to clone
- * @olen_aligned: Block-aligned value of olen, extent_same uses
- *               identical values here
+ * @olen_aligned: Block-aligned value of olen
  * @destoff: Offset within @inode to start clone
+ * @no_time_update: Whether to update mtime/ctime on the target inode
  */
 static int btrfs_clone(struct inode *src, struct inode *inode,
 		       const u64 off, const u64 olen, const u64 olen_aligned,
-		       const u64 destoff)
+		       const u64 destoff, int no_time_update)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_path *path = NULL;
@@ -3521,7 +3655,8 @@ process_slot:
 					      root->sectorsize);
 			ret = clone_finish_inode_update(trans, inode,
 							last_dest_end,
-							destoff, olen);
+							destoff, olen,
+							no_time_update);
 			if (ret)
 				goto out;
 			if (new_key.offset + datal >= destoff + len)
@@ -3559,7 +3694,7 @@ process_slot:
 		clone_update_extent_map(inode, trans, NULL, last_dest_end,
 					destoff + len - last_dest_end);
 		ret = clone_finish_inode_update(trans, inode, destoff + len,
-						destoff, olen);
+						destoff, olen, no_time_update);
 	}
 
 out:
@@ -3696,7 +3831,7 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 		lock_extent_range(inode, destoff, len);
 	}
 
-	ret = btrfs_clone(src, inode, off, olen, len, destoff);
+	ret = btrfs_clone(src, inode, off, olen, len, destoff, 0);
 
 	if (same_inode) {
 		u64 lock_start = min_t(u64, off, destoff);
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index 89656d799ff6..52170cf1757e 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -552,6 +552,10 @@ void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
 	trace_btrfs_ordered_extent_put(entry->inode, entry);
 
 	if (atomic_dec_and_test(&entry->refs)) {
+		ASSERT(list_empty(&entry->log_list));
+		ASSERT(list_empty(&entry->trans_list));
+		ASSERT(list_empty(&entry->root_extent_list));
+		ASSERT(RB_EMPTY_NODE(&entry->rb_node));
 		if (entry->inode)
 			btrfs_add_delayed_iput(entry->inode);
 		while (!list_empty(&entry->list)) {
@@ -579,6 +583,7 @@ void btrfs_remove_ordered_extent(struct inode *inode,
 	spin_lock_irq(&tree->lock);
 	node = &entry->rb_node;
 	rb_erase(node, &tree->tree);
+	RB_CLEAR_NODE(node);
 	if (tree->last == node)
 		tree->last = NULL;
 	set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags);
diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index d5f1f033b7a0..e9ace099162c 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -1349,6 +1349,11 @@ int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
 	struct btrfs_root *quota_root;
 	struct btrfs_qgroup *qgroup;
 	int ret = 0;
+	/* Sometimes we would want to clear the limit on this qgroup.
+	 * To meet this requirement, we treat the -1 as a special value
+	 * which tell kernel to clear the limit on this qgroup.
+	 */
+	const u64 CLEAR_VALUE = -1;
 
 	mutex_lock(&fs_info->qgroup_ioctl_lock);
 	quota_root = fs_info->quota_root;
@@ -1364,14 +1369,42 @@ int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
 	}
 
 	spin_lock(&fs_info->qgroup_lock);
-	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_RFER)
-		qgroup->max_rfer = limit->max_rfer;
-	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_EXCL)
-		qgroup->max_excl = limit->max_excl;
-	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_RFER)
-		qgroup->rsv_rfer = limit->rsv_rfer;
-	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_EXCL)
-		qgroup->rsv_excl = limit->rsv_excl;
+	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_RFER) {
+		if (limit->max_rfer == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_MAX_RFER;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_MAX_RFER;
+			qgroup->max_rfer = 0;
+		} else {
+			qgroup->max_rfer = limit->max_rfer;
+		}
+	}
+	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_EXCL) {
+		if (limit->max_excl == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_MAX_EXCL;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_MAX_EXCL;
+			qgroup->max_excl = 0;
+		} else {
+			qgroup->max_excl = limit->max_excl;
+		}
+	}
+	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_RFER) {
+		if (limit->rsv_rfer == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_RSV_RFER;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_RSV_RFER;
+			qgroup->rsv_rfer = 0;
+		} else {
+			qgroup->rsv_rfer = limit->rsv_rfer;
+		}
+	}
+	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_EXCL) {
+		if (limit->rsv_excl == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_RSV_EXCL;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_RSV_EXCL;
+			qgroup->rsv_excl = 0;
+		} else {
+			qgroup->rsv_excl = limit->rsv_excl;
+		}
+	}
 	qgroup->lim_flags |= limit->flags;
 
 	spin_unlock(&fs_info->qgroup_lock);
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 827951fbf7fc..88cbb5995667 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -4049,7 +4049,7 @@ restart:
 	if (trans && progress && err == -ENOSPC) {
 		ret = btrfs_force_chunk_alloc(trans, rc->extent_root,
 					      rc->block_group->flags);
-		if (ret == 0) {
+		if (ret == 1) {
 			err = 0;
 			progress = 0;
 			goto restart;
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 9f2feabe99f2..94db0fa5225a 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -3571,7 +3571,6 @@ static noinline_for_stack int scrub_supers(struct scrub_ctx *sctx,
 static noinline_for_stack int scrub_workers_get(struct btrfs_fs_info *fs_info,
 						int is_dev_replace)
 {
-	int ret = 0;
 	unsigned int flags = WQ_FREEZABLE | WQ_UNBOUND;
 	int max_active = fs_info->thread_pool_size;
 
@@ -3584,34 +3583,36 @@ static noinline_for_stack int scrub_workers_get(struct btrfs_fs_info *fs_info,
 			fs_info->scrub_workers =
 				btrfs_alloc_workqueue("btrfs-scrub", flags,
 						      max_active, 4);
-		if (!fs_info->scrub_workers) {
-			ret = -ENOMEM;
-			goto out;
-		}
+		if (!fs_info->scrub_workers)
+			goto fail_scrub_workers;
+
 		fs_info->scrub_wr_completion_workers =
 			btrfs_alloc_workqueue("btrfs-scrubwrc", flags,
 					      max_active, 2);
-		if (!fs_info->scrub_wr_completion_workers) {
-			ret = -ENOMEM;
-			goto out;
-		}
+		if (!fs_info->scrub_wr_completion_workers)
+			goto fail_scrub_wr_completion_workers;
+
 		fs_info->scrub_nocow_workers =
 			btrfs_alloc_workqueue("btrfs-scrubnc", flags, 1, 0);
-		if (!fs_info->scrub_nocow_workers) {
-			ret = -ENOMEM;
-			goto out;
-		}
+		if (!fs_info->scrub_nocow_workers)
+			goto fail_scrub_nocow_workers;
 		fs_info->scrub_parity_workers =
 			btrfs_alloc_workqueue("btrfs-scrubparity", flags,
 					      max_active, 2);
-		if (!fs_info->scrub_parity_workers) {
-			ret = -ENOMEM;
-			goto out;
-		}
+		if (!fs_info->scrub_parity_workers)
+			goto fail_scrub_parity_workers;
 	}
 	++fs_info->scrub_workers_refcnt;
-out:
-	return ret;
+	return 0;
+
+fail_scrub_parity_workers:
+	btrfs_destroy_workqueue(fs_info->scrub_nocow_workers);
+fail_scrub_nocow_workers:
+	btrfs_destroy_workqueue(fs_info->scrub_wr_completion_workers);
+fail_scrub_wr_completion_workers:
+	btrfs_destroy_workqueue(fs_info->scrub_workers);
+fail_scrub_workers:
+	return -ENOMEM;
 }
 
 static noinline_for_stack void scrub_workers_put(struct btrfs_fs_info *fs_info)
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index 1ce80c1c4eb6..9c45431e69ab 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -4117,6 +4117,187 @@ static int logged_inode_size(struct btrfs_root *log, struct inode *inode,
 	return 0;
 }
 
+/*
+ * At the moment we always log all xattrs. This is to figure out at log replay
+ * time which xattrs must have their deletion replayed. If a xattr is missing
+ * in the log tree and exists in the fs/subvol tree, we delete it. This is
+ * because if a xattr is deleted, the inode is fsynced and a power failure
+ * happens, causing the log to be replayed the next time the fs is mounted,
+ * we want the xattr to not exist anymore (same behaviour as other filesystems
+ * with a journal, ext3/4, xfs, f2fs, etc).
+ */
+static int btrfs_log_all_xattrs(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root,
+				struct inode *inode,
+				struct btrfs_path *path,
+				struct btrfs_path *dst_path)
+{
+	int ret;
+	struct btrfs_key key;
+	const u64 ino = btrfs_ino(inode);
+	int ins_nr = 0;
+	int start_slot = 0;
+
+	key.objectid = ino;
+	key.type = BTRFS_XATTR_ITEM_KEY;
+	key.offset = 0;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (ret < 0)
+		return ret;
+
+	while (true) {
+		int slot = path->slots[0];
+		struct extent_buffer *leaf = path->nodes[0];
+		int nritems = btrfs_header_nritems(leaf);
+
+		if (slot >= nritems) {
+			if (ins_nr > 0) {
+				u64 last_extent = 0;
+
+				ret = copy_items(trans, inode, dst_path, path,
+						 &last_extent, start_slot,
+						 ins_nr, 1, 0);
+				/* can't be 1, extent items aren't processed */
+				ASSERT(ret <= 0);
+				if (ret < 0)
+					return ret;
+				ins_nr = 0;
+			}
+			ret = btrfs_next_leaf(root, path);
+			if (ret < 0)
+				return ret;
+			else if (ret > 0)
+				break;
+			continue;
+		}
+
+		btrfs_item_key_to_cpu(leaf, &key, slot);
+		if (key.objectid != ino || key.type != BTRFS_XATTR_ITEM_KEY)
+			break;
+
+		if (ins_nr == 0)
+			start_slot = slot;
+		ins_nr++;
+		path->slots[0]++;
+		cond_resched();
+	}
+	if (ins_nr > 0) {
+		u64 last_extent = 0;
+
+		ret = copy_items(trans, inode, dst_path, path,
+				 &last_extent, start_slot,
+				 ins_nr, 1, 0);
+		/* can't be 1, extent items aren't processed */
+		ASSERT(ret <= 0);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+/*
+ * If the no holes feature is enabled we need to make sure any hole between the
+ * last extent and the i_size of our inode is explicitly marked in the log. This
+ * is to make sure that doing something like:
+ *
+ *      1) create file with 128Kb of data
+ *      2) truncate file to 64Kb
+ *      3) truncate file to 256Kb
+ *      4) fsync file
+ *      5) <crash/power failure>
+ *      6) mount fs and trigger log replay
+ *
+ * Will give us a file with a size of 256Kb, the first 64Kb of data match what
+ * the file had in its first 64Kb of data at step 1 and the last 192Kb of the
+ * file correspond to a hole. The presence of explicit holes in a log tree is
+ * what guarantees that log replay will remove/adjust file extent items in the
+ * fs/subvol tree.
+ *
+ * Here we do not need to care about holes between extents, that is already done
+ * by copy_items(). We also only need to do this in the full sync path, where we
+ * lookup for extents from the fs/subvol tree only. In the fast path case, we
+ * lookup the list of modified extent maps and if any represents a hole, we
+ * insert a corresponding extent representing a hole in the log tree.
+ */
+static int btrfs_log_trailing_hole(struct btrfs_trans_handle *trans,
+				   struct btrfs_root *root,
+				   struct inode *inode,
+				   struct btrfs_path *path)
+{
+	int ret;
+	struct btrfs_key key;
+	u64 hole_start;
+	u64 hole_size;
+	struct extent_buffer *leaf;
+	struct btrfs_root *log = root->log_root;
+	const u64 ino = btrfs_ino(inode);
+	const u64 i_size = i_size_read(inode);
+
+	if (!btrfs_fs_incompat(root->fs_info, NO_HOLES))
+		return 0;
+
+	key.objectid = ino;
+	key.type = BTRFS_EXTENT_DATA_KEY;
+	key.offset = (u64)-1;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	ASSERT(ret != 0);
+	if (ret < 0)
+		return ret;
+
+	ASSERT(path->slots[0] > 0);
+	path->slots[0]--;
+	leaf = path->nodes[0];
+	btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+
+	if (key.objectid != ino || key.type != BTRFS_EXTENT_DATA_KEY) {
+		/* inode does not have any extents */
+		hole_start = 0;
+		hole_size = i_size;
+	} else {
+		struct btrfs_file_extent_item *extent;
+		u64 len;
+
+		/*
+		 * If there's an extent beyond i_size, an explicit hole was
+		 * already inserted by copy_items().
+		 */
+		if (key.offset >= i_size)
+			return 0;
+
+		extent = btrfs_item_ptr(leaf, path->slots[0],
+					struct btrfs_file_extent_item);
+
+		if (btrfs_file_extent_type(leaf, extent) ==
+		    BTRFS_FILE_EXTENT_INLINE) {
+			len = btrfs_file_extent_inline_len(leaf,
+							   path->slots[0],
+							   extent);
+			ASSERT(len == i_size);
+			return 0;
+		}
+
+		len = btrfs_file_extent_num_bytes(leaf, extent);
+		/* Last extent goes beyond i_size, no need to log a hole. */
+		if (key.offset + len > i_size)
+			return 0;
+		hole_start = key.offset + len;
+		hole_size = i_size - hole_start;
+	}
+	btrfs_release_path(path);
+
+	/* Last extent ends at i_size. */
+	if (hole_size == 0)
+		return 0;
+
+	hole_size = ALIGN(hole_size, root->sectorsize);
+	ret = btrfs_insert_file_extent(trans, log, ino, hole_start, 0, 0,
+				       hole_size, 0, hole_size, 0, 0, 0);
+	return ret;
+}
+
 /* log a single inode in the tree log.
  * At least one parent directory for this inode must exist in the tree
  * or be logged already.
@@ -4155,6 +4336,7 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 	u64 ino = btrfs_ino(inode);
 	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	u64 logged_isize = 0;
+	bool need_log_inode_item = true;
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -4263,11 +4445,6 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 		} else {
 			if (inode_only == LOG_INODE_ALL)
 				fast_search = true;
-			ret = log_inode_item(trans, log, dst_path, inode);
-			if (ret) {
-				err = ret;
-				goto out_unlock;
-			}
 			goto log_extents;
 		}
 
@@ -4290,6 +4467,28 @@ again:
 		if (min_key.type > max_key.type)
 			break;
 
+		if (min_key.type == BTRFS_INODE_ITEM_KEY)
+			need_log_inode_item = false;
+
+		/* Skip xattrs, we log them later with btrfs_log_all_xattrs() */
+		if (min_key.type == BTRFS_XATTR_ITEM_KEY) {
+			if (ins_nr == 0)
+				goto next_slot;
+			ret = copy_items(trans, inode, dst_path, path,
+					 &last_extent, ins_start_slot,
+					 ins_nr, inode_only, logged_isize);
+			if (ret < 0) {
+				err = ret;
+				goto out_unlock;
+			}
+			ins_nr = 0;
+			if (ret) {
+				btrfs_release_path(path);
+				continue;
+			}
+			goto next_slot;
+		}
+
 		src = path->nodes[0];
 		if (ins_nr && ins_start_slot + ins_nr == path->slots[0]) {
 			ins_nr++;
@@ -4357,9 +4556,26 @@ next_slot:
 		ins_nr = 0;
 	}
 
+	btrfs_release_path(path);
+	btrfs_release_path(dst_path);
+	err = btrfs_log_all_xattrs(trans, root, inode, path, dst_path);
+	if (err)
+		goto out_unlock;
+	if (max_key.type >= BTRFS_EXTENT_DATA_KEY && !fast_search) {
+		btrfs_release_path(path);
+		btrfs_release_path(dst_path);
+		err = btrfs_log_trailing_hole(trans, root, inode, path);
+		if (err)
+			goto out_unlock;
+	}
 log_extents:
 	btrfs_release_path(path);
 	btrfs_release_path(dst_path);
+	if (need_log_inode_item) {
+		err = log_inode_item(trans, log, dst_path, inode);
+		if (err)
+			goto out_unlock;
+	}
 	if (fast_search) {
 		/*
 		 * Some ordered extents started by fsync might have completed
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 4b438b4c8c91..fbe7c104531c 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2766,6 +2766,20 @@ static int btrfs_relocate_chunk(struct btrfs_root *root,
 	root = root->fs_info->chunk_root;
 	extent_root = root->fs_info->extent_root;
 
+	/*
+	 * Prevent races with automatic removal of unused block groups.
+	 * After we relocate and before we remove the chunk with offset
+	 * chunk_offset, automatic removal of the block group can kick in,
+	 * resulting in a failure when calling btrfs_remove_chunk() below.
+	 *
+	 * Make sure to acquire this mutex before doing a tree search (dev
+	 * or chunk trees) to find chunks. Otherwise the cleaner kthread might
+	 * call btrfs_remove_chunk() (through btrfs_delete_unused_bgs()) after
+	 * we release the path used to search the chunk/dev tree and before
+	 * the current task acquires this mutex and calls us.
+	 */
+	ASSERT(mutex_is_locked(&root->fs_info->delete_unused_bgs_mutex));
+
 	ret = btrfs_can_relocate(extent_root, chunk_offset);
 	if (ret)
 		return -ENOSPC;
@@ -2814,13 +2828,18 @@ again:
 	key.type = BTRFS_CHUNK_ITEM_KEY;
 
 	while (1) {
+		mutex_lock(&root->fs_info->delete_unused_bgs_mutex);
 		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
-		if (ret < 0)
+		if (ret < 0) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			goto error;
+		}
 		BUG_ON(ret == 0); /* Corruption */
 
 		ret = btrfs_previous_item(chunk_root, path, key.objectid,
 					  key.type);
+		if (ret)
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		if (ret < 0)
 			goto error;
 		if (ret > 0)
@@ -2843,6 +2862,7 @@ again:
 			else
 				BUG_ON(ret);
 		}
+		mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 
 		if (found_key.offset == 0)
 			break;
@@ -3299,9 +3319,12 @@ again:
 			goto error;
 		}
 
+		mutex_lock(&fs_info->delete_unused_bgs_mutex);
 		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
-		if (ret < 0)
+		if (ret < 0) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			goto error;
+		}
 
 		/*
 		 * this shouldn't happen, it means the last relocate
@@ -3313,6 +3336,7 @@ again:
 		ret = btrfs_previous_item(chunk_root, path, 0,
 					  BTRFS_CHUNK_ITEM_KEY);
 		if (ret) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			ret = 0;
 			break;
 		}
@@ -3321,8 +3345,10 @@ again:
 		slot = path->slots[0];
 		btrfs_item_key_to_cpu(leaf, &found_key, slot);
 
-		if (found_key.objectid != key.objectid)
+		if (found_key.objectid != key.objectid) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			break;
+		}
 
 		chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);
 
@@ -3335,10 +3361,13 @@ again:
 		ret = should_balance_chunk(chunk_root, leaf, chunk,
 					   found_key.offset);
 		btrfs_release_path(path);
-		if (!ret)
+		if (!ret) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			goto loop;
+		}
 
 		if (counting) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			spin_lock(&fs_info->balance_lock);
 			bctl->stat.expected++;
 			spin_unlock(&fs_info->balance_lock);
@@ -3348,6 +3377,7 @@ again:
 		ret = btrfs_relocate_chunk(chunk_root,
 					   found_key.objectid,
 					   found_key.offset);
+		mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 		if (ret && ret != -ENOSPC)
 			goto error;
 		if (ret == -ENOSPC) {
@@ -4087,11 +4117,16 @@ again:
 	key.type = BTRFS_DEV_EXTENT_KEY;
 
 	do {
+		mutex_lock(&root->fs_info->delete_unused_bgs_mutex);
 		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-		if (ret < 0)
+		if (ret < 0) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			goto done;
+		}
 
 		ret = btrfs_previous_item(root, path, 0, key.type);
+		if (ret)
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		if (ret < 0)
 			goto done;
 		if (ret) {
@@ -4105,6 +4140,7 @@ again:
 		btrfs_item_key_to_cpu(l, &key, path->slots[0]);
 
 		if (key.objectid != device->devid) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			btrfs_release_path(path);
 			break;
 		}
@@ -4113,6 +4149,7 @@ again:
 		length = btrfs_dev_extent_length(l, dev_extent);
 
 		if (key.offset + length <= new_size) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			btrfs_release_path(path);
 			break;
 		}
@@ -4122,6 +4159,7 @@ again:
 		btrfs_release_path(path);
 
 		ret = btrfs_relocate_chunk(root, chunk_objectid, chunk_offset);
+		mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		if (ret && ret != -ENOSPC)
 			goto done;
 		if (ret == -ENOSPC)
@@ -5715,7 +5753,6 @@ static inline void btrfs_end_bbio(struct btrfs_bio *bbio, struct bio *bio, int e
 static void btrfs_end_bio(struct bio *bio, int err)
 {
 	struct btrfs_bio *bbio = bio->bi_private;
-	struct btrfs_device *dev = bbio->stripes[0].dev;
 	int is_orig_bio = 0;
 
 	if (err) {
@@ -5723,6 +5760,7 @@ static void btrfs_end_bio(struct bio *bio, int err)
 		if (err == -EIO || err == -EREMOTEIO) {
 			unsigned int stripe_index =
 				btrfs_io_bio(bio)->stripe_index;
+			struct btrfs_device *dev;
 
 			BUG_ON(stripe_index >= bbio->num_stripes);
 			dev = bbio->stripes[stripe_index].dev;
diff --git a/fs/cachefiles/internal.h b/fs/cachefiles/internal.h
index 8c52472d2efa..aecd0859eacb 100644
--- a/fs/cachefiles/internal.h
+++ b/fs/cachefiles/internal.h
@@ -43,7 +43,6 @@ struct cachefiles_object {
 	loff_t				i_size;		/* object size */
 	unsigned long			flags;
 #define CACHEFILES_OBJECT_ACTIVE	0		/* T if marked active */
-#define CACHEFILES_OBJECT_BURIED	1		/* T if preemptively buried */
 	atomic_t			usage;		/* object usage count */
 	uint8_t				type;		/* object type */
 	uint8_t				new;		/* T if object new */
diff --git a/fs/cachefiles/namei.c b/fs/cachefiles/namei.c
index ab857ab9f40d..fc1056f5c96a 100644
--- a/fs/cachefiles/namei.c
+++ b/fs/cachefiles/namei.c
@@ -97,7 +97,8 @@ static noinline void cachefiles_printk_object(struct cachefiles_object *object,
  *   call vfs_unlink(), vfs_rmdir() or vfs_rename()
  */
 static void cachefiles_mark_object_buried(struct cachefiles_cache *cache,
-					  struct dentry *dentry)
+					  struct dentry *dentry,
+					  enum fscache_why_object_killed why)
 {
 	struct cachefiles_object *object;
 	struct rb_node *p;
@@ -132,8 +133,9 @@ found_dentry:
 		pr_err("\n");
 		pr_err("Error: Can't preemptively bury live object\n");
 		cachefiles_printk_object(object, NULL);
-	} else if (test_and_set_bit(CACHEFILES_OBJECT_BURIED, &object->flags)) {
-		pr_err("Error: Object already preemptively buried\n");
+	} else {
+		if (why != FSCACHE_OBJECT_IS_STALE)
+			fscache_object_mark_killed(&object->fscache, why);
 	}
 
 	write_unlock(&cache->active_lock);
@@ -265,7 +267,8 @@ requeue:
 static int cachefiles_bury_object(struct cachefiles_cache *cache,
 				  struct dentry *dir,
 				  struct dentry *rep,
-				  bool preemptive)
+				  bool preemptive,
+				  enum fscache_why_object_killed why)
 {
 	struct dentry *grave, *trap;
 	struct path path, path_to_graveyard;
@@ -289,7 +292,7 @@ static int cachefiles_bury_object(struct cachefiles_cache *cache,
 			ret = vfs_unlink(d_inode(dir), rep, NULL);
 
 			if (preemptive)
-				cachefiles_mark_object_buried(cache, rep);
+				cachefiles_mark_object_buried(cache, rep, why);
 		}
 
 		mutex_unlock(&d_inode(dir)->i_mutex);
@@ -394,7 +397,7 @@ try_again:
 					    "Rename failed with error %d", ret);
 
 		if (preemptive)
-			cachefiles_mark_object_buried(cache, rep);
+			cachefiles_mark_object_buried(cache, rep, why);
 	}
 
 	unlock_rename(cache->graveyard, dir);
@@ -422,7 +425,7 @@ int cachefiles_delete_object(struct cachefiles_cache *cache,
 
 	mutex_lock_nested(&d_inode(dir)->i_mutex, I_MUTEX_PARENT);
 
-	if (test_bit(CACHEFILES_OBJECT_BURIED, &object->flags)) {
+	if (test_bit(FSCACHE_OBJECT_KILLED_BY_CACHE, &object->fscache.flags)) {
 		/* object allocation for the same key preemptively deleted this
 		 * object's file so that it could create its own file */
 		_debug("object preemptively buried");
@@ -433,7 +436,8 @@ int cachefiles_delete_object(struct cachefiles_cache *cache,
 		 * may have been renamed */
 		if (dir == object->dentry->d_parent) {
 			ret = cachefiles_bury_object(cache, dir,
-						     object->dentry, false);
+						     object->dentry, false,
+						     FSCACHE_OBJECT_WAS_RETIRED);
 		} else {
 			/* it got moved, presumably by cachefilesd culling it,
 			 * so it's no longer in the key path and we can ignore
@@ -522,7 +526,7 @@ lookup_again:
 		if (d_is_negative(next)) {
 			ret = cachefiles_has_space(cache, 1, 0);
 			if (ret < 0)
-				goto create_error;
+				goto no_space_error;
 
 			path.dentry = dir;
 			ret = security_path_mkdir(&path, next, 0);
@@ -551,7 +555,7 @@ lookup_again:
 		if (d_is_negative(next)) {
 			ret = cachefiles_has_space(cache, 1, 0);
 			if (ret < 0)
-				goto create_error;
+				goto no_space_error;
 
 			path.dentry = dir;
 			ret = security_path_mknod(&path, next, S_IFREG, 0);
@@ -602,7 +606,8 @@ lookup_again:
 			 * mutex) */
 			object->dentry = NULL;
 
-			ret = cachefiles_bury_object(cache, dir, next, true);
+			ret = cachefiles_bury_object(cache, dir, next, true,
+						     FSCACHE_OBJECT_IS_STALE);
 			dput(next);
 			next = NULL;
 
@@ -610,6 +615,7 @@ lookup_again:
 				goto delete_error;
 
 			_debug("redo lookup");
+			fscache_object_retrying_stale(&object->fscache);
 			goto lookup_again;
 		}
 	}
@@ -662,6 +668,8 @@ lookup_again:
 	_leave(" = 0 [%lu]", d_backing_inode(object->dentry)->i_ino);
 	return 0;
 
+no_space_error:
+	fscache_object_mark_killed(&object->fscache, FSCACHE_OBJECT_NO_SPACE);
 create_error:
 	_debug("create error %d", ret);
 	if (ret == -EIO)
@@ -927,7 +935,8 @@ int cachefiles_cull(struct cachefiles_cache *cache, struct dentry *dir,
 	/*  actually remove the victim (drops the dir mutex) */
 	_debug("bury");
 
-	ret = cachefiles_bury_object(cache, dir, victim, false);
+	ret = cachefiles_bury_object(cache, dir, victim, false,
+				     FSCACHE_OBJECT_WAS_CULLED);
 	if (ret < 0)
 		goto error;
 
diff --git a/fs/ceph/acl.c b/fs/ceph/acl.c
index 64fa248343f6..8f84646f10e9 100644
--- a/fs/ceph/acl.c
+++ b/fs/ceph/acl.c
@@ -187,10 +187,10 @@ int ceph_pre_init_acls(struct inode *dir, umode_t *mode,
 		val_size2 = posix_acl_xattr_size(default_acl->a_count);
 
 	err = -ENOMEM;
-	tmp_buf = kmalloc(max(val_size1, val_size2), GFP_NOFS);
+	tmp_buf = kmalloc(max(val_size1, val_size2), GFP_KERNEL);
 	if (!tmp_buf)
 		goto out_err;
-	pagelist = kmalloc(sizeof(struct ceph_pagelist), GFP_NOFS);
+	pagelist = kmalloc(sizeof(struct ceph_pagelist), GFP_KERNEL);
 	if (!pagelist)
 		goto out_err;
 	ceph_pagelist_init(pagelist);
diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index e162bcd105ee..890c50971a69 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -87,17 +87,21 @@ static int ceph_set_page_dirty(struct page *page)
 	inode = mapping->host;
 	ci = ceph_inode(inode);
 
-	/*
-	 * Note that we're grabbing a snapc ref here without holding
-	 * any locks!
-	 */
-	snapc = ceph_get_snap_context(ci->i_snap_realm->cached_context);
-
 	/* dirty the head */
 	spin_lock(&ci->i_ceph_lock);
-	if (ci->i_head_snapc == NULL)
-		ci->i_head_snapc = ceph_get_snap_context(snapc);
-	++ci->i_wrbuffer_ref_head;
+	BUG_ON(ci->i_wr_ref == 0); // caller should hold Fw reference
+	if (__ceph_have_pending_cap_snap(ci)) {
+		struct ceph_cap_snap *capsnap =
+				list_last_entry(&ci->i_cap_snaps,
+						struct ceph_cap_snap,
+						ci_item);
+		snapc = ceph_get_snap_context(capsnap->context);
+		capsnap->dirty_pages++;
+	} else {
+		BUG_ON(!ci->i_head_snapc);
+		snapc = ceph_get_snap_context(ci->i_head_snapc);
+		++ci->i_wrbuffer_ref_head;
+	}
 	if (ci->i_wrbuffer_ref == 0)
 		ihold(inode);
 	++ci->i_wrbuffer_ref;
@@ -346,7 +350,7 @@ static int start_read(struct inode *inode, struct list_head *page_list, int max)
 
 	/* build page vector */
 	nr_pages = calc_pages_for(0, len);
-	pages = kmalloc(sizeof(*pages) * nr_pages, GFP_NOFS);
+	pages = kmalloc(sizeof(*pages) * nr_pages, GFP_KERNEL);
 	ret = -ENOMEM;
 	if (!pages)
 		goto out;
@@ -358,7 +362,7 @@ static int start_read(struct inode *inode, struct list_head *page_list, int max)
  		dout("start_read %p adding %p idx %lu\n", inode, page,
 		     page->index);
 		if (add_to_page_cache_lru(page, &inode->i_data, page->index,
-					  GFP_NOFS)) {
+					  GFP_KERNEL)) {
 			ceph_fscache_uncache_page(inode, page);
 			page_cache_release(page);
 			dout("start_read %p add_to_page_cache failed %p\n",
@@ -436,7 +440,7 @@ out:
  * only snap context we are allowed to write back.
  */
 static struct ceph_snap_context *get_oldest_context(struct inode *inode,
-						    u64 *snap_size)
+						    loff_t *snap_size)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_snap_context *snapc = NULL;
@@ -476,8 +480,9 @@ static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
 	struct ceph_osd_client *osdc;
 	struct ceph_snap_context *snapc, *oldest;
 	loff_t page_off = page_offset(page);
+	loff_t snap_size = -1;
 	long writeback_stat;
-	u64 truncate_size, snap_size = 0;
+	u64 truncate_size;
 	u32 truncate_seq;
 	int err = 0, len = PAGE_CACHE_SIZE;
 
@@ -512,7 +517,7 @@ static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
 	spin_lock(&ci->i_ceph_lock);
 	truncate_seq = ci->i_truncate_seq;
 	truncate_size = ci->i_truncate_size;
-	if (!snap_size)
+	if (snap_size == -1)
 		snap_size = i_size_read(inode);
 	spin_unlock(&ci->i_ceph_lock);
 
@@ -695,7 +700,8 @@ static int ceph_writepages_start(struct address_space *mapping,
 	unsigned wsize = 1 << inode->i_blkbits;
 	struct ceph_osd_request *req = NULL;
 	int do_sync = 0;
-	u64 truncate_size, snap_size;
+	loff_t snap_size, i_size;
+	u64 truncate_size;
 	u32 truncate_seq;
 
 	/*
@@ -741,7 +747,7 @@ static int ceph_writepages_start(struct address_space *mapping,
 retry:
 	/* find oldest snap context with dirty data */
 	ceph_put_snap_context(snapc);
-	snap_size = 0;
+	snap_size = -1;
 	snapc = get_oldest_context(inode, &snap_size);
 	if (!snapc) {
 		/* hmm, why does writepages get called when there
@@ -749,16 +755,13 @@ retry:
 		dout(" no snap context with dirty data?\n");
 		goto out;
 	}
-	if (snap_size == 0)
-		snap_size = i_size_read(inode);
 	dout(" oldest snapc is %p seq %lld (%d snaps)\n",
 	     snapc, snapc->seq, snapc->num_snaps);
 
 	spin_lock(&ci->i_ceph_lock);
 	truncate_seq = ci->i_truncate_seq;
 	truncate_size = ci->i_truncate_size;
-	if (!snap_size)
-		snap_size = i_size_read(inode);
+	i_size = i_size_read(inode);
 	spin_unlock(&ci->i_ceph_lock);
 
 	if (last_snapc && snapc != last_snapc) {
@@ -828,8 +831,10 @@ get_more_pages:
 				dout("waiting on writeback %p\n", page);
 				wait_on_page_writeback(page);
 			}
-			if (page_offset(page) >= snap_size) {
-				dout("%p page eof %llu\n", page, snap_size);
+			if (page_offset(page) >=
+			    (snap_size == -1 ? i_size : snap_size)) {
+				dout("%p page eof %llu\n", page,
+				     (snap_size == -1 ? i_size : snap_size));
 				done = 1;
 				unlock_page(page);
 				break;
@@ -884,7 +889,8 @@ get_more_pages:
 				}
 
 				if (do_sync)
-					osd_req_op_init(req, 1, CEPH_OSD_OP_STARTSYNC);
+					osd_req_op_init(req, 1,
+							CEPH_OSD_OP_STARTSYNC, 0);
 
 				req->r_callback = writepages_finish;
 				req->r_inode = inode;
@@ -944,10 +950,18 @@ get_more_pages:
 		}
 
 		/* Format the osd request message and submit the write */
-
 		offset = page_offset(pages[0]);
-		len = min(snap_size - offset,
-			  (u64)locked_pages << PAGE_CACHE_SHIFT);
+		len = (u64)locked_pages << PAGE_CACHE_SHIFT;
+		if (snap_size == -1) {
+			len = min(len, (u64)i_size_read(inode) - offset);
+			 /* writepages_finish() clears writeback pages
+			  * according to the data length, so make sure
+			  * data length covers all locked pages */
+			len = max(len, 1 +
+				((u64)(locked_pages - 1) << PAGE_CACHE_SHIFT));
+		} else {
+			len = min(len, snap_size - offset);
+		}
 		dout("writepages got %d pages at %llu~%llu\n",
 		     locked_pages, offset, len);
 
@@ -1032,7 +1046,6 @@ static int ceph_update_writeable_page(struct file *file,
 {
 	struct inode *inode = file_inode(file);
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
 	loff_t page_off = pos & PAGE_CACHE_MASK;
 	int pos_in_page = pos & ~PAGE_CACHE_MASK;
 	int end_in_page = pos_in_page + len;
@@ -1044,10 +1057,6 @@ retry_locked:
 	/* writepages currently holds page lock, but if we change that later, */
 	wait_on_page_writeback(page);
 
-	/* check snap context */
-	BUG_ON(!ci->i_snap_realm);
-	down_read(&mdsc->snap_rwsem);
-	BUG_ON(!ci->i_snap_realm->cached_context);
 	snapc = page_snap_context(page);
 	if (snapc && snapc != ci->i_head_snapc) {
 		/*
@@ -1055,7 +1064,6 @@ retry_locked:
 		 * context!  is it writeable now?
 		 */
 		oldest = get_oldest_context(inode, NULL);
-		up_read(&mdsc->snap_rwsem);
 
 		if (snapc->seq > oldest->seq) {
 			ceph_put_snap_context(oldest);
@@ -1112,7 +1120,6 @@ retry_locked:
 	}
 
 	/* we need to read it. */
-	up_read(&mdsc->snap_rwsem);
 	r = readpage_nounlock(file, page);
 	if (r < 0)
 		goto fail_nosnap;
@@ -1157,16 +1164,13 @@ static int ceph_write_begin(struct file *file, struct address_space *mapping,
 
 /*
  * we don't do anything in here that simple_write_end doesn't do
- * except adjust dirty page accounting and drop read lock on
- * mdsc->snap_rwsem.
+ * except adjust dirty page accounting
  */
 static int ceph_write_end(struct file *file, struct address_space *mapping,
 			  loff_t pos, unsigned len, unsigned copied,
 			  struct page *page, void *fsdata)
 {
 	struct inode *inode = file_inode(file);
-	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
-	struct ceph_mds_client *mdsc = fsc->mdsc;
 	unsigned from = pos & (PAGE_CACHE_SIZE - 1);
 	int check_cap = 0;
 
@@ -1188,7 +1192,6 @@ static int ceph_write_end(struct file *file, struct address_space *mapping,
 	set_page_dirty(page);
 
 	unlock_page(page);
-	up_read(&mdsc->snap_rwsem);
 	page_cache_release(page);
 
 	if (check_cap)
@@ -1314,13 +1317,17 @@ static int ceph_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 	struct inode *inode = file_inode(vma->vm_file);
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_file_info *fi = vma->vm_file->private_data;
-	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
+	struct ceph_cap_flush *prealloc_cf;
 	struct page *page = vmf->page;
 	loff_t off = page_offset(page);
 	loff_t size = i_size_read(inode);
 	size_t len;
 	int want, got, ret;
 
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		return VM_FAULT_SIGBUS;
+
 	if (ci->i_inline_version != CEPH_INLINE_NONE) {
 		struct page *locked_page = NULL;
 		if (off == 0) {
@@ -1330,8 +1337,10 @@ static int ceph_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 		ret = ceph_uninline_data(vma->vm_file, locked_page);
 		if (locked_page)
 			unlock_page(locked_page);
-		if (ret < 0)
-			return VM_FAULT_SIGBUS;
+		if (ret < 0) {
+			ret = VM_FAULT_SIGBUS;
+			goto out_free;
+		}
 	}
 
 	if (off + PAGE_CACHE_SIZE <= size)
@@ -1353,7 +1362,8 @@ static int ceph_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 			break;
 		if (ret != -ERESTARTSYS) {
 			WARN_ON(1);
-			return VM_FAULT_SIGBUS;
+			ret = VM_FAULT_SIGBUS;
+			goto out_free;
 		}
 	}
 	dout("page_mkwrite %p %llu~%zd got cap refs on %s\n",
@@ -1373,7 +1383,6 @@ static int ceph_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 	if (ret == 0) {
 		/* success.  we'll keep the page locked. */
 		set_page_dirty(page);
-		up_read(&mdsc->snap_rwsem);
 		ret = VM_FAULT_LOCKED;
 	} else {
 		if (ret == -ENOMEM)
@@ -1389,7 +1398,8 @@ out:
 		int dirty;
 		spin_lock(&ci->i_ceph_lock);
 		ci->i_inline_version = CEPH_INLINE_NONE;
-		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR);
+		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
+					       &prealloc_cf);
 		spin_unlock(&ci->i_ceph_lock);
 		if (dirty)
 			__mark_inode_dirty(inode, dirty);
@@ -1398,6 +1408,8 @@ out:
 	dout("page_mkwrite %p %llu~%zd dropping cap refs on %s ret %d\n",
 	     inode, off, len, ceph_cap_string(got), ret);
 	ceph_put_cap_refs(ci, got);
+out_free:
+	ceph_free_cap_flush(prealloc_cf);
 
 	return ret;
 }
@@ -1509,8 +1521,7 @@ int ceph_uninline_data(struct file *filp, struct page *locked_page)
 				    ceph_vino(inode), 0, &len, 0, 1,
 				    CEPH_OSD_OP_CREATE,
 				    CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE,
-				    ci->i_snap_realm->cached_context,
-				    0, 0, false);
+				    ceph_empty_snapc, 0, 0, false);
 	if (IS_ERR(req)) {
 		err = PTR_ERR(req);
 		goto out;
@@ -1528,7 +1539,7 @@ int ceph_uninline_data(struct file *filp, struct page *locked_page)
 				    ceph_vino(inode), 0, &len, 1, 3,
 				    CEPH_OSD_OP_WRITE,
 				    CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE,
-				    ci->i_snap_realm->cached_context,
+				    ceph_empty_snapc,
 				    ci->i_truncate_seq, ci->i_truncate_size,
 				    false);
 	if (IS_ERR(req)) {
@@ -1597,3 +1608,206 @@ int ceph_mmap(struct file *file, struct vm_area_struct *vma)
 	vma->vm_ops = &ceph_vmops;
 	return 0;
 }
+
+enum {
+	POOL_READ	= 1,
+	POOL_WRITE	= 2,
+};
+
+static int __ceph_pool_perm_get(struct ceph_inode_info *ci, u32 pool)
+{
+	struct ceph_fs_client *fsc = ceph_inode_to_client(&ci->vfs_inode);
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_osd_request *rd_req = NULL, *wr_req = NULL;
+	struct rb_node **p, *parent;
+	struct ceph_pool_perm *perm;
+	struct page **pages;
+	int err = 0, err2 = 0, have = 0;
+
+	down_read(&mdsc->pool_perm_rwsem);
+	p = &mdsc->pool_perm_tree.rb_node;
+	while (*p) {
+		perm = rb_entry(*p, struct ceph_pool_perm, node);
+		if (pool < perm->pool)
+			p = &(*p)->rb_left;
+		else if (pool > perm->pool)
+			p = &(*p)->rb_right;
+		else {
+			have = perm->perm;
+			break;
+		}
+	}
+	up_read(&mdsc->pool_perm_rwsem);
+	if (*p)
+		goto out;
+
+	dout("__ceph_pool_perm_get pool %u no perm cached\n", pool);
+
+	down_write(&mdsc->pool_perm_rwsem);
+	parent = NULL;
+	while (*p) {
+		parent = *p;
+		perm = rb_entry(parent, struct ceph_pool_perm, node);
+		if (pool < perm->pool)
+			p = &(*p)->rb_left;
+		else if (pool > perm->pool)
+			p = &(*p)->rb_right;
+		else {
+			have = perm->perm;
+			break;
+		}
+	}
+	if (*p) {
+		up_write(&mdsc->pool_perm_rwsem);
+		goto out;
+	}
+
+	rd_req = ceph_osdc_alloc_request(&fsc->client->osdc,
+					 ceph_empty_snapc,
+					 1, false, GFP_NOFS);
+	if (!rd_req) {
+		err = -ENOMEM;
+		goto out_unlock;
+	}
+
+	rd_req->r_flags = CEPH_OSD_FLAG_READ;
+	osd_req_op_init(rd_req, 0, CEPH_OSD_OP_STAT, 0);
+	rd_req->r_base_oloc.pool = pool;
+	snprintf(rd_req->r_base_oid.name, sizeof(rd_req->r_base_oid.name),
+		 "%llx.00000000", ci->i_vino.ino);
+	rd_req->r_base_oid.name_len = strlen(rd_req->r_base_oid.name);
+
+	wr_req = ceph_osdc_alloc_request(&fsc->client->osdc,
+					 ceph_empty_snapc,
+					 1, false, GFP_NOFS);
+	if (!wr_req) {
+		err = -ENOMEM;
+		goto out_unlock;
+	}
+
+	wr_req->r_flags = CEPH_OSD_FLAG_WRITE |
+			  CEPH_OSD_FLAG_ACK | CEPH_OSD_FLAG_ONDISK;
+	osd_req_op_init(wr_req, 0, CEPH_OSD_OP_CREATE, CEPH_OSD_OP_FLAG_EXCL);
+	wr_req->r_base_oloc.pool = pool;
+	wr_req->r_base_oid = rd_req->r_base_oid;
+
+	/* one page should be large enough for STAT data */
+	pages = ceph_alloc_page_vector(1, GFP_KERNEL);
+	if (IS_ERR(pages)) {
+		err = PTR_ERR(pages);
+		goto out_unlock;
+	}
+
+	osd_req_op_raw_data_in_pages(rd_req, 0, pages, PAGE_SIZE,
+				     0, false, true);
+	ceph_osdc_build_request(rd_req, 0, NULL, CEPH_NOSNAP,
+				&ci->vfs_inode.i_mtime);
+	err = ceph_osdc_start_request(&fsc->client->osdc, rd_req, false);
+
+	ceph_osdc_build_request(wr_req, 0, NULL, CEPH_NOSNAP,
+				&ci->vfs_inode.i_mtime);
+	err2 = ceph_osdc_start_request(&fsc->client->osdc, wr_req, false);
+
+	if (!err)
+		err = ceph_osdc_wait_request(&fsc->client->osdc, rd_req);
+	if (!err2)
+		err2 = ceph_osdc_wait_request(&fsc->client->osdc, wr_req);
+
+	if (err >= 0 || err == -ENOENT)
+		have |= POOL_READ;
+	else if (err != -EPERM)
+		goto out_unlock;
+
+	if (err2 == 0 || err2 == -EEXIST)
+		have |= POOL_WRITE;
+	else if (err2 != -EPERM) {
+		err = err2;
+		goto out_unlock;
+	}
+
+	perm = kmalloc(sizeof(*perm), GFP_NOFS);
+	if (!perm) {
+		err = -ENOMEM;
+		goto out_unlock;
+	}
+
+	perm->pool = pool;
+	perm->perm = have;
+	rb_link_node(&perm->node, parent, p);
+	rb_insert_color(&perm->node, &mdsc->pool_perm_tree);
+	err = 0;
+out_unlock:
+	up_write(&mdsc->pool_perm_rwsem);
+
+	if (rd_req)
+		ceph_osdc_put_request(rd_req);
+	if (wr_req)
+		ceph_osdc_put_request(wr_req);
+out:
+	if (!err)
+		err = have;
+	dout("__ceph_pool_perm_get pool %u result = %d\n", pool, err);
+	return err;
+}
+
+int ceph_pool_perm_check(struct ceph_inode_info *ci, int need)
+{
+	u32 pool;
+	int ret, flags;
+
+	if (ceph_test_mount_opt(ceph_inode_to_client(&ci->vfs_inode),
+				NOPOOLPERM))
+		return 0;
+
+	spin_lock(&ci->i_ceph_lock);
+	flags = ci->i_ceph_flags;
+	pool = ceph_file_layout_pg_pool(ci->i_layout);
+	spin_unlock(&ci->i_ceph_lock);
+check:
+	if (flags & CEPH_I_POOL_PERM) {
+		if ((need & CEPH_CAP_FILE_RD) && !(flags & CEPH_I_POOL_RD)) {
+			dout("ceph_pool_perm_check pool %u no read perm\n",
+			     pool);
+			return -EPERM;
+		}
+		if ((need & CEPH_CAP_FILE_WR) && !(flags & CEPH_I_POOL_WR)) {
+			dout("ceph_pool_perm_check pool %u no write perm\n",
+			     pool);
+			return -EPERM;
+		}
+		return 0;
+	}
+
+	ret = __ceph_pool_perm_get(ci, pool);
+	if (ret < 0)
+		return ret;
+
+	flags = CEPH_I_POOL_PERM;
+	if (ret & POOL_READ)
+		flags |= CEPH_I_POOL_RD;
+	if (ret & POOL_WRITE)
+		flags |= CEPH_I_POOL_WR;
+
+	spin_lock(&ci->i_ceph_lock);
+	if (pool == ceph_file_layout_pg_pool(ci->i_layout)) {
+		ci->i_ceph_flags = flags;
+        } else {
+		pool = ceph_file_layout_pg_pool(ci->i_layout);
+		flags = ci->i_ceph_flags;
+	}
+	spin_unlock(&ci->i_ceph_lock);
+	goto check;
+}
+
+void ceph_pool_perm_destroy(struct ceph_mds_client *mdsc)
+{
+	struct ceph_pool_perm *perm;
+	struct rb_node *n;
+
+	while (!RB_EMPTY_ROOT(&mdsc->pool_perm_tree)) {
+		n = rb_first(&mdsc->pool_perm_tree);
+		perm = rb_entry(n, struct ceph_pool_perm, node);
+		rb_erase(n, &mdsc->pool_perm_tree);
+		kfree(perm);
+	}
+}
diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index be5ea6af8366..dc10c9dd36c1 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -833,7 +833,9 @@ int __ceph_caps_used(struct ceph_inode_info *ci)
 		used |= CEPH_CAP_PIN;
 	if (ci->i_rd_ref)
 		used |= CEPH_CAP_FILE_RD;
-	if (ci->i_rdcache_ref || ci->vfs_inode.i_data.nrpages)
+	if (ci->i_rdcache_ref ||
+	    (!S_ISDIR(ci->vfs_inode.i_mode) && /* ignore readdir cache */
+	     ci->vfs_inode.i_data.nrpages))
 		used |= CEPH_CAP_FILE_CACHE;
 	if (ci->i_wr_ref)
 		used |= CEPH_CAP_FILE_WR;
@@ -926,16 +928,6 @@ void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release)
 
 	/* remove from session list */
 	spin_lock(&session->s_cap_lock);
-	/*
-	 * s_cap_reconnect is protected by s_cap_lock. no one changes
-	 * s_cap_gen while session is in the reconnect state.
-	 */
-	if (queue_release &&
-	    (!session->s_cap_reconnect ||
-	     cap->cap_gen == session->s_cap_gen))
-		__queue_cap_release(session, ci->i_vino.ino, cap->cap_id,
-				    cap->mseq, cap->issue_seq);
-
 	if (session->s_cap_iterator == cap) {
 		/* not yet, we are iterating over this very cap */
 		dout("__ceph_remove_cap  delaying %p removal from session %p\n",
@@ -948,6 +940,25 @@ void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release)
 	}
 	/* protect backpointer with s_cap_lock: see iterate_session_caps */
 	cap->ci = NULL;
+
+	/*
+	 * s_cap_reconnect is protected by s_cap_lock. no one changes
+	 * s_cap_gen while session is in the reconnect state.
+	 */
+	if (queue_release &&
+	    (!session->s_cap_reconnect || cap->cap_gen == session->s_cap_gen)) {
+		cap->queue_release = 1;
+		if (removed) {
+			list_add_tail(&cap->session_caps,
+				      &session->s_cap_releases);
+			session->s_num_cap_releases++;
+			removed = 0;
+		}
+	} else {
+		cap->queue_release = 0;
+	}
+	cap->cap_ino = ci->i_vino.ino;
+
 	spin_unlock(&session->s_cap_lock);
 
 	/* remove from inode list */
@@ -977,8 +988,8 @@ void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release)
 static int send_cap_msg(struct ceph_mds_session *session,
 			u64 ino, u64 cid, int op,
 			int caps, int wanted, int dirty,
-			u32 seq, u64 flush_tid, u32 issue_seq, u32 mseq,
-			u64 size, u64 max_size,
+			u32 seq, u64 flush_tid, u64 oldest_flush_tid,
+			u32 issue_seq, u32 mseq, u64 size, u64 max_size,
 			struct timespec *mtime, struct timespec *atime,
 			u64 time_warp_seq,
 			kuid_t uid, kgid_t gid, umode_t mode,
@@ -992,20 +1003,23 @@ static int send_cap_msg(struct ceph_mds_session *session,
 	size_t extra_len;
 
 	dout("send_cap_msg %s %llx %llx caps %s wanted %s dirty %s"
-	     " seq %u/%u mseq %u follows %lld size %llu/%llu"
+	     " seq %u/%u tid %llu/%llu mseq %u follows %lld size %llu/%llu"
 	     " xattr_ver %llu xattr_len %d\n", ceph_cap_op_name(op),
 	     cid, ino, ceph_cap_string(caps), ceph_cap_string(wanted),
 	     ceph_cap_string(dirty),
-	     seq, issue_seq, mseq, follows, size, max_size,
+	     seq, issue_seq, flush_tid, oldest_flush_tid,
+	     mseq, follows, size, max_size,
 	     xattr_version, xattrs_buf ? (int)xattrs_buf->vec.iov_len : 0);
 
-	/* flock buffer size + inline version + inline data size */
-	extra_len = 4 + 8 + 4;
+	/* flock buffer size + inline version + inline data size +
+	 * osd_epoch_barrier + oldest_flush_tid */
+	extra_len = 4 + 8 + 4 + 4 + 8;
 	msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPS, sizeof(*fc) + extra_len,
 			   GFP_NOFS, false);
 	if (!msg)
 		return -ENOMEM;
 
+	msg->hdr.version = cpu_to_le16(6);
 	msg->hdr.tid = cpu_to_le64(flush_tid);
 
 	fc = msg->front.iov_base;
@@ -1041,6 +1055,10 @@ static int send_cap_msg(struct ceph_mds_session *session,
 	ceph_encode_64(&p, inline_data ? 0 : CEPH_INLINE_NONE);
 	/* inline data size */
 	ceph_encode_32(&p, 0);
+	/* osd_epoch_barrier */
+	ceph_encode_32(&p, 0);
+	/* oldest_flush_tid */
+	ceph_encode_64(&p, oldest_flush_tid);
 
 	fc->xattr_version = cpu_to_le64(xattr_version);
 	if (xattrs_buf) {
@@ -1053,44 +1071,6 @@ static int send_cap_msg(struct ceph_mds_session *session,
 	return 0;
 }
 
-void __queue_cap_release(struct ceph_mds_session *session,
-			 u64 ino, u64 cap_id, u32 migrate_seq,
-			 u32 issue_seq)
-{
-	struct ceph_msg *msg;
-	struct ceph_mds_cap_release *head;
-	struct ceph_mds_cap_item *item;
-
-	BUG_ON(!session->s_num_cap_releases);
-	msg = list_first_entry(&session->s_cap_releases,
-			       struct ceph_msg, list_head);
-
-	dout(" adding %llx release to mds%d msg %p (%d left)\n",
-	     ino, session->s_mds, msg, session->s_num_cap_releases);
-
-	BUG_ON(msg->front.iov_len + sizeof(*item) > PAGE_CACHE_SIZE);
-	head = msg->front.iov_base;
-	le32_add_cpu(&head->num, 1);
-	item = msg->front.iov_base + msg->front.iov_len;
-	item->ino = cpu_to_le64(ino);
-	item->cap_id = cpu_to_le64(cap_id);
-	item->migrate_seq = cpu_to_le32(migrate_seq);
-	item->seq = cpu_to_le32(issue_seq);
-
-	session->s_num_cap_releases--;
-
-	msg->front.iov_len += sizeof(*item);
-	if (le32_to_cpu(head->num) == CEPH_CAPS_PER_RELEASE) {
-		dout(" release msg %p full\n", msg);
-		list_move_tail(&msg->list_head, &session->s_cap_releases_done);
-	} else {
-		dout(" release msg %p at %d/%d (%d)\n", msg,
-		     (int)le32_to_cpu(head->num),
-		     (int)CEPH_CAPS_PER_RELEASE,
-		     (int)msg->front.iov_len);
-	}
-}
-
 /*
  * Queue cap releases when an inode is dropped from our cache.  Since
  * inode is about to be destroyed, there is no need for i_ceph_lock.
@@ -1127,7 +1107,7 @@ void ceph_queue_caps_release(struct inode *inode)
  */
 static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap,
 		      int op, int used, int want, int retain, int flushing,
-		      unsigned *pflush_tid)
+		      u64 flush_tid, u64 oldest_flush_tid)
 	__releases(cap->ci->i_ceph_lock)
 {
 	struct ceph_inode_info *ci = cap->ci;
@@ -1145,8 +1125,6 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap,
 	u64 xattr_version = 0;
 	struct ceph_buffer *xattr_blob = NULL;
 	int delayed = 0;
-	u64 flush_tid = 0;
-	int i;
 	int ret;
 	bool inline_data;
 
@@ -1190,26 +1168,7 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap,
 	cap->implemented &= cap->issued | used;
 	cap->mds_wanted = want;
 
-	if (flushing) {
-		/*
-		 * assign a tid for flush operations so we can avoid
-		 * flush1 -> dirty1 -> flush2 -> flushack1 -> mark
-		 * clean type races.  track latest tid for every bit
-		 * so we can handle flush AxFw, flush Fw, and have the
-		 * first ack clean Ax.
-		 */
-		flush_tid = ++ci->i_cap_flush_last_tid;
-		if (pflush_tid)
-			*pflush_tid = flush_tid;
-		dout(" cap_flush_tid %d\n", (int)flush_tid);
-		for (i = 0; i < CEPH_CAP_BITS; i++)
-			if (flushing & (1 << i))
-				ci->i_cap_flush_tid[i] = flush_tid;
-
-		follows = ci->i_head_snapc->seq;
-	} else {
-		follows = 0;
-	}
+	follows = flushing ? ci->i_head_snapc->seq : 0;
 
 	keep = cap->implemented;
 	seq = cap->seq;
@@ -1237,7 +1196,8 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap,
 	spin_unlock(&ci->i_ceph_lock);
 
 	ret = send_cap_msg(session, ceph_vino(inode).ino, cap_id,
-		op, keep, want, flushing, seq, flush_tid, issue_seq, mseq,
+		op, keep, want, flushing, seq,
+		flush_tid, oldest_flush_tid, issue_seq, mseq,
 		size, max_size, &mtime, &atime, time_warp_seq,
 		uid, gid, mode, xattr_version, xattr_blob,
 		follows, inline_data);
@@ -1259,14 +1219,14 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap,
  * asynchronously back to the MDS once sync writes complete and dirty
  * data is written out.
  *
- * Unless @again is true, skip cap_snaps that were already sent to
+ * Unless @kick is true, skip cap_snaps that were already sent to
  * the MDS (i.e., during this session).
  *
  * Called under i_ceph_lock.  Takes s_mutex as needed.
  */
 void __ceph_flush_snaps(struct ceph_inode_info *ci,
 			struct ceph_mds_session **psession,
-			int again)
+			int kick)
 		__releases(ci->i_ceph_lock)
 		__acquires(ci->i_ceph_lock)
 {
@@ -1297,11 +1257,8 @@ retry:
 		if (capsnap->dirty_pages || capsnap->writing)
 			break;
 
-		/*
-		 * if cap writeback already occurred, we should have dropped
-		 * the capsnap in ceph_put_wrbuffer_cap_refs.
-		 */
-		BUG_ON(capsnap->dirty == 0);
+		/* should be removed by ceph_try_drop_cap_snap() */
+		BUG_ON(!capsnap->need_flush);
 
 		/* pick mds, take s_mutex */
 		if (ci->i_auth_cap == NULL) {
@@ -1310,7 +1267,7 @@ retry:
 		}
 
 		/* only flush each capsnap once */
-		if (!again && !list_empty(&capsnap->flushing_item)) {
+		if (!kick && !list_empty(&capsnap->flushing_item)) {
 			dout("already flushed %p, skipping\n", capsnap);
 			continue;
 		}
@@ -1320,6 +1277,9 @@ retry:
 
 		if (session && session->s_mds != mds) {
 			dout("oops, wrong session %p mutex\n", session);
+			if (kick)
+				goto out;
+
 			mutex_unlock(&session->s_mutex);
 			ceph_put_mds_session(session);
 			session = NULL;
@@ -1343,20 +1303,22 @@ retry:
 			goto retry;
 		}
 
-		capsnap->flush_tid = ++ci->i_cap_flush_last_tid;
+		spin_lock(&mdsc->cap_dirty_lock);
+		capsnap->flush_tid = ++mdsc->last_cap_flush_tid;
+		spin_unlock(&mdsc->cap_dirty_lock);
+
 		atomic_inc(&capsnap->nref);
-		if (!list_empty(&capsnap->flushing_item))
-			list_del_init(&capsnap->flushing_item);
-		list_add_tail(&capsnap->flushing_item,
-			      &session->s_cap_snaps_flushing);
+		if (list_empty(&capsnap->flushing_item))
+			list_add_tail(&capsnap->flushing_item,
+				      &session->s_cap_snaps_flushing);
 		spin_unlock(&ci->i_ceph_lock);
 
 		dout("flush_snaps %p cap_snap %p follows %lld tid %llu\n",
 		     inode, capsnap, capsnap->follows, capsnap->flush_tid);
 		send_cap_msg(session, ceph_vino(inode).ino, 0,
 			     CEPH_CAP_OP_FLUSHSNAP, capsnap->issued, 0,
-			     capsnap->dirty, 0, capsnap->flush_tid, 0, mseq,
-			     capsnap->size, 0,
+			     capsnap->dirty, 0, capsnap->flush_tid, 0,
+			     0, mseq, capsnap->size, 0,
 			     &capsnap->mtime, &capsnap->atime,
 			     capsnap->time_warp_seq,
 			     capsnap->uid, capsnap->gid, capsnap->mode,
@@ -1396,7 +1358,8 @@ static void ceph_flush_snaps(struct ceph_inode_info *ci)
  * Caller is then responsible for calling __mark_inode_dirty with the
  * returned flags value.
  */
-int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask)
+int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
+			   struct ceph_cap_flush **pcf)
 {
 	struct ceph_mds_client *mdsc =
 		ceph_sb_to_client(ci->vfs_inode.i_sb)->mdsc;
@@ -1416,9 +1379,14 @@ int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask)
 	     ceph_cap_string(was | mask));
 	ci->i_dirty_caps |= mask;
 	if (was == 0) {
-		if (!ci->i_head_snapc)
+		WARN_ON_ONCE(ci->i_prealloc_cap_flush);
+		swap(ci->i_prealloc_cap_flush, *pcf);
+
+		if (!ci->i_head_snapc) {
+			WARN_ON_ONCE(!rwsem_is_locked(&mdsc->snap_rwsem));
 			ci->i_head_snapc = ceph_get_snap_context(
 				ci->i_snap_realm->cached_context);
+		}
 		dout(" inode %p now dirty snapc %p auth cap %p\n",
 		     &ci->vfs_inode, ci->i_head_snapc, ci->i_auth_cap);
 		BUG_ON(!list_empty(&ci->i_dirty_item));
@@ -1429,6 +1397,8 @@ int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask)
 			ihold(inode);
 			dirty |= I_DIRTY_SYNC;
 		}
+	} else {
+		WARN_ON_ONCE(!ci->i_prealloc_cap_flush);
 	}
 	BUG_ON(list_empty(&ci->i_dirty_item));
 	if (((was | ci->i_flushing_caps) & CEPH_CAP_FILE_BUFFER) &&
@@ -1438,6 +1408,74 @@ int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask)
 	return dirty;
 }
 
+static void __add_cap_flushing_to_inode(struct ceph_inode_info *ci,
+					struct ceph_cap_flush *cf)
+{
+	struct rb_node **p = &ci->i_cap_flush_tree.rb_node;
+	struct rb_node *parent = NULL;
+	struct ceph_cap_flush *other = NULL;
+
+	while (*p) {
+		parent = *p;
+		other = rb_entry(parent, struct ceph_cap_flush, i_node);
+
+		if (cf->tid < other->tid)
+			p = &(*p)->rb_left;
+		else if (cf->tid > other->tid)
+			p = &(*p)->rb_right;
+		else
+			BUG();
+	}
+
+	rb_link_node(&cf->i_node, parent, p);
+	rb_insert_color(&cf->i_node, &ci->i_cap_flush_tree);
+}
+
+static void __add_cap_flushing_to_mdsc(struct ceph_mds_client *mdsc,
+				       struct ceph_cap_flush *cf)
+{
+	struct rb_node **p = &mdsc->cap_flush_tree.rb_node;
+	struct rb_node *parent = NULL;
+	struct ceph_cap_flush *other = NULL;
+
+	while (*p) {
+		parent = *p;
+		other = rb_entry(parent, struct ceph_cap_flush, g_node);
+
+		if (cf->tid < other->tid)
+			p = &(*p)->rb_left;
+		else if (cf->tid > other->tid)
+			p = &(*p)->rb_right;
+		else
+			BUG();
+	}
+
+	rb_link_node(&cf->g_node, parent, p);
+	rb_insert_color(&cf->g_node, &mdsc->cap_flush_tree);
+}
+
+struct ceph_cap_flush *ceph_alloc_cap_flush(void)
+{
+	return kmem_cache_alloc(ceph_cap_flush_cachep, GFP_KERNEL);
+}
+
+void ceph_free_cap_flush(struct ceph_cap_flush *cf)
+{
+	if (cf)
+		kmem_cache_free(ceph_cap_flush_cachep, cf);
+}
+
+static u64 __get_oldest_flush_tid(struct ceph_mds_client *mdsc)
+{
+	struct rb_node *n = rb_first(&mdsc->cap_flush_tree);
+	if (n) {
+		struct ceph_cap_flush *cf =
+			rb_entry(n, struct ceph_cap_flush, g_node);
+		return cf->tid;
+	}
+	return 0;
+}
+
 /*
  * Add dirty inode to the flushing list.  Assigned a seq number so we
  * can wait for caps to flush without starving.
@@ -1445,14 +1483,17 @@ int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask)
  * Called under i_ceph_lock.
  */
 static int __mark_caps_flushing(struct inode *inode,
-				 struct ceph_mds_session *session)
+				struct ceph_mds_session *session,
+				u64 *flush_tid, u64 *oldest_flush_tid)
 {
 	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
 	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_cap_flush *cf = NULL;
 	int flushing;
 
 	BUG_ON(ci->i_dirty_caps == 0);
 	BUG_ON(list_empty(&ci->i_dirty_item));
+	BUG_ON(!ci->i_prealloc_cap_flush);
 
 	flushing = ci->i_dirty_caps;
 	dout("__mark_caps_flushing flushing %s, flushing_caps %s -> %s\n",
@@ -1463,22 +1504,31 @@ static int __mark_caps_flushing(struct inode *inode,
 	ci->i_dirty_caps = 0;
 	dout(" inode %p now !dirty\n", inode);
 
+	swap(cf, ci->i_prealloc_cap_flush);
+	cf->caps = flushing;
+	cf->kick = false;
+
 	spin_lock(&mdsc->cap_dirty_lock);
 	list_del_init(&ci->i_dirty_item);
 
+	cf->tid = ++mdsc->last_cap_flush_tid;
+	__add_cap_flushing_to_mdsc(mdsc, cf);
+	*oldest_flush_tid = __get_oldest_flush_tid(mdsc);
+
 	if (list_empty(&ci->i_flushing_item)) {
-		ci->i_cap_flush_seq = ++mdsc->cap_flush_seq;
 		list_add_tail(&ci->i_flushing_item, &session->s_cap_flushing);
 		mdsc->num_cap_flushing++;
-		dout(" inode %p now flushing seq %lld\n", inode,
-		     ci->i_cap_flush_seq);
+		dout(" inode %p now flushing tid %llu\n", inode, cf->tid);
 	} else {
 		list_move_tail(&ci->i_flushing_item, &session->s_cap_flushing);
-		dout(" inode %p now flushing (more) seq %lld\n", inode,
-		     ci->i_cap_flush_seq);
+		dout(" inode %p now flushing (more) tid %llu\n",
+		     inode, cf->tid);
 	}
 	spin_unlock(&mdsc->cap_dirty_lock);
 
+	__add_cap_flushing_to_inode(ci, cf);
+
+	*flush_tid = cf->tid;
 	return flushing;
 }
 
@@ -1524,6 +1574,7 @@ void ceph_check_caps(struct ceph_inode_info *ci, int flags,
 	struct ceph_mds_client *mdsc = fsc->mdsc;
 	struct inode *inode = &ci->vfs_inode;
 	struct ceph_cap *cap;
+	u64 flush_tid, oldest_flush_tid;
 	int file_wanted, used, cap_used;
 	int took_snap_rwsem = 0;             /* true if mdsc->snap_rwsem held */
 	int issued, implemented, want, retain, revoking, flushing = 0;
@@ -1553,13 +1604,13 @@ retry:
 retry_locked:
 	file_wanted = __ceph_caps_file_wanted(ci);
 	used = __ceph_caps_used(ci);
-	want = file_wanted | used;
 	issued = __ceph_caps_issued(ci, &implemented);
 	revoking = implemented & ~issued;
 
-	retain = want | CEPH_CAP_PIN;
+	want = file_wanted;
+	retain = file_wanted | used | CEPH_CAP_PIN;
 	if (!mdsc->stopping && inode->i_nlink > 0) {
-		if (want) {
+		if (file_wanted) {
 			retain |= CEPH_CAP_ANY;       /* be greedy */
 		} else if (S_ISDIR(inode->i_mode) &&
 			   (issued & CEPH_CAP_FILE_SHARED) &&
@@ -1602,9 +1653,10 @@ retry_locked:
 	 * If we fail, it's because pages are locked.... try again later.
 	 */
 	if ((!is_delayed || mdsc->stopping) &&
-	    ci->i_wrbuffer_ref == 0 &&               /* no dirty pages... */
-	    inode->i_data.nrpages &&                 /* have cached pages */
-	    (file_wanted == 0 ||                     /* no open files */
+	    !S_ISDIR(inode->i_mode) &&		/* ignore readdir cache */
+	    ci->i_wrbuffer_ref == 0 &&		/* no dirty pages... */
+	    inode->i_data.nrpages &&		/* have cached pages */
+	    (file_wanted == 0 ||		/* no open files */
 	     (revoking & (CEPH_CAP_FILE_CACHE|
 			  CEPH_CAP_FILE_LAZYIO))) && /*  or revoking cache */
 	    !tried_invalidate) {
@@ -1742,17 +1794,25 @@ ack:
 			took_snap_rwsem = 1;
 		}
 
-		if (cap == ci->i_auth_cap && ci->i_dirty_caps)
-			flushing = __mark_caps_flushing(inode, session);
-		else
+		if (cap == ci->i_auth_cap && ci->i_dirty_caps) {
+			flushing = __mark_caps_flushing(inode, session,
+							&flush_tid,
+							&oldest_flush_tid);
+		} else {
 			flushing = 0;
+			flush_tid = 0;
+			spin_lock(&mdsc->cap_dirty_lock);
+			oldest_flush_tid = __get_oldest_flush_tid(mdsc);
+			spin_unlock(&mdsc->cap_dirty_lock);
+		}
 
 		mds = cap->mds;  /* remember mds, so we don't repeat */
 		sent++;
 
 		/* __send_cap drops i_ceph_lock */
 		delayed += __send_cap(mdsc, cap, CEPH_CAP_OP_UPDATE, cap_used,
-				      want, retain, flushing, NULL);
+				      want, retain, flushing,
+				      flush_tid, oldest_flush_tid);
 		goto retry; /* retake i_ceph_lock and restart our cap scan. */
 	}
 
@@ -1781,12 +1841,13 @@ ack:
 /*
  * Try to flush dirty caps back to the auth mds.
  */
-static int try_flush_caps(struct inode *inode, unsigned *flush_tid)
+static int try_flush_caps(struct inode *inode, u64 *ptid)
 {
 	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	int flushing = 0;
 	struct ceph_mds_session *session = NULL;
+	int flushing = 0;
+	u64 flush_tid = 0, oldest_flush_tid = 0;
 
 retry:
 	spin_lock(&ci->i_ceph_lock);
@@ -1811,42 +1872,54 @@ retry:
 		if (cap->session->s_state < CEPH_MDS_SESSION_OPEN)
 			goto out;
 
-		flushing = __mark_caps_flushing(inode, session);
+		flushing = __mark_caps_flushing(inode, session, &flush_tid,
+						&oldest_flush_tid);
 
 		/* __send_cap drops i_ceph_lock */
 		delayed = __send_cap(mdsc, cap, CEPH_CAP_OP_FLUSH, used, want,
-				     cap->issued | cap->implemented, flushing,
-				     flush_tid);
-		if (!delayed)
-			goto out_unlocked;
+				     (cap->issued | cap->implemented),
+				     flushing, flush_tid, oldest_flush_tid);
 
-		spin_lock(&ci->i_ceph_lock);
-		__cap_delay_requeue(mdsc, ci);
+		if (delayed) {
+			spin_lock(&ci->i_ceph_lock);
+			__cap_delay_requeue(mdsc, ci);
+			spin_unlock(&ci->i_ceph_lock);
+		}
+	} else {
+		struct rb_node *n = rb_last(&ci->i_cap_flush_tree);
+		if (n) {
+			struct ceph_cap_flush *cf =
+				rb_entry(n, struct ceph_cap_flush, i_node);
+			flush_tid = cf->tid;
+		}
+		flushing = ci->i_flushing_caps;
+		spin_unlock(&ci->i_ceph_lock);
 	}
 out:
-	spin_unlock(&ci->i_ceph_lock);
-out_unlocked:
 	if (session)
 		mutex_unlock(&session->s_mutex);
+
+	*ptid = flush_tid;
 	return flushing;
 }
 
 /*
  * Return true if we've flushed caps through the given flush_tid.
  */
-static int caps_are_flushed(struct inode *inode, unsigned tid)
+static int caps_are_flushed(struct inode *inode, u64 flush_tid)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	int i, ret = 1;
+	struct ceph_cap_flush *cf;
+	struct rb_node *n;
+	int ret = 1;
 
 	spin_lock(&ci->i_ceph_lock);
-	for (i = 0; i < CEPH_CAP_BITS; i++)
-		if ((ci->i_flushing_caps & (1 << i)) &&
-		    ci->i_cap_flush_tid[i] <= tid) {
-			/* still flushing this bit */
+	n = rb_first(&ci->i_cap_flush_tree);
+	if (n) {
+		cf = rb_entry(n, struct ceph_cap_flush, i_node);
+		if (cf->tid <= flush_tid)
 			ret = 0;
-			break;
-		}
+	}
 	spin_unlock(&ci->i_ceph_lock);
 	return ret;
 }
@@ -1864,13 +1937,16 @@ static void sync_write_wait(struct inode *inode)
 	struct ceph_osd_request *req;
 	u64 last_tid;
 
+	if (!S_ISREG(inode->i_mode))
+		return;
+
 	spin_lock(&ci->i_unsafe_lock);
 	if (list_empty(head))
 		goto out;
 
 	/* set upper bound as _last_ entry in chain */
-	req = list_entry(head->prev, struct ceph_osd_request,
-			 r_unsafe_item);
+	req = list_last_entry(head, struct ceph_osd_request,
+			      r_unsafe_item);
 	last_tid = req->r_tid;
 
 	do {
@@ -1888,18 +1964,64 @@ static void sync_write_wait(struct inode *inode)
 		 */
 		if (list_empty(head))
 			break;
-		req = list_entry(head->next, struct ceph_osd_request,
-				 r_unsafe_item);
+		req = list_first_entry(head, struct ceph_osd_request,
+				       r_unsafe_item);
 	} while (req->r_tid < last_tid);
 out:
 	spin_unlock(&ci->i_unsafe_lock);
 }
 
+/*
+ * wait for any uncommitted directory operations to commit.
+ */
+static int unsafe_dirop_wait(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct list_head *head = &ci->i_unsafe_dirops;
+	struct ceph_mds_request *req;
+	u64 last_tid;
+	int ret = 0;
+
+	if (!S_ISDIR(inode->i_mode))
+		return 0;
+
+	spin_lock(&ci->i_unsafe_lock);
+	if (list_empty(head))
+		goto out;
+
+	req = list_last_entry(head, struct ceph_mds_request,
+			      r_unsafe_dir_item);
+	last_tid = req->r_tid;
+
+	do {
+		ceph_mdsc_get_request(req);
+		spin_unlock(&ci->i_unsafe_lock);
+
+		dout("unsafe_dirop_wait %p wait on tid %llu (until %llu)\n",
+		     inode, req->r_tid, last_tid);
+		ret = !wait_for_completion_timeout(&req->r_safe_completion,
+					ceph_timeout_jiffies(req->r_timeout));
+		if (ret)
+			ret = -EIO;  /* timed out */
+
+		ceph_mdsc_put_request(req);
+
+		spin_lock(&ci->i_unsafe_lock);
+		if (ret || list_empty(head))
+			break;
+		req = list_first_entry(head, struct ceph_mds_request,
+				       r_unsafe_dir_item);
+	} while (req->r_tid < last_tid);
+out:
+	spin_unlock(&ci->i_unsafe_lock);
+	return ret;
+}
+
 int ceph_fsync(struct file *file, loff_t start, loff_t end, int datasync)
 {
 	struct inode *inode = file->f_mapping->host;
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	unsigned flush_tid;
+	u64 flush_tid;
 	int ret;
 	int dirty;
 
@@ -1908,25 +2030,30 @@ int ceph_fsync(struct file *file, loff_t start, loff_t end, int datasync)
 
 	ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
 	if (ret < 0)
-		return ret;
+		goto out;
+
+	if (datasync)
+		goto out;
+
 	mutex_lock(&inode->i_mutex);
 
 	dirty = try_flush_caps(inode, &flush_tid);
 	dout("fsync dirty caps are %s\n", ceph_cap_string(dirty));
 
+	ret = unsafe_dirop_wait(inode);
+
 	/*
 	 * only wait on non-file metadata writeback (the mds
 	 * can recover size and mtime, so we don't need to
 	 * wait for that)
 	 */
-	if (!datasync && (dirty & ~CEPH_CAP_ANY_FILE_WR)) {
-		dout("fsync waiting for flush_tid %u\n", flush_tid);
+	if (!ret && (dirty & ~CEPH_CAP_ANY_FILE_WR)) {
 		ret = wait_event_interruptible(ci->i_cap_wq,
-				       caps_are_flushed(inode, flush_tid));
+					caps_are_flushed(inode, flush_tid));
 	}
-
-	dout("fsync %p%s done\n", inode, datasync ? " datasync" : "");
 	mutex_unlock(&inode->i_mutex);
+out:
+	dout("fsync %p%s result=%d\n", inode, datasync ? " datasync" : "", ret);
 	return ret;
 }
 
@@ -1939,7 +2066,7 @@ int ceph_fsync(struct file *file, loff_t start, loff_t end, int datasync)
 int ceph_write_inode(struct inode *inode, struct writeback_control *wbc)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	unsigned flush_tid;
+	u64 flush_tid;
 	int err = 0;
 	int dirty;
 	int wait = wbc->sync_mode == WB_SYNC_ALL;
@@ -1994,6 +2121,104 @@ static void kick_flushing_capsnaps(struct ceph_mds_client *mdsc,
 	}
 }
 
+static int __kick_flushing_caps(struct ceph_mds_client *mdsc,
+				struct ceph_mds_session *session,
+				struct ceph_inode_info *ci,
+				bool kick_all)
+{
+	struct inode *inode = &ci->vfs_inode;
+	struct ceph_cap *cap;
+	struct ceph_cap_flush *cf;
+	struct rb_node *n;
+	int delayed = 0;
+	u64 first_tid = 0;
+	u64 oldest_flush_tid;
+
+	spin_lock(&mdsc->cap_dirty_lock);
+	oldest_flush_tid = __get_oldest_flush_tid(mdsc);
+	spin_unlock(&mdsc->cap_dirty_lock);
+
+	while (true) {
+		spin_lock(&ci->i_ceph_lock);
+		cap = ci->i_auth_cap;
+		if (!(cap && cap->session == session)) {
+			pr_err("%p auth cap %p not mds%d ???\n", inode,
+					cap, session->s_mds);
+			spin_unlock(&ci->i_ceph_lock);
+			break;
+		}
+
+		for (n = rb_first(&ci->i_cap_flush_tree); n; n = rb_next(n)) {
+			cf = rb_entry(n, struct ceph_cap_flush, i_node);
+			if (cf->tid < first_tid)
+				continue;
+			if (kick_all || cf->kick)
+				break;
+		}
+		if (!n) {
+			spin_unlock(&ci->i_ceph_lock);
+			break;
+		}
+
+		cf = rb_entry(n, struct ceph_cap_flush, i_node);
+		cf->kick = false;
+
+		first_tid = cf->tid + 1;
+
+		dout("kick_flushing_caps %p cap %p tid %llu %s\n", inode,
+		     cap, cf->tid, ceph_cap_string(cf->caps));
+		delayed |= __send_cap(mdsc, cap, CEPH_CAP_OP_FLUSH,
+				      __ceph_caps_used(ci),
+				      __ceph_caps_wanted(ci),
+				      cap->issued | cap->implemented,
+				      cf->caps, cf->tid, oldest_flush_tid);
+	}
+	return delayed;
+}
+
+void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
+				   struct ceph_mds_session *session)
+{
+	struct ceph_inode_info *ci;
+	struct ceph_cap *cap;
+	struct ceph_cap_flush *cf;
+	struct rb_node *n;
+
+	dout("early_kick_flushing_caps mds%d\n", session->s_mds);
+	list_for_each_entry(ci, &session->s_cap_flushing, i_flushing_item) {
+		spin_lock(&ci->i_ceph_lock);
+		cap = ci->i_auth_cap;
+		if (!(cap && cap->session == session)) {
+			pr_err("%p auth cap %p not mds%d ???\n",
+				&ci->vfs_inode, cap, session->s_mds);
+			spin_unlock(&ci->i_ceph_lock);
+			continue;
+		}
+
+
+		/*
+		 * if flushing caps were revoked, we re-send the cap flush
+		 * in client reconnect stage. This guarantees MDS * processes
+		 * the cap flush message before issuing the flushing caps to
+		 * other client.
+		 */
+		if ((cap->issued & ci->i_flushing_caps) !=
+		    ci->i_flushing_caps) {
+			spin_unlock(&ci->i_ceph_lock);
+			if (!__kick_flushing_caps(mdsc, session, ci, true))
+				continue;
+			spin_lock(&ci->i_ceph_lock);
+		}
+
+		for (n = rb_first(&ci->i_cap_flush_tree); n; n = rb_next(n)) {
+			cf = rb_entry(n, struct ceph_cap_flush, i_node);
+			cf->kick = true;
+		}
+
+		spin_unlock(&ci->i_ceph_lock);
+	}
+}
+
 void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
 			     struct ceph_mds_session *session)
 {
@@ -2003,28 +2228,10 @@ void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
 
 	dout("kick_flushing_caps mds%d\n", session->s_mds);
 	list_for_each_entry(ci, &session->s_cap_flushing, i_flushing_item) {
-		struct inode *inode = &ci->vfs_inode;
-		struct ceph_cap *cap;
-		int delayed = 0;
-
-		spin_lock(&ci->i_ceph_lock);
-		cap = ci->i_auth_cap;
-		if (cap && cap->session == session) {
-			dout("kick_flushing_caps %p cap %p %s\n", inode,
-			     cap, ceph_cap_string(ci->i_flushing_caps));
-			delayed = __send_cap(mdsc, cap, CEPH_CAP_OP_FLUSH,
-					     __ceph_caps_used(ci),
-					     __ceph_caps_wanted(ci),
-					     cap->issued | cap->implemented,
-					     ci->i_flushing_caps, NULL);
-			if (delayed) {
-				spin_lock(&ci->i_ceph_lock);
-				__cap_delay_requeue(mdsc, ci);
-				spin_unlock(&ci->i_ceph_lock);
-			}
-		} else {
-			pr_err("%p auth cap %p not mds%d ???\n", inode,
-			       cap, session->s_mds);
+		int delayed = __kick_flushing_caps(mdsc, session, ci, false);
+		if (delayed) {
+			spin_lock(&ci->i_ceph_lock);
+			__cap_delay_requeue(mdsc, ci);
 			spin_unlock(&ci->i_ceph_lock);
 		}
 	}
@@ -2036,26 +2243,25 @@ static void kick_flushing_inode_caps(struct ceph_mds_client *mdsc,
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_cap *cap;
-	int delayed = 0;
 
 	spin_lock(&ci->i_ceph_lock);
 	cap = ci->i_auth_cap;
-	dout("kick_flushing_inode_caps %p flushing %s flush_seq %lld\n", inode,
-	     ceph_cap_string(ci->i_flushing_caps), ci->i_cap_flush_seq);
+	dout("kick_flushing_inode_caps %p flushing %s\n", inode,
+	     ceph_cap_string(ci->i_flushing_caps));
 
 	__ceph_flush_snaps(ci, &session, 1);
 
 	if (ci->i_flushing_caps) {
+		int delayed;
+
 		spin_lock(&mdsc->cap_dirty_lock);
 		list_move_tail(&ci->i_flushing_item,
 			       &cap->session->s_cap_flushing);
 		spin_unlock(&mdsc->cap_dirty_lock);
 
-		delayed = __send_cap(mdsc, cap, CEPH_CAP_OP_FLUSH,
-				     __ceph_caps_used(ci),
-				     __ceph_caps_wanted(ci),
-				     cap->issued | cap->implemented,
-				     ci->i_flushing_caps, NULL);
+		spin_unlock(&ci->i_ceph_lock);
+
+		delayed = __kick_flushing_caps(mdsc, session, ci, true);
 		if (delayed) {
 			spin_lock(&ci->i_ceph_lock);
 			__cap_delay_requeue(mdsc, ci);
@@ -2073,7 +2279,8 @@ static void kick_flushing_inode_caps(struct ceph_mds_client *mdsc,
  *
  * Protected by i_ceph_lock.
  */
-static void __take_cap_refs(struct ceph_inode_info *ci, int got)
+static void __take_cap_refs(struct ceph_inode_info *ci, int got,
+			    bool snap_rwsem_locked)
 {
 	if (got & CEPH_CAP_PIN)
 		ci->i_pin_ref++;
@@ -2081,8 +2288,14 @@ static void __take_cap_refs(struct ceph_inode_info *ci, int got)
 		ci->i_rd_ref++;
 	if (got & CEPH_CAP_FILE_CACHE)
 		ci->i_rdcache_ref++;
-	if (got & CEPH_CAP_FILE_WR)
+	if (got & CEPH_CAP_FILE_WR) {
+		if (ci->i_wr_ref == 0 && !ci->i_head_snapc) {
+			BUG_ON(!snap_rwsem_locked);
+			ci->i_head_snapc = ceph_get_snap_context(
+					ci->i_snap_realm->cached_context);
+		}
 		ci->i_wr_ref++;
+	}
 	if (got & CEPH_CAP_FILE_BUFFER) {
 		if (ci->i_wb_ref == 0)
 			ihold(&ci->vfs_inode);
@@ -2100,16 +2313,19 @@ static void __take_cap_refs(struct ceph_inode_info *ci, int got)
  * requested from the MDS.
  */
 static int try_get_cap_refs(struct ceph_inode_info *ci, int need, int want,
-			    loff_t endoff, int *got, int *check_max, int *err)
+			    loff_t endoff, bool nonblock, int *got, int *err)
 {
 	struct inode *inode = &ci->vfs_inode;
+	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
 	int ret = 0;
 	int have, implemented;
 	int file_wanted;
+	bool snap_rwsem_locked = false;
 
 	dout("get_cap_refs %p need %s want %s\n", inode,
 	     ceph_cap_string(need), ceph_cap_string(want));
 
+again:
 	spin_lock(&ci->i_ceph_lock);
 
 	/* make sure file is actually open */
@@ -2125,6 +2341,10 @@ static int try_get_cap_refs(struct ceph_inode_info *ci, int need, int want,
 	/* finish pending truncate */
 	while (ci->i_truncate_pending) {
 		spin_unlock(&ci->i_ceph_lock);
+		if (snap_rwsem_locked) {
+			up_read(&mdsc->snap_rwsem);
+			snap_rwsem_locked = false;
+		}
 		__ceph_do_pending_vmtruncate(inode);
 		spin_lock(&ci->i_ceph_lock);
 	}
@@ -2136,7 +2356,7 @@ static int try_get_cap_refs(struct ceph_inode_info *ci, int need, int want,
 			dout("get_cap_refs %p endoff %llu > maxsize %llu\n",
 			     inode, endoff, ci->i_max_size);
 			if (endoff > ci->i_requested_max_size) {
-				*check_max = 1;
+				*err = -EAGAIN;
 				ret = 1;
 			}
 			goto out_unlock;
@@ -2164,8 +2384,29 @@ static int try_get_cap_refs(struct ceph_inode_info *ci, int need, int want,
 		     inode, ceph_cap_string(have), ceph_cap_string(not),
 		     ceph_cap_string(revoking));
 		if ((revoking & not) == 0) {
+			if (!snap_rwsem_locked &&
+			    !ci->i_head_snapc &&
+			    (need & CEPH_CAP_FILE_WR)) {
+				if (!down_read_trylock(&mdsc->snap_rwsem)) {
+					/*
+					 * we can not call down_read() when
+					 * task isn't in TASK_RUNNING state
+					 */
+					if (nonblock) {
+						*err = -EAGAIN;
+						ret = 1;
+						goto out_unlock;
+					}
+
+					spin_unlock(&ci->i_ceph_lock);
+					down_read(&mdsc->snap_rwsem);
+					snap_rwsem_locked = true;
+					goto again;
+				}
+				snap_rwsem_locked = true;
+			}
 			*got = need | (have & want);
-			__take_cap_refs(ci, *got);
+			__take_cap_refs(ci, *got, true);
 			ret = 1;
 		}
 	} else {
@@ -2189,6 +2430,8 @@ static int try_get_cap_refs(struct ceph_inode_info *ci, int need, int want,
 	}
 out_unlock:
 	spin_unlock(&ci->i_ceph_lock);
+	if (snap_rwsem_locked)
+		up_read(&mdsc->snap_rwsem);
 
 	dout("get_cap_refs %p ret %d got %s\n", inode,
 	     ret, ceph_cap_string(*got));
@@ -2231,50 +2474,70 @@ static void check_max_size(struct inode *inode, loff_t endoff)
 int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
 		  loff_t endoff, int *got, struct page **pinned_page)
 {
-	int _got, check_max, ret, err = 0;
+	int _got, ret, err = 0;
 
-retry:
-	if (endoff > 0)
-		check_max_size(&ci->vfs_inode, endoff);
-	_got = 0;
-	check_max = 0;
-	ret = wait_event_interruptible(ci->i_cap_wq,
-				try_get_cap_refs(ci, need, want, endoff,
-						 &_got, &check_max, &err));
-	if (err)
-		ret = err;
+	ret = ceph_pool_perm_check(ci, need);
 	if (ret < 0)
 		return ret;
 
-	if (check_max)
-		goto retry;
+	while (true) {
+		if (endoff > 0)
+			check_max_size(&ci->vfs_inode, endoff);
 
-	if (ci->i_inline_version != CEPH_INLINE_NONE &&
-	    (_got & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) &&
-	    i_size_read(&ci->vfs_inode) > 0) {
-		struct page *page = find_get_page(ci->vfs_inode.i_mapping, 0);
-		if (page) {
-			if (PageUptodate(page)) {
-				*pinned_page = page;
-				goto out;
-			}
-			page_cache_release(page);
-		}
-		/*
-		 * drop cap refs first because getattr while holding
-		 * caps refs can cause deadlock.
-		 */
-		ceph_put_cap_refs(ci, _got);
+		err = 0;
 		_got = 0;
+		ret = try_get_cap_refs(ci, need, want, endoff,
+				       false, &_got, &err);
+		if (ret) {
+			if (err == -EAGAIN)
+				continue;
+			if (err < 0)
+				return err;
+		} else {
+			ret = wait_event_interruptible(ci->i_cap_wq,
+					try_get_cap_refs(ci, need, want, endoff,
+							 true, &_got, &err));
+			if (err == -EAGAIN)
+				continue;
+			if (err < 0)
+				ret = err;
+			if (ret < 0)
+				return ret;
+		}
 
-		/* getattr request will bring inline data into page cache */
-		ret = __ceph_do_getattr(&ci->vfs_inode, NULL,
-					CEPH_STAT_CAP_INLINE_DATA, true);
-		if (ret < 0)
-			return ret;
-		goto retry;
+		if (ci->i_inline_version != CEPH_INLINE_NONE &&
+		    (_got & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) &&
+		    i_size_read(&ci->vfs_inode) > 0) {
+			struct page *page =
+				find_get_page(ci->vfs_inode.i_mapping, 0);
+			if (page) {
+				if (PageUptodate(page)) {
+					*pinned_page = page;
+					break;
+				}
+				page_cache_release(page);
+			}
+			/*
+			 * drop cap refs first because getattr while
+			 * holding * caps refs can cause deadlock.
+			 */
+			ceph_put_cap_refs(ci, _got);
+			_got = 0;
+
+			/*
+			 * getattr request will bring inline data into
+			 * page cache
+			 */
+			ret = __ceph_do_getattr(&ci->vfs_inode, NULL,
+						CEPH_STAT_CAP_INLINE_DATA,
+						true);
+			if (ret < 0)
+				return ret;
+			continue;
+		}
+		break;
 	}
-out:
+
 	*got = _got;
 	return 0;
 }
@@ -2286,10 +2549,31 @@ out:
 void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps)
 {
 	spin_lock(&ci->i_ceph_lock);
-	__take_cap_refs(ci, caps);
+	__take_cap_refs(ci, caps, false);
 	spin_unlock(&ci->i_ceph_lock);
 }
 
+
+/*
+ * drop cap_snap that is not associated with any snapshot.
+ * we don't need to send FLUSHSNAP message for it.
+ */
+static int ceph_try_drop_cap_snap(struct ceph_cap_snap *capsnap)
+{
+	if (!capsnap->need_flush &&
+	    !capsnap->writing && !capsnap->dirty_pages) {
+
+		dout("dropping cap_snap %p follows %llu\n",
+		     capsnap, capsnap->follows);
+		ceph_put_snap_context(capsnap->context);
+		list_del(&capsnap->ci_item);
+		list_del(&capsnap->flushing_item);
+		ceph_put_cap_snap(capsnap);
+		return 1;
+	}
+	return 0;
+}
+
 /*
  * Release cap refs.
  *
@@ -2303,7 +2587,6 @@ void ceph_put_cap_refs(struct ceph_inode_info *ci, int had)
 {
 	struct inode *inode = &ci->vfs_inode;
 	int last = 0, put = 0, flushsnaps = 0, wake = 0;
-	struct ceph_cap_snap *capsnap;
 
 	spin_lock(&ci->i_ceph_lock);
 	if (had & CEPH_CAP_PIN)
@@ -2325,17 +2608,24 @@ void ceph_put_cap_refs(struct ceph_inode_info *ci, int had)
 	if (had & CEPH_CAP_FILE_WR)
 		if (--ci->i_wr_ref == 0) {
 			last++;
-			if (!list_empty(&ci->i_cap_snaps)) {
-				capsnap = list_first_entry(&ci->i_cap_snaps,
-						     struct ceph_cap_snap,
-						     ci_item);
-				if (capsnap->writing) {
-					capsnap->writing = 0;
-					flushsnaps =
-						__ceph_finish_cap_snap(ci,
-								       capsnap);
-					wake = 1;
-				}
+			if (__ceph_have_pending_cap_snap(ci)) {
+				struct ceph_cap_snap *capsnap =
+					list_last_entry(&ci->i_cap_snaps,
+							struct ceph_cap_snap,
+							ci_item);
+				capsnap->writing = 0;
+				if (ceph_try_drop_cap_snap(capsnap))
+					put++;
+				else if (__ceph_finish_cap_snap(ci, capsnap))
+					flushsnaps = 1;
+				wake = 1;
+			}
+			if (ci->i_wrbuffer_ref_head == 0 &&
+			    ci->i_dirty_caps == 0 &&
+			    ci->i_flushing_caps == 0) {
+				BUG_ON(!ci->i_head_snapc);
+				ceph_put_snap_context(ci->i_head_snapc);
+				ci->i_head_snapc = NULL;
 			}
 			/* see comment in __ceph_remove_cap() */
 			if (!__ceph_is_any_caps(ci) && ci->i_snap_realm)
@@ -2352,7 +2642,7 @@ void ceph_put_cap_refs(struct ceph_inode_info *ci, int had)
 		ceph_flush_snaps(ci);
 	if (wake)
 		wake_up_all(&ci->i_cap_wq);
-	if (put)
+	while (put-- > 0)
 		iput(inode);
 }
 
@@ -2380,7 +2670,9 @@ void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
 	if (ci->i_head_snapc == snapc) {
 		ci->i_wrbuffer_ref_head -= nr;
 		if (ci->i_wrbuffer_ref_head == 0 &&
-		    ci->i_dirty_caps == 0 && ci->i_flushing_caps == 0) {
+		    ci->i_wr_ref == 0 &&
+		    ci->i_dirty_caps == 0 &&
+		    ci->i_flushing_caps == 0) {
 			BUG_ON(!ci->i_head_snapc);
 			ceph_put_snap_context(ci->i_head_snapc);
 			ci->i_head_snapc = NULL;
@@ -2401,25 +2693,15 @@ void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
 		capsnap->dirty_pages -= nr;
 		if (capsnap->dirty_pages == 0) {
 			complete_capsnap = 1;
-			if (capsnap->dirty == 0)
-				/* cap writeback completed before we created
-				 * the cap_snap; no FLUSHSNAP is needed */
-				drop_capsnap = 1;
+			drop_capsnap = ceph_try_drop_cap_snap(capsnap);
 		}
 		dout("put_wrbuffer_cap_refs on %p cap_snap %p "
-		     " snap %lld %d/%d -> %d/%d %s%s%s\n",
+		     " snap %lld %d/%d -> %d/%d %s%s\n",
 		     inode, capsnap, capsnap->context->seq,
 		     ci->i_wrbuffer_ref+nr, capsnap->dirty_pages + nr,
 		     ci->i_wrbuffer_ref, capsnap->dirty_pages,
 		     last ? " (wrbuffer last)" : "",
-		     complete_capsnap ? " (complete capsnap)" : "",
-		     drop_capsnap ? " (drop capsnap)" : "");
-		if (drop_capsnap) {
-			ceph_put_snap_context(capsnap->context);
-			list_del(&capsnap->ci_item);
-			list_del(&capsnap->flushing_item);
-			ceph_put_cap_snap(capsnap);
-		}
+		     complete_capsnap ? " (complete capsnap)" : "");
 	}
 
 	spin_unlock(&ci->i_ceph_lock);
@@ -2526,7 +2808,8 @@ static void handle_cap_grant(struct ceph_mds_client *mdsc,
 	 * try to invalidate (once).  (If there are dirty buffers, we
 	 * will invalidate _after_ writeback.)
 	 */
-	if (((cap->issued & ~newcaps) & CEPH_CAP_FILE_CACHE) &&
+	if (!S_ISDIR(inode->i_mode) && /* don't invalidate readdir cache */
+	    ((cap->issued & ~newcaps) & CEPH_CAP_FILE_CACHE) &&
 	    (newcaps & CEPH_CAP_FILE_LAZYIO) == 0 &&
 	    !ci->i_wrbuffer_ref) {
 		if (try_nonblocking_invalidate(inode)) {
@@ -2732,16 +3015,29 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid,
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
+	struct ceph_cap_flush *cf;
+	struct rb_node *n;
+	LIST_HEAD(to_remove);
 	unsigned seq = le32_to_cpu(m->seq);
 	int dirty = le32_to_cpu(m->dirty);
 	int cleaned = 0;
 	int drop = 0;
-	int i;
 
-	for (i = 0; i < CEPH_CAP_BITS; i++)
-		if ((dirty & (1 << i)) &&
-		    (u16)flush_tid == ci->i_cap_flush_tid[i])
-			cleaned |= 1 << i;
+	n = rb_first(&ci->i_cap_flush_tree);
+	while (n) {
+		cf = rb_entry(n, struct ceph_cap_flush, i_node);
+		n = rb_next(&cf->i_node);
+		if (cf->tid == flush_tid)
+			cleaned = cf->caps;
+		if (cf->tid <= flush_tid) {
+			rb_erase(&cf->i_node, &ci->i_cap_flush_tree);
+			list_add_tail(&cf->list, &to_remove);
+		} else {
+			cleaned &= ~cf->caps;
+			if (!cleaned)
+				break;
+		}
+	}
 
 	dout("handle_cap_flush_ack inode %p mds%d seq %d on %s cleaned %s,"
 	     " flushing %s -> %s\n",
@@ -2749,12 +3045,23 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid,
 	     ceph_cap_string(cleaned), ceph_cap_string(ci->i_flushing_caps),
 	     ceph_cap_string(ci->i_flushing_caps & ~cleaned));
 
-	if (ci->i_flushing_caps == (ci->i_flushing_caps & ~cleaned))
+	if (list_empty(&to_remove) && !cleaned)
 		goto out;
 
 	ci->i_flushing_caps &= ~cleaned;
 
 	spin_lock(&mdsc->cap_dirty_lock);
+
+	if (!list_empty(&to_remove)) {
+		list_for_each_entry(cf, &to_remove, list)
+			rb_erase(&cf->g_node, &mdsc->cap_flush_tree);
+
+		n = rb_first(&mdsc->cap_flush_tree);
+		cf = n ? rb_entry(n, struct ceph_cap_flush, g_node) : NULL;
+		if (!cf || cf->tid > flush_tid)
+			wake_up_all(&mdsc->cap_flushing_wq);
+	}
+
 	if (ci->i_flushing_caps == 0) {
 		list_del_init(&ci->i_flushing_item);
 		if (!list_empty(&session->s_cap_flushing))
@@ -2764,14 +3071,14 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid,
 					 struct ceph_inode_info,
 					 i_flushing_item)->vfs_inode);
 		mdsc->num_cap_flushing--;
-		wake_up_all(&mdsc->cap_flushing_wq);
 		dout(" inode %p now !flushing\n", inode);
 
 		if (ci->i_dirty_caps == 0) {
 			dout(" inode %p now clean\n", inode);
 			BUG_ON(!list_empty(&ci->i_dirty_item));
 			drop = 1;
-			if (ci->i_wrbuffer_ref_head == 0) {
+			if (ci->i_wr_ref == 0 &&
+			    ci->i_wrbuffer_ref_head == 0) {
 				BUG_ON(!ci->i_head_snapc);
 				ceph_put_snap_context(ci->i_head_snapc);
 				ci->i_head_snapc = NULL;
@@ -2785,6 +3092,13 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid,
 
 out:
 	spin_unlock(&ci->i_ceph_lock);
+
+	while (!list_empty(&to_remove)) {
+		cf = list_first_entry(&to_remove,
+				      struct ceph_cap_flush, list);
+		list_del(&cf->list);
+		ceph_free_cap_flush(cf);
+	}
 	if (drop)
 		iput(inode);
 }
@@ -2800,6 +3114,7 @@ static void handle_cap_flushsnap_ack(struct inode *inode, u64 flush_tid,
 				     struct ceph_mds_session *session)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
 	u64 follows = le64_to_cpu(m->snap_follows);
 	struct ceph_cap_snap *capsnap;
 	int drop = 0;
@@ -2823,6 +3138,7 @@ static void handle_cap_flushsnap_ack(struct inode *inode, u64 flush_tid,
 			list_del(&capsnap->ci_item);
 			list_del(&capsnap->flushing_item);
 			ceph_put_cap_snap(capsnap);
+			wake_up_all(&mdsc->cap_flushing_wq);
 			drop = 1;
 			break;
 		} else {
@@ -2971,7 +3287,6 @@ retry:
 			mutex_lock_nested(&session->s_mutex,
 					  SINGLE_DEPTH_NESTING);
 		}
-		ceph_add_cap_releases(mdsc, tsession);
 		new_cap = ceph_get_cap(mdsc, NULL);
 	} else {
 		WARN_ON(1);
@@ -3167,16 +3482,20 @@ void ceph_handle_caps(struct ceph_mds_session *session,
 	dout(" mds%d seq %lld cap seq %u\n", session->s_mds, session->s_seq,
 	     (unsigned)seq);
 
-	if (op == CEPH_CAP_OP_IMPORT)
-		ceph_add_cap_releases(mdsc, session);
-
 	if (!inode) {
 		dout(" i don't have ino %llx\n", vino.ino);
 
 		if (op == CEPH_CAP_OP_IMPORT) {
+			cap = ceph_get_cap(mdsc, NULL);
+			cap->cap_ino = vino.ino;
+			cap->queue_release = 1;
+			cap->cap_id = cap_id;
+			cap->mseq = mseq;
+			cap->seq = seq;
 			spin_lock(&session->s_cap_lock);
-			__queue_cap_release(session, vino.ino, cap_id,
-					    mseq, seq);
+			list_add_tail(&cap->session_caps,
+					&session->s_cap_releases);
+			session->s_num_cap_releases++;
 			spin_unlock(&session->s_cap_lock);
 		}
 		goto flush_cap_releases;
@@ -3252,11 +3571,10 @@ void ceph_handle_caps(struct ceph_mds_session *session,
 
 flush_cap_releases:
 	/*
-	 * send any full release message to try to move things
+	 * send any cap release message to try to move things
 	 * along for the mds (who clearly thinks we still have this
 	 * cap).
 	 */
-	ceph_add_cap_releases(mdsc, session);
 	ceph_send_cap_releases(mdsc, session);
 
 done:
diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index 4248307fea90..9314b4ea2375 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -38,7 +38,7 @@ int ceph_init_dentry(struct dentry *dentry)
 	if (dentry->d_fsdata)
 		return 0;
 
-	di = kmem_cache_alloc(ceph_dentry_cachep, GFP_NOFS | __GFP_ZERO);
+	di = kmem_cache_alloc(ceph_dentry_cachep, GFP_KERNEL | __GFP_ZERO);
 	if (!di)
 		return -ENOMEM;          /* oh well */
 
@@ -107,6 +107,27 @@ static int fpos_cmp(loff_t l, loff_t r)
 }
 
 /*
+ * make note of the last dentry we read, so we can
+ * continue at the same lexicographical point,
+ * regardless of what dir changes take place on the
+ * server.
+ */
+static int note_last_dentry(struct ceph_file_info *fi, const char *name,
+		            int len, unsigned next_offset)
+{
+	char *buf = kmalloc(len+1, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+	kfree(fi->last_name);
+	fi->last_name = buf;
+	memcpy(fi->last_name, name, len);
+	fi->last_name[len] = 0;
+	fi->next_offset = next_offset;
+	dout("note_last_dentry '%s'\n", fi->last_name);
+	return 0;
+}
+
+/*
  * When possible, we try to satisfy a readdir by peeking at the
  * dcache.  We make this work by carefully ordering dentries on
  * d_child when we initially get results back from the MDS, and
@@ -123,123 +144,113 @@ static int __dcache_readdir(struct file *file,  struct dir_context *ctx,
 	struct ceph_file_info *fi = file->private_data;
 	struct dentry *parent = file->f_path.dentry;
 	struct inode *dir = d_inode(parent);
-	struct list_head *p;
-	struct dentry *dentry, *last;
+	struct dentry *dentry, *last = NULL;
 	struct ceph_dentry_info *di;
+	unsigned nsize = PAGE_CACHE_SIZE / sizeof(struct dentry *);
 	int err = 0;
+	loff_t ptr_pos = 0;
+	struct ceph_readdir_cache_control cache_ctl = {};
 
-	/* claim ref on last dentry we returned */
-	last = fi->dentry;
-	fi->dentry = NULL;
-
-	dout("__dcache_readdir %p v%u at %llu (last %p)\n",
-	     dir, shared_gen, ctx->pos, last);
+	dout("__dcache_readdir %p v%u at %llu\n", dir, shared_gen, ctx->pos);
 
-	spin_lock(&parent->d_lock);
-
-	/* start at beginning? */
-	if (ctx->pos == 2 || last == NULL ||
-	    fpos_cmp(ctx->pos, ceph_dentry(last)->offset) < 0) {
-		if (list_empty(&parent->d_subdirs))
-			goto out_unlock;
-		p = parent->d_subdirs.prev;
-		dout(" initial p %p/%p\n", p->prev, p->next);
-	} else {
-		p = last->d_child.prev;
+	/* we can calculate cache index for the first dirfrag */
+	if (ceph_frag_is_leftmost(fpos_frag(ctx->pos))) {
+		cache_ctl.index = fpos_off(ctx->pos) - 2;
+		BUG_ON(cache_ctl.index < 0);
+		ptr_pos = cache_ctl.index * sizeof(struct dentry *);
 	}
 
-more:
-	dentry = list_entry(p, struct dentry, d_child);
-	di = ceph_dentry(dentry);
-	while (1) {
-		dout(" p %p/%p %s d_subdirs %p/%p\n", p->prev, p->next,
-		     d_unhashed(dentry) ? "!hashed" : "hashed",
-		     parent->d_subdirs.prev, parent->d_subdirs.next);
-		if (p == &parent->d_subdirs) {
+	while (true) {
+		pgoff_t pgoff;
+		bool emit_dentry;
+
+		if (ptr_pos >= i_size_read(dir)) {
 			fi->flags |= CEPH_F_ATEND;
-			goto out_unlock;
+			err = 0;
+			break;
+		}
+
+		err = -EAGAIN;
+		pgoff = ptr_pos >> PAGE_CACHE_SHIFT;
+		if (!cache_ctl.page || pgoff != page_index(cache_ctl.page)) {
+			ceph_readdir_cache_release(&cache_ctl);
+			cache_ctl.page = find_lock_page(&dir->i_data, pgoff);
+			if (!cache_ctl.page) {
+				dout(" page %lu not found\n", pgoff);
+				break;
+			}
+			/* reading/filling the cache are serialized by
+			 * i_mutex, no need to use page lock */
+			unlock_page(cache_ctl.page);
+			cache_ctl.dentries = kmap(cache_ctl.page);
 		}
-		spin_lock_nested(&dentry->d_lock, DENTRY_D_LOCK_NESTED);
+
+		rcu_read_lock();
+		spin_lock(&parent->d_lock);
+		/* check i_size again here, because empty directory can be
+		 * marked as complete while not holding the i_mutex. */
+		if (ceph_dir_is_complete_ordered(dir) &&
+		    ptr_pos < i_size_read(dir))
+			dentry = cache_ctl.dentries[cache_ctl.index % nsize];
+		else
+			dentry = NULL;
+		spin_unlock(&parent->d_lock);
+		if (dentry && !lockref_get_not_dead(&dentry->d_lockref))
+			dentry = NULL;
+		rcu_read_unlock();
+		if (!dentry)
+			break;
+
+		emit_dentry = false;
+		di = ceph_dentry(dentry);
+		spin_lock(&dentry->d_lock);
 		if (di->lease_shared_gen == shared_gen &&
-		    !d_unhashed(dentry) && d_really_is_positive(dentry) &&
+		    d_really_is_positive(dentry) &&
 		    ceph_snap(d_inode(dentry)) != CEPH_SNAPDIR &&
 		    ceph_ino(d_inode(dentry)) != CEPH_INO_CEPH &&
-		    fpos_cmp(ctx->pos, di->offset) <= 0)
-			break;
-		dout(" skipping %p %pd at %llu (%llu)%s%s\n", dentry,
-		     dentry, di->offset,
-		     ctx->pos, d_unhashed(dentry) ? " unhashed" : "",
-		     !d_inode(dentry) ? " null" : "");
+		    fpos_cmp(ctx->pos, di->offset) <= 0) {
+			emit_dentry = true;
+		}
 		spin_unlock(&dentry->d_lock);
-		p = p->prev;
-		dentry = list_entry(p, struct dentry, d_child);
-		di = ceph_dentry(dentry);
-	}
-
-	dget_dlock(dentry);
-	spin_unlock(&dentry->d_lock);
-	spin_unlock(&parent->d_lock);
 
-	/* make sure a dentry wasn't dropped while we didn't have parent lock */
-	if (!ceph_dir_is_complete_ordered(dir)) {
-		dout(" lost dir complete on %p; falling back to mds\n", dir);
-		dput(dentry);
-		err = -EAGAIN;
-		goto out;
-	}
+		if (emit_dentry) {
+			dout(" %llu (%llu) dentry %p %pd %p\n", di->offset, ctx->pos,
+			     dentry, dentry, d_inode(dentry));
+			ctx->pos = di->offset;
+			if (!dir_emit(ctx, dentry->d_name.name,
+				      dentry->d_name.len,
+				      ceph_translate_ino(dentry->d_sb,
+							 d_inode(dentry)->i_ino),
+				      d_inode(dentry)->i_mode >> 12)) {
+				dput(dentry);
+				err = 0;
+				break;
+			}
+			ctx->pos++;
 
-	dout(" %llu (%llu) dentry %p %pd %p\n", di->offset, ctx->pos,
-	     dentry, dentry, d_inode(dentry));
-	if (!dir_emit(ctx, dentry->d_name.name,
-		      dentry->d_name.len,
-		      ceph_translate_ino(dentry->d_sb, d_inode(dentry)->i_ino),
-		      d_inode(dentry)->i_mode >> 12)) {
-		if (last) {
-			/* remember our position */
-			fi->dentry = last;
-			fi->next_offset = fpos_off(di->offset);
+			if (last)
+				dput(last);
+			last = dentry;
+		} else {
+			dput(dentry);
 		}
-		dput(dentry);
-		return 0;
-	}
-
-	ctx->pos = di->offset + 1;
 
-	if (last)
-		dput(last);
-	last = dentry;
-
-	spin_lock(&parent->d_lock);
-	p = p->prev;	/* advance to next dentry */
-	goto more;
-
-out_unlock:
-	spin_unlock(&parent->d_lock);
-out:
-	if (last)
+		cache_ctl.index++;
+		ptr_pos += sizeof(struct dentry *);
+	}
+	ceph_readdir_cache_release(&cache_ctl);
+	if (last) {
+		int ret;
+		di = ceph_dentry(last);
+		ret = note_last_dentry(fi, last->d_name.name, last->d_name.len,
+				       fpos_off(di->offset) + 1);
+		if (ret < 0)
+			err = ret;
 		dput(last);
+	}
 	return err;
 }
 
-/*
- * make note of the last dentry we read, so we can
- * continue at the same lexicographical point,
- * regardless of what dir changes take place on the
- * server.
- */
-static int note_last_dentry(struct ceph_file_info *fi, const char *name,
-			    int len)
-{
-	kfree(fi->last_name);
-	fi->last_name = kmalloc(len+1, GFP_NOFS);
-	if (!fi->last_name)
-		return -ENOMEM;
-	memcpy(fi->last_name, name, len);
-	fi->last_name[len] = 0;
-	dout("note_last_dentry '%s'\n", fi->last_name);
-	return 0;
-}
-
 static int ceph_readdir(struct file *file, struct dir_context *ctx)
 {
 	struct ceph_file_info *fi = file->private_data;
@@ -280,8 +291,7 @@ static int ceph_readdir(struct file *file, struct dir_context *ctx)
 
 	/* can we use the dcache? */
 	spin_lock(&ci->i_ceph_lock);
-	if ((ctx->pos == 2 || fi->dentry) &&
-	    ceph_test_mount_opt(fsc, DCACHE) &&
+	if (ceph_test_mount_opt(fsc, DCACHE) &&
 	    !ceph_test_mount_opt(fsc, NOASYNCREADDIR) &&
 	    ceph_snap(inode) != CEPH_SNAPDIR &&
 	    __ceph_dir_is_complete_ordered(ci) &&
@@ -296,24 +306,8 @@ static int ceph_readdir(struct file *file, struct dir_context *ctx)
 	} else {
 		spin_unlock(&ci->i_ceph_lock);
 	}
-	if (fi->dentry) {
-		err = note_last_dentry(fi, fi->dentry->d_name.name,
-				       fi->dentry->d_name.len);
-		if (err)
-			return err;
-		dput(fi->dentry);
-		fi->dentry = NULL;
-	}
 
 	/* proceed with a normal readdir */
-
-	if (ctx->pos == 2) {
-		/* note dir version at start of readdir so we can tell
-		 * if any dentries get dropped */
-		fi->dir_release_count = atomic_read(&ci->i_release_count);
-		fi->dir_ordered_count = ci->i_ordered_count;
-	}
-
 more:
 	/* do we have the correct frag content buffered? */
 	if (fi->frag != frag || fi->last_readdir == NULL) {
@@ -342,12 +336,15 @@ more:
 		req->r_direct_hash = ceph_frag_value(frag);
 		req->r_direct_is_hash = true;
 		if (fi->last_name) {
-			req->r_path2 = kstrdup(fi->last_name, GFP_NOFS);
+			req->r_path2 = kstrdup(fi->last_name, GFP_KERNEL);
 			if (!req->r_path2) {
 				ceph_mdsc_put_request(req);
 				return -ENOMEM;
 			}
 		}
+		req->r_dir_release_cnt = fi->dir_release_count;
+		req->r_dir_ordered_cnt = fi->dir_ordered_count;
+		req->r_readdir_cache_idx = fi->readdir_cache_idx;
 		req->r_readdir_offset = fi->next_offset;
 		req->r_args.readdir.frag = cpu_to_le32(frag);
 
@@ -364,26 +361,38 @@ more:
 		     (int)req->r_reply_info.dir_end,
 		     (int)req->r_reply_info.dir_complete);
 
-		if (!req->r_did_prepopulate) {
-			dout("readdir !did_prepopulate");
-			/* preclude from marking dir complete */
-			fi->dir_release_count--;
-		}
 
 		/* note next offset and last dentry name */
 		rinfo = &req->r_reply_info;
 		if (le32_to_cpu(rinfo->dir_dir->frag) != frag) {
 			frag = le32_to_cpu(rinfo->dir_dir->frag);
-			if (ceph_frag_is_leftmost(frag))
-				fi->next_offset = 2;
-			else
-				fi->next_offset = 0;
-			off = fi->next_offset;
+			off = req->r_readdir_offset;
+			fi->next_offset = off;
 		}
+
 		fi->frag = frag;
 		fi->offset = fi->next_offset;
 		fi->last_readdir = req;
 
+		if (req->r_did_prepopulate) {
+			fi->readdir_cache_idx = req->r_readdir_cache_idx;
+			if (fi->readdir_cache_idx < 0) {
+				/* preclude from marking dir ordered */
+				fi->dir_ordered_count = 0;
+			} else if (ceph_frag_is_leftmost(frag) && off == 2) {
+				/* note dir version at start of readdir so
+				 * we can tell if any dentries get dropped */
+				fi->dir_release_count = req->r_dir_release_cnt;
+				fi->dir_ordered_count = req->r_dir_ordered_cnt;
+			}
+		} else {
+			dout("readdir !did_prepopulate");
+			/* disable readdir cache */
+			fi->readdir_cache_idx = -1;
+			/* preclude from marking dir complete */
+			fi->dir_release_count = 0;
+		}
+
 		if (req->r_reply_info.dir_end) {
 			kfree(fi->last_name);
 			fi->last_name = NULL;
@@ -394,10 +403,10 @@ more:
 		} else {
 			err = note_last_dentry(fi,
 				       rinfo->dir_dname[rinfo->dir_nr-1],
-				       rinfo->dir_dname_len[rinfo->dir_nr-1]);
+				       rinfo->dir_dname_len[rinfo->dir_nr-1],
+				       fi->next_offset + rinfo->dir_nr);
 			if (err)
 				return err;
-			fi->next_offset += rinfo->dir_nr;
 		}
 	}
 
@@ -453,16 +462,22 @@ more:
 	 * were released during the whole readdir, and we should have
 	 * the complete dir contents in our cache.
 	 */
-	spin_lock(&ci->i_ceph_lock);
-	if (atomic_read(&ci->i_release_count) == fi->dir_release_count) {
-		if (ci->i_ordered_count == fi->dir_ordered_count)
+	if (atomic64_read(&ci->i_release_count) == fi->dir_release_count) {
+		spin_lock(&ci->i_ceph_lock);
+		if (fi->dir_ordered_count == atomic64_read(&ci->i_ordered_count)) {
 			dout(" marking %p complete and ordered\n", inode);
-		else
+			/* use i_size to track number of entries in
+			 * readdir cache */
+			BUG_ON(fi->readdir_cache_idx < 0);
+			i_size_write(inode, fi->readdir_cache_idx *
+				     sizeof(struct dentry*));
+		} else {
 			dout(" marking %p complete\n", inode);
+		}
 		__ceph_dir_set_complete(ci, fi->dir_release_count,
 					fi->dir_ordered_count);
+		spin_unlock(&ci->i_ceph_lock);
 	}
-	spin_unlock(&ci->i_ceph_lock);
 
 	dout("readdir %p file %p done.\n", inode, file);
 	return 0;
@@ -476,14 +491,12 @@ static void reset_readdir(struct ceph_file_info *fi, unsigned frag)
 	}
 	kfree(fi->last_name);
 	fi->last_name = NULL;
+	fi->dir_release_count = 0;
+	fi->readdir_cache_idx = -1;
 	if (ceph_frag_is_leftmost(frag))
 		fi->next_offset = 2;  /* compensate for . and .. */
 	else
 		fi->next_offset = 0;
-	if (fi->dentry) {
-		dput(fi->dentry);
-		fi->dentry = NULL;
-	}
 	fi->flags &= ~CEPH_F_ATEND;
 }
 
@@ -497,13 +510,12 @@ static loff_t ceph_dir_llseek(struct file *file, loff_t offset, int whence)
 	mutex_lock(&inode->i_mutex);
 	retval = -EINVAL;
 	switch (whence) {
-	case SEEK_END:
-		offset += inode->i_size + 2;   /* FIXME */
-		break;
 	case SEEK_CUR:
 		offset += file->f_pos;
 	case SEEK_SET:
 		break;
+	case SEEK_END:
+		retval = -EOPNOTSUPP;
 	default:
 		goto out;
 	}
@@ -516,20 +528,18 @@ static loff_t ceph_dir_llseek(struct file *file, loff_t offset, int whence)
 		}
 		retval = offset;
 
-		/*
-		 * discard buffered readdir content on seekdir(0), or
-		 * seek to new frag, or seek prior to current chunk.
-		 */
 		if (offset == 0 ||
 		    fpos_frag(offset) != fi->frag ||
 		    fpos_off(offset) < fi->offset) {
+			/* discard buffered readdir content on seekdir(0), or
+			 * seek to new frag, or seek prior to current chunk */
 			dout("dir_llseek dropping %p content\n", file);
 			reset_readdir(fi, fpos_frag(offset));
+		} else if (fpos_cmp(offset, old_offset) > 0) {
+			/* reset dir_release_count if we did a forward seek */
+			fi->dir_release_count = 0;
+			fi->readdir_cache_idx = -1;
 		}
-
-		/* bump dir_release_count if we did a forward seek */
-		if (fpos_cmp(offset, old_offset) > 0)
-			fi->dir_release_count--;
 	}
 out:
 	mutex_unlock(&inode->i_mutex);
@@ -764,7 +774,7 @@ static int ceph_symlink(struct inode *dir, struct dentry *dentry,
 		err = PTR_ERR(req);
 		goto out;
 	}
-	req->r_path2 = kstrdup(dest, GFP_NOFS);
+	req->r_path2 = kstrdup(dest, GFP_KERNEL);
 	if (!req->r_path2) {
 		err = -ENOMEM;
 		ceph_mdsc_put_request(req);
@@ -985,16 +995,15 @@ static int ceph_rename(struct inode *old_dir, struct dentry *old_dentry,
 		 * to do it here.
 		 */
 
+		/* d_move screws up sibling dentries' offsets */
+		ceph_dir_clear_complete(old_dir);
+		ceph_dir_clear_complete(new_dir);
+
 		d_move(old_dentry, new_dentry);
 
 		/* ensure target dentry is invalidated, despite
 		   rehashing bug in vfs_rename_dir */
 		ceph_invalidate_dentry_lease(new_dentry);
-
-		/* d_move screws up sibling dentries' offsets */
-		ceph_dir_clear_complete(old_dir);
-		ceph_dir_clear_complete(new_dir);
-
 	}
 	ceph_mdsc_put_request(req);
 	return err;
@@ -1189,7 +1198,7 @@ static ssize_t ceph_read_dir(struct file *file, char __user *buf, size_t size,
 		return -EISDIR;
 
 	if (!cf->dir_info) {
-		cf->dir_info = kmalloc(bufsize, GFP_NOFS);
+		cf->dir_info = kmalloc(bufsize, GFP_KERNEL);
 		if (!cf->dir_info)
 			return -ENOMEM;
 		cf->dir_info_len =
@@ -1224,66 +1233,6 @@ static ssize_t ceph_read_dir(struct file *file, char __user *buf, size_t size,
 }
 
 /*
- * an fsync() on a dir will wait for any uncommitted directory
- * operations to commit.
- */
-static int ceph_dir_fsync(struct file *file, loff_t start, loff_t end,
-			  int datasync)
-{
-	struct inode *inode = file_inode(file);
-	struct ceph_inode_info *ci = ceph_inode(inode);
-	struct list_head *head = &ci->i_unsafe_dirops;
-	struct ceph_mds_request *req;
-	u64 last_tid;
-	int ret = 0;
-
-	dout("dir_fsync %p\n", inode);
-	ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
-	if (ret)
-		return ret;
-	mutex_lock(&inode->i_mutex);
-
-	spin_lock(&ci->i_unsafe_lock);
-	if (list_empty(head))
-		goto out;
-
-	req = list_entry(head->prev,
-			 struct ceph_mds_request, r_unsafe_dir_item);
-	last_tid = req->r_tid;
-
-	do {
-		ceph_mdsc_get_request(req);
-		spin_unlock(&ci->i_unsafe_lock);
-
-		dout("dir_fsync %p wait on tid %llu (until %llu)\n",
-		     inode, req->r_tid, last_tid);
-		if (req->r_timeout) {
-			unsigned long time_left = wait_for_completion_timeout(
-							&req->r_safe_completion,
-							req->r_timeout);
-			if (time_left > 0)
-				ret = 0;
-			else
-				ret = -EIO;  /* timed out */
-		} else {
-			wait_for_completion(&req->r_safe_completion);
-		}
-		ceph_mdsc_put_request(req);
-
-		spin_lock(&ci->i_unsafe_lock);
-		if (ret || list_empty(head))
-			break;
-		req = list_entry(head->next,
-				 struct ceph_mds_request, r_unsafe_dir_item);
-	} while (req->r_tid < last_tid);
-out:
-	spin_unlock(&ci->i_unsafe_lock);
-	mutex_unlock(&inode->i_mutex);
-
-	return ret;
-}
-
-/*
  * We maintain a private dentry LRU.
  *
  * FIXME: this needs to be changed to a per-mds lru to be useful.
@@ -1353,7 +1302,7 @@ const struct file_operations ceph_dir_fops = {
 	.open = ceph_open,
 	.release = ceph_release,
 	.unlocked_ioctl = ceph_ioctl,
-	.fsync = ceph_dir_fsync,
+	.fsync = ceph_fsync,
 };
 
 const struct file_operations ceph_snapdir_fops = {
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index 3b6b522b4b31..8b79d87eaf46 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -89,13 +89,14 @@ static int ceph_init_file(struct inode *inode, struct file *file, int fmode)
 	case S_IFDIR:
 		dout("init_file %p %p 0%o (regular)\n", inode, file,
 		     inode->i_mode);
-		cf = kmem_cache_alloc(ceph_file_cachep, GFP_NOFS | __GFP_ZERO);
+		cf = kmem_cache_alloc(ceph_file_cachep, GFP_KERNEL | __GFP_ZERO);
 		if (cf == NULL) {
 			ceph_put_fmode(ceph_inode(inode), fmode); /* clean up */
 			return -ENOMEM;
 		}
 		cf->fmode = fmode;
 		cf->next_offset = 2;
+		cf->readdir_cache_idx = -1;
 		file->private_data = cf;
 		BUG_ON(inode->i_fop->release != ceph_release);
 		break;
@@ -324,7 +325,6 @@ int ceph_release(struct inode *inode, struct file *file)
 		ceph_mdsc_put_request(cf->last_readdir);
 	kfree(cf->last_name);
 	kfree(cf->dir_info);
-	dput(cf->dentry);
 	kmem_cache_free(ceph_file_cachep, cf);
 
 	/* wake up anyone waiting for caps on this inode */
@@ -483,7 +483,7 @@ static ssize_t ceph_sync_read(struct kiocb *iocb, struct iov_iter *i,
 		}
 	} else {
 		num_pages = calc_pages_for(off, len);
-		pages = ceph_alloc_page_vector(num_pages, GFP_NOFS);
+		pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL);
 		if (IS_ERR(pages))
 			return PTR_ERR(pages);
 		ret = striped_read(inode, off, len, pages,
@@ -557,13 +557,13 @@ static void ceph_sync_write_unsafe(struct ceph_osd_request *req, bool unsafe)
  * objects, rollback on failure, etc.)
  */
 static ssize_t
-ceph_sync_direct_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
+ceph_sync_direct_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos,
+		       struct ceph_snap_context *snapc)
 {
 	struct file *file = iocb->ki_filp;
 	struct inode *inode = file_inode(file);
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
-	struct ceph_snap_context *snapc;
 	struct ceph_vino vino;
 	struct ceph_osd_request *req;
 	struct page **pages;
@@ -600,7 +600,6 @@ ceph_sync_direct_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
 		size_t start;
 		ssize_t n;
 
-		snapc = ci->i_snap_realm->cached_context;
 		vino = ceph_vino(inode);
 		req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
 					    vino, pos, &len, 0,
@@ -614,7 +613,7 @@ ceph_sync_direct_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
 			break;
 		}
 
-		osd_req_op_init(req, 1, CEPH_OSD_OP_STARTSYNC);
+		osd_req_op_init(req, 1, CEPH_OSD_OP_STARTSYNC, 0);
 
 		n = iov_iter_get_pages_alloc(from, &pages, len, &start);
 		if (unlikely(n < 0)) {
@@ -674,13 +673,13 @@ ceph_sync_direct_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
  * objects, rollback on failure, etc.)
  */
 static ssize_t
-ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
+ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos,
+		struct ceph_snap_context *snapc)
 {
 	struct file *file = iocb->ki_filp;
 	struct inode *inode = file_inode(file);
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
-	struct ceph_snap_context *snapc;
 	struct ceph_vino vino;
 	struct ceph_osd_request *req;
 	struct page **pages;
@@ -717,7 +716,6 @@ ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
 		size_t left;
 		int n;
 
-		snapc = ci->i_snap_realm->cached_context;
 		vino = ceph_vino(inode);
 		req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
 					    vino, pos, &len, 0, 1,
@@ -736,7 +734,7 @@ ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
 		 */
 		num_pages = (len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
 
-		pages = ceph_alloc_page_vector(num_pages, GFP_NOFS);
+		pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL);
 		if (IS_ERR(pages)) {
 			ret = PTR_ERR(pages);
 			goto out;
@@ -860,7 +858,7 @@ again:
 		struct page *page = NULL;
 		loff_t i_size;
 		if (retry_op == READ_INLINE) {
-			page = __page_cache_alloc(GFP_NOFS);
+			page = __page_cache_alloc(GFP_KERNEL);
 			if (!page)
 				return -ENOMEM;
 		}
@@ -941,6 +939,7 @@ static ssize_t ceph_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_osd_client *osdc =
 		&ceph_sb_to_client(inode->i_sb)->client->osdc;
+	struct ceph_cap_flush *prealloc_cf;
 	ssize_t count, written = 0;
 	int err, want, got;
 	loff_t pos;
@@ -948,6 +947,10 @@ static ssize_t ceph_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	if (ceph_snap(inode) != CEPH_NOSNAP)
 		return -EROFS;
 
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		return -ENOMEM;
+
 	mutex_lock(&inode->i_mutex);
 
 	/* We can write back this queue in page reclaim */
@@ -959,7 +962,7 @@ static ssize_t ceph_write_iter(struct kiocb *iocb, struct iov_iter *from)
 
 	pos = iocb->ki_pos;
 	count = iov_iter_count(from);
-	err = file_remove_suid(file);
+	err = file_remove_privs(file);
 	if (err)
 		goto out;
 
@@ -996,14 +999,30 @@ retry_snap:
 
 	if ((got & (CEPH_CAP_FILE_BUFFER|CEPH_CAP_FILE_LAZYIO)) == 0 ||
 	    (iocb->ki_flags & IOCB_DIRECT) || (fi->flags & CEPH_F_SYNC)) {
+		struct ceph_snap_context *snapc;
 		struct iov_iter data;
 		mutex_unlock(&inode->i_mutex);
+
+		spin_lock(&ci->i_ceph_lock);
+		if (__ceph_have_pending_cap_snap(ci)) {
+			struct ceph_cap_snap *capsnap =
+					list_last_entry(&ci->i_cap_snaps,
+							struct ceph_cap_snap,
+							ci_item);
+			snapc = ceph_get_snap_context(capsnap->context);
+		} else {
+			BUG_ON(!ci->i_head_snapc);
+			snapc = ceph_get_snap_context(ci->i_head_snapc);
+		}
+		spin_unlock(&ci->i_ceph_lock);
+
 		/* we might need to revert back to that point */
 		data = *from;
 		if (iocb->ki_flags & IOCB_DIRECT)
-			written = ceph_sync_direct_write(iocb, &data, pos);
+			written = ceph_sync_direct_write(iocb, &data, pos,
+							 snapc);
 		else
-			written = ceph_sync_write(iocb, &data, pos);
+			written = ceph_sync_write(iocb, &data, pos, snapc);
 		if (written == -EOLDSNAPC) {
 			dout("aio_write %p %llx.%llx %llu~%u"
 				"got EOLDSNAPC, retrying\n",
@@ -1014,6 +1033,7 @@ retry_snap:
 		}
 		if (written > 0)
 			iov_iter_advance(from, written);
+		ceph_put_snap_context(snapc);
 	} else {
 		loff_t old_size = inode->i_size;
 		/*
@@ -1035,7 +1055,8 @@ retry_snap:
 		int dirty;
 		spin_lock(&ci->i_ceph_lock);
 		ci->i_inline_version = CEPH_INLINE_NONE;
-		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR);
+		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
+					       &prealloc_cf);
 		spin_unlock(&ci->i_ceph_lock);
 		if (dirty)
 			__mark_inode_dirty(inode, dirty);
@@ -1059,6 +1080,7 @@ retry_snap:
 out:
 	mutex_unlock(&inode->i_mutex);
 out_unlocked:
+	ceph_free_cap_flush(prealloc_cf);
 	current->backing_dev_info = NULL;
 	return written ? written : err;
 }
@@ -1255,6 +1277,7 @@ static long ceph_fallocate(struct file *file, int mode,
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_osd_client *osdc =
 		&ceph_inode_to_client(inode)->client->osdc;
+	struct ceph_cap_flush *prealloc_cf;
 	int want, got = 0;
 	int dirty;
 	int ret = 0;
@@ -1267,6 +1290,10 @@ static long ceph_fallocate(struct file *file, int mode,
 	if (!S_ISREG(inode->i_mode))
 		return -EOPNOTSUPP;
 
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		return -ENOMEM;
+
 	mutex_lock(&inode->i_mutex);
 
 	if (ceph_snap(inode) != CEPH_NOSNAP) {
@@ -1313,7 +1340,8 @@ static long ceph_fallocate(struct file *file, int mode,
 	if (!ret) {
 		spin_lock(&ci->i_ceph_lock);
 		ci->i_inline_version = CEPH_INLINE_NONE;
-		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR);
+		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
+					       &prealloc_cf);
 		spin_unlock(&ci->i_ceph_lock);
 		if (dirty)
 			__mark_inode_dirty(inode, dirty);
@@ -1322,6 +1350,7 @@ static long ceph_fallocate(struct file *file, int mode,
 	ceph_put_cap_refs(ci, got);
 unlock:
 	mutex_unlock(&inode->i_mutex);
+	ceph_free_cap_flush(prealloc_cf);
 	return ret;
 }
 
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index 571acd88606c..96d2bd829902 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -389,9 +389,10 @@ struct inode *ceph_alloc_inode(struct super_block *sb)
 	ci->i_inline_version = 0;
 	ci->i_time_warp_seq = 0;
 	ci->i_ceph_flags = 0;
-	ci->i_ordered_count = 0;
-	atomic_set(&ci->i_release_count, 1);
-	atomic_set(&ci->i_complete_count, 0);
+	atomic64_set(&ci->i_ordered_count, 1);
+	atomic64_set(&ci->i_release_count, 1);
+	atomic64_set(&ci->i_complete_seq[0], 0);
+	atomic64_set(&ci->i_complete_seq[1], 0);
 	ci->i_symlink = NULL;
 
 	memset(&ci->i_dir_layout, 0, sizeof(ci->i_dir_layout));
@@ -415,9 +416,8 @@ struct inode *ceph_alloc_inode(struct super_block *sb)
 	ci->i_flushing_caps = 0;
 	INIT_LIST_HEAD(&ci->i_dirty_item);
 	INIT_LIST_HEAD(&ci->i_flushing_item);
-	ci->i_cap_flush_seq = 0;
-	ci->i_cap_flush_last_tid = 0;
-	memset(&ci->i_cap_flush_tid, 0, sizeof(ci->i_cap_flush_tid));
+	ci->i_prealloc_cap_flush = NULL;
+	ci->i_cap_flush_tree = RB_ROOT;
 	init_waitqueue_head(&ci->i_cap_wq);
 	ci->i_hold_caps_min = 0;
 	ci->i_hold_caps_max = 0;
@@ -752,7 +752,10 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 
 	if (new_version ||
 	    (new_issued & (CEPH_CAP_ANY_FILE_RD | CEPH_CAP_ANY_FILE_WR))) {
+		if (ci->i_layout.fl_pg_pool != info->layout.fl_pg_pool)
+			ci->i_ceph_flags &= ~CEPH_I_POOL_PERM;
 		ci->i_layout = info->layout;
+
 		queue_trunc = ceph_fill_file_size(inode, issued,
 					le32_to_cpu(info->truncate_seq),
 					le64_to_cpu(info->truncate_size),
@@ -858,9 +861,10 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 			    (issued & CEPH_CAP_FILE_EXCL) == 0 &&
 			    !__ceph_dir_is_complete(ci)) {
 				dout(" marking %p complete (empty)\n", inode);
+				i_size_write(inode, 0);
 				__ceph_dir_set_complete(ci,
-					atomic_read(&ci->i_release_count),
-					ci->i_ordered_count);
+					atomic64_read(&ci->i_release_count),
+					atomic64_read(&ci->i_ordered_count));
 			}
 
 			wake = true;
@@ -1212,6 +1216,10 @@ retry_lookup:
 			dout("fill_trace doing d_move %p -> %p\n",
 			     req->r_old_dentry, dn);
 
+			/* d_move screws up sibling dentries' offsets */
+			ceph_dir_clear_ordered(dir);
+			ceph_dir_clear_ordered(olddir);
+
 			d_move(req->r_old_dentry, dn);
 			dout(" src %p '%pd' dst %p '%pd'\n",
 			     req->r_old_dentry,
@@ -1222,10 +1230,6 @@ retry_lookup:
 			   rehashing bug in vfs_rename_dir */
 			ceph_invalidate_dentry_lease(dn);
 
-			/* d_move screws up sibling dentries' offsets */
-			ceph_dir_clear_ordered(dir);
-			ceph_dir_clear_ordered(olddir);
-
 			dout("dn %p gets new offset %lld\n", req->r_old_dentry,
 			     ceph_dentry(req->r_old_dentry)->offset);
 
@@ -1333,6 +1337,49 @@ static int readdir_prepopulate_inodes_only(struct ceph_mds_request *req,
 	return err;
 }
 
+void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl)
+{
+	if (ctl->page) {
+		kunmap(ctl->page);
+		page_cache_release(ctl->page);
+		ctl->page = NULL;
+	}
+}
+
+static int fill_readdir_cache(struct inode *dir, struct dentry *dn,
+			      struct ceph_readdir_cache_control *ctl,
+			      struct ceph_mds_request *req)
+{
+	struct ceph_inode_info *ci = ceph_inode(dir);
+	unsigned nsize = PAGE_CACHE_SIZE / sizeof(struct dentry*);
+	unsigned idx = ctl->index % nsize;
+	pgoff_t pgoff = ctl->index / nsize;
+
+	if (!ctl->page || pgoff != page_index(ctl->page)) {
+		ceph_readdir_cache_release(ctl);
+		ctl->page  = grab_cache_page(&dir->i_data, pgoff);
+		if (!ctl->page) {
+			ctl->index = -1;
+			return -ENOMEM;
+		}
+		/* reading/filling the cache are serialized by
+		 * i_mutex, no need to use page lock */
+		unlock_page(ctl->page);
+		ctl->dentries = kmap(ctl->page);
+	}
+
+	if (req->r_dir_release_cnt == atomic64_read(&ci->i_release_count) &&
+	    req->r_dir_ordered_cnt == atomic64_read(&ci->i_ordered_count)) {
+		dout("readdir cache dn %p idx %d\n", dn, ctl->index);
+		ctl->dentries[idx] = dn;
+		ctl->index++;
+	} else {
+		dout("disable readdir cache\n");
+		ctl->index = -1;
+	}
+	return 0;
+}
+
 int ceph_readdir_prepopulate(struct ceph_mds_request *req,
 			     struct ceph_mds_session *session)
 {
@@ -1345,8 +1392,11 @@ int ceph_readdir_prepopulate(struct ceph_mds_request *req,
 	struct inode *snapdir = NULL;
 	struct ceph_mds_request_head *rhead = req->r_request->front.iov_base;
 	struct ceph_dentry_info *di;
-	u64 r_readdir_offset = req->r_readdir_offset;
 	u32 frag = le32_to_cpu(rhead->args.readdir.frag);
+	struct ceph_readdir_cache_control cache_ctl = {};
+
+	if (req->r_aborted)
+		return readdir_prepopulate_inodes_only(req, session);
 
 	if (rinfo->dir_dir &&
 	    le32_to_cpu(rinfo->dir_dir->frag) != frag) {
@@ -1354,14 +1404,11 @@ int ceph_readdir_prepopulate(struct ceph_mds_request *req,
 		     frag, le32_to_cpu(rinfo->dir_dir->frag));
 		frag = le32_to_cpu(rinfo->dir_dir->frag);
 		if (ceph_frag_is_leftmost(frag))
-			r_readdir_offset = 2;
+			req->r_readdir_offset = 2;
 		else
-			r_readdir_offset = 0;
+			req->r_readdir_offset = 0;
 	}
 
-	if (req->r_aborted)
-		return readdir_prepopulate_inodes_only(req, session);
-
 	if (le32_to_cpu(rinfo->head->op) == CEPH_MDS_OP_LSSNAP) {
 		snapdir = ceph_get_snapdir(d_inode(parent));
 		parent = d_find_alias(snapdir);
@@ -1374,6 +1421,17 @@ int ceph_readdir_prepopulate(struct ceph_mds_request *req,
 			ceph_fill_dirfrag(d_inode(parent), rinfo->dir_dir);
 	}
 
+	if (ceph_frag_is_leftmost(frag) && req->r_readdir_offset == 2) {
+		/* note dir version at start of readdir so we can tell
+		 * if any dentries get dropped */
+		struct ceph_inode_info *ci = ceph_inode(d_inode(parent));
+		req->r_dir_release_cnt = atomic64_read(&ci->i_release_count);
+		req->r_dir_ordered_cnt = atomic64_read(&ci->i_ordered_count);
+		req->r_readdir_cache_idx = 0;
+	}
+
+	cache_ctl.index = req->r_readdir_cache_idx;
+
 	/* FIXME: release caps/leases if error occurs */
 	for (i = 0; i < rinfo->dir_nr; i++) {
 		struct ceph_vino vino;
@@ -1413,13 +1471,6 @@ retry_lookup:
 			d_delete(dn);
 			dput(dn);
 			goto retry_lookup;
-		} else {
-			/* reorder parent's d_subdirs */
-			spin_lock(&parent->d_lock);
-			spin_lock_nested(&dn->d_lock, DENTRY_D_LOCK_NESTED);
-			list_move(&dn->d_child, &parent->d_subdirs);
-			spin_unlock(&dn->d_lock);
-			spin_unlock(&parent->d_lock);
 		}
 
 		/* inode */
@@ -1436,13 +1487,15 @@ retry_lookup:
 			}
 		}
 
-		if (fill_inode(in, NULL, &rinfo->dir_in[i], NULL, session,
-			       req->r_request_started, -1,
-			       &req->r_caps_reservation) < 0) {
+		ret = fill_inode(in, NULL, &rinfo->dir_in[i], NULL, session,
+				 req->r_request_started, -1,
+				 &req->r_caps_reservation);
+		if (ret < 0) {
 			pr_err("fill_inode badness on %p\n", in);
 			if (d_really_is_negative(dn))
 				iput(in);
 			d_drop(dn);
+			err = ret;
 			goto next_item;
 		}
 
@@ -1458,19 +1511,28 @@ retry_lookup:
 		}
 
 		di = dn->d_fsdata;
-		di->offset = ceph_make_fpos(frag, i + r_readdir_offset);
+		di->offset = ceph_make_fpos(frag, i + req->r_readdir_offset);
 
 		update_dentry_lease(dn, rinfo->dir_dlease[i],
 				    req->r_session,
 				    req->r_request_started);
+
+		if (err == 0 && cache_ctl.index >= 0) {
+			ret = fill_readdir_cache(d_inode(parent), dn,
+						 &cache_ctl, req);
+			if (ret < 0)
+				err = ret;
+		}
 next_item:
 		if (dn)
 			dput(dn);
 	}
-	if (err == 0)
-		req->r_did_prepopulate = true;
-
 out:
+	if (err == 0) {
+		req->r_did_prepopulate = true;
+		req->r_readdir_cache_idx = cache_ctl.index;
+	}
+	ceph_readdir_cache_release(&cache_ctl);
 	if (snapdir) {
 		iput(snapdir);
 		dput(parent);
@@ -1712,11 +1774,13 @@ int ceph_setattr(struct dentry *dentry, struct iattr *attr)
 	const unsigned int ia_valid = attr->ia_valid;
 	struct ceph_mds_request *req;
 	struct ceph_mds_client *mdsc = ceph_sb_to_client(dentry->d_sb)->mdsc;
+	struct ceph_cap_flush *prealloc_cf;
 	int issued;
 	int release = 0, dirtied = 0;
 	int mask = 0;
 	int err = 0;
 	int inode_dirty_flags = 0;
+	bool lock_snap_rwsem = false;
 
 	if (ceph_snap(inode) != CEPH_NOSNAP)
 		return -EROFS;
@@ -1725,13 +1789,31 @@ int ceph_setattr(struct dentry *dentry, struct iattr *attr)
 	if (err != 0)
 		return err;
 
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		return -ENOMEM;
+
 	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETATTR,
 				       USE_AUTH_MDS);
-	if (IS_ERR(req))
+	if (IS_ERR(req)) {
+		ceph_free_cap_flush(prealloc_cf);
 		return PTR_ERR(req);
+	}
 
 	spin_lock(&ci->i_ceph_lock);
 	issued = __ceph_caps_issued(ci, NULL);
+
+	if (!ci->i_head_snapc &&
+	    (issued & (CEPH_CAP_ANY_EXCL | CEPH_CAP_FILE_WR))) {
+		lock_snap_rwsem = true;
+		if (!down_read_trylock(&mdsc->snap_rwsem)) {
+			spin_unlock(&ci->i_ceph_lock);
+			down_read(&mdsc->snap_rwsem);
+			spin_lock(&ci->i_ceph_lock);
+			issued = __ceph_caps_issued(ci, NULL);
+		}
+	}
+
 	dout("setattr %p issued %s\n", inode, ceph_cap_string(issued));
 
 	if (ia_valid & ATTR_UID) {
@@ -1874,12 +1956,15 @@ int ceph_setattr(struct dentry *dentry, struct iattr *attr)
 		dout("setattr %p ATTR_FILE ... hrm!\n", inode);
 
 	if (dirtied) {
-		inode_dirty_flags = __ceph_mark_dirty_caps(ci, dirtied);
+		inode_dirty_flags = __ceph_mark_dirty_caps(ci, dirtied,
+							   &prealloc_cf);
 		inode->i_ctime = CURRENT_TIME;
 	}
 
 	release &= issued;
 	spin_unlock(&ci->i_ceph_lock);
+	if (lock_snap_rwsem)
+		up_read(&mdsc->snap_rwsem);
 
 	if (inode_dirty_flags)
 		__mark_inode_dirty(inode, inode_dirty_flags);
@@ -1904,9 +1989,11 @@ int ceph_setattr(struct dentry *dentry, struct iattr *attr)
 	ceph_mdsc_put_request(req);
 	if (mask & CEPH_SETATTR_SIZE)
 		__ceph_do_pending_vmtruncate(inode);
+	ceph_free_cap_flush(prealloc_cf);
 	return err;
 out_put:
 	ceph_mdsc_put_request(req);
+	ceph_free_cap_flush(prealloc_cf);
 	return err;
 }
 
diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 84f37f34f9aa..6aa07af67603 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -8,6 +8,7 @@
 #include <linux/debugfs.h>
 #include <linux/seq_file.h>
 #include <linux/utsname.h>
+#include <linux/ratelimit.h>
 
 #include "super.h"
 #include "mds_client.h"
@@ -458,7 +459,6 @@ static struct ceph_mds_session *register_session(struct ceph_mds_client *mdsc,
 	s->s_cap_reconnect = 0;
 	s->s_cap_iterator = NULL;
 	INIT_LIST_HEAD(&s->s_cap_releases);
-	INIT_LIST_HEAD(&s->s_cap_releases_done);
 	INIT_LIST_HEAD(&s->s_cap_flushing);
 	INIT_LIST_HEAD(&s->s_cap_snaps_flushing);
 
@@ -629,6 +629,9 @@ static void __register_request(struct ceph_mds_client *mdsc,
 	req->r_uid = current_fsuid();
 	req->r_gid = current_fsgid();
 
+	if (mdsc->oldest_tid == 0 && req->r_op != CEPH_MDS_OP_SETFILELOCK)
+		mdsc->oldest_tid = req->r_tid;
+
 	if (dir) {
 		struct ceph_inode_info *ci = ceph_inode(dir);
 
@@ -644,6 +647,21 @@ static void __unregister_request(struct ceph_mds_client *mdsc,
 				 struct ceph_mds_request *req)
 {
 	dout("__unregister_request %p tid %lld\n", req, req->r_tid);
+
+	if (req->r_tid == mdsc->oldest_tid) {
+		struct rb_node *p = rb_next(&req->r_node);
+		mdsc->oldest_tid = 0;
+		while (p) {
+			struct ceph_mds_request *next_req =
+				rb_entry(p, struct ceph_mds_request, r_node);
+			if (next_req->r_op != CEPH_MDS_OP_SETFILELOCK) {
+				mdsc->oldest_tid = next_req->r_tid;
+				break;
+			}
+			p = rb_next(p);
+		}
+	}
+
 	rb_erase(&req->r_node, &mdsc->request_tree);
 	RB_CLEAR_NODE(&req->r_node);
 
@@ -998,27 +1016,25 @@ void ceph_mdsc_open_export_target_sessions(struct ceph_mds_client *mdsc,
  * session caps
  */
 
-/*
- * Free preallocated cap messages assigned to this session
- */
-static void cleanup_cap_releases(struct ceph_mds_session *session)
+/* caller holds s_cap_lock, we drop it */
+static void cleanup_cap_releases(struct ceph_mds_client *mdsc,
+				 struct ceph_mds_session *session)
+	__releases(session->s_cap_lock)
 {
-	struct ceph_msg *msg;
+	LIST_HEAD(tmp_list);
+	list_splice_init(&session->s_cap_releases, &tmp_list);
+	session->s_num_cap_releases = 0;
+	spin_unlock(&session->s_cap_lock);
 
-	spin_lock(&session->s_cap_lock);
-	while (!list_empty(&session->s_cap_releases)) {
-		msg = list_first_entry(&session->s_cap_releases,
-				       struct ceph_msg, list_head);
-		list_del_init(&msg->list_head);
-		ceph_msg_put(msg);
-	}
-	while (!list_empty(&session->s_cap_releases_done)) {
-		msg = list_first_entry(&session->s_cap_releases_done,
-				       struct ceph_msg, list_head);
-		list_del_init(&msg->list_head);
-		ceph_msg_put(msg);
+	dout("cleanup_cap_releases mds%d\n", session->s_mds);
+	while (!list_empty(&tmp_list)) {
+		struct ceph_cap *cap;
+		/* zero out the in-progress message */
+		cap = list_first_entry(&tmp_list,
+					struct ceph_cap, session_caps);
+		list_del(&cap->session_caps);
+		ceph_put_cap(mdsc, cap);
 	}
-	spin_unlock(&session->s_cap_lock);
 }
 
 static void cleanup_session_requests(struct ceph_mds_client *mdsc,
@@ -1033,7 +1049,8 @@ static void cleanup_session_requests(struct ceph_mds_client *mdsc,
 		req = list_first_entry(&session->s_unsafe,
 				       struct ceph_mds_request, r_unsafe_item);
 		list_del_init(&req->r_unsafe_item);
-		pr_info(" dropping unsafe request %llu\n", req->r_tid);
+		pr_warn_ratelimited(" dropping unsafe request %llu\n",
+				    req->r_tid);
 		__unregister_request(mdsc, req);
 	}
 	/* zero r_attempts, so kick_requests() will re-send requests */
@@ -1095,10 +1112,16 @@ static int iterate_session_caps(struct ceph_mds_session *session,
 			dout("iterate_session_caps  finishing cap %p removal\n",
 			     cap);
 			BUG_ON(cap->session != session);
+			cap->session = NULL;
 			list_del_init(&cap->session_caps);
 			session->s_nr_caps--;
-			cap->session = NULL;
-			old_cap = cap;  /* put_cap it w/o locks held */
+			if (cap->queue_release) {
+				list_add_tail(&cap->session_caps,
+					      &session->s_cap_releases);
+				session->s_num_cap_releases++;
+			} else {
+				old_cap = cap;  /* put_cap it w/o locks held */
+			}
 		}
 		if (ret < 0)
 			goto out;
@@ -1119,6 +1142,7 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 				  void *arg)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
+	LIST_HEAD(to_remove);
 	int drop = 0;
 
 	dout("removing cap %p, ci is %p, inode is %p\n",
@@ -1126,12 +1150,27 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 	spin_lock(&ci->i_ceph_lock);
 	__ceph_remove_cap(cap, false);
 	if (!ci->i_auth_cap) {
+		struct ceph_cap_flush *cf;
 		struct ceph_mds_client *mdsc =
 			ceph_sb_to_client(inode->i_sb)->mdsc;
 
+		while (true) {
+			struct rb_node *n = rb_first(&ci->i_cap_flush_tree);
+			if (!n)
+				break;
+			cf = rb_entry(n, struct ceph_cap_flush, i_node);
+			rb_erase(&cf->i_node, &ci->i_cap_flush_tree);
+			list_add(&cf->list, &to_remove);
+		}
+
 		spin_lock(&mdsc->cap_dirty_lock);
+
+		list_for_each_entry(cf, &to_remove, list)
+			rb_erase(&cf->g_node, &mdsc->cap_flush_tree);
+
 		if (!list_empty(&ci->i_dirty_item)) {
-			pr_info(" dropping dirty %s state for %p %lld\n",
+			pr_warn_ratelimited(
+				" dropping dirty %s state for %p %lld\n",
 				ceph_cap_string(ci->i_dirty_caps),
 				inode, ceph_ino(inode));
 			ci->i_dirty_caps = 0;
@@ -1139,7 +1178,8 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 			drop = 1;
 		}
 		if (!list_empty(&ci->i_flushing_item)) {
-			pr_info(" dropping dirty+flushing %s state for %p %lld\n",
+			pr_warn_ratelimited(
+				" dropping dirty+flushing %s state for %p %lld\n",
 				ceph_cap_string(ci->i_flushing_caps),
 				inode, ceph_ino(inode));
 			ci->i_flushing_caps = 0;
@@ -1148,8 +1188,20 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 			drop = 1;
 		}
 		spin_unlock(&mdsc->cap_dirty_lock);
+
+		if (!ci->i_dirty_caps && ci->i_prealloc_cap_flush) {
+			list_add(&ci->i_prealloc_cap_flush->list, &to_remove);
+			ci->i_prealloc_cap_flush = NULL;
+		}
 	}
 	spin_unlock(&ci->i_ceph_lock);
+	while (!list_empty(&to_remove)) {
+		struct ceph_cap_flush *cf;
+		cf = list_first_entry(&to_remove,
+				      struct ceph_cap_flush, list);
+		list_del(&cf->list);
+		ceph_free_cap_flush(cf);
+	}
 	while (drop--)
 		iput(inode);
 	return 0;
@@ -1191,11 +1243,12 @@ static void remove_session_caps(struct ceph_mds_session *session)
 			spin_lock(&session->s_cap_lock);
 		}
 	}
-	spin_unlock(&session->s_cap_lock);
+
+	// drop cap expires and unlock s_cap_lock
+	cleanup_cap_releases(session->s_mdsc, session);
 
 	BUG_ON(session->s_nr_caps > 0);
 	BUG_ON(!list_empty(&session->s_cap_flushing));
-	cleanup_cap_releases(session);
 }
 
 /*
@@ -1371,7 +1424,8 @@ static int trim_caps_cb(struct inode *inode, struct ceph_cap *cap, void *arg)
 	     inode, cap, ceph_cap_string(mine), ceph_cap_string(oissued),
 	     ceph_cap_string(used), ceph_cap_string(wanted));
 	if (cap == ci->i_auth_cap) {
-		if (ci->i_dirty_caps | ci->i_flushing_caps)
+		if (ci->i_dirty_caps || ci->i_flushing_caps ||
+		    !list_empty(&ci->i_cap_snaps))
 			goto out;
 		if ((used | wanted) & CEPH_CAP_ANY_WR)
 			goto out;
@@ -1417,121 +1471,80 @@ static int trim_caps(struct ceph_mds_client *mdsc,
 		session->s_trim_caps = 0;
 	}
 
-	ceph_add_cap_releases(mdsc, session);
 	ceph_send_cap_releases(mdsc, session);
 	return 0;
 }
 
-/*
- * Allocate cap_release messages.  If there is a partially full message
- * in the queue, try to allocate enough to cover it's remainder, so that
- * we can send it immediately.
- *
- * Called under s_mutex.
- */
-int ceph_add_cap_releases(struct ceph_mds_client *mdsc,
-			  struct ceph_mds_session *session)
+static int check_capsnap_flush(struct ceph_inode_info *ci,
+			       u64 want_snap_seq)
 {
-	struct ceph_msg *msg, *partial = NULL;
-	struct ceph_mds_cap_release *head;
-	int err = -ENOMEM;
-	int extra = mdsc->fsc->mount_options->cap_release_safety;
-	int num;
-
-	dout("add_cap_releases %p mds%d extra %d\n", session, session->s_mds,
-	     extra);
-
-	spin_lock(&session->s_cap_lock);
-
-	if (!list_empty(&session->s_cap_releases)) {
-		msg = list_first_entry(&session->s_cap_releases,
-				       struct ceph_msg,
-				 list_head);
-		head = msg->front.iov_base;
-		num = le32_to_cpu(head->num);
-		if (num) {
-			dout(" partial %p with (%d/%d)\n", msg, num,
-			     (int)CEPH_CAPS_PER_RELEASE);
-			extra += CEPH_CAPS_PER_RELEASE - num;
-			partial = msg;
-		}
-	}
-	while (session->s_num_cap_releases < session->s_nr_caps + extra) {
-		spin_unlock(&session->s_cap_lock);
-		msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPRELEASE, PAGE_CACHE_SIZE,
-				   GFP_NOFS, false);
-		if (!msg)
-			goto out_unlocked;
-		dout("add_cap_releases %p msg %p now %d\n", session, msg,
-		     (int)msg->front.iov_len);
-		head = msg->front.iov_base;
-		head->num = cpu_to_le32(0);
-		msg->front.iov_len = sizeof(*head);
-		spin_lock(&session->s_cap_lock);
-		list_add(&msg->list_head, &session->s_cap_releases);
-		session->s_num_cap_releases += CEPH_CAPS_PER_RELEASE;
-	}
-
-	if (partial) {
-		head = partial->front.iov_base;
-		num = le32_to_cpu(head->num);
-		dout(" queueing partial %p with %d/%d\n", partial, num,
-		     (int)CEPH_CAPS_PER_RELEASE);
-		list_move_tail(&partial->list_head,
-			       &session->s_cap_releases_done);
-		session->s_num_cap_releases -= CEPH_CAPS_PER_RELEASE - num;
+	int ret = 1;
+	spin_lock(&ci->i_ceph_lock);
+	if (want_snap_seq > 0 && !list_empty(&ci->i_cap_snaps)) {
+		struct ceph_cap_snap *capsnap =
+			list_first_entry(&ci->i_cap_snaps,
+					 struct ceph_cap_snap, ci_item);
+		ret = capsnap->follows >= want_snap_seq;
 	}
-	err = 0;
-	spin_unlock(&session->s_cap_lock);
-out_unlocked:
-	return err;
+	spin_unlock(&ci->i_ceph_lock);
+	return ret;
 }
 
-static int check_cap_flush(struct inode *inode, u64 want_flush_seq)
+static int check_caps_flush(struct ceph_mds_client *mdsc,
+			    u64 want_flush_tid)
 {
-	struct ceph_inode_info *ci = ceph_inode(inode);
-	int ret;
-	spin_lock(&ci->i_ceph_lock);
-	if (ci->i_flushing_caps)
-		ret = ci->i_cap_flush_seq >= want_flush_seq;
-	else
-		ret = 1;
-	spin_unlock(&ci->i_ceph_lock);
+	struct rb_node *n;
+	struct ceph_cap_flush *cf;
+	int ret = 1;
+
+	spin_lock(&mdsc->cap_dirty_lock);
+	n = rb_first(&mdsc->cap_flush_tree);
+	cf = n ? rb_entry(n, struct ceph_cap_flush, g_node) : NULL;
+	if (cf && cf->tid <= want_flush_tid) {
+		dout("check_caps_flush still flushing tid %llu <= %llu\n",
+		     cf->tid, want_flush_tid);
+		ret = 0;
+	}
+	spin_unlock(&mdsc->cap_dirty_lock);
 	return ret;
 }
 
 /*
  * flush all dirty inode data to disk.
  *
- * returns true if we've flushed through want_flush_seq
+ * returns true if we've flushed through want_flush_tid
  */
-static void wait_caps_flush(struct ceph_mds_client *mdsc, u64 want_flush_seq)
+static void wait_caps_flush(struct ceph_mds_client *mdsc,
+			    u64 want_flush_tid, u64 want_snap_seq)
 {
 	int mds;
 
-	dout("check_cap_flush want %lld\n", want_flush_seq);
+	dout("check_caps_flush want %llu snap want %llu\n",
+	     want_flush_tid, want_snap_seq);
 	mutex_lock(&mdsc->mutex);
-	for (mds = 0; mds < mdsc->max_sessions; mds++) {
+	for (mds = 0; mds < mdsc->max_sessions; ) {
 		struct ceph_mds_session *session = mdsc->sessions[mds];
 		struct inode *inode = NULL;
 
-		if (!session)
+		if (!session) {
+			mds++;
 			continue;
+		}
 		get_session(session);
 		mutex_unlock(&mdsc->mutex);
 
 		mutex_lock(&session->s_mutex);
-		if (!list_empty(&session->s_cap_flushing)) {
-			struct ceph_inode_info *ci =
-				list_entry(session->s_cap_flushing.next,
-					   struct ceph_inode_info,
-					   i_flushing_item);
-
-			if (!check_cap_flush(&ci->vfs_inode, want_flush_seq)) {
-				dout("check_cap_flush still flushing %p "
-				     "seq %lld <= %lld to mds%d\n",
-				     &ci->vfs_inode, ci->i_cap_flush_seq,
-				     want_flush_seq, session->s_mds);
+		if (!list_empty(&session->s_cap_snaps_flushing)) {
+			struct ceph_cap_snap *capsnap =
+				list_first_entry(&session->s_cap_snaps_flushing,
+						 struct ceph_cap_snap,
+						 flushing_item);
+			struct ceph_inode_info *ci = capsnap->ci;
+			if (!check_capsnap_flush(ci, want_snap_seq)) {
+				dout("check_cap_flush still flushing snap %p "
+				     "follows %lld <= %lld to mds%d\n",
+				     &ci->vfs_inode, capsnap->follows,
+				     want_snap_seq, mds);
 				inode = igrab(&ci->vfs_inode);
 			}
 		}
@@ -1540,15 +1553,21 @@ static void wait_caps_flush(struct ceph_mds_client *mdsc, u64 want_flush_seq)
 
 		if (inode) {
 			wait_event(mdsc->cap_flushing_wq,
-				   check_cap_flush(inode, want_flush_seq));
+				   check_capsnap_flush(ceph_inode(inode),
+						       want_snap_seq));
 			iput(inode);
+		} else {
+			mds++;
 		}
 
 		mutex_lock(&mdsc->mutex);
 	}
-
 	mutex_unlock(&mdsc->mutex);
-	dout("check_cap_flush ok, flushed thru %lld\n", want_flush_seq);
+
+	wait_event(mdsc->cap_flushing_wq,
+		   check_caps_flush(mdsc, want_flush_tid));
+
+	dout("check_caps_flush ok, flushed thru %llu\n", want_flush_tid);
 }
 
 /*
@@ -1557,60 +1576,74 @@ static void wait_caps_flush(struct ceph_mds_client *mdsc, u64 want_flush_seq)
 void ceph_send_cap_releases(struct ceph_mds_client *mdsc,
 			    struct ceph_mds_session *session)
 {
-	struct ceph_msg *msg;
+	struct ceph_msg *msg = NULL;
+	struct ceph_mds_cap_release *head;
+	struct ceph_mds_cap_item *item;
+	struct ceph_cap *cap;
+	LIST_HEAD(tmp_list);
+	int num_cap_releases;
 
-	dout("send_cap_releases mds%d\n", session->s_mds);
 	spin_lock(&session->s_cap_lock);
-	while (!list_empty(&session->s_cap_releases_done)) {
-		msg = list_first_entry(&session->s_cap_releases_done,
-				 struct ceph_msg, list_head);
-		list_del_init(&msg->list_head);
-		spin_unlock(&session->s_cap_lock);
-		msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
-		dout("send_cap_releases mds%d %p\n", session->s_mds, msg);
-		ceph_con_send(&session->s_con, msg);
-		spin_lock(&session->s_cap_lock);
-	}
+again:
+	list_splice_init(&session->s_cap_releases, &tmp_list);
+	num_cap_releases = session->s_num_cap_releases;
+	session->s_num_cap_releases = 0;
 	spin_unlock(&session->s_cap_lock);
-}
 
-static void discard_cap_releases(struct ceph_mds_client *mdsc,
-				 struct ceph_mds_session *session)
-{
-	struct ceph_msg *msg;
-	struct ceph_mds_cap_release *head;
-	unsigned num;
-
-	dout("discard_cap_releases mds%d\n", session->s_mds);
+	while (!list_empty(&tmp_list)) {
+		if (!msg) {
+			msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPRELEASE,
+					PAGE_CACHE_SIZE, GFP_NOFS, false);
+			if (!msg)
+				goto out_err;
+			head = msg->front.iov_base;
+			head->num = cpu_to_le32(0);
+			msg->front.iov_len = sizeof(*head);
+		}
+		cap = list_first_entry(&tmp_list, struct ceph_cap,
+					session_caps);
+		list_del(&cap->session_caps);
+		num_cap_releases--;
 
-	if (!list_empty(&session->s_cap_releases)) {
-		/* zero out the in-progress message */
-		msg = list_first_entry(&session->s_cap_releases,
-					struct ceph_msg, list_head);
 		head = msg->front.iov_base;
-		num = le32_to_cpu(head->num);
-		dout("discard_cap_releases mds%d %p %u\n",
-		     session->s_mds, msg, num);
-		head->num = cpu_to_le32(0);
-		msg->front.iov_len = sizeof(*head);
-		session->s_num_cap_releases += num;
+		le32_add_cpu(&head->num, 1);
+		item = msg->front.iov_base + msg->front.iov_len;
+		item->ino = cpu_to_le64(cap->cap_ino);
+		item->cap_id = cpu_to_le64(cap->cap_id);
+		item->migrate_seq = cpu_to_le32(cap->mseq);
+		item->seq = cpu_to_le32(cap->issue_seq);
+		msg->front.iov_len += sizeof(*item);
+
+		ceph_put_cap(mdsc, cap);
+
+		if (le32_to_cpu(head->num) == CEPH_CAPS_PER_RELEASE) {
+			msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
+			dout("send_cap_releases mds%d %p\n", session->s_mds, msg);
+			ceph_con_send(&session->s_con, msg);
+			msg = NULL;
+		}
 	}
 
-	/* requeue completed messages */
-	while (!list_empty(&session->s_cap_releases_done)) {
-		msg = list_first_entry(&session->s_cap_releases_done,
-				 struct ceph_msg, list_head);
-		list_del_init(&msg->list_head);
+	BUG_ON(num_cap_releases != 0);
 
-		head = msg->front.iov_base;
-		num = le32_to_cpu(head->num);
-		dout("discard_cap_releases mds%d %p %u\n", session->s_mds, msg,
-		     num);
-		session->s_num_cap_releases += num;
-		head->num = cpu_to_le32(0);
-		msg->front.iov_len = sizeof(*head);
-		list_add(&msg->list_head, &session->s_cap_releases);
+	spin_lock(&session->s_cap_lock);
+	if (!list_empty(&session->s_cap_releases))
+		goto again;
+	spin_unlock(&session->s_cap_lock);
+
+	if (msg) {
+		msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
+		dout("send_cap_releases mds%d %p\n", session->s_mds, msg);
+		ceph_con_send(&session->s_con, msg);
 	}
+	return;
+out_err:
+	pr_err("send_cap_releases mds%d, failed to allocate message\n",
+		session->s_mds);
+	spin_lock(&session->s_cap_lock);
+	list_splice(&tmp_list, &session->s_cap_releases);
+	session->s_num_cap_releases += num_cap_releases;
+	spin_unlock(&session->s_cap_lock);
 }
 
 /*
@@ -1635,7 +1668,8 @@ int ceph_alloc_readdir_reply_buffer(struct ceph_mds_request *req,
 
 	order = get_order(size * num_entries);
 	while (order >= 0) {
-		rinfo->dir_in = (void*)__get_free_pages(GFP_NOFS | __GFP_NOWARN,
+		rinfo->dir_in = (void*)__get_free_pages(GFP_KERNEL |
+							__GFP_NOWARN,
 							order);
 		if (rinfo->dir_in)
 			break;
@@ -1697,13 +1731,9 @@ static struct ceph_mds_request *__get_oldest_req(struct ceph_mds_client *mdsc)
 			struct ceph_mds_request, r_node);
 }
 
-static u64 __get_oldest_tid(struct ceph_mds_client *mdsc)
+static inline  u64 __get_oldest_tid(struct ceph_mds_client *mdsc)
 {
-	struct ceph_mds_request *req = __get_oldest_req(mdsc);
-
-	if (req)
-		return req->r_tid;
-	return 0;
+	return mdsc->oldest_tid;
 }
 
 /*
@@ -2267,15 +2297,18 @@ int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
 	/* wait */
 	mutex_unlock(&mdsc->mutex);
 	dout("do_request waiting\n");
-	if (req->r_timeout) {
-		err = (long)wait_for_completion_killable_timeout(
-			&req->r_completion, req->r_timeout);
-		if (err == 0)
-			err = -EIO;
-	} else if (req->r_wait_for_completion) {
+	if (!req->r_timeout && req->r_wait_for_completion) {
 		err = req->r_wait_for_completion(mdsc, req);
 	} else {
-		err = wait_for_completion_killable(&req->r_completion);
+		long timeleft = wait_for_completion_killable_timeout(
+					&req->r_completion,
+					ceph_timeout_jiffies(req->r_timeout));
+		if (timeleft > 0)
+			err = 0;
+		else if (!timeleft)
+			err = -EIO;  /* timed out */
+		else
+			err = timeleft;  /* killed */
 	}
 	dout("do_request waited, got %d\n", err);
 	mutex_lock(&mdsc->mutex);
@@ -2496,7 +2529,6 @@ out_err:
 	}
 	mutex_unlock(&mdsc->mutex);
 
-	ceph_add_cap_releases(mdsc, req->r_session);
 	mutex_unlock(&session->s_mutex);
 
 	/* kick calling process */
@@ -2888,8 +2920,7 @@ static void send_mds_reconnect(struct ceph_mds_client *mdsc,
 	 */
 	session->s_cap_reconnect = 1;
 	/* drop old cap expires; we're about to reestablish that state */
-	discard_cap_releases(mdsc, session);
-	spin_unlock(&session->s_cap_lock);
+	cleanup_cap_releases(mdsc, session);
 
 	/* trim unused caps to reduce MDS's cache rejoin time */
 	if (mdsc->fsc->sb->s_root)
@@ -2956,6 +2987,9 @@ static void send_mds_reconnect(struct ceph_mds_client *mdsc,
 
 	reply->hdr.data_len = cpu_to_le32(pagelist->length);
 	ceph_msg_data_add_pagelist(reply, pagelist);
+
+	ceph_early_kick_flushing_caps(mdsc, session);
+
 	ceph_con_send(&session->s_con, reply);
 
 	mutex_unlock(&session->s_mutex);
@@ -3352,7 +3386,6 @@ static void delayed_work(struct work_struct *work)
 			send_renew_caps(mdsc, s);
 		else
 			ceph_con_keepalive(&s->s_con);
-		ceph_add_cap_releases(mdsc, s);
 		if (s->s_state == CEPH_MDS_SESSION_OPEN ||
 		    s->s_state == CEPH_MDS_SESSION_HUNG)
 			ceph_send_cap_releases(mdsc, s);
@@ -3390,11 +3423,13 @@ int ceph_mdsc_init(struct ceph_fs_client *fsc)
 	atomic_set(&mdsc->num_sessions, 0);
 	mdsc->max_sessions = 0;
 	mdsc->stopping = 0;
+	mdsc->last_snap_seq = 0;
 	init_rwsem(&mdsc->snap_rwsem);
 	mdsc->snap_realms = RB_ROOT;
 	INIT_LIST_HEAD(&mdsc->snap_empty);
 	spin_lock_init(&mdsc->snap_empty_lock);
 	mdsc->last_tid = 0;
+	mdsc->oldest_tid = 0;
 	mdsc->request_tree = RB_ROOT;
 	INIT_DELAYED_WORK(&mdsc->delayed_work, delayed_work);
 	mdsc->last_renew_caps = jiffies;
@@ -3402,7 +3437,8 @@ int ceph_mdsc_init(struct ceph_fs_client *fsc)
 	spin_lock_init(&mdsc->cap_delay_lock);
 	INIT_LIST_HEAD(&mdsc->snap_flush_list);
 	spin_lock_init(&mdsc->snap_flush_lock);
-	mdsc->cap_flush_seq = 0;
+	mdsc->last_cap_flush_tid = 1;
+	mdsc->cap_flush_tree = RB_ROOT;
 	INIT_LIST_HEAD(&mdsc->cap_dirty);
 	INIT_LIST_HEAD(&mdsc->cap_dirty_migrating);
 	mdsc->num_cap_flushing = 0;
@@ -3414,6 +3450,9 @@ int ceph_mdsc_init(struct ceph_fs_client *fsc)
 	ceph_caps_init(mdsc);
 	ceph_adjust_min_caps(mdsc, fsc->min_caps);
 
+	init_rwsem(&mdsc->pool_perm_rwsem);
+	mdsc->pool_perm_tree = RB_ROOT;
+
 	return 0;
 }
 
@@ -3423,8 +3462,8 @@ int ceph_mdsc_init(struct ceph_fs_client *fsc)
  */
 static void wait_requests(struct ceph_mds_client *mdsc)
 {
+	struct ceph_options *opts = mdsc->fsc->client->options;
 	struct ceph_mds_request *req;
-	struct ceph_fs_client *fsc = mdsc->fsc;
 
 	mutex_lock(&mdsc->mutex);
 	if (__get_oldest_req(mdsc)) {
@@ -3432,7 +3471,7 @@ static void wait_requests(struct ceph_mds_client *mdsc)
 
 		dout("wait_requests waiting for requests\n");
 		wait_for_completion_timeout(&mdsc->safe_umount_waiters,
-				    fsc->client->options->mount_timeout * HZ);
+				    ceph_timeout_jiffies(opts->mount_timeout));
 
 		/* tear down remaining requests */
 		mutex_lock(&mdsc->mutex);
@@ -3485,7 +3524,8 @@ restart:
 			nextreq = rb_entry(n, struct ceph_mds_request, r_node);
 		else
 			nextreq = NULL;
-		if ((req->r_op & CEPH_MDS_OP_WRITE)) {
+		if (req->r_op != CEPH_MDS_OP_SETFILELOCK &&
+		    (req->r_op & CEPH_MDS_OP_WRITE)) {
 			/* write op */
 			ceph_mdsc_get_request(req);
 			if (nextreq)
@@ -3513,7 +3553,7 @@ restart:
 
 void ceph_mdsc_sync(struct ceph_mds_client *mdsc)
 {
-	u64 want_tid, want_flush;
+	u64 want_tid, want_flush, want_snap;
 
 	if (mdsc->fsc->mount_state == CEPH_MOUNT_SHUTDOWN)
 		return;
@@ -3525,13 +3565,18 @@ void ceph_mdsc_sync(struct ceph_mds_client *mdsc)
 
 	ceph_flush_dirty_caps(mdsc);
 	spin_lock(&mdsc->cap_dirty_lock);
-	want_flush = mdsc->cap_flush_seq;
+	want_flush = mdsc->last_cap_flush_tid;
 	spin_unlock(&mdsc->cap_dirty_lock);
 
-	dout("sync want tid %lld flush_seq %lld\n", want_tid, want_flush);
+	down_read(&mdsc->snap_rwsem);
+	want_snap = mdsc->last_snap_seq;
+	up_read(&mdsc->snap_rwsem);
+
+	dout("sync want tid %lld flush_seq %lld snap_seq %lld\n",
+	     want_tid, want_flush, want_snap);
 
 	wait_unsafe_requests(mdsc, want_tid);
-	wait_caps_flush(mdsc, want_flush);
+	wait_caps_flush(mdsc, want_flush, want_snap);
 }
 
 /*
@@ -3549,10 +3594,9 @@ static bool done_closing_sessions(struct ceph_mds_client *mdsc)
  */
 void ceph_mdsc_close_sessions(struct ceph_mds_client *mdsc)
 {
+	struct ceph_options *opts = mdsc->fsc->client->options;
 	struct ceph_mds_session *session;
 	int i;
-	struct ceph_fs_client *fsc = mdsc->fsc;
-	unsigned long timeout = fsc->client->options->mount_timeout * HZ;
 
 	dout("close_sessions\n");
 
@@ -3573,7 +3617,7 @@ void ceph_mdsc_close_sessions(struct ceph_mds_client *mdsc)
 
 	dout("waiting for sessions to close\n");
 	wait_event_timeout(mdsc->session_close_wq, done_closing_sessions(mdsc),
-			   timeout);
+			   ceph_timeout_jiffies(opts->mount_timeout));
 
 	/* tear down remaining sessions */
 	mutex_lock(&mdsc->mutex);
@@ -3607,6 +3651,7 @@ static void ceph_mdsc_stop(struct ceph_mds_client *mdsc)
 		ceph_mdsmap_destroy(mdsc->mdsmap);
 	kfree(mdsc->sessions);
 	ceph_caps_finalize(mdsc);
+	ceph_pool_perm_destroy(mdsc);
 }
 
 void ceph_mdsc_destroy(struct ceph_fs_client *fsc)
diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h
index 1875b5d985c6..762757e6cebf 100644
--- a/fs/ceph/mds_client.h
+++ b/fs/ceph/mds_client.h
@@ -139,7 +139,6 @@ struct ceph_mds_session {
 	int		  s_cap_reconnect;
 	int		  s_readonly;
 	struct list_head  s_cap_releases; /* waiting cap_release messages */
-	struct list_head  s_cap_releases_done; /* ready to send */
 	struct ceph_cap  *s_cap_iterator;
 
 	/* protected by mutex */
@@ -228,7 +227,7 @@ struct ceph_mds_request {
 	int r_err;
 	bool r_aborted;
 
-	unsigned long r_timeout;  /* optional.  jiffies */
+	unsigned long r_timeout;  /* optional.  jiffies, 0 is "wait forever" */
 	unsigned long r_started;  /* start time to measure timeout against */
 	unsigned long r_request_started; /* start time for mds request only,
 					    used to measure lease durations */
@@ -254,12 +253,21 @@ struct ceph_mds_request {
 	bool		  r_got_unsafe, r_got_safe, r_got_result;
 
 	bool              r_did_prepopulate;
+	long long	  r_dir_release_cnt;
+	long long	  r_dir_ordered_cnt;
+	int		  r_readdir_cache_idx;
 	u32               r_readdir_offset;
 
 	struct ceph_cap_reservation r_caps_reservation;
 	int r_num_caps;
 };
 
+struct ceph_pool_perm {
+	struct rb_node node;
+	u32 pool;
+	int perm;
+};
+
 /*
  * mds client state
  */
@@ -284,12 +292,15 @@ struct ceph_mds_client {
 	 * references (implying they contain no inodes with caps) that
 	 * should be destroyed.
 	 */
+	u64			last_snap_seq;
 	struct rw_semaphore     snap_rwsem;
 	struct rb_root          snap_realms;
 	struct list_head        snap_empty;
 	spinlock_t              snap_empty_lock;  /* protect snap_empty */
 
 	u64                    last_tid;      /* most recent mds request */
+	u64                    oldest_tid;    /* oldest incomplete mds request,
+						 excluding setfilelock requests */
 	struct rb_root         request_tree;  /* pending mds requests */
 	struct delayed_work    delayed_work;  /* delayed work */
 	unsigned long    last_renew_caps;  /* last time we renewed our caps */
@@ -298,7 +309,8 @@ struct ceph_mds_client {
 	struct list_head snap_flush_list;  /* cap_snaps ready to flush */
 	spinlock_t       snap_flush_lock;
 
-	u64               cap_flush_seq;
+	u64               last_cap_flush_tid;
+	struct rb_root    cap_flush_tree;
 	struct list_head  cap_dirty;        /* inodes with dirty caps */
 	struct list_head  cap_dirty_migrating; /* ...that are migration... */
 	int               num_cap_flushing; /* # caps we are flushing */
@@ -328,6 +340,9 @@ struct ceph_mds_client {
 	spinlock_t	  dentry_lru_lock;
 	struct list_head  dentry_lru;
 	int		  num_dentry;
+
+	struct rw_semaphore     pool_perm_rwsem;
+	struct rb_root		pool_perm_tree;
 };
 
 extern const char *ceph_mds_op_name(int op);
@@ -379,8 +394,6 @@ static inline void ceph_mdsc_put_request(struct ceph_mds_request *req)
 	kref_put(&req->r_kref, ceph_mdsc_release_request);
 }
 
-extern int ceph_add_cap_releases(struct ceph_mds_client *mdsc,
-				 struct ceph_mds_session *session);
 extern void ceph_send_cap_releases(struct ceph_mds_client *mdsc,
 				   struct ceph_mds_session *session);
 
diff --git a/fs/ceph/snap.c b/fs/ceph/snap.c
index a97e39f09ba6..233d906aec02 100644
--- a/fs/ceph/snap.c
+++ b/fs/ceph/snap.c
@@ -296,7 +296,7 @@ static int cmpu64_rev(const void *a, const void *b)
 }
 
 
-static struct ceph_snap_context *empty_snapc;
+struct ceph_snap_context *ceph_empty_snapc;
 
 /*
  * build the snap context for a given realm.
@@ -338,9 +338,9 @@ static int build_snap_context(struct ceph_snap_realm *realm)
 		return 0;
 	}
 
-	if (num == 0 && realm->seq == empty_snapc->seq) {
-		ceph_get_snap_context(empty_snapc);
-		snapc = empty_snapc;
+	if (num == 0 && realm->seq == ceph_empty_snapc->seq) {
+		ceph_get_snap_context(ceph_empty_snapc);
+		snapc = ceph_empty_snapc;
 		goto done;
 	}
 
@@ -436,6 +436,14 @@ static int dup_array(u64 **dst, __le64 *src, u32 num)
 	return 0;
 }
 
+static bool has_new_snaps(struct ceph_snap_context *o,
+			  struct ceph_snap_context *n)
+{
+	if (n->num_snaps == 0)
+		return false;
+	/* snaps are in descending order */
+	return n->snaps[0] > o->seq;
+}
 
 /*
  * When a snapshot is applied, the size/mtime inode metadata is queued
@@ -455,6 +463,7 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci)
 {
 	struct inode *inode = &ci->vfs_inode;
 	struct ceph_cap_snap *capsnap;
+	struct ceph_snap_context *old_snapc, *new_snapc;
 	int used, dirty;
 
 	capsnap = kzalloc(sizeof(*capsnap), GFP_NOFS);
@@ -467,6 +476,9 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci)
 	used = __ceph_caps_used(ci);
 	dirty = __ceph_caps_dirty(ci);
 
+	old_snapc = ci->i_head_snapc;
+	new_snapc = ci->i_snap_realm->cached_context;
+
 	/*
 	 * If there is a write in progress, treat that as a dirty Fw,
 	 * even though it hasn't completed yet; by the time we finish
@@ -481,76 +493,95 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci)
 		   writes in progress now were started before the previous
 		   cap_snap.  lucky us. */
 		dout("queue_cap_snap %p already pending\n", inode);
-		kfree(capsnap);
-	} else if (ci->i_snap_realm->cached_context == empty_snapc) {
-		dout("queue_cap_snap %p empty snapc\n", inode);
-		kfree(capsnap);
-	} else if (dirty & (CEPH_CAP_AUTH_EXCL|CEPH_CAP_XATTR_EXCL|
-			    CEPH_CAP_FILE_EXCL|CEPH_CAP_FILE_WR)) {
-		struct ceph_snap_context *snapc = ci->i_head_snapc;
-
-		/*
-		 * if we are a sync write, we may need to go to the snaprealm
-		 * to get the current snapc.
-		 */
-		if (!snapc)
-			snapc = ci->i_snap_realm->cached_context;
+		goto update_snapc;
+	}
+	if (ci->i_wrbuffer_ref_head == 0 &&
+	    !(dirty & (CEPH_CAP_ANY_EXCL|CEPH_CAP_FILE_WR))) {
+		dout("queue_cap_snap %p nothing dirty|writing\n", inode);
+		goto update_snapc;
+	}
 
-		dout("queue_cap_snap %p cap_snap %p queuing under %p %s\n",
-		     inode, capsnap, snapc, ceph_cap_string(dirty));
-		ihold(inode);
+	BUG_ON(!old_snapc);
 
-		atomic_set(&capsnap->nref, 1);
-		capsnap->ci = ci;
-		INIT_LIST_HEAD(&capsnap->ci_item);
-		INIT_LIST_HEAD(&capsnap->flushing_item);
-
-		capsnap->follows = snapc->seq;
-		capsnap->issued = __ceph_caps_issued(ci, NULL);
-		capsnap->dirty = dirty;
-
-		capsnap->mode = inode->i_mode;
-		capsnap->uid = inode->i_uid;
-		capsnap->gid = inode->i_gid;
-
-		if (dirty & CEPH_CAP_XATTR_EXCL) {
-			__ceph_build_xattrs_blob(ci);
-			capsnap->xattr_blob =
-				ceph_buffer_get(ci->i_xattrs.blob);
-			capsnap->xattr_version = ci->i_xattrs.version;
-		} else {
-			capsnap->xattr_blob = NULL;
-			capsnap->xattr_version = 0;
+	/*
+	 * There is no need to send FLUSHSNAP message to MDS if there is
+	 * no new snapshot. But when there is dirty pages or on-going
+	 * writes, we still need to create cap_snap. cap_snap is needed
+	 * by the write path and page writeback path.
+	 *
+	 * also see ceph_try_drop_cap_snap()
+	 */
+	if (has_new_snaps(old_snapc, new_snapc)) {
+		if (dirty & (CEPH_CAP_ANY_EXCL|CEPH_CAP_FILE_WR))
+			capsnap->need_flush = true;
+	} else {
+		if (!(used & CEPH_CAP_FILE_WR) &&
+		    ci->i_wrbuffer_ref_head == 0) {
+			dout("queue_cap_snap %p "
+			     "no new_snap|dirty_page|writing\n", inode);
+			goto update_snapc;
 		}
+	}
 
-		capsnap->inline_data = ci->i_inline_version != CEPH_INLINE_NONE;
-
-		/* dirty page count moved from _head to this cap_snap;
-		   all subsequent writes page dirties occur _after_ this
-		   snapshot. */
-		capsnap->dirty_pages = ci->i_wrbuffer_ref_head;
-		ci->i_wrbuffer_ref_head = 0;
-		capsnap->context = snapc;
-		ci->i_head_snapc =
-			ceph_get_snap_context(ci->i_snap_realm->cached_context);
-		dout(" new snapc is %p\n", ci->i_head_snapc);
-		list_add_tail(&capsnap->ci_item, &ci->i_cap_snaps);
-
-		if (used & CEPH_CAP_FILE_WR) {
-			dout("queue_cap_snap %p cap_snap %p snapc %p"
-			     " seq %llu used WR, now pending\n", inode,
-			     capsnap, snapc, snapc->seq);
-			capsnap->writing = 1;
-		} else {
-			/* note mtime, size NOW. */
-			__ceph_finish_cap_snap(ci, capsnap);
-		}
+	dout("queue_cap_snap %p cap_snap %p queuing under %p %s %s\n",
+	     inode, capsnap, old_snapc, ceph_cap_string(dirty),
+	     capsnap->need_flush ? "" : "no_flush");
+	ihold(inode);
+
+	atomic_set(&capsnap->nref, 1);
+	capsnap->ci = ci;
+	INIT_LIST_HEAD(&capsnap->ci_item);
+	INIT_LIST_HEAD(&capsnap->flushing_item);
+
+	capsnap->follows = old_snapc->seq;
+	capsnap->issued = __ceph_caps_issued(ci, NULL);
+	capsnap->dirty = dirty;
+
+	capsnap->mode = inode->i_mode;
+	capsnap->uid = inode->i_uid;
+	capsnap->gid = inode->i_gid;
+
+	if (dirty & CEPH_CAP_XATTR_EXCL) {
+		__ceph_build_xattrs_blob(ci);
+		capsnap->xattr_blob =
+			ceph_buffer_get(ci->i_xattrs.blob);
+		capsnap->xattr_version = ci->i_xattrs.version;
 	} else {
-		dout("queue_cap_snap %p nothing dirty|writing\n", inode);
-		kfree(capsnap);
+		capsnap->xattr_blob = NULL;
+		capsnap->xattr_version = 0;
 	}
 
+	capsnap->inline_data = ci->i_inline_version != CEPH_INLINE_NONE;
+
+	/* dirty page count moved from _head to this cap_snap;
+	   all subsequent writes page dirties occur _after_ this
+	   snapshot. */
+	capsnap->dirty_pages = ci->i_wrbuffer_ref_head;
+	ci->i_wrbuffer_ref_head = 0;
+	capsnap->context = old_snapc;
+	list_add_tail(&capsnap->ci_item, &ci->i_cap_snaps);
+	old_snapc = NULL;
+
+	if (used & CEPH_CAP_FILE_WR) {
+		dout("queue_cap_snap %p cap_snap %p snapc %p"
+		     " seq %llu used WR, now pending\n", inode,
+		     capsnap, old_snapc, old_snapc->seq);
+		capsnap->writing = 1;
+	} else {
+		/* note mtime, size NOW. */
+		__ceph_finish_cap_snap(ci, capsnap);
+	}
+	capsnap = NULL;
+
+update_snapc:
+	if (ci->i_head_snapc) {
+		ci->i_head_snapc = ceph_get_snap_context(new_snapc);
+		dout(" new snapc is %p\n", new_snapc);
+	}
 	spin_unlock(&ci->i_ceph_lock);
+
+	kfree(capsnap);
+	ceph_put_snap_context(old_snapc);
 }
 
 /*
@@ -699,6 +730,8 @@ more:
 
 		/* queue realm for cap_snap creation */
 		list_add(&realm->dirty_item, &dirty_realms);
+		if (realm->seq > mdsc->last_snap_seq)
+			mdsc->last_snap_seq = realm->seq;
 
 		invalidate = 1;
 	} else if (!realm->cached_context) {
@@ -964,14 +997,14 @@ out:
 
 int __init ceph_snap_init(void)
 {
-	empty_snapc = ceph_create_snap_context(0, GFP_NOFS);
-	if (!empty_snapc)
+	ceph_empty_snapc = ceph_create_snap_context(0, GFP_NOFS);
+	if (!ceph_empty_snapc)
 		return -ENOMEM;
-	empty_snapc->seq = 1;
+	ceph_empty_snapc->seq = 1;
 	return 0;
 }
 
 void ceph_snap_exit(void)
 {
-	ceph_put_snap_context(empty_snapc);
+	ceph_put_snap_context(ceph_empty_snapc);
 }
diff --git a/fs/ceph/super.c b/fs/ceph/super.c
index 4e9905374078..d1c833c321b9 100644
--- a/fs/ceph/super.c
+++ b/fs/ceph/super.c
@@ -134,10 +134,12 @@ enum {
 	Opt_noino32,
 	Opt_fscache,
 	Opt_nofscache,
+	Opt_poolperm,
+	Opt_nopoolperm,
 #ifdef CONFIG_CEPH_FS_POSIX_ACL
 	Opt_acl,
 #endif
-	Opt_noacl
+	Opt_noacl,
 };
 
 static match_table_t fsopt_tokens = {
@@ -165,6 +167,8 @@ static match_table_t fsopt_tokens = {
 	{Opt_noino32, "noino32"},
 	{Opt_fscache, "fsc"},
 	{Opt_nofscache, "nofsc"},
+	{Opt_poolperm, "poolperm"},
+	{Opt_nopoolperm, "nopoolperm"},
 #ifdef CONFIG_CEPH_FS_POSIX_ACL
 	{Opt_acl, "acl"},
 #endif
@@ -268,6 +272,13 @@ static int parse_fsopt_token(char *c, void *private)
 	case Opt_nofscache:
 		fsopt->flags &= ~CEPH_MOUNT_OPT_FSCACHE;
 		break;
+	case Opt_poolperm:
+		fsopt->flags &= ~CEPH_MOUNT_OPT_NOPOOLPERM;
+		printk ("pool perm");
+		break;
+	case Opt_nopoolperm:
+		fsopt->flags |= CEPH_MOUNT_OPT_NOPOOLPERM;
+		break;
 #ifdef CONFIG_CEPH_FS_POSIX_ACL
 	case Opt_acl:
 		fsopt->sb_flags |= MS_POSIXACL;
@@ -436,6 +447,8 @@ static int ceph_show_options(struct seq_file *m, struct dentry *root)
 		seq_puts(m, ",nodcache");
 	if (fsopt->flags & CEPH_MOUNT_OPT_FSCACHE)
 		seq_puts(m, ",fsc");
+	if (fsopt->flags & CEPH_MOUNT_OPT_NOPOOLPERM)
+		seq_puts(m, ",nopoolperm");
 
 #ifdef CONFIG_CEPH_FS_POSIX_ACL
 	if (fsopt->sb_flags & MS_POSIXACL)
@@ -609,6 +622,7 @@ static void destroy_fs_client(struct ceph_fs_client *fsc)
  */
 struct kmem_cache *ceph_inode_cachep;
 struct kmem_cache *ceph_cap_cachep;
+struct kmem_cache *ceph_cap_flush_cachep;
 struct kmem_cache *ceph_dentry_cachep;
 struct kmem_cache *ceph_file_cachep;
 
@@ -634,6 +648,10 @@ static int __init init_caches(void)
 				     SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD);
 	if (ceph_cap_cachep == NULL)
 		goto bad_cap;
+	ceph_cap_flush_cachep = KMEM_CACHE(ceph_cap_flush,
+					   SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD);
+	if (ceph_cap_flush_cachep == NULL)
+		goto bad_cap_flush;
 
 	ceph_dentry_cachep = KMEM_CACHE(ceph_dentry_info,
 					SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD);
@@ -652,6 +670,8 @@ static int __init init_caches(void)
 bad_file:
 	kmem_cache_destroy(ceph_dentry_cachep);
 bad_dentry:
+	kmem_cache_destroy(ceph_cap_flush_cachep);
+bad_cap_flush:
 	kmem_cache_destroy(ceph_cap_cachep);
 bad_cap:
 	kmem_cache_destroy(ceph_inode_cachep);
@@ -668,6 +688,7 @@ static void destroy_caches(void)
 
 	kmem_cache_destroy(ceph_inode_cachep);
 	kmem_cache_destroy(ceph_cap_cachep);
+	kmem_cache_destroy(ceph_cap_flush_cachep);
 	kmem_cache_destroy(ceph_dentry_cachep);
 	kmem_cache_destroy(ceph_file_cachep);
 
@@ -729,7 +750,7 @@ static struct dentry *open_root_dentry(struct ceph_fs_client *fsc,
 	req->r_ino1.ino = CEPH_INO_ROOT;
 	req->r_ino1.snap = CEPH_NOSNAP;
 	req->r_started = started;
-	req->r_timeout = fsc->client->options->mount_timeout * HZ;
+	req->r_timeout = fsc->client->options->mount_timeout;
 	req->r_args.getattr.mask = cpu_to_le32(CEPH_STAT_CAP_INODE);
 	req->r_num_caps = 2;
 	err = ceph_mdsc_do_request(mdsc, NULL, req);
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index fa20e1318939..860cc016e70d 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -35,6 +35,7 @@
 #define CEPH_MOUNT_OPT_INO32           (1<<8) /* 32 bit inos */
 #define CEPH_MOUNT_OPT_DCACHE          (1<<9) /* use dcache for readdir etc */
 #define CEPH_MOUNT_OPT_FSCACHE         (1<<10) /* use fscache */
+#define CEPH_MOUNT_OPT_NOPOOLPERM      (1<<11) /* no pool permission check */
 
 #define CEPH_MOUNT_OPT_DEFAULT    (CEPH_MOUNT_OPT_RBYTES | \
 				   CEPH_MOUNT_OPT_DCACHE)
@@ -121,11 +122,21 @@ struct ceph_cap {
 	struct rb_node ci_node;          /* per-ci cap tree */
 	struct ceph_mds_session *session;
 	struct list_head session_caps;   /* per-session caplist */
-	int mds;
 	u64 cap_id;       /* unique cap id (mds provided) */
-	int issued;       /* latest, from the mds */
-	int implemented;  /* implemented superset of issued (for revocation) */
-	int mds_wanted;
+	union {
+		/* in-use caps */
+		struct {
+			int issued;       /* latest, from the mds */
+			int implemented;  /* implemented superset of
+					     issued (for revocation) */
+			int mds, mds_wanted;
+		};
+		/* caps to release */
+		struct {
+			u64 cap_ino;
+			int queue_release;
+		};
+	};
 	u32 seq, issue_seq, mseq;
 	u32 cap_gen;      /* active/stale cycle */
 	unsigned long last_used;
@@ -163,6 +174,7 @@ struct ceph_cap_snap {
 	int writing;   /* a sync write is still in progress */
 	int dirty_pages;     /* dirty pages awaiting writeback */
 	bool inline_data;
+	bool need_flush;
 };
 
 static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
@@ -174,6 +186,17 @@ static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
 	}
 }
 
+struct ceph_cap_flush {
+	u64 tid;
+	int caps;
+	bool kick;
+	struct rb_node g_node; // global
+	union {
+		struct rb_node i_node; // inode
+		struct list_head list;
+	};
+};
+
 /*
  * The frag tree describes how a directory is fragmented, potentially across
  * multiple metadata servers.  It is also used to indicate points where
@@ -259,9 +282,9 @@ struct ceph_inode_info {
 	u32 i_time_warp_seq;
 
 	unsigned i_ceph_flags;
-	int i_ordered_count;
-	atomic_t i_release_count;
-	atomic_t i_complete_count;
+	atomic64_t i_release_count;
+	atomic64_t i_ordered_count;
+	atomic64_t i_complete_seq[2];
 
 	struct ceph_dir_layout i_dir_layout;
 	struct ceph_file_layout i_layout;
@@ -283,11 +306,11 @@ struct ceph_inode_info {
 	struct ceph_cap *i_auth_cap;     /* authoritative cap, if any */
 	unsigned i_dirty_caps, i_flushing_caps;     /* mask of dirtied fields */
 	struct list_head i_dirty_item, i_flushing_item;
-	u64 i_cap_flush_seq;
 	/* we need to track cap writeback on a per-cap-bit basis, to allow
 	 * overlapping, pipelined cap flushes to the mds.  we can probably
 	 * reduce the tid to 8 bits if we're concerned about inode size. */
-	u16 i_cap_flush_last_tid, i_cap_flush_tid[CEPH_CAP_BITS];
+	struct ceph_cap_flush *i_prealloc_cap_flush;
+	struct rb_root i_cap_flush_tree;
 	wait_queue_head_t i_cap_wq;      /* threads waiting on a capability */
 	unsigned long i_hold_caps_min; /* jiffies */
 	unsigned long i_hold_caps_max; /* jiffies */
@@ -438,36 +461,46 @@ static inline struct inode *ceph_find_inode(struct super_block *sb,
 /*
  * Ceph inode.
  */
-#define CEPH_I_DIR_ORDERED	1  /* dentries in dir are ordered */
-#define CEPH_I_NODELAY		4  /* do not delay cap release */
-#define CEPH_I_FLUSH		8  /* do not delay flush of dirty metadata */
-#define CEPH_I_NOFLUSH		16 /* do not flush dirty caps */
+#define CEPH_I_DIR_ORDERED	(1 << 0)  /* dentries in dir are ordered */
+#define CEPH_I_NODELAY		(1 << 1)  /* do not delay cap release */
+#define CEPH_I_FLUSH		(1 << 2)  /* do not delay flush of dirty metadata */
+#define CEPH_I_NOFLUSH		(1 << 3)  /* do not flush dirty caps */
+#define CEPH_I_POOL_PERM	(1 << 4)  /* pool rd/wr bits are valid */
+#define CEPH_I_POOL_RD		(1 << 5)  /* can read from pool */
+#define CEPH_I_POOL_WR		(1 << 6)  /* can write to pool */
+
 
 static inline void __ceph_dir_set_complete(struct ceph_inode_info *ci,
-					   int release_count, int ordered_count)
+					   long long release_count,
+					   long long ordered_count)
 {
-	atomic_set(&ci->i_complete_count, release_count);
-	if (ci->i_ordered_count == ordered_count)
-		ci->i_ceph_flags |= CEPH_I_DIR_ORDERED;
-	else
-		ci->i_ceph_flags &= ~CEPH_I_DIR_ORDERED;
+	smp_mb__before_atomic();
+	atomic64_set(&ci->i_complete_seq[0], release_count);
+	atomic64_set(&ci->i_complete_seq[1], ordered_count);
 }
 
 static inline void __ceph_dir_clear_complete(struct ceph_inode_info *ci)
 {
-	atomic_inc(&ci->i_release_count);
+	atomic64_inc(&ci->i_release_count);
+}
+
+static inline void __ceph_dir_clear_ordered(struct ceph_inode_info *ci)
+{
+	atomic64_inc(&ci->i_ordered_count);
 }
 
 static inline bool __ceph_dir_is_complete(struct ceph_inode_info *ci)
 {
-	return atomic_read(&ci->i_complete_count) ==
-		atomic_read(&ci->i_release_count);
+	return atomic64_read(&ci->i_complete_seq[0]) ==
+		atomic64_read(&ci->i_release_count);
 }
 
 static inline bool __ceph_dir_is_complete_ordered(struct ceph_inode_info *ci)
 {
-	return __ceph_dir_is_complete(ci) &&
-		(ci->i_ceph_flags & CEPH_I_DIR_ORDERED);
+	return  atomic64_read(&ci->i_complete_seq[0]) ==
+		atomic64_read(&ci->i_release_count) &&
+		atomic64_read(&ci->i_complete_seq[1]) ==
+		atomic64_read(&ci->i_ordered_count);
 }
 
 static inline void ceph_dir_clear_complete(struct inode *inode)
@@ -477,20 +510,13 @@ static inline void ceph_dir_clear_complete(struct inode *inode)
 
 static inline void ceph_dir_clear_ordered(struct inode *inode)
 {
-	struct ceph_inode_info *ci = ceph_inode(inode);
-	spin_lock(&ci->i_ceph_lock);
-	ci->i_ordered_count++;
-	ci->i_ceph_flags &= ~CEPH_I_DIR_ORDERED;
-	spin_unlock(&ci->i_ceph_lock);
+	__ceph_dir_clear_ordered(ceph_inode(inode));
 }
 
 static inline bool ceph_dir_is_complete_ordered(struct inode *inode)
 {
-	struct ceph_inode_info *ci = ceph_inode(inode);
-	bool ret;
-	spin_lock(&ci->i_ceph_lock);
-	ret = __ceph_dir_is_complete_ordered(ci);
-	spin_unlock(&ci->i_ceph_lock);
+	bool ret = __ceph_dir_is_complete_ordered(ceph_inode(inode));
+	smp_rmb();
 	return ret;
 }
 
@@ -552,7 +578,10 @@ static inline int __ceph_caps_dirty(struct ceph_inode_info *ci)
 {
 	return ci->i_dirty_caps | ci->i_flushing_caps;
 }
-extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask);
+extern struct ceph_cap_flush *ceph_alloc_cap_flush(void);
+extern void ceph_free_cap_flush(struct ceph_cap_flush *cf);
+extern int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask,
+				  struct ceph_cap_flush **pcf);
 
 extern int __ceph_caps_revoking_other(struct ceph_inode_info *ci,
 				      struct ceph_cap *ocap, int mask);
@@ -606,16 +635,20 @@ struct ceph_file_info {
 	unsigned offset;       /* offset of last chunk, adjusted for . and .. */
 	unsigned next_offset;  /* offset of next chunk (last_name's + 1) */
 	char *last_name;       /* last entry in previous chunk */
-	struct dentry *dentry; /* next dentry (for dcache readdir) */
-	int dir_release_count;
-	int dir_ordered_count;
+	long long dir_release_count;
+	long long dir_ordered_count;
+	int readdir_cache_idx;
 
 	/* used for -o dirstat read() on directory thing */
 	char *dir_info;
 	int dir_info_len;
 };
 
-
+struct ceph_readdir_cache_control {
+	struct page  *page;
+	struct dentry **dentries;
+	int index;
+};
 
 /*
  * A "snap realm" describes a subset of the file hierarchy sharing
@@ -687,6 +720,7 @@ static inline int default_congestion_kb(void)
 
 
 /* snap.c */
+extern struct ceph_snap_context *ceph_empty_snapc;
 struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
 					       u64 ino);
 extern void ceph_get_snap_realm(struct ceph_mds_client *mdsc,
@@ -713,8 +747,8 @@ extern void ceph_snap_exit(void);
 static inline bool __ceph_have_pending_cap_snap(struct ceph_inode_info *ci)
 {
 	return !list_empty(&ci->i_cap_snaps) &&
-		list_entry(ci->i_cap_snaps.prev, struct ceph_cap_snap,
-			   ci_item)->writing;
+	       list_last_entry(&ci->i_cap_snaps, struct ceph_cap_snap,
+			       ci_item)->writing;
 }
 
 /* inode.c */
@@ -838,12 +872,12 @@ extern void ceph_put_cap(struct ceph_mds_client *mdsc,
 			 struct ceph_cap *cap);
 extern int ceph_is_any_caps(struct inode *inode);
 
-extern void __queue_cap_release(struct ceph_mds_session *session, u64 ino,
-				u64 cap_id, u32 migrate_seq, u32 issue_seq);
 extern void ceph_queue_caps_release(struct inode *inode);
 extern int ceph_write_inode(struct inode *inode, struct writeback_control *wbc);
 extern int ceph_fsync(struct file *file, loff_t start, loff_t end,
 		      int datasync);
+extern void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc,
+					  struct ceph_mds_session *session);
 extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
 				    struct ceph_mds_session *session);
 extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
@@ -879,6 +913,9 @@ extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
 /* addr.c */
 extern const struct address_space_operations ceph_aops;
 extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
+extern int ceph_uninline_data(struct file *filp, struct page *locked_page);
+extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need);
+extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
 
 /* file.c */
 extern const struct file_operations ceph_file_fops;
@@ -890,7 +927,6 @@ extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
 extern int ceph_release(struct inode *inode, struct file *filp);
 extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
 				  char *data, size_t len);
-int ceph_uninline_data(struct file *filp, struct page *locked_page);
 /* dir.c */
 extern const struct file_operations ceph_dir_fops;
 extern const struct file_operations ceph_snapdir_fops;
@@ -911,6 +947,7 @@ extern void ceph_dentry_lru_del(struct dentry *dn);
 extern void ceph_invalidate_dentry_lease(struct dentry *dentry);
 extern unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn);
 extern struct inode *ceph_get_dentry_parent_inode(struct dentry *dentry);
+extern void ceph_readdir_cache_release(struct ceph_readdir_cache_control *ctl);
 
 /*
  * our d_ops vary depending on whether the inode is live,
diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c
index cd7ffad4041d..819163d8313b 100644
--- a/fs/ceph/xattr.c
+++ b/fs/ceph/xattr.c
@@ -911,6 +911,8 @@ int __ceph_setxattr(struct dentry *dentry, const char *name,
 	struct inode *inode = d_inode(dentry);
 	struct ceph_vxattr *vxattr;
 	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(dentry->d_sb)->mdsc;
+	struct ceph_cap_flush *prealloc_cf = NULL;
 	int issued;
 	int err;
 	int dirty = 0;
@@ -920,6 +922,7 @@ int __ceph_setxattr(struct dentry *dentry, const char *name,
 	char *newval = NULL;
 	struct ceph_inode_xattr *xattr = NULL;
 	int required_blob_size;
+	bool lock_snap_rwsem = false;
 
 	if (!ceph_is_valid_xattr(name))
 		return -EOPNOTSUPP;
@@ -948,12 +951,27 @@ int __ceph_setxattr(struct dentry *dentry, const char *name,
 	if (!xattr)
 		goto out;
 
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		goto out;
+
 	spin_lock(&ci->i_ceph_lock);
 retry:
 	issued = __ceph_caps_issued(ci, NULL);
-	dout("setxattr %p issued %s\n", inode, ceph_cap_string(issued));
 	if (ci->i_xattrs.version == 0 || !(issued & CEPH_CAP_XATTR_EXCL))
 		goto do_sync;
+
+	if (!lock_snap_rwsem && !ci->i_head_snapc) {
+		lock_snap_rwsem = true;
+		if (!down_read_trylock(&mdsc->snap_rwsem)) {
+			spin_unlock(&ci->i_ceph_lock);
+			down_read(&mdsc->snap_rwsem);
+			spin_lock(&ci->i_ceph_lock);
+			goto retry;
+		}
+	}
+
+	dout("setxattr %p issued %s\n", inode, ceph_cap_string(issued));
 	__build_xattrs(inode);
 
 	required_blob_size = __get_required_blob_size(ci, name_len, val_len);
@@ -966,7 +984,7 @@ retry:
 		dout(" preaallocating new blob size=%d\n", required_blob_size);
 		blob = ceph_buffer_new(required_blob_size, GFP_NOFS);
 		if (!blob)
-			goto out;
+			goto do_sync_unlocked;
 		spin_lock(&ci->i_ceph_lock);
 		if (ci->i_xattrs.prealloc_blob)
 			ceph_buffer_put(ci->i_xattrs.prealloc_blob);
@@ -978,21 +996,28 @@ retry:
 			  flags, value ? 1 : -1, &xattr);
 
 	if (!err) {
-		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_XATTR_EXCL);
+		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_XATTR_EXCL,
+					       &prealloc_cf);
 		ci->i_xattrs.dirty = true;
 		inode->i_ctime = CURRENT_TIME;
 	}
 
 	spin_unlock(&ci->i_ceph_lock);
+	if (lock_snap_rwsem)
+		up_read(&mdsc->snap_rwsem);
 	if (dirty)
 		__mark_inode_dirty(inode, dirty);
+	ceph_free_cap_flush(prealloc_cf);
 	return err;
 
 do_sync:
 	spin_unlock(&ci->i_ceph_lock);
 do_sync_unlocked:
+	if (lock_snap_rwsem)
+		up_read(&mdsc->snap_rwsem);
 	err = ceph_sync_setxattr(dentry, name, value, size, flags);
 out:
+	ceph_free_cap_flush(prealloc_cf);
 	kfree(newname);
 	kfree(newval);
 	kfree(xattr);
@@ -1044,10 +1069,13 @@ int __ceph_removexattr(struct dentry *dentry, const char *name)
 	struct inode *inode = d_inode(dentry);
 	struct ceph_vxattr *vxattr;
 	struct ceph_inode_info *ci = ceph_inode(inode);
+	struct ceph_mds_client *mdsc = ceph_sb_to_client(dentry->d_sb)->mdsc;
+	struct ceph_cap_flush *prealloc_cf = NULL;
 	int issued;
 	int err;
 	int required_blob_size;
 	int dirty;
+	bool lock_snap_rwsem = false;
 
 	if (!ceph_is_valid_xattr(name))
 		return -EOPNOTSUPP;
@@ -1060,14 +1088,29 @@ int __ceph_removexattr(struct dentry *dentry, const char *name)
 	if (!strncmp(name, XATTR_CEPH_PREFIX, XATTR_CEPH_PREFIX_LEN))
 		goto do_sync_unlocked;
 
+	prealloc_cf = ceph_alloc_cap_flush();
+	if (!prealloc_cf)
+		return -ENOMEM;
+
 	err = -ENOMEM;
 	spin_lock(&ci->i_ceph_lock);
 retry:
 	issued = __ceph_caps_issued(ci, NULL);
-	dout("removexattr %p issued %s\n", inode, ceph_cap_string(issued));
-
 	if (ci->i_xattrs.version == 0 || !(issued & CEPH_CAP_XATTR_EXCL))
 		goto do_sync;
+
+	if (!lock_snap_rwsem && !ci->i_head_snapc) {
+		lock_snap_rwsem = true;
+		if (!down_read_trylock(&mdsc->snap_rwsem)) {
+			spin_unlock(&ci->i_ceph_lock);
+			down_read(&mdsc->snap_rwsem);
+			spin_lock(&ci->i_ceph_lock);
+			goto retry;
+		}
+	}
+
+	dout("removexattr %p issued %s\n", inode, ceph_cap_string(issued));
+
 	__build_xattrs(inode);
 
 	required_blob_size = __get_required_blob_size(ci, 0, 0);
@@ -1080,7 +1123,7 @@ retry:
 		dout(" preaallocating new blob size=%d\n", required_blob_size);
 		blob = ceph_buffer_new(required_blob_size, GFP_NOFS);
 		if (!blob)
-			goto out;
+			goto do_sync_unlocked;
 		spin_lock(&ci->i_ceph_lock);
 		if (ci->i_xattrs.prealloc_blob)
 			ceph_buffer_put(ci->i_xattrs.prealloc_blob);
@@ -1090,18 +1133,24 @@ retry:
 
 	err = __remove_xattr_by_name(ceph_inode(inode), name);
 
-	dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_XATTR_EXCL);
+	dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_XATTR_EXCL,
+				       &prealloc_cf);
 	ci->i_xattrs.dirty = true;
 	inode->i_ctime = CURRENT_TIME;
 	spin_unlock(&ci->i_ceph_lock);
+	if (lock_snap_rwsem)
+		up_read(&mdsc->snap_rwsem);
 	if (dirty)
 		__mark_inode_dirty(inode, dirty);
+	ceph_free_cap_flush(prealloc_cf);
 	return err;
 do_sync:
 	spin_unlock(&ci->i_ceph_lock);
 do_sync_unlocked:
+	if (lock_snap_rwsem)
+		up_read(&mdsc->snap_rwsem);
+	ceph_free_cap_flush(prealloc_cf);
 	err = ceph_send_removexattr(dentry, name);
-out:
 	return err;
 }
 
diff --git a/fs/coda/coda_linux.h b/fs/coda/coda_linux.h
index d6f7a76a1f5b..f829fe963f5b 100644
--- a/fs/coda/coda_linux.h
+++ b/fs/coda/coda_linux.h
@@ -79,7 +79,7 @@ void coda_sysctl_clean(void);
 
 static inline struct coda_inode_info *ITOC(struct inode *inode)
 {
-	return list_entry(inode, struct coda_inode_info, vfs_inode);
+	return container_of(inode, struct coda_inode_info, vfs_inode);
 }
 
 static __inline__ struct CodaFid *coda_i2f(struct inode *inode)
diff --git a/fs/compat_ioctl.c b/fs/compat_ioctl.c
index 6b8e2f091f5b..48851f6ea6ec 100644
--- a/fs/compat_ioctl.c
+++ b/fs/compat_ioctl.c
@@ -896,6 +896,7 @@ COMPATIBLE_IOCTL(FIGETBSZ)
 /* 'X' - originally XFS but some now in the VFS */
 COMPATIBLE_IOCTL(FIFREEZE)
 COMPATIBLE_IOCTL(FITHAW)
+COMPATIBLE_IOCTL(FITRIM)
 COMPATIBLE_IOCTL(KDGETKEYCODE)
 COMPATIBLE_IOCTL(KDSETKEYCODE)
 COMPATIBLE_IOCTL(KDGKBTYPE)
diff --git a/fs/configfs/inode.c b/fs/configfs/inode.c
index 8d89f5fd0331..eae87575e681 100644
--- a/fs/configfs/inode.c
+++ b/fs/configfs/inode.c
@@ -236,7 +236,7 @@ void configfs_drop_dentry(struct configfs_dirent * sd, struct dentry * parent)
 
 	if (dentry) {
 		spin_lock(&dentry->d_lock);
-		if (!d_unhashed(dentry) && d_really_is_positive(dentry)) {
+		if (simple_positive(dentry)) {
 			dget_dlock(dentry);
 			__d_drop(dentry);
 			spin_unlock(&dentry->d_lock);
diff --git a/fs/configfs/mount.c b/fs/configfs/mount.c
index 537356742091..a8f3b589a2df 100644
--- a/fs/configfs/mount.c
+++ b/fs/configfs/mount.c
@@ -129,8 +129,6 @@ void configfs_release_fs(void)
 }
 
 
-static struct kobject *config_kobj;
-
 static int __init configfs_init(void)
 {
 	int err = -ENOMEM;
@@ -141,8 +139,8 @@ static int __init configfs_init(void)
 	if (!configfs_dir_cachep)
 		goto out;
 
-	config_kobj = kobject_create_and_add("config", kernel_kobj);
-	if (!config_kobj)
+	err = sysfs_create_mount_point(kernel_kobj, "config");
+	if (err)
 		goto out2;
 
 	err = register_filesystem(&configfs_fs_type);
@@ -152,7 +150,7 @@ static int __init configfs_init(void)
 	return 0;
 out3:
 	pr_err("Unable to register filesystem!\n");
-	kobject_put(config_kobj);
+	sysfs_remove_mount_point(kernel_kobj, "config");
 out2:
 	kmem_cache_destroy(configfs_dir_cachep);
 	configfs_dir_cachep = NULL;
@@ -163,7 +161,7 @@ out:
 static void __exit configfs_exit(void)
 {
 	unregister_filesystem(&configfs_fs_type);
-	kobject_put(config_kobj);
+	sysfs_remove_mount_point(kernel_kobj, "config");
 	kmem_cache_destroy(configfs_dir_cachep);
 	configfs_dir_cachep = NULL;
 }
diff --git a/fs/coredump.c b/fs/coredump.c
index e52e0064feac..c5ecde6f3eed 100644
--- a/fs/coredump.c
+++ b/fs/coredump.c
@@ -140,7 +140,7 @@ static int cn_print_exe_file(struct core_name *cn)
 		goto put_exe_file;
 	}
 
-	path = d_path(&exe_file->f_path, pathbuf, PATH_MAX);
+	path = file_path(exe_file, pathbuf, PATH_MAX);
 	if (IS_ERR(path)) {
 		ret = PTR_ERR(path);
 		goto free_buf;
diff --git a/fs/dax.c b/fs/dax.c
index 99b5fbc38992..c3e21ccfc358 100644
--- a/fs/dax.c
+++ b/fs/dax.c
@@ -155,7 +155,7 @@ static ssize_t dax_io(struct inode *inode, struct iov_iter *iter,
 		}
 
 		if (iov_iter_rw(iter) == WRITE)
-			len = copy_from_iter(addr, max - pos, iter);
+			len = copy_from_iter_nocache(addr, max - pos, iter);
 		else if (!hole)
 			len = copy_to_iter(addr, max - pos, iter);
 		else
@@ -209,7 +209,8 @@ ssize_t dax_do_io(struct kiocb *iocb, struct inode *inode,
 	}
 
 	/* Protects against truncate */
-	inode_dio_begin(inode);
+	if (!(flags & DIO_SKIP_DIO_COUNT))
+		inode_dio_begin(inode);
 
 	retval = dax_io(inode, iter, pos, end, get_block, &bh);
 
@@ -219,7 +220,8 @@ ssize_t dax_do_io(struct kiocb *iocb, struct inode *inode,
 	if ((retval > 0) && end_io)
 		end_io(iocb, pos, retval, bh.b_private);
 
-	inode_dio_end(inode);
+	if (!(flags & DIO_SKIP_DIO_COUNT))
+		inode_dio_end(inode);
  out:
 	return retval;
 }
diff --git a/fs/dcache.c b/fs/dcache.c
index 592c4b582495..5c8ea15e73a5 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -642,7 +642,7 @@ static inline bool fast_dput(struct dentry *dentry)
 
 	/*
 	 * If we have a d_op->d_delete() operation, we sould not
-	 * let the dentry count go to zero, so use "put__or_lock".
+	 * let the dentry count go to zero, so use "put_or_lock".
 	 */
 	if (unlikely(dentry->d_flags & DCACHE_OP_DELETE))
 		return lockref_put_or_lock(&dentry->d_lockref);
@@ -697,7 +697,7 @@ static inline bool fast_dput(struct dentry *dentry)
 	 */
 	smp_rmb();
 	d_flags = ACCESS_ONCE(dentry->d_flags);
-	d_flags &= DCACHE_REFERENCED | DCACHE_LRU_LIST;
+	d_flags &= DCACHE_REFERENCED | DCACHE_LRU_LIST | DCACHE_DISCONNECTED;
 
 	/* Nothing to do? Dropping the reference was all we needed? */
 	if (d_flags == (DCACHE_REFERENCED | DCACHE_LRU_LIST) && !d_unhashed(dentry))
@@ -776,6 +776,9 @@ repeat:
 	if (unlikely(d_unhashed(dentry)))
 		goto kill_it;
 
+	if (unlikely(dentry->d_flags & DCACHE_DISCONNECTED))
+		goto kill_it;
+
 	if (unlikely(dentry->d_flags & DCACHE_OP_DELETE)) {
 		if (dentry->d_op->d_delete(dentry))
 			goto kill_it;
@@ -1673,7 +1676,8 @@ void d_set_d_op(struct dentry *dentry, const struct dentry_operations *op)
 				DCACHE_OP_COMPARE	|
 				DCACHE_OP_REVALIDATE	|
 				DCACHE_OP_WEAK_REVALIDATE	|
-				DCACHE_OP_DELETE ));
+				DCACHE_OP_DELETE	|
+				DCACHE_OP_SELECT_INODE));
 	dentry->d_op = op;
 	if (!op)
 		return;
@@ -1689,6 +1693,8 @@ void d_set_d_op(struct dentry *dentry, const struct dentry_operations *op)
 		dentry->d_flags |= DCACHE_OP_DELETE;
 	if (op->d_prune)
 		dentry->d_flags |= DCACHE_OP_PRUNE;
+	if (op->d_select_inode)
+		dentry->d_flags |= DCACHE_OP_SELECT_INODE;
 
 }
 EXPORT_SYMBOL(d_set_d_op);
@@ -2927,17 +2933,6 @@ restart:
 				vfsmnt = &mnt->mnt;
 				continue;
 			}
-			/*
-			 * Filesystems needing to implement special "root names"
-			 * should do so with ->d_dname()
-			 */
-			if (IS_ROOT(dentry) &&
-			   (dentry->d_name.len != 1 ||
-			    dentry->d_name.name[0] != '/')) {
-				WARN(1, "Root dentry has weird name <%.*s>\n",
-				     (int) dentry->d_name.len,
-				     dentry->d_name.name);
-			}
 			if (!error)
 				error = is_mounted(vfsmnt) ? 1 : 2;
 			break;
diff --git a/fs/debugfs/inode.c b/fs/debugfs/inode.c
index 7eaec88ea970..c711be8d6a3c 100644
--- a/fs/debugfs/inode.c
+++ b/fs/debugfs/inode.c
@@ -44,11 +44,6 @@ static struct inode *debugfs_get_inode(struct super_block *sb)
 	return inode;
 }
 
-static inline int debugfs_positive(struct dentry *dentry)
-{
-	return d_really_is_positive(dentry) && !d_unhashed(dentry);
-}
-
 struct debugfs_mount_opts {
 	kuid_t uid;
 	kgid_t gid;
@@ -522,7 +517,7 @@ static int __debugfs_remove(struct dentry *dentry, struct dentry *parent)
 {
 	int ret = 0;
 
-	if (debugfs_positive(dentry)) {
+	if (simple_positive(dentry)) {
 		dget(dentry);
 		if (d_is_dir(dentry))
 			ret = simple_rmdir(d_inode(parent), dentry);
@@ -602,7 +597,7 @@ void debugfs_remove_recursive(struct dentry *dentry)
 	 */
 	spin_lock(&parent->d_lock);
 	list_for_each_entry(child, &parent->d_subdirs, d_child) {
-		if (!debugfs_positive(child))
+		if (!simple_positive(child))
 			continue;
 
 		/* perhaps simple_empty(child) makes more sense */
@@ -623,7 +618,7 @@ void debugfs_remove_recursive(struct dentry *dentry)
 		 * from d_subdirs. When releasing the parent->d_lock we can
 		 * no longer trust that the next pointer is valid.
 		 * Restart the loop. We'll skip this one with the
-		 * debugfs_positive() check.
+		 * simple_positive() check.
 		 */
 		goto loop;
 	}
@@ -716,20 +711,17 @@ bool debugfs_initialized(void)
 }
 EXPORT_SYMBOL_GPL(debugfs_initialized);
 
-
-static struct kobject *debug_kobj;
-
 static int __init debugfs_init(void)
 {
 	int retval;
 
-	debug_kobj = kobject_create_and_add("debug", kernel_kobj);
-	if (!debug_kobj)
-		return -EINVAL;
+	retval = sysfs_create_mount_point(kernel_kobj, "debug");
+	if (retval)
+		return retval;
 
 	retval = register_filesystem(&debug_fs_type);
 	if (retval)
-		kobject_put(debug_kobj);
+		sysfs_remove_mount_point(kernel_kobj, "debug");
 	else
 		debugfs_registered = true;
 
diff --git a/fs/ecryptfs/file.c b/fs/ecryptfs/file.c
index 72afcc629d7b..feef8a9c4de7 100644
--- a/fs/ecryptfs/file.c
+++ b/fs/ecryptfs/file.c
@@ -325,7 +325,6 @@ ecryptfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
 		return rc;
 
 	switch (cmd) {
-	case FITRIM:
 	case FS_IOC32_GETFLAGS:
 	case FS_IOC32_SETFLAGS:
 	case FS_IOC32_GETVERSION:
diff --git a/fs/exofs/dir.c b/fs/exofs/dir.c
index 4deb0b05b011..e5bb2abf77f9 100644
--- a/fs/exofs/dir.c
+++ b/fs/exofs/dir.c
@@ -44,12 +44,6 @@ static inline void exofs_put_page(struct page *page)
 	page_cache_release(page);
 }
 
-/* Accesses dir's inode->i_size must be called under inode lock */
-static inline unsigned long dir_pages(struct inode *inode)
-{
-	return (inode->i_size + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
-}
-
 static unsigned exofs_last_byte(struct inode *inode, unsigned long page_nr)
 {
 	loff_t last_byte = inode->i_size;
diff --git a/fs/ext2/dir.c b/fs/ext2/dir.c
index 796b491e6978..0c6638b40f21 100644
--- a/fs/ext2/dir.c
+++ b/fs/ext2/dir.c
@@ -70,11 +70,6 @@ static inline void ext2_put_page(struct page *page)
 	page_cache_release(page);
 }
 
-static inline unsigned long dir_pages(struct inode *inode)
-{
-	return (inode->i_size+PAGE_CACHE_SIZE-1)>>PAGE_CACHE_SHIFT;
-}
-
 /*
  * Return the offset into page `page_nr' of the last valid
  * byte in that page, plus one.
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index aadb72828834..2553aa8b608d 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -504,7 +504,7 @@ __read_extent_tree_block(const char *function, unsigned int line,
 	struct buffer_head		*bh;
 	int				err;
 
-	bh = sb_getblk(inode->i_sb, pblk);
+	bh = sb_getblk_gfp(inode->i_sb, pblk, __GFP_MOVABLE | GFP_NOFS);
 	if (unlikely(!bh))
 		return ERR_PTR(-ENOMEM);
 
@@ -1089,7 +1089,7 @@ static int ext4_ext_split(handle_t *handle, struct inode *inode,
 		err = -EIO;
 		goto cleanup;
 	}
-	bh = sb_getblk(inode->i_sb, newblock);
+	bh = sb_getblk_gfp(inode->i_sb, newblock, __GFP_MOVABLE | GFP_NOFS);
 	if (unlikely(!bh)) {
 		err = -ENOMEM;
 		goto cleanup;
@@ -1283,7 +1283,7 @@ static int ext4_ext_grow_indepth(handle_t *handle, struct inode *inode,
 	if (newblock == 0)
 		return err;
 
-	bh = sb_getblk(inode->i_sb, newblock);
+	bh = sb_getblk_gfp(inode->i_sb, newblock, __GFP_MOVABLE | GFP_NOFS);
 	if (unlikely(!bh))
 		return -ENOMEM;
 	lock_buffer(bh);
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 41f8e55afcd1..cecf9aa10811 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1323,7 +1323,7 @@ static void ext4_da_page_release_reservation(struct page *page,
 					     unsigned int offset,
 					     unsigned int length)
 {
-	int to_release = 0;
+	int to_release = 0, contiguous_blks = 0;
 	struct buffer_head *head, *bh;
 	unsigned int curr_off = 0;
 	struct inode *inode = page->mapping->host;
@@ -1344,14 +1344,23 @@ static void ext4_da_page_release_reservation(struct page *page,
 
 		if ((offset <= curr_off) && (buffer_delay(bh))) {
 			to_release++;
+			contiguous_blks++;
 			clear_buffer_delay(bh);
+		} else if (contiguous_blks) {
+			lblk = page->index <<
+			       (PAGE_CACHE_SHIFT - inode->i_blkbits);
+			lblk += (curr_off >> inode->i_blkbits) -
+				contiguous_blks;
+			ext4_es_remove_extent(inode, lblk, contiguous_blks);
+			contiguous_blks = 0;
 		}
 		curr_off = next_off;
 	} while ((bh = bh->b_this_page) != head);
 
-	if (to_release) {
+	if (contiguous_blks) {
 		lblk = page->index << (PAGE_CACHE_SHIFT - inode->i_blkbits);
-		ext4_es_remove_extent(inode, lblk, to_release);
+		lblk += (curr_off >> inode->i_blkbits) - contiguous_blks;
+		ext4_es_remove_extent(inode, lblk, contiguous_blks);
 	}
 
 	/* If we have released all the blocks belonging to a cluster, then we
@@ -4344,7 +4353,12 @@ static void ext4_update_other_inodes_time(struct super_block *sb,
 	int inode_size = EXT4_INODE_SIZE(sb);
 
 	oi.orig_ino = orig_ino;
-	ino = (orig_ino & ~(inodes_per_block - 1)) + 1;
+	/*
+	 * Calculate the first inode in the inode table block.  Inode
+	 * numbers are one-based.  That is, the first inode in a block
+	 * (assuming 4k blocks and 256 byte inodes) is (n*16 + 1).
+	 */
+	ino = ((orig_ino - 1) & ~(inodes_per_block - 1)) + 1;
 	for (i = 0; i < inodes_per_block; i++, ino++, buf += inode_size) {
 		if (ino == orig_ino)
 			continue;
diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c
index cb8451246b30..1346cfa355d0 100644
--- a/fs/ext4/ioctl.c
+++ b/fs/ext4/ioctl.c
@@ -755,7 +755,6 @@ long ext4_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
 		return err;
 	}
 	case EXT4_IOC_MOVE_EXT:
-	case FITRIM:
 	case EXT4_IOC_RESIZE_FS:
 	case EXT4_IOC_PRECACHE_EXTENTS:
 	case EXT4_IOC_SET_ENCRYPTION_POLICY:
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index f6aedf88da43..34b610ea5030 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -4816,18 +4816,12 @@ do_more:
 		/*
 		 * blocks being freed are metadata. these blocks shouldn't
 		 * be used until this transaction is committed
+		 *
+		 * We use __GFP_NOFAIL because ext4_free_blocks() is not allowed
+		 * to fail.
 		 */
-	retry:
-		new_entry = kmem_cache_alloc(ext4_free_data_cachep, GFP_NOFS);
-		if (!new_entry) {
-			/*
-			 * We use a retry loop because
-			 * ext4_free_blocks() is not allowed to fail.
-			 */
-			cond_resched();
-			congestion_wait(BLK_RW_ASYNC, HZ/50);
-			goto retry;
-		}
+		new_entry = kmem_cache_alloc(ext4_free_data_cachep,
+				GFP_NOFS|__GFP_NOFAIL);
 		new_entry->efd_start_cluster = bit;
 		new_entry->efd_group = block_group;
 		new_entry->efd_count = count_clusters;
diff --git a/fs/ext4/migrate.c b/fs/ext4/migrate.c
index b52374e42102..6163ad21cb0e 100644
--- a/fs/ext4/migrate.c
+++ b/fs/ext4/migrate.c
@@ -620,6 +620,7 @@ int ext4_ind_migrate(struct inode *inode)
 	struct ext4_inode_info		*ei = EXT4_I(inode);
 	struct ext4_extent		*ex;
 	unsigned int			i, len;
+	ext4_lblk_t			start, end;
 	ext4_fsblk_t			blk;
 	handle_t			*handle;
 	int				ret;
@@ -633,6 +634,14 @@ int ext4_ind_migrate(struct inode *inode)
 				       EXT4_FEATURE_RO_COMPAT_BIGALLOC))
 		return -EOPNOTSUPP;
 
+	/*
+	 * In order to get correct extent info, force all delayed allocation
+	 * blocks to be allocated, otherwise delayed allocation blocks may not
+	 * be reflected and bypass the checks on extent header.
+	 */
+	if (test_opt(inode->i_sb, DELALLOC))
+		ext4_alloc_da_blocks(inode);
+
 	handle = ext4_journal_start(inode, EXT4_HT_MIGRATE, 1);
 	if (IS_ERR(handle))
 		return PTR_ERR(handle);
@@ -650,11 +659,13 @@ int ext4_ind_migrate(struct inode *inode)
 		goto errout;
 	}
 	if (eh->eh_entries == 0)
-		blk = len = 0;
+		blk = len = start = end = 0;
 	else {
 		len = le16_to_cpu(ex->ee_len);
 		blk = ext4_ext_pblock(ex);
-		if (len > EXT4_NDIR_BLOCKS) {
+		start = le32_to_cpu(ex->ee_block);
+		end = start + len - 1;
+		if (end >= EXT4_NDIR_BLOCKS) {
 			ret = -EOPNOTSUPP;
 			goto errout;
 		}
@@ -662,7 +673,7 @@ int ext4_ind_migrate(struct inode *inode)
 
 	ext4_clear_inode_flag(inode, EXT4_INODE_EXTENTS);
 	memset(ei->i_data, 0, sizeof(ei->i_data));
-	for (i=0; i < len; i++)
+	for (i = start; i <= end; i++)
 		ei->i_data[i] = cpu_to_le32(blk++);
 	ext4_mark_inode_dirty(handle, inode);
 errout:
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 5c787647afe2..58987b5c514b 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -452,7 +452,7 @@ void __ext4_error_file(struct file *file, const char *function,
 	es = EXT4_SB(inode->i_sb)->s_es;
 	es->s_last_error_ino = cpu_to_le32(inode->i_ino);
 	if (ext4_error_ratelimit(inode->i_sb)) {
-		path = d_path(&(file->f_path), pathname, sizeof(pathname));
+		path = file_path(file, pathname, sizeof(pathname));
 		if (IS_ERR(path))
 			path = "(unknown)";
 		va_start(args, fmt);
diff --git a/fs/file.c b/fs/file.c
index 93c5f89c248b..6c672ad329e9 100644
--- a/fs/file.c
+++ b/fs/file.c
@@ -147,6 +147,13 @@ static int expand_fdtable(struct files_struct *files, int nr)
 
 	spin_unlock(&files->file_lock);
 	new_fdt = alloc_fdtable(nr);
+
+	/* make sure all __fd_install() have seen resize_in_progress
+	 * or have finished their rcu_read_lock_sched() section.
+	 */
+	if (atomic_read(&files->count) > 1)
+		synchronize_sched();
+
 	spin_lock(&files->file_lock);
 	if (!new_fdt)
 		return -ENOMEM;
@@ -158,21 +165,14 @@ static int expand_fdtable(struct files_struct *files, int nr)
 		__free_fdtable(new_fdt);
 		return -EMFILE;
 	}
-	/*
-	 * Check again since another task may have expanded the fd table while
-	 * we dropped the lock
-	 */
 	cur_fdt = files_fdtable(files);
-	if (nr >= cur_fdt->max_fds) {
-		/* Continue as planned */
-		copy_fdtable(new_fdt, cur_fdt);
-		rcu_assign_pointer(files->fdt, new_fdt);
-		if (cur_fdt != &files->fdtab)
-			call_rcu(&cur_fdt->rcu, free_fdtable_rcu);
-	} else {
-		/* Somebody else expanded, so undo our attempt */
-		__free_fdtable(new_fdt);
-	}
+	BUG_ON(nr < cur_fdt->max_fds);
+	copy_fdtable(new_fdt, cur_fdt);
+	rcu_assign_pointer(files->fdt, new_fdt);
+	if (cur_fdt != &files->fdtab)
+		call_rcu(&cur_fdt->rcu, free_fdtable_rcu);
+	/* coupled with smp_rmb() in __fd_install() */
+	smp_wmb();
 	return 1;
 }
 
@@ -185,21 +185,38 @@ static int expand_fdtable(struct files_struct *files, int nr)
  * The files->file_lock should be held on entry, and will be held on exit.
  */
 static int expand_files(struct files_struct *files, int nr)
+	__releases(files->file_lock)
+	__acquires(files->file_lock)
 {
 	struct fdtable *fdt;
+	int expanded = 0;
 
+repeat:
 	fdt = files_fdtable(files);
 
 	/* Do we need to expand? */
 	if (nr < fdt->max_fds)
-		return 0;
+		return expanded;
 
 	/* Can we expand? */
 	if (nr >= sysctl_nr_open)
 		return -EMFILE;
 
+	if (unlikely(files->resize_in_progress)) {
+		spin_unlock(&files->file_lock);
+		expanded = 1;
+		wait_event(files->resize_wait, !files->resize_in_progress);
+		spin_lock(&files->file_lock);
+		goto repeat;
+	}
+
 	/* All good, so we try */
-	return expand_fdtable(files, nr);
+	files->resize_in_progress = true;
+	expanded = expand_fdtable(files, nr);
+	files->resize_in_progress = false;
+
+	wake_up_all(&files->resize_wait);
+	return expanded;
 }
 
 static inline void __set_close_on_exec(int fd, struct fdtable *fdt)
@@ -256,6 +273,8 @@ struct files_struct *dup_fd(struct files_struct *oldf, int *errorp)
 	atomic_set(&newf->count, 1);
 
 	spin_lock_init(&newf->file_lock);
+	newf->resize_in_progress = false;
+	init_waitqueue_head(&newf->resize_wait);
 	newf->next_fd = 0;
 	new_fdt = &newf->fdtab;
 	new_fdt->max_fds = NR_OPEN_DEFAULT;
@@ -553,11 +572,21 @@ void __fd_install(struct files_struct *files, unsigned int fd,
 		struct file *file)
 {
 	struct fdtable *fdt;
-	spin_lock(&files->file_lock);
-	fdt = files_fdtable(files);
+
+	might_sleep();
+	rcu_read_lock_sched();
+
+	while (unlikely(files->resize_in_progress)) {
+		rcu_read_unlock_sched();
+		wait_event(files->resize_wait, !files->resize_in_progress);
+		rcu_read_lock_sched();
+	}
+	/* coupled with smp_wmb() in expand_fdtable() */
+	smp_rmb();
+	fdt = rcu_dereference_sched(files->fdt);
 	BUG_ON(fdt->fd[fd] != NULL);
 	rcu_assign_pointer(fdt->fd[fd], file);
-	spin_unlock(&files->file_lock);
+	rcu_read_unlock_sched();
 }
 
 void fd_install(unsigned int fd, struct file *file)
@@ -635,11 +664,17 @@ static struct file *__fget(unsigned int fd, fmode_t mask)
 	struct file *file;
 
 	rcu_read_lock();
+loop:
 	file = fcheck_files(files, fd);
 	if (file) {
-		/* File object ref couldn't be taken */
-		if ((file->f_mode & mask) || !get_file_rcu(file))
+		/* File object ref couldn't be taken.
+		 * dup2() atomicity guarantee is the reason
+		 * we loop to catch the new file (or NULL pointer)
+		 */
+		if (file->f_mode & mask)
 			file = NULL;
+		else if (!get_file_rcu(file))
+			goto loop;
 	}
 	rcu_read_unlock();
 
diff --git a/fs/file_table.c b/fs/file_table.c
index 294174dcc226..7f9d407c7595 100644
--- a/fs/file_table.c
+++ b/fs/file_table.c
@@ -20,7 +20,6 @@
 #include <linux/cdev.h>
 #include <linux/fsnotify.h>
 #include <linux/sysctl.h>
-#include <linux/lglock.h>
 #include <linux/percpu_counter.h>
 #include <linux/percpu.h>
 #include <linux/hardirq.h>
diff --git a/fs/freevxfs/vxfs_lookup.c b/fs/freevxfs/vxfs_lookup.c
index 99c7f0a37af4..484b32d3234a 100644
--- a/fs/freevxfs/vxfs_lookup.c
+++ b/fs/freevxfs/vxfs_lookup.c
@@ -61,13 +61,6 @@ const struct file_operations vxfs_dir_operations = {
 	.iterate =		vxfs_readdir,
 };
 
- 
-static inline u_long
-dir_pages(struct inode *inode)
-{
-	return (inode->i_size + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
-}
- 
 static inline u_long
 dir_blocks(struct inode *ip)
 {
diff --git a/fs/fscache/cookie.c b/fs/fscache/cookie.c
index 89acec742e0b..d403c69bee08 100644
--- a/fs/fscache/cookie.c
+++ b/fs/fscache/cookie.c
@@ -327,7 +327,8 @@ static int fscache_alloc_object(struct fscache_cache *cache,
 
 object_already_extant:
 	ret = -ENOBUFS;
-	if (fscache_object_is_dead(object)) {
+	if (fscache_object_is_dying(object) ||
+	    fscache_cache_is_broken(object)) {
 		spin_unlock(&cookie->lock);
 		goto error;
 	}
@@ -671,7 +672,7 @@ int __fscache_check_consistency(struct fscache_cookie *cookie)
 	if (!op)
 		return -ENOMEM;
 
-	fscache_operation_init(op, NULL, NULL);
+	fscache_operation_init(op, NULL, NULL, NULL);
 	op->flags = FSCACHE_OP_MYTHREAD |
 		(1 << FSCACHE_OP_WAITING) |
 		(1 << FSCACHE_OP_UNUSE_COOKIE);
@@ -695,8 +696,7 @@ int __fscache_check_consistency(struct fscache_cookie *cookie)
 	/* the work queue now carries its own ref on the object */
 	spin_unlock(&cookie->lock);
 
-	ret = fscache_wait_for_operation_activation(object, op,
-						    NULL, NULL, NULL);
+	ret = fscache_wait_for_operation_activation(object, op, NULL, NULL);
 	if (ret == 0) {
 		/* ask the cache to honour the operation */
 		ret = object->cache->ops->check_consistency(op);
diff --git a/fs/fscache/internal.h b/fs/fscache/internal.h
index 7872a62ef30c..97ec45110957 100644
--- a/fs/fscache/internal.h
+++ b/fs/fscache/internal.h
@@ -124,8 +124,7 @@ extern int fscache_submit_exclusive_op(struct fscache_object *,
 				       struct fscache_operation *);
 extern int fscache_submit_op(struct fscache_object *,
 			     struct fscache_operation *);
-extern int fscache_cancel_op(struct fscache_operation *,
-			     void (*)(struct fscache_operation *));
+extern int fscache_cancel_op(struct fscache_operation *, bool);
 extern void fscache_cancel_all_ops(struct fscache_object *);
 extern void fscache_abort_object(struct fscache_object *);
 extern void fscache_start_operations(struct fscache_object *);
@@ -138,8 +137,7 @@ extern int fscache_wait_for_deferred_lookup(struct fscache_cookie *);
 extern int fscache_wait_for_operation_activation(struct fscache_object *,
 						 struct fscache_operation *,
 						 atomic_t *,
-						 atomic_t *,
-						 void (*)(struct fscache_operation *));
+						 atomic_t *);
 extern void fscache_invalidate_writes(struct fscache_cookie *);
 
 /*
@@ -164,6 +162,7 @@ extern atomic_t fscache_n_op_pend;
 extern atomic_t fscache_n_op_run;
 extern atomic_t fscache_n_op_enqueue;
 extern atomic_t fscache_n_op_deferred_release;
+extern atomic_t fscache_n_op_initialised;
 extern atomic_t fscache_n_op_release;
 extern atomic_t fscache_n_op_gc;
 extern atomic_t fscache_n_op_cancelled;
@@ -271,6 +270,11 @@ extern atomic_t fscache_n_cop_write_page;
 extern atomic_t fscache_n_cop_uncache_page;
 extern atomic_t fscache_n_cop_dissociate_pages;
 
+extern atomic_t fscache_n_cache_no_space_reject;
+extern atomic_t fscache_n_cache_stale_objects;
+extern atomic_t fscache_n_cache_retired_objects;
+extern atomic_t fscache_n_cache_culled_objects;
+
 static inline void fscache_stat(atomic_t *stat)
 {
 	atomic_inc(stat);
diff --git a/fs/fscache/object.c b/fs/fscache/object.c
index da032daf0e0d..9e792e30f4db 100644
--- a/fs/fscache/object.c
+++ b/fs/fscache/object.c
@@ -328,6 +328,17 @@ void fscache_object_init(struct fscache_object *object,
 EXPORT_SYMBOL(fscache_object_init);
 
 /*
+ * Mark the object as no longer being live, making sure that we synchronise
+ * against op submission.
+ */
+static inline void fscache_mark_object_dead(struct fscache_object *object)
+{
+	spin_lock(&object->lock);
+	clear_bit(FSCACHE_OBJECT_IS_LIVE, &object->flags);
+	spin_unlock(&object->lock);
+}
+
+/*
  * Abort object initialisation before we start it.
  */
 static const struct fscache_state *fscache_abort_initialisation(struct fscache_object *object,
@@ -610,6 +621,8 @@ static const struct fscache_state *fscache_lookup_failure(struct fscache_object
 	object->cache->ops->lookup_complete(object);
 	fscache_stat_d(&fscache_n_cop_lookup_complete);
 
+	set_bit(FSCACHE_OBJECT_KILLED_BY_CACHE, &object->flags);
+
 	cookie = object->cookie;
 	set_bit(FSCACHE_COOKIE_UNAVAILABLE, &cookie->flags);
 	if (test_and_clear_bit(FSCACHE_COOKIE_LOOKING_UP, &cookie->flags))
@@ -629,7 +642,7 @@ static const struct fscache_state *fscache_kill_object(struct fscache_object *ob
 	_enter("{OBJ%x,%d,%d},%d",
 	       object->debug_id, object->n_ops, object->n_children, event);
 
-	clear_bit(FSCACHE_OBJECT_IS_LIVE, &object->flags);
+	fscache_mark_object_dead(object);
 	object->oob_event_mask = 0;
 
 	if (list_empty(&object->dependents) &&
@@ -948,7 +961,8 @@ static const struct fscache_state *_fscache_invalidate_object(struct fscache_obj
 	if (!op)
 		goto nomem;
 
-	fscache_operation_init(op, object->cache->ops->invalidate_object, NULL);
+	fscache_operation_init(op, object->cache->ops->invalidate_object,
+			       NULL, NULL);
 	op->flags = FSCACHE_OP_ASYNC |
 		(1 << FSCACHE_OP_EXCLUSIVE) |
 		(1 << FSCACHE_OP_UNUSE_COOKIE);
@@ -974,13 +988,13 @@ static const struct fscache_state *_fscache_invalidate_object(struct fscache_obj
 	return transit_to(UPDATE_OBJECT);
 
 nomem:
-	clear_bit(FSCACHE_OBJECT_IS_LIVE, &object->flags);
+	fscache_mark_object_dead(object);
 	fscache_unuse_cookie(object);
 	_leave(" [ENOMEM]");
 	return transit_to(KILL_OBJECT);
 
 submit_op_failed:
-	clear_bit(FSCACHE_OBJECT_IS_LIVE, &object->flags);
+	fscache_mark_object_dead(object);
 	spin_unlock(&cookie->lock);
 	fscache_unuse_cookie(object);
 	kfree(op);
@@ -1016,3 +1030,50 @@ static const struct fscache_state *fscache_update_object(struct fscache_object *
 	_leave("");
 	return transit_to(WAIT_FOR_CMD);
 }
+
+/**
+ * fscache_object_retrying_stale - Note retrying stale object
+ * @object: The object that will be retried
+ *
+ * Note that an object lookup found an on-disk object that was adjudged to be
+ * stale and has been deleted.  The lookup will be retried.
+ */
+void fscache_object_retrying_stale(struct fscache_object *object)
+{
+	fscache_stat(&fscache_n_cache_no_space_reject);
+}
+EXPORT_SYMBOL(fscache_object_retrying_stale);
+
+/**
+ * fscache_object_mark_killed - Note that an object was killed
+ * @object: The object that was culled
+ * @why: The reason the object was killed.
+ *
+ * Note that an object was killed.  Returns true if the object was
+ * already marked killed, false if it wasn't.
+ */
+void fscache_object_mark_killed(struct fscache_object *object,
+				enum fscache_why_object_killed why)
+{
+	if (test_and_set_bit(FSCACHE_OBJECT_KILLED_BY_CACHE, &object->flags)) {
+		pr_err("Error: Object already killed by cache [%s]\n",
+		       object->cache->identifier);
+		return;
+	}
+
+	switch (why) {
+	case FSCACHE_OBJECT_NO_SPACE:
+		fscache_stat(&fscache_n_cache_no_space_reject);
+		break;
+	case FSCACHE_OBJECT_IS_STALE:
+		fscache_stat(&fscache_n_cache_stale_objects);
+		break;
+	case FSCACHE_OBJECT_WAS_RETIRED:
+		fscache_stat(&fscache_n_cache_retired_objects);
+		break;
+	case FSCACHE_OBJECT_WAS_CULLED:
+		fscache_stat(&fscache_n_cache_culled_objects);
+		break;
+	}
+}
+EXPORT_SYMBOL(fscache_object_mark_killed);
diff --git a/fs/fscache/operation.c b/fs/fscache/operation.c
index e7b87a0e5185..de67745e1cd7 100644
--- a/fs/fscache/operation.c
+++ b/fs/fscache/operation.c
@@ -20,6 +20,35 @@
 atomic_t fscache_op_debug_id;
 EXPORT_SYMBOL(fscache_op_debug_id);
 
+static void fscache_operation_dummy_cancel(struct fscache_operation *op)
+{
+}
+
+/**
+ * fscache_operation_init - Do basic initialisation of an operation
+ * @op: The operation to initialise
+ * @release: The release function to assign
+ *
+ * Do basic initialisation of an operation.  The caller must still set flags,
+ * object and processor if needed.
+ */
+void fscache_operation_init(struct fscache_operation *op,
+			    fscache_operation_processor_t processor,
+			    fscache_operation_cancel_t cancel,
+			    fscache_operation_release_t release)
+{
+	INIT_WORK(&op->work, fscache_op_work_func);
+	atomic_set(&op->usage, 1);
+	op->state = FSCACHE_OP_ST_INITIALISED;
+	op->debug_id = atomic_inc_return(&fscache_op_debug_id);
+	op->processor = processor;
+	op->cancel = cancel ?: fscache_operation_dummy_cancel;
+	op->release = release;
+	INIT_LIST_HEAD(&op->pend_link);
+	fscache_stat(&fscache_n_op_initialised);
+}
+EXPORT_SYMBOL(fscache_operation_init);
+
 /**
  * fscache_enqueue_operation - Enqueue an operation for processing
  * @op: The operation to enqueue
@@ -76,6 +105,43 @@ static void fscache_run_op(struct fscache_object *object,
 }
 
 /*
+ * report an unexpected submission
+ */
+static void fscache_report_unexpected_submission(struct fscache_object *object,
+						 struct fscache_operation *op,
+						 const struct fscache_state *ostate)
+{
+	static bool once_only;
+	struct fscache_operation *p;
+	unsigned n;
+
+	if (once_only)
+		return;
+	once_only = true;
+
+	kdebug("unexpected submission OP%x [OBJ%x %s]",
+	       op->debug_id, object->debug_id, object->state->name);
+	kdebug("objstate=%s [%s]", object->state->name, ostate->name);
+	kdebug("objflags=%lx", object->flags);
+	kdebug("objevent=%lx [%lx]", object->events, object->event_mask);
+	kdebug("ops=%u inp=%u exc=%u",
+	       object->n_ops, object->n_in_progress, object->n_exclusive);
+
+	if (!list_empty(&object->pending_ops)) {
+		n = 0;
+		list_for_each_entry(p, &object->pending_ops, pend_link) {
+			ASSERTCMP(p->object, ==, object);
+			kdebug("%p %p", op->processor, op->release);
+			n++;
+		}
+
+		kdebug("n=%u", n);
+	}
+
+	dump_stack();
+}
+
+/*
  * submit an exclusive operation for an object
  * - other ops are excluded from running simultaneously with this one
  * - this gets any extra refs it needs on an op
@@ -83,6 +149,8 @@ static void fscache_run_op(struct fscache_object *object,
 int fscache_submit_exclusive_op(struct fscache_object *object,
 				struct fscache_operation *op)
 {
+	const struct fscache_state *ostate;
+	unsigned long flags;
 	int ret;
 
 	_enter("{OBJ%x OP%x},", object->debug_id, op->debug_id);
@@ -95,8 +163,21 @@ int fscache_submit_exclusive_op(struct fscache_object *object,
 	ASSERTCMP(object->n_ops, >=, object->n_exclusive);
 	ASSERT(list_empty(&op->pend_link));
 
+	ostate = object->state;
+	smp_rmb();
+
 	op->state = FSCACHE_OP_ST_PENDING;
-	if (fscache_object_is_active(object)) {
+	flags = READ_ONCE(object->flags);
+	if (unlikely(!(flags & BIT(FSCACHE_OBJECT_IS_LIVE)))) {
+		fscache_stat(&fscache_n_op_rejected);
+		op->cancel(op);
+		op->state = FSCACHE_OP_ST_CANCELLED;
+		ret = -ENOBUFS;
+	} else if (unlikely(fscache_cache_is_broken(object))) {
+		op->cancel(op);
+		op->state = FSCACHE_OP_ST_CANCELLED;
+		ret = -EIO;
+	} else if (flags & BIT(FSCACHE_OBJECT_IS_AVAILABLE)) {
 		op->object = object;
 		object->n_ops++;
 		object->n_exclusive++;	/* reads and writes must wait */
@@ -118,7 +199,7 @@ int fscache_submit_exclusive_op(struct fscache_object *object,
 		/* need to issue a new write op after this */
 		clear_bit(FSCACHE_OBJECT_PENDING_WRITE, &object->flags);
 		ret = 0;
-	} else if (test_bit(FSCACHE_OBJECT_IS_LOOKED_UP, &object->flags)) {
+	} else if (flags & BIT(FSCACHE_OBJECT_IS_LOOKED_UP)) {
 		op->object = object;
 		object->n_ops++;
 		object->n_exclusive++;	/* reads and writes must wait */
@@ -126,12 +207,15 @@ int fscache_submit_exclusive_op(struct fscache_object *object,
 		list_add_tail(&op->pend_link, &object->pending_ops);
 		fscache_stat(&fscache_n_op_pend);
 		ret = 0;
+	} else if (flags & BIT(FSCACHE_OBJECT_KILLED_BY_CACHE)) {
+		op->cancel(op);
+		op->state = FSCACHE_OP_ST_CANCELLED;
+		ret = -ENOBUFS;
 	} else {
-		/* If we're in any other state, there must have been an I/O
-		 * error of some nature.
-		 */
-		ASSERT(test_bit(FSCACHE_IOERROR, &object->cache->flags));
-		ret = -EIO;
+		fscache_report_unexpected_submission(object, op, ostate);
+		op->cancel(op);
+		op->state = FSCACHE_OP_ST_CANCELLED;
+		ret = -ENOBUFS;
 	}
 
 	spin_unlock(&object->lock);
@@ -139,43 +223,6 @@ int fscache_submit_exclusive_op(struct fscache_object *object,
 }
 
 /*
- * report an unexpected submission
- */
-static void fscache_report_unexpected_submission(struct fscache_object *object,
-						 struct fscache_operation *op,
-						 const struct fscache_state *ostate)
-{
-	static bool once_only;
-	struct fscache_operation *p;
-	unsigned n;
-
-	if (once_only)
-		return;
-	once_only = true;
-
-	kdebug("unexpected submission OP%x [OBJ%x %s]",
-	       op->debug_id, object->debug_id, object->state->name);
-	kdebug("objstate=%s [%s]", object->state->name, ostate->name);
-	kdebug("objflags=%lx", object->flags);
-	kdebug("objevent=%lx [%lx]", object->events, object->event_mask);
-	kdebug("ops=%u inp=%u exc=%u",
-	       object->n_ops, object->n_in_progress, object->n_exclusive);
-
-	if (!list_empty(&object->pending_ops)) {
-		n = 0;
-		list_for_each_entry(p, &object->pending_ops, pend_link) {
-			ASSERTCMP(p->object, ==, object);
-			kdebug("%p %p", op->processor, op->release);
-			n++;
-		}
-
-		kdebug("n=%u", n);
-	}
-
-	dump_stack();
-}
-
-/*
  * submit an operation for an object
  * - objects may be submitted only in the following states:
  *   - during object creation (write ops may be submitted)
@@ -187,6 +234,7 @@ int fscache_submit_op(struct fscache_object *object,
 		      struct fscache_operation *op)
 {
 	const struct fscache_state *ostate;
+	unsigned long flags;
 	int ret;
 
 	_enter("{OBJ%x OP%x},{%u}",
@@ -204,7 +252,17 @@ int fscache_submit_op(struct fscache_object *object,
 	smp_rmb();
 
 	op->state = FSCACHE_OP_ST_PENDING;
-	if (fscache_object_is_active(object)) {
+	flags = READ_ONCE(object->flags);
+	if (unlikely(!(flags & BIT(FSCACHE_OBJECT_IS_LIVE)))) {
+		fscache_stat(&fscache_n_op_rejected);
+		op->cancel(op);
+		op->state = FSCACHE_OP_ST_CANCELLED;
+		ret = -ENOBUFS;
+	} else if (unlikely(fscache_cache_is_broken(object))) {
+		op->cancel(op);
+		op->state = FSCACHE_OP_ST_CANCELLED;
+		ret = -EIO;
+	} else if (flags & BIT(FSCACHE_OBJECT_IS_AVAILABLE)) {
 		op->object = object;
 		object->n_ops++;
 
@@ -222,23 +280,21 @@ int fscache_submit_op(struct fscache_object *object,
 			fscache_run_op(object, op);
 		}
 		ret = 0;
-	} else if (test_bit(FSCACHE_OBJECT_IS_LOOKED_UP, &object->flags)) {
+	} else if (flags & BIT(FSCACHE_OBJECT_IS_LOOKED_UP)) {
 		op->object = object;
 		object->n_ops++;
 		atomic_inc(&op->usage);
 		list_add_tail(&op->pend_link, &object->pending_ops);
 		fscache_stat(&fscache_n_op_pend);
 		ret = 0;
-	} else if (fscache_object_is_dying(object)) {
-		fscache_stat(&fscache_n_op_rejected);
+	} else if (flags & BIT(FSCACHE_OBJECT_KILLED_BY_CACHE)) {
+		op->cancel(op);
 		op->state = FSCACHE_OP_ST_CANCELLED;
 		ret = -ENOBUFS;
-	} else if (!test_bit(FSCACHE_IOERROR, &object->cache->flags)) {
+	} else {
 		fscache_report_unexpected_submission(object, op, ostate);
 		ASSERT(!fscache_object_is_active(object));
-		op->state = FSCACHE_OP_ST_CANCELLED;
-		ret = -ENOBUFS;
-	} else {
+		op->cancel(op);
 		op->state = FSCACHE_OP_ST_CANCELLED;
 		ret = -ENOBUFS;
 	}
@@ -293,9 +349,10 @@ void fscache_start_operations(struct fscache_object *object)
  * cancel an operation that's pending on an object
  */
 int fscache_cancel_op(struct fscache_operation *op,
-		      void (*do_cancel)(struct fscache_operation *))
+		      bool cancel_in_progress_op)
 {
 	struct fscache_object *object = op->object;
+	bool put = false;
 	int ret;
 
 	_enter("OBJ%x OP%x}", op->object->debug_id, op->debug_id);
@@ -309,19 +366,37 @@ int fscache_cancel_op(struct fscache_operation *op,
 	ret = -EBUSY;
 	if (op->state == FSCACHE_OP_ST_PENDING) {
 		ASSERT(!list_empty(&op->pend_link));
-		fscache_stat(&fscache_n_op_cancelled);
 		list_del_init(&op->pend_link);
-		if (do_cancel)
-			do_cancel(op);
+		put = true;
+
+		fscache_stat(&fscache_n_op_cancelled);
+		op->cancel(op);
+		op->state = FSCACHE_OP_ST_CANCELLED;
+		if (test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags))
+			object->n_exclusive--;
+		if (test_and_clear_bit(FSCACHE_OP_WAITING, &op->flags))
+			wake_up_bit(&op->flags, FSCACHE_OP_WAITING);
+		ret = 0;
+	} else if (op->state == FSCACHE_OP_ST_IN_PROGRESS && cancel_in_progress_op) {
+		ASSERTCMP(object->n_in_progress, >, 0);
+		if (test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags))
+			object->n_exclusive--;
+		object->n_in_progress--;
+		if (object->n_in_progress == 0)
+			fscache_start_operations(object);
+
+		fscache_stat(&fscache_n_op_cancelled);
+		op->cancel(op);
 		op->state = FSCACHE_OP_ST_CANCELLED;
 		if (test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags))
 			object->n_exclusive--;
 		if (test_and_clear_bit(FSCACHE_OP_WAITING, &op->flags))
 			wake_up_bit(&op->flags, FSCACHE_OP_WAITING);
-		fscache_put_operation(op);
 		ret = 0;
 	}
 
+	if (put)
+		fscache_put_operation(op);
 	spin_unlock(&object->lock);
 	_leave(" = %d", ret);
 	return ret;
@@ -345,6 +420,7 @@ void fscache_cancel_all_ops(struct fscache_object *object)
 		list_del_init(&op->pend_link);
 
 		ASSERTCMP(op->state, ==, FSCACHE_OP_ST_PENDING);
+		op->cancel(op);
 		op->state = FSCACHE_OP_ST_CANCELLED;
 
 		if (test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags))
@@ -377,8 +453,12 @@ void fscache_op_complete(struct fscache_operation *op, bool cancelled)
 
 	spin_lock(&object->lock);
 
-	op->state = cancelled ?
-		FSCACHE_OP_ST_CANCELLED : FSCACHE_OP_ST_COMPLETE;
+	if (!cancelled) {
+		op->state = FSCACHE_OP_ST_COMPLETE;
+	} else {
+		op->cancel(op);
+		op->state = FSCACHE_OP_ST_CANCELLED;
+	}
 
 	if (test_bit(FSCACHE_OP_EXCLUSIVE, &op->flags))
 		object->n_exclusive--;
@@ -409,9 +489,9 @@ void fscache_put_operation(struct fscache_operation *op)
 		return;
 
 	_debug("PUT OP");
-	ASSERTIFCMP(op->state != FSCACHE_OP_ST_COMPLETE,
+	ASSERTIFCMP(op->state != FSCACHE_OP_ST_INITIALISED &&
+		    op->state != FSCACHE_OP_ST_COMPLETE,
 		    op->state, ==, FSCACHE_OP_ST_CANCELLED);
-	op->state = FSCACHE_OP_ST_DEAD;
 
 	fscache_stat(&fscache_n_op_release);
 
@@ -419,37 +499,39 @@ void fscache_put_operation(struct fscache_operation *op)
 		op->release(op);
 		op->release = NULL;
 	}
+	op->state = FSCACHE_OP_ST_DEAD;
 
 	object = op->object;
+	if (likely(object)) {
+		if (test_bit(FSCACHE_OP_DEC_READ_CNT, &op->flags))
+			atomic_dec(&object->n_reads);
+		if (test_bit(FSCACHE_OP_UNUSE_COOKIE, &op->flags))
+			fscache_unuse_cookie(object);
+
+		/* now... we may get called with the object spinlock held, so we
+		 * complete the cleanup here only if we can immediately acquire the
+		 * lock, and defer it otherwise */
+		if (!spin_trylock(&object->lock)) {
+			_debug("defer put");
+			fscache_stat(&fscache_n_op_deferred_release);
+
+			cache = object->cache;
+			spin_lock(&cache->op_gc_list_lock);
+			list_add_tail(&op->pend_link, &cache->op_gc_list);
+			spin_unlock(&cache->op_gc_list_lock);
+			schedule_work(&cache->op_gc);
+			_leave(" [defer]");
+			return;
+		}
 
-	if (test_bit(FSCACHE_OP_DEC_READ_CNT, &op->flags))
-		atomic_dec(&object->n_reads);
-	if (test_bit(FSCACHE_OP_UNUSE_COOKIE, &op->flags))
-		fscache_unuse_cookie(object);
-
-	/* now... we may get called with the object spinlock held, so we
-	 * complete the cleanup here only if we can immediately acquire the
-	 * lock, and defer it otherwise */
-	if (!spin_trylock(&object->lock)) {
-		_debug("defer put");
-		fscache_stat(&fscache_n_op_deferred_release);
+		ASSERTCMP(object->n_ops, >, 0);
+		object->n_ops--;
+		if (object->n_ops == 0)
+			fscache_raise_event(object, FSCACHE_OBJECT_EV_CLEARED);
 
-		cache = object->cache;
-		spin_lock(&cache->op_gc_list_lock);
-		list_add_tail(&op->pend_link, &cache->op_gc_list);
-		spin_unlock(&cache->op_gc_list_lock);
-		schedule_work(&cache->op_gc);
-		_leave(" [defer]");
-		return;
+		spin_unlock(&object->lock);
 	}
 
-	ASSERTCMP(object->n_ops, >, 0);
-	object->n_ops--;
-	if (object->n_ops == 0)
-		fscache_raise_event(object, FSCACHE_OBJECT_EV_CLEARED);
-
-	spin_unlock(&object->lock);
-
 	kfree(op);
 	_leave(" [done]");
 }
diff --git a/fs/fscache/page.c b/fs/fscache/page.c
index de33b3fccca6..483bbc613bf0 100644
--- a/fs/fscache/page.c
+++ b/fs/fscache/page.c
@@ -213,7 +213,7 @@ int __fscache_attr_changed(struct fscache_cookie *cookie)
 		return -ENOMEM;
 	}
 
-	fscache_operation_init(op, fscache_attr_changed_op, NULL);
+	fscache_operation_init(op, fscache_attr_changed_op, NULL, NULL);
 	op->flags = FSCACHE_OP_ASYNC |
 		(1 << FSCACHE_OP_EXCLUSIVE) |
 		(1 << FSCACHE_OP_UNUSE_COOKIE);
@@ -239,7 +239,7 @@ nobufs_dec:
 	wake_cookie = __fscache_unuse_cookie(cookie);
 nobufs:
 	spin_unlock(&cookie->lock);
-	kfree(op);
+	fscache_put_operation(op);
 	if (wake_cookie)
 		__fscache_wake_unused_cookie(cookie);
 	fscache_stat(&fscache_n_attr_changed_nobufs);
@@ -249,6 +249,17 @@ nobufs:
 EXPORT_SYMBOL(__fscache_attr_changed);
 
 /*
+ * Handle cancellation of a pending retrieval op
+ */
+static void fscache_do_cancel_retrieval(struct fscache_operation *_op)
+{
+	struct fscache_retrieval *op =
+		container_of(_op, struct fscache_retrieval, op);
+
+	atomic_set(&op->n_pages, 0);
+}
+
+/*
  * release a retrieval op reference
  */
 static void fscache_release_retrieval_op(struct fscache_operation *_op)
@@ -258,11 +269,12 @@ static void fscache_release_retrieval_op(struct fscache_operation *_op)
 
 	_enter("{OP%x}", op->op.debug_id);
 
-	ASSERTCMP(atomic_read(&op->n_pages), ==, 0);
+	ASSERTIFCMP(op->op.state != FSCACHE_OP_ST_INITIALISED,
+		    atomic_read(&op->n_pages), ==, 0);
 
 	fscache_hist(fscache_retrieval_histogram, op->start_time);
 	if (op->context)
-		fscache_put_context(op->op.object->cookie, op->context);
+		fscache_put_context(op->cookie, op->context);
 
 	_leave("");
 }
@@ -285,15 +297,24 @@ static struct fscache_retrieval *fscache_alloc_retrieval(
 		return NULL;
 	}
 
-	fscache_operation_init(&op->op, NULL, fscache_release_retrieval_op);
+	fscache_operation_init(&op->op, NULL,
+			       fscache_do_cancel_retrieval,
+			       fscache_release_retrieval_op);
 	op->op.flags	= FSCACHE_OP_MYTHREAD |
 		(1UL << FSCACHE_OP_WAITING) |
 		(1UL << FSCACHE_OP_UNUSE_COOKIE);
+	op->cookie	= cookie;
 	op->mapping	= mapping;
 	op->end_io_func	= end_io_func;
 	op->context	= context;
 	op->start_time	= jiffies;
 	INIT_LIST_HEAD(&op->to_do);
+
+	/* Pin the netfs read context in case we need to do the actual netfs
+	 * read because we've encountered a cache read failure.
+	 */
+	if (context)
+		fscache_get_context(op->cookie, context);
 	return op;
 }
 
@@ -330,24 +351,12 @@ int fscache_wait_for_deferred_lookup(struct fscache_cookie *cookie)
 }
 
 /*
- * Handle cancellation of a pending retrieval op
- */
-static void fscache_do_cancel_retrieval(struct fscache_operation *_op)
-{
-	struct fscache_retrieval *op =
-		container_of(_op, struct fscache_retrieval, op);
-
-	atomic_set(&op->n_pages, 0);
-}
-
-/*
  * wait for an object to become active (or dead)
  */
 int fscache_wait_for_operation_activation(struct fscache_object *object,
 					  struct fscache_operation *op,
 					  atomic_t *stat_op_waits,
-					  atomic_t *stat_object_dead,
-					  void (*do_cancel)(struct fscache_operation *))
+					  atomic_t *stat_object_dead)
 {
 	int ret;
 
@@ -359,7 +368,7 @@ int fscache_wait_for_operation_activation(struct fscache_object *object,
 		fscache_stat(stat_op_waits);
 	if (wait_on_bit(&op->flags, FSCACHE_OP_WAITING,
 			TASK_INTERRUPTIBLE) != 0) {
-		ret = fscache_cancel_op(op, do_cancel);
+		ret = fscache_cancel_op(op, false);
 		if (ret == 0)
 			return -ERESTARTSYS;
 
@@ -377,11 +386,13 @@ check_if_dead:
 		_leave(" = -ENOBUFS [cancelled]");
 		return -ENOBUFS;
 	}
-	if (unlikely(fscache_object_is_dead(object))) {
-		pr_err("%s() = -ENOBUFS [obj dead %d]\n", __func__, op->state);
-		fscache_cancel_op(op, do_cancel);
+	if (unlikely(fscache_object_is_dying(object) ||
+		     fscache_cache_is_broken(object))) {
+		enum fscache_operation_state state = op->state;
+		fscache_cancel_op(op, true);
 		if (stat_object_dead)
 			fscache_stat(stat_object_dead);
+		_leave(" = -ENOBUFS [obj dead %d]", state);
 		return -ENOBUFS;
 	}
 	return 0;
@@ -453,17 +464,12 @@ int __fscache_read_or_alloc_page(struct fscache_cookie *cookie,
 
 	fscache_stat(&fscache_n_retrieval_ops);
 
-	/* pin the netfs read context in case we need to do the actual netfs
-	 * read because we've encountered a cache read failure */
-	fscache_get_context(object->cookie, op->context);
-
 	/* we wait for the operation to become active, and then process it
 	 * *here*, in this thread, and not in the thread pool */
 	ret = fscache_wait_for_operation_activation(
 		object, &op->op,
 		__fscache_stat(&fscache_n_retrieval_op_waits),
-		__fscache_stat(&fscache_n_retrievals_object_dead),
-		fscache_do_cancel_retrieval);
+		__fscache_stat(&fscache_n_retrievals_object_dead));
 	if (ret < 0)
 		goto error;
 
@@ -503,7 +509,7 @@ nobufs_unlock:
 	spin_unlock(&cookie->lock);
 	if (wake_cookie)
 		__fscache_wake_unused_cookie(cookie);
-	kfree(op);
+	fscache_put_retrieval(op);
 nobufs:
 	fscache_stat(&fscache_n_retrievals_nobufs);
 	_leave(" = -ENOBUFS");
@@ -584,17 +590,12 @@ int __fscache_read_or_alloc_pages(struct fscache_cookie *cookie,
 
 	fscache_stat(&fscache_n_retrieval_ops);
 
-	/* pin the netfs read context in case we need to do the actual netfs
-	 * read because we've encountered a cache read failure */
-	fscache_get_context(object->cookie, op->context);
-
 	/* we wait for the operation to become active, and then process it
 	 * *here*, in this thread, and not in the thread pool */
 	ret = fscache_wait_for_operation_activation(
 		object, &op->op,
 		__fscache_stat(&fscache_n_retrieval_op_waits),
-		__fscache_stat(&fscache_n_retrievals_object_dead),
-		fscache_do_cancel_retrieval);
+		__fscache_stat(&fscache_n_retrievals_object_dead));
 	if (ret < 0)
 		goto error;
 
@@ -632,7 +633,7 @@ nobufs_unlock_dec:
 	wake_cookie = __fscache_unuse_cookie(cookie);
 nobufs_unlock:
 	spin_unlock(&cookie->lock);
-	kfree(op);
+	fscache_put_retrieval(op);
 	if (wake_cookie)
 		__fscache_wake_unused_cookie(cookie);
 nobufs:
@@ -700,8 +701,7 @@ int __fscache_alloc_page(struct fscache_cookie *cookie,
 	ret = fscache_wait_for_operation_activation(
 		object, &op->op,
 		__fscache_stat(&fscache_n_alloc_op_waits),
-		__fscache_stat(&fscache_n_allocs_object_dead),
-		fscache_do_cancel_retrieval);
+		__fscache_stat(&fscache_n_allocs_object_dead));
 	if (ret < 0)
 		goto error;
 
@@ -726,7 +726,7 @@ nobufs_unlock_dec:
 	wake_cookie = __fscache_unuse_cookie(cookie);
 nobufs_unlock:
 	spin_unlock(&cookie->lock);
-	kfree(op);
+	fscache_put_retrieval(op);
 	if (wake_cookie)
 		__fscache_wake_unused_cookie(cookie);
 nobufs:
@@ -944,7 +944,7 @@ int __fscache_write_page(struct fscache_cookie *cookie,
 	if (!op)
 		goto nomem;
 
-	fscache_operation_init(&op->op, fscache_write_op,
+	fscache_operation_init(&op->op, fscache_write_op, NULL,
 			       fscache_release_write_op);
 	op->op.flags = FSCACHE_OP_ASYNC |
 		(1 << FSCACHE_OP_WAITING) |
@@ -1016,7 +1016,7 @@ already_pending:
 	spin_unlock(&object->lock);
 	spin_unlock(&cookie->lock);
 	radix_tree_preload_end();
-	kfree(op);
+	fscache_put_operation(&op->op);
 	fscache_stat(&fscache_n_stores_ok);
 	_leave(" = 0");
 	return 0;
@@ -1036,7 +1036,7 @@ nobufs_unlock_obj:
 nobufs:
 	spin_unlock(&cookie->lock);
 	radix_tree_preload_end();
-	kfree(op);
+	fscache_put_operation(&op->op);
 	if (wake_cookie)
 		__fscache_wake_unused_cookie(cookie);
 	fscache_stat(&fscache_n_stores_nobufs);
@@ -1044,7 +1044,7 @@ nobufs:
 	return -ENOBUFS;
 
 nomem_free:
-	kfree(op);
+	fscache_put_operation(&op->op);
 nomem:
 	fscache_stat(&fscache_n_stores_oom);
 	_leave(" = -ENOMEM");
diff --git a/fs/fscache/stats.c b/fs/fscache/stats.c
index 40d13c70ef51..7cfa0aacdf6d 100644
--- a/fs/fscache/stats.c
+++ b/fs/fscache/stats.c
@@ -23,6 +23,7 @@ atomic_t fscache_n_op_run;
 atomic_t fscache_n_op_enqueue;
 atomic_t fscache_n_op_requeue;
 atomic_t fscache_n_op_deferred_release;
+atomic_t fscache_n_op_initialised;
 atomic_t fscache_n_op_release;
 atomic_t fscache_n_op_gc;
 atomic_t fscache_n_op_cancelled;
@@ -130,6 +131,11 @@ atomic_t fscache_n_cop_write_page;
 atomic_t fscache_n_cop_uncache_page;
 atomic_t fscache_n_cop_dissociate_pages;
 
+atomic_t fscache_n_cache_no_space_reject;
+atomic_t fscache_n_cache_stale_objects;
+atomic_t fscache_n_cache_retired_objects;
+atomic_t fscache_n_cache_culled_objects;
+
 /*
  * display the general statistics
  */
@@ -246,7 +252,8 @@ static int fscache_stats_show(struct seq_file *m, void *v)
 		   atomic_read(&fscache_n_op_enqueue),
 		   atomic_read(&fscache_n_op_cancelled),
 		   atomic_read(&fscache_n_op_rejected));
-	seq_printf(m, "Ops    : dfr=%u rel=%u gc=%u\n",
+	seq_printf(m, "Ops    : ini=%u dfr=%u rel=%u gc=%u\n",
+		   atomic_read(&fscache_n_op_initialised),
 		   atomic_read(&fscache_n_op_deferred_release),
 		   atomic_read(&fscache_n_op_release),
 		   atomic_read(&fscache_n_op_gc));
@@ -271,6 +278,11 @@ static int fscache_stats_show(struct seq_file *m, void *v)
 		   atomic_read(&fscache_n_cop_write_page),
 		   atomic_read(&fscache_n_cop_uncache_page),
 		   atomic_read(&fscache_n_cop_dissociate_pages));
+	seq_printf(m, "CacheEv: nsp=%d stl=%d rtr=%d cul=%d\n",
+		   atomic_read(&fscache_n_cache_no_space_reject),
+		   atomic_read(&fscache_n_cache_stale_objects),
+		   atomic_read(&fscache_n_cache_retired_objects),
+		   atomic_read(&fscache_n_cache_culled_objects));
 	return 0;
 }
 
diff --git a/fs/fuse/cuse.c b/fs/fuse/cuse.c
index e5bbf748b698..eae2c11268bc 100644
--- a/fs/fuse/cuse.c
+++ b/fs/fuse/cuse.c
@@ -489,6 +489,7 @@ static void cuse_fc_release(struct fuse_conn *fc)
  */
 static int cuse_channel_open(struct inode *inode, struct file *file)
 {
+	struct fuse_dev *fud;
 	struct cuse_conn *cc;
 	int rc;
 
@@ -499,17 +500,22 @@ static int cuse_channel_open(struct inode *inode, struct file *file)
 
 	fuse_conn_init(&cc->fc);
 
+	fud = fuse_dev_alloc(&cc->fc);
+	if (!fud) {
+		kfree(cc);
+		return -ENOMEM;
+	}
+
 	INIT_LIST_HEAD(&cc->list);
 	cc->fc.release = cuse_fc_release;
 
-	cc->fc.connected = 1;
 	cc->fc.initialized = 1;
 	rc = cuse_send_init(cc);
 	if (rc) {
-		fuse_conn_put(&cc->fc);
+		fuse_dev_free(fud);
 		return rc;
 	}
-	file->private_data = &cc->fc;	/* channel owns base reference to cc */
+	file->private_data = fud;
 
 	return 0;
 }
@@ -527,7 +533,8 @@ static int cuse_channel_open(struct inode *inode, struct file *file)
  */
 static int cuse_channel_release(struct inode *inode, struct file *file)
 {
-	struct cuse_conn *cc = fc_to_cc(file->private_data);
+	struct fuse_dev *fud = file->private_data;
+	struct cuse_conn *cc = fc_to_cc(fud->fc);
 	int rc;
 
 	/* remove from the conntbl, no more access from this point on */
diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c
index c8b68ab2e574..80cc1b35d460 100644
--- a/fs/fuse/dev.c
+++ b/fs/fuse/dev.c
@@ -25,13 +25,13 @@ MODULE_ALIAS("devname:fuse");
 
 static struct kmem_cache *fuse_req_cachep;
 
-static struct fuse_conn *fuse_get_conn(struct file *file)
+static struct fuse_dev *fuse_get_dev(struct file *file)
 {
 	/*
 	 * Lockless access is OK, because file->private data is set
 	 * once during mount and is valid until the file is released.
 	 */
-	return file->private_data;
+	return ACCESS_ONCE(file->private_data);
 }
 
 static void fuse_request_init(struct fuse_req *req, struct page **pages,
@@ -48,6 +48,7 @@ static void fuse_request_init(struct fuse_req *req, struct page **pages,
 	req->pages = pages;
 	req->page_descs = page_descs;
 	req->max_pages = npages;
+	__set_bit(FR_PENDING, &req->flags);
 }
 
 static struct fuse_req *__fuse_request_alloc(unsigned npages, gfp_t flags)
@@ -168,6 +169,10 @@ static struct fuse_req *__fuse_get_req(struct fuse_conn *fc, unsigned npages,
 	if (!fc->connected)
 		goto out;
 
+	err = -ECONNREFUSED;
+	if (fc->conn_error)
+		goto out;
+
 	req = fuse_request_alloc(npages);
 	err = -ENOMEM;
 	if (!req) {
@@ -177,8 +182,10 @@ static struct fuse_req *__fuse_get_req(struct fuse_conn *fc, unsigned npages,
 	}
 
 	fuse_req_init_context(req);
-	req->waiting = 1;
-	req->background = for_background;
+	__set_bit(FR_WAITING, &req->flags);
+	if (for_background)
+		__set_bit(FR_BACKGROUND, &req->flags);
+
 	return req;
 
  out:
@@ -268,15 +275,15 @@ struct fuse_req *fuse_get_req_nofail_nopages(struct fuse_conn *fc,
 		req = get_reserved_req(fc, file);
 
 	fuse_req_init_context(req);
-	req->waiting = 1;
-	req->background = 0;
+	__set_bit(FR_WAITING, &req->flags);
+	__clear_bit(FR_BACKGROUND, &req->flags);
 	return req;
 }
 
 void fuse_put_request(struct fuse_conn *fc, struct fuse_req *req)
 {
 	if (atomic_dec_and_test(&req->count)) {
-		if (unlikely(req->background)) {
+		if (test_bit(FR_BACKGROUND, &req->flags)) {
 			/*
 			 * We get here in the unlikely case that a background
 			 * request was allocated but not sent
@@ -287,8 +294,10 @@ void fuse_put_request(struct fuse_conn *fc, struct fuse_req *req)
 			spin_unlock(&fc->lock);
 		}
 
-		if (req->waiting)
+		if (test_bit(FR_WAITING, &req->flags)) {
+			__clear_bit(FR_WAITING, &req->flags);
 			atomic_dec(&fc->num_waiting);
+		}
 
 		if (req->stolen_file)
 			put_reserved_req(fc, req);
@@ -309,46 +318,38 @@ static unsigned len_args(unsigned numargs, struct fuse_arg *args)
 	return nbytes;
 }
 
-static u64 fuse_get_unique(struct fuse_conn *fc)
+static u64 fuse_get_unique(struct fuse_iqueue *fiq)
 {
-	fc->reqctr++;
-	/* zero is special */
-	if (fc->reqctr == 0)
-		fc->reqctr = 1;
-
-	return fc->reqctr;
+	return ++fiq->reqctr;
 }
 
-static void queue_request(struct fuse_conn *fc, struct fuse_req *req)
+static void queue_request(struct fuse_iqueue *fiq, struct fuse_req *req)
 {
 	req->in.h.len = sizeof(struct fuse_in_header) +
 		len_args(req->in.numargs, (struct fuse_arg *) req->in.args);
-	list_add_tail(&req->list, &fc->pending);
-	req->state = FUSE_REQ_PENDING;
-	if (!req->waiting) {
-		req->waiting = 1;
-		atomic_inc(&fc->num_waiting);
-	}
-	wake_up(&fc->waitq);
-	kill_fasync(&fc->fasync, SIGIO, POLL_IN);
+	list_add_tail(&req->list, &fiq->pending);
+	wake_up_locked(&fiq->waitq);
+	kill_fasync(&fiq->fasync, SIGIO, POLL_IN);
 }
 
 void fuse_queue_forget(struct fuse_conn *fc, struct fuse_forget_link *forget,
 		       u64 nodeid, u64 nlookup)
 {
+	struct fuse_iqueue *fiq = &fc->iq;
+
 	forget->forget_one.nodeid = nodeid;
 	forget->forget_one.nlookup = nlookup;
 
-	spin_lock(&fc->lock);
-	if (fc->connected) {
-		fc->forget_list_tail->next = forget;
-		fc->forget_list_tail = forget;
-		wake_up(&fc->waitq);
-		kill_fasync(&fc->fasync, SIGIO, POLL_IN);
+	spin_lock(&fiq->waitq.lock);
+	if (fiq->connected) {
+		fiq->forget_list_tail->next = forget;
+		fiq->forget_list_tail = forget;
+		wake_up_locked(&fiq->waitq);
+		kill_fasync(&fiq->fasync, SIGIO, POLL_IN);
 	} else {
 		kfree(forget);
 	}
-	spin_unlock(&fc->lock);
+	spin_unlock(&fiq->waitq.lock);
 }
 
 static void flush_bg_queue(struct fuse_conn *fc)
@@ -356,12 +357,15 @@ static void flush_bg_queue(struct fuse_conn *fc)
 	while (fc->active_background < fc->max_background &&
 	       !list_empty(&fc->bg_queue)) {
 		struct fuse_req *req;
+		struct fuse_iqueue *fiq = &fc->iq;
 
 		req = list_entry(fc->bg_queue.next, struct fuse_req, list);
 		list_del(&req->list);
 		fc->active_background++;
-		req->in.h.unique = fuse_get_unique(fc);
-		queue_request(fc, req);
+		spin_lock(&fiq->waitq.lock);
+		req->in.h.unique = fuse_get_unique(fiq);
+		queue_request(fiq, req);
+		spin_unlock(&fiq->waitq.lock);
 	}
 }
 
@@ -372,20 +376,22 @@ static void flush_bg_queue(struct fuse_conn *fc)
  * was closed.  The requester thread is woken up (if still waiting),
  * the 'end' callback is called if given, else the reference to the
  * request is released
- *
- * Called with fc->lock, unlocks it
  */
 static void request_end(struct fuse_conn *fc, struct fuse_req *req)
-__releases(fc->lock)
 {
-	void (*end) (struct fuse_conn *, struct fuse_req *) = req->end;
-	req->end = NULL;
-	list_del(&req->list);
-	list_del(&req->intr_entry);
-	req->state = FUSE_REQ_FINISHED;
-	if (req->background) {
-		req->background = 0;
+	struct fuse_iqueue *fiq = &fc->iq;
+
+	if (test_and_set_bit(FR_FINISHED, &req->flags))
+		return;
 
+	spin_lock(&fiq->waitq.lock);
+	list_del_init(&req->intr_entry);
+	spin_unlock(&fiq->waitq.lock);
+	WARN_ON(test_bit(FR_PENDING, &req->flags));
+	WARN_ON(test_bit(FR_SENT, &req->flags));
+	if (test_bit(FR_BACKGROUND, &req->flags)) {
+		spin_lock(&fc->lock);
+		clear_bit(FR_BACKGROUND, &req->flags);
 		if (fc->num_background == fc->max_background)
 			fc->blocked = 0;
 
@@ -401,122 +407,105 @@ __releases(fc->lock)
 		fc->num_background--;
 		fc->active_background--;
 		flush_bg_queue(fc);
+		spin_unlock(&fc->lock);
 	}
-	spin_unlock(&fc->lock);
 	wake_up(&req->waitq);
-	if (end)
-		end(fc, req);
+	if (req->end)
+		req->end(fc, req);
 	fuse_put_request(fc, req);
 }
 
-static void wait_answer_interruptible(struct fuse_conn *fc,
-				      struct fuse_req *req)
-__releases(fc->lock)
-__acquires(fc->lock)
-{
-	if (signal_pending(current))
-		return;
-
-	spin_unlock(&fc->lock);
-	wait_event_interruptible(req->waitq, req->state == FUSE_REQ_FINISHED);
-	spin_lock(&fc->lock);
-}
-
-static void queue_interrupt(struct fuse_conn *fc, struct fuse_req *req)
+static void queue_interrupt(struct fuse_iqueue *fiq, struct fuse_req *req)
 {
-	list_add_tail(&req->intr_entry, &fc->interrupts);
-	wake_up(&fc->waitq);
-	kill_fasync(&fc->fasync, SIGIO, POLL_IN);
+	spin_lock(&fiq->waitq.lock);
+	if (list_empty(&req->intr_entry)) {
+		list_add_tail(&req->intr_entry, &fiq->interrupts);
+		wake_up_locked(&fiq->waitq);
+	}
+	spin_unlock(&fiq->waitq.lock);
+	kill_fasync(&fiq->fasync, SIGIO, POLL_IN);
 }
 
 static void request_wait_answer(struct fuse_conn *fc, struct fuse_req *req)
-__releases(fc->lock)
-__acquires(fc->lock)
 {
+	struct fuse_iqueue *fiq = &fc->iq;
+	int err;
+
 	if (!fc->no_interrupt) {
 		/* Any signal may interrupt this */
-		wait_answer_interruptible(fc, req);
-
-		if (req->aborted)
-			goto aborted;
-		if (req->state == FUSE_REQ_FINISHED)
+		err = wait_event_interruptible(req->waitq,
+					test_bit(FR_FINISHED, &req->flags));
+		if (!err)
 			return;
 
-		req->interrupted = 1;
-		if (req->state == FUSE_REQ_SENT)
-			queue_interrupt(fc, req);
+		set_bit(FR_INTERRUPTED, &req->flags);
+		/* matches barrier in fuse_dev_do_read() */
+		smp_mb__after_atomic();
+		if (test_bit(FR_SENT, &req->flags))
+			queue_interrupt(fiq, req);
 	}
 
-	if (!req->force) {
+	if (!test_bit(FR_FORCE, &req->flags)) {
 		sigset_t oldset;
 
 		/* Only fatal signals may interrupt this */
 		block_sigs(&oldset);
-		wait_answer_interruptible(fc, req);
+		err = wait_event_interruptible(req->waitq,
+					test_bit(FR_FINISHED, &req->flags));
 		restore_sigs(&oldset);
 
-		if (req->aborted)
-			goto aborted;
-		if (req->state == FUSE_REQ_FINISHED)
+		if (!err)
 			return;
 
+		spin_lock(&fiq->waitq.lock);
 		/* Request is not yet in userspace, bail out */
-		if (req->state == FUSE_REQ_PENDING) {
+		if (test_bit(FR_PENDING, &req->flags)) {
 			list_del(&req->list);
+			spin_unlock(&fiq->waitq.lock);
 			__fuse_put_request(req);
 			req->out.h.error = -EINTR;
 			return;
 		}
+		spin_unlock(&fiq->waitq.lock);
 	}
 
 	/*
 	 * Either request is already in userspace, or it was forced.
 	 * Wait it out.
 	 */
-	spin_unlock(&fc->lock);
-	wait_event(req->waitq, req->state == FUSE_REQ_FINISHED);
-	spin_lock(&fc->lock);
-
-	if (!req->aborted)
-		return;
-
- aborted:
-	BUG_ON(req->state != FUSE_REQ_FINISHED);
-	if (req->locked) {
-		/* This is uninterruptible sleep, because data is
-		   being copied to/from the buffers of req.  During
-		   locked state, there mustn't be any filesystem
-		   operation (e.g. page fault), since that could lead
-		   to deadlock */
-		spin_unlock(&fc->lock);
-		wait_event(req->waitq, !req->locked);
-		spin_lock(&fc->lock);
-	}
+	wait_event(req->waitq, test_bit(FR_FINISHED, &req->flags));
 }
 
 static void __fuse_request_send(struct fuse_conn *fc, struct fuse_req *req)
 {
-	BUG_ON(req->background);
-	spin_lock(&fc->lock);
-	if (!fc->connected)
+	struct fuse_iqueue *fiq = &fc->iq;
+
+	BUG_ON(test_bit(FR_BACKGROUND, &req->flags));
+	spin_lock(&fiq->waitq.lock);
+	if (!fiq->connected) {
+		spin_unlock(&fiq->waitq.lock);
 		req->out.h.error = -ENOTCONN;
-	else if (fc->conn_error)
-		req->out.h.error = -ECONNREFUSED;
-	else {
-		req->in.h.unique = fuse_get_unique(fc);
-		queue_request(fc, req);
+	} else {
+		req->in.h.unique = fuse_get_unique(fiq);
+		queue_request(fiq, req);
 		/* acquire extra reference, since request is still needed
 		   after request_end() */
 		__fuse_get_request(req);
+		spin_unlock(&fiq->waitq.lock);
 
 		request_wait_answer(fc, req);
+		/* Pairs with smp_wmb() in request_end() */
+		smp_rmb();
 	}
-	spin_unlock(&fc->lock);
 }
 
 void fuse_request_send(struct fuse_conn *fc, struct fuse_req *req)
 {
-	req->isreply = 1;
+	__set_bit(FR_ISREPLY, &req->flags);
+	if (!test_bit(FR_WAITING, &req->flags)) {
+		__set_bit(FR_WAITING, &req->flags);
+		atomic_inc(&fc->num_waiting);
+	}
 	__fuse_request_send(fc, req);
 }
 EXPORT_SYMBOL_GPL(fuse_request_send);
@@ -586,10 +575,20 @@ ssize_t fuse_simple_request(struct fuse_conn *fc, struct fuse_args *args)
 	return ret;
 }
 
-static void fuse_request_send_nowait_locked(struct fuse_conn *fc,
-					    struct fuse_req *req)
+/*
+ * Called under fc->lock
+ *
+ * fc->connected must have been checked previously
+ */
+void fuse_request_send_background_locked(struct fuse_conn *fc,
+					 struct fuse_req *req)
 {
-	BUG_ON(!req->background);
+	BUG_ON(!test_bit(FR_BACKGROUND, &req->flags));
+	if (!test_bit(FR_WAITING, &req->flags)) {
+		__set_bit(FR_WAITING, &req->flags);
+		atomic_inc(&fc->num_waiting);
+	}
+	__set_bit(FR_ISREPLY, &req->flags);
 	fc->num_background++;
 	if (fc->num_background == fc->max_background)
 		fc->blocked = 1;
@@ -602,54 +601,40 @@ static void fuse_request_send_nowait_locked(struct fuse_conn *fc,
 	flush_bg_queue(fc);
 }
 
-static void fuse_request_send_nowait(struct fuse_conn *fc, struct fuse_req *req)
+void fuse_request_send_background(struct fuse_conn *fc, struct fuse_req *req)
 {
+	BUG_ON(!req->end);
 	spin_lock(&fc->lock);
 	if (fc->connected) {
-		fuse_request_send_nowait_locked(fc, req);
+		fuse_request_send_background_locked(fc, req);
 		spin_unlock(&fc->lock);
 	} else {
+		spin_unlock(&fc->lock);
 		req->out.h.error = -ENOTCONN;
-		request_end(fc, req);
+		req->end(fc, req);
+		fuse_put_request(fc, req);
 	}
 }
-
-void fuse_request_send_background(struct fuse_conn *fc, struct fuse_req *req)
-{
-	req->isreply = 1;
-	fuse_request_send_nowait(fc, req);
-}
 EXPORT_SYMBOL_GPL(fuse_request_send_background);
 
 static int fuse_request_send_notify_reply(struct fuse_conn *fc,
 					  struct fuse_req *req, u64 unique)
 {
 	int err = -ENODEV;
+	struct fuse_iqueue *fiq = &fc->iq;
 
-	req->isreply = 0;
+	__clear_bit(FR_ISREPLY, &req->flags);
 	req->in.h.unique = unique;
-	spin_lock(&fc->lock);
-	if (fc->connected) {
-		queue_request(fc, req);
+	spin_lock(&fiq->waitq.lock);
+	if (fiq->connected) {
+		queue_request(fiq, req);
 		err = 0;
 	}
-	spin_unlock(&fc->lock);
+	spin_unlock(&fiq->waitq.lock);
 
 	return err;
 }
 
-/*
- * Called under fc->lock
- *
- * fc->connected must have been checked previously
- */
-void fuse_request_send_background_locked(struct fuse_conn *fc,
-					 struct fuse_req *req)
-{
-	req->isreply = 1;
-	fuse_request_send_nowait_locked(fc, req);
-}
-
 void fuse_force_forget(struct file *file, u64 nodeid)
 {
 	struct inode *inode = file_inode(file);
@@ -665,7 +650,7 @@ void fuse_force_forget(struct file *file, u64 nodeid)
 	req->in.numargs = 1;
 	req->in.args[0].size = sizeof(inarg);
 	req->in.args[0].value = &inarg;
-	req->isreply = 0;
+	__clear_bit(FR_ISREPLY, &req->flags);
 	__fuse_request_send(fc, req);
 	/* ignore errors */
 	fuse_put_request(fc, req);
@@ -676,38 +661,39 @@ void fuse_force_forget(struct file *file, u64 nodeid)
  * anything that could cause a page-fault.  If the request was already
  * aborted bail out.
  */
-static int lock_request(struct fuse_conn *fc, struct fuse_req *req)
+static int lock_request(struct fuse_req *req)
 {
 	int err = 0;
 	if (req) {
-		spin_lock(&fc->lock);
-		if (req->aborted)
+		spin_lock(&req->waitq.lock);
+		if (test_bit(FR_ABORTED, &req->flags))
 			err = -ENOENT;
 		else
-			req->locked = 1;
-		spin_unlock(&fc->lock);
+			set_bit(FR_LOCKED, &req->flags);
+		spin_unlock(&req->waitq.lock);
 	}
 	return err;
 }
 
 /*
- * Unlock request.  If it was aborted during being locked, the
- * requester thread is currently waiting for it to be unlocked, so
- * wake it up.
+ * Unlock request.  If it was aborted while locked, caller is responsible
+ * for unlocking and ending the request.
  */
-static void unlock_request(struct fuse_conn *fc, struct fuse_req *req)
+static int unlock_request(struct fuse_req *req)
 {
+	int err = 0;
 	if (req) {
-		spin_lock(&fc->lock);
-		req->locked = 0;
-		if (req->aborted)
-			wake_up(&req->waitq);
-		spin_unlock(&fc->lock);
+		spin_lock(&req->waitq.lock);
+		if (test_bit(FR_ABORTED, &req->flags))
+			err = -ENOENT;
+		else
+			clear_bit(FR_LOCKED, &req->flags);
+		spin_unlock(&req->waitq.lock);
 	}
+	return err;
 }
 
 struct fuse_copy_state {
-	struct fuse_conn *fc;
 	int write;
 	struct fuse_req *req;
 	struct iov_iter *iter;
@@ -721,13 +707,10 @@ struct fuse_copy_state {
 	unsigned move_pages:1;
 };
 
-static void fuse_copy_init(struct fuse_copy_state *cs,
-			   struct fuse_conn *fc,
-			   int write,
+static void fuse_copy_init(struct fuse_copy_state *cs, int write,
 			   struct iov_iter *iter)
 {
 	memset(cs, 0, sizeof(*cs));
-	cs->fc = fc;
 	cs->write = write;
 	cs->iter = iter;
 }
@@ -760,7 +743,10 @@ static int fuse_copy_fill(struct fuse_copy_state *cs)
 	struct page *page;
 	int err;
 
-	unlock_request(cs->fc, cs->req);
+	err = unlock_request(cs->req);
+	if (err)
+		return err;
+
 	fuse_copy_finish(cs);
 	if (cs->pipebufs) {
 		struct pipe_buffer *buf = cs->pipebufs;
@@ -809,7 +795,7 @@ static int fuse_copy_fill(struct fuse_copy_state *cs)
 		iov_iter_advance(cs->iter, err);
 	}
 
-	return lock_request(cs->fc, cs->req);
+	return lock_request(cs->req);
 }
 
 /* Do as much copy to/from userspace buffer as we can */
@@ -860,7 +846,10 @@ static int fuse_try_move_page(struct fuse_copy_state *cs, struct page **pagep)
 	struct page *newpage;
 	struct pipe_buffer *buf = cs->pipebufs;
 
-	unlock_request(cs->fc, cs->req);
+	err = unlock_request(cs->req);
+	if (err)
+		return err;
+
 	fuse_copy_finish(cs);
 
 	err = buf->ops->confirm(cs->pipe, buf);
@@ -914,12 +903,12 @@ static int fuse_try_move_page(struct fuse_copy_state *cs, struct page **pagep)
 		lru_cache_add_file(newpage);
 
 	err = 0;
-	spin_lock(&cs->fc->lock);
-	if (cs->req->aborted)
+	spin_lock(&cs->req->waitq.lock);
+	if (test_bit(FR_ABORTED, &cs->req->flags))
 		err = -ENOENT;
 	else
 		*pagep = newpage;
-	spin_unlock(&cs->fc->lock);
+	spin_unlock(&cs->req->waitq.lock);
 
 	if (err) {
 		unlock_page(newpage);
@@ -939,7 +928,7 @@ out_fallback:
 	cs->pg = buf->page;
 	cs->offset = buf->offset;
 
-	err = lock_request(cs->fc, cs->req);
+	err = lock_request(cs->req);
 	if (err)
 		return err;
 
@@ -950,11 +939,15 @@ static int fuse_ref_page(struct fuse_copy_state *cs, struct page *page,
 			 unsigned offset, unsigned count)
 {
 	struct pipe_buffer *buf;
+	int err;
 
 	if (cs->nr_segs == cs->pipe->buffers)
 		return -EIO;
 
-	unlock_request(cs->fc, cs->req);
+	err = unlock_request(cs->req);
+	if (err)
+		return err;
+
 	fuse_copy_finish(cs);
 
 	buf = cs->pipebufs;
@@ -1065,36 +1058,15 @@ static int fuse_copy_args(struct fuse_copy_state *cs, unsigned numargs,
 	return err;
 }
 
-static int forget_pending(struct fuse_conn *fc)
+static int forget_pending(struct fuse_iqueue *fiq)
 {
-	return fc->forget_list_head.next != NULL;
+	return fiq->forget_list_head.next != NULL;
 }
 
-static int request_pending(struct fuse_conn *fc)
+static int request_pending(struct fuse_iqueue *fiq)
 {
-	return !list_empty(&fc->pending) || !list_empty(&fc->interrupts) ||
-		forget_pending(fc);
-}
-
-/* Wait until a request is available on the pending list */
-static void request_wait(struct fuse_conn *fc)
-__releases(fc->lock)
-__acquires(fc->lock)
-{
-	DECLARE_WAITQUEUE(wait, current);
-
-	add_wait_queue_exclusive(&fc->waitq, &wait);
-	while (fc->connected && !request_pending(fc)) {
-		set_current_state(TASK_INTERRUPTIBLE);
-		if (signal_pending(current))
-			break;
-
-		spin_unlock(&fc->lock);
-		schedule();
-		spin_lock(&fc->lock);
-	}
-	set_current_state(TASK_RUNNING);
-	remove_wait_queue(&fc->waitq, &wait);
+	return !list_empty(&fiq->pending) || !list_empty(&fiq->interrupts) ||
+		forget_pending(fiq);
 }
 
 /*
@@ -1103,11 +1075,12 @@ __acquires(fc->lock)
  * Unlike other requests this is assembled on demand, without a need
  * to allocate a separate fuse_req structure.
  *
- * Called with fc->lock held, releases it
+ * Called with fiq->waitq.lock held, releases it
  */
-static int fuse_read_interrupt(struct fuse_conn *fc, struct fuse_copy_state *cs,
+static int fuse_read_interrupt(struct fuse_iqueue *fiq,
+			       struct fuse_copy_state *cs,
 			       size_t nbytes, struct fuse_req *req)
-__releases(fc->lock)
+__releases(fiq->waitq.lock)
 {
 	struct fuse_in_header ih;
 	struct fuse_interrupt_in arg;
@@ -1115,7 +1088,7 @@ __releases(fc->lock)
 	int err;
 
 	list_del_init(&req->intr_entry);
-	req->intr_unique = fuse_get_unique(fc);
+	req->intr_unique = fuse_get_unique(fiq);
 	memset(&ih, 0, sizeof(ih));
 	memset(&arg, 0, sizeof(arg));
 	ih.len = reqsize;
@@ -1123,7 +1096,7 @@ __releases(fc->lock)
 	ih.unique = req->intr_unique;
 	arg.unique = req->in.h.unique;
 
-	spin_unlock(&fc->lock);
+	spin_unlock(&fiq->waitq.lock);
 	if (nbytes < reqsize)
 		return -EINVAL;
 
@@ -1135,21 +1108,21 @@ __releases(fc->lock)
 	return err ? err : reqsize;
 }
 
-static struct fuse_forget_link *dequeue_forget(struct fuse_conn *fc,
+static struct fuse_forget_link *dequeue_forget(struct fuse_iqueue *fiq,
 					       unsigned max,
 					       unsigned *countp)
 {
-	struct fuse_forget_link *head = fc->forget_list_head.next;
+	struct fuse_forget_link *head = fiq->forget_list_head.next;
 	struct fuse_forget_link **newhead = &head;
 	unsigned count;
 
 	for (count = 0; *newhead != NULL && count < max; count++)
 		newhead = &(*newhead)->next;
 
-	fc->forget_list_head.next = *newhead;
+	fiq->forget_list_head.next = *newhead;
 	*newhead = NULL;
-	if (fc->forget_list_head.next == NULL)
-		fc->forget_list_tail = &fc->forget_list_head;
+	if (fiq->forget_list_head.next == NULL)
+		fiq->forget_list_tail = &fiq->forget_list_head;
 
 	if (countp != NULL)
 		*countp = count;
@@ -1157,24 +1130,24 @@ static struct fuse_forget_link *dequeue_forget(struct fuse_conn *fc,
 	return head;
 }
 
-static int fuse_read_single_forget(struct fuse_conn *fc,
+static int fuse_read_single_forget(struct fuse_iqueue *fiq,
 				   struct fuse_copy_state *cs,
 				   size_t nbytes)
-__releases(fc->lock)
+__releases(fiq->waitq.lock)
 {
 	int err;
-	struct fuse_forget_link *forget = dequeue_forget(fc, 1, NULL);
+	struct fuse_forget_link *forget = dequeue_forget(fiq, 1, NULL);
 	struct fuse_forget_in arg = {
 		.nlookup = forget->forget_one.nlookup,
 	};
 	struct fuse_in_header ih = {
 		.opcode = FUSE_FORGET,
 		.nodeid = forget->forget_one.nodeid,
-		.unique = fuse_get_unique(fc),
+		.unique = fuse_get_unique(fiq),
 		.len = sizeof(ih) + sizeof(arg),
 	};
 
-	spin_unlock(&fc->lock);
+	spin_unlock(&fiq->waitq.lock);
 	kfree(forget);
 	if (nbytes < ih.len)
 		return -EINVAL;
@@ -1190,9 +1163,9 @@ __releases(fc->lock)
 	return ih.len;
 }
 
-static int fuse_read_batch_forget(struct fuse_conn *fc,
+static int fuse_read_batch_forget(struct fuse_iqueue *fiq,
 				   struct fuse_copy_state *cs, size_t nbytes)
-__releases(fc->lock)
+__releases(fiq->waitq.lock)
 {
 	int err;
 	unsigned max_forgets;
@@ -1201,18 +1174,18 @@ __releases(fc->lock)
 	struct fuse_batch_forget_in arg = { .count = 0 };
 	struct fuse_in_header ih = {
 		.opcode = FUSE_BATCH_FORGET,
-		.unique = fuse_get_unique(fc),
+		.unique = fuse_get_unique(fiq),
 		.len = sizeof(ih) + sizeof(arg),
 	};
 
 	if (nbytes < ih.len) {
-		spin_unlock(&fc->lock);
+		spin_unlock(&fiq->waitq.lock);
 		return -EINVAL;
 	}
 
 	max_forgets = (nbytes - ih.len) / sizeof(struct fuse_forget_one);
-	head = dequeue_forget(fc, max_forgets, &count);
-	spin_unlock(&fc->lock);
+	head = dequeue_forget(fiq, max_forgets, &count);
+	spin_unlock(&fiq->waitq.lock);
 
 	arg.count = count;
 	ih.len += count * sizeof(struct fuse_forget_one);
@@ -1239,14 +1212,15 @@ __releases(fc->lock)
 	return ih.len;
 }
 
-static int fuse_read_forget(struct fuse_conn *fc, struct fuse_copy_state *cs,
+static int fuse_read_forget(struct fuse_conn *fc, struct fuse_iqueue *fiq,
+			    struct fuse_copy_state *cs,
 			    size_t nbytes)
-__releases(fc->lock)
+__releases(fiq->waitq.lock)
 {
-	if (fc->minor < 16 || fc->forget_list_head.next->next == NULL)
-		return fuse_read_single_forget(fc, cs, nbytes);
+	if (fc->minor < 16 || fiq->forget_list_head.next->next == NULL)
+		return fuse_read_single_forget(fiq, cs, nbytes);
 	else
-		return fuse_read_batch_forget(fc, cs, nbytes);
+		return fuse_read_batch_forget(fiq, cs, nbytes);
 }
 
 /*
@@ -1258,46 +1232,51 @@ __releases(fc->lock)
  * request_end().  Otherwise add it to the processing list, and set
  * the 'sent' flag.
  */
-static ssize_t fuse_dev_do_read(struct fuse_conn *fc, struct file *file,
+static ssize_t fuse_dev_do_read(struct fuse_dev *fud, struct file *file,
 				struct fuse_copy_state *cs, size_t nbytes)
 {
-	int err;
+	ssize_t err;
+	struct fuse_conn *fc = fud->fc;
+	struct fuse_iqueue *fiq = &fc->iq;
+	struct fuse_pqueue *fpq = &fud->pq;
 	struct fuse_req *req;
 	struct fuse_in *in;
 	unsigned reqsize;
 
  restart:
-	spin_lock(&fc->lock);
+	spin_lock(&fiq->waitq.lock);
 	err = -EAGAIN;
-	if ((file->f_flags & O_NONBLOCK) && fc->connected &&
-	    !request_pending(fc))
+	if ((file->f_flags & O_NONBLOCK) && fiq->connected &&
+	    !request_pending(fiq))
 		goto err_unlock;
 
-	request_wait(fc);
-	err = -ENODEV;
-	if (!fc->connected)
+	err = wait_event_interruptible_exclusive_locked(fiq->waitq,
+				!fiq->connected || request_pending(fiq));
+	if (err)
 		goto err_unlock;
-	err = -ERESTARTSYS;
-	if (!request_pending(fc))
+
+	err = -ENODEV;
+	if (!fiq->connected)
 		goto err_unlock;
 
-	if (!list_empty(&fc->interrupts)) {
-		req = list_entry(fc->interrupts.next, struct fuse_req,
+	if (!list_empty(&fiq->interrupts)) {
+		req = list_entry(fiq->interrupts.next, struct fuse_req,
 				 intr_entry);
-		return fuse_read_interrupt(fc, cs, nbytes, req);
+		return fuse_read_interrupt(fiq, cs, nbytes, req);
 	}
 
-	if (forget_pending(fc)) {
-		if (list_empty(&fc->pending) || fc->forget_batch-- > 0)
-			return fuse_read_forget(fc, cs, nbytes);
+	if (forget_pending(fiq)) {
+		if (list_empty(&fiq->pending) || fiq->forget_batch-- > 0)
+			return fuse_read_forget(fc, fiq, cs, nbytes);
 
-		if (fc->forget_batch <= -8)
-			fc->forget_batch = 16;
+		if (fiq->forget_batch <= -8)
+			fiq->forget_batch = 16;
 	}
 
-	req = list_entry(fc->pending.next, struct fuse_req, list);
-	req->state = FUSE_REQ_READING;
-	list_move(&req->list, &fc->io);
+	req = list_entry(fiq->pending.next, struct fuse_req, list);
+	clear_bit(FR_PENDING, &req->flags);
+	list_del_init(&req->list);
+	spin_unlock(&fiq->waitq.lock);
 
 	in = &req->in;
 	reqsize = in->h.len;
@@ -1310,37 +1289,48 @@ static ssize_t fuse_dev_do_read(struct fuse_conn *fc, struct file *file,
 		request_end(fc, req);
 		goto restart;
 	}
-	spin_unlock(&fc->lock);
+	spin_lock(&fpq->lock);
+	list_add(&req->list, &fpq->io);
+	spin_unlock(&fpq->lock);
 	cs->req = req;
 	err = fuse_copy_one(cs, &in->h, sizeof(in->h));
 	if (!err)
 		err = fuse_copy_args(cs, in->numargs, in->argpages,
 				     (struct fuse_arg *) in->args, 0);
 	fuse_copy_finish(cs);
-	spin_lock(&fc->lock);
-	req->locked = 0;
-	if (req->aborted) {
-		request_end(fc, req);
-		return -ENODEV;
+	spin_lock(&fpq->lock);
+	clear_bit(FR_LOCKED, &req->flags);
+	if (!fpq->connected) {
+		err = -ENODEV;
+		goto out_end;
 	}
 	if (err) {
 		req->out.h.error = -EIO;
-		request_end(fc, req);
-		return err;
+		goto out_end;
 	}
-	if (!req->isreply)
-		request_end(fc, req);
-	else {
-		req->state = FUSE_REQ_SENT;
-		list_move_tail(&req->list, &fc->processing);
-		if (req->interrupted)
-			queue_interrupt(fc, req);
-		spin_unlock(&fc->lock);
+	if (!test_bit(FR_ISREPLY, &req->flags)) {
+		err = reqsize;
+		goto out_end;
 	}
+	list_move_tail(&req->list, &fpq->processing);
+	spin_unlock(&fpq->lock);
+	set_bit(FR_SENT, &req->flags);
+	/* matches barrier in request_wait_answer() */
+	smp_mb__after_atomic();
+	if (test_bit(FR_INTERRUPTED, &req->flags))
+		queue_interrupt(fiq, req);
+
 	return reqsize;
 
+out_end:
+	if (!test_bit(FR_PRIVATE, &req->flags))
+		list_del_init(&req->list);
+	spin_unlock(&fpq->lock);
+	request_end(fc, req);
+	return err;
+
  err_unlock:
-	spin_unlock(&fc->lock);
+	spin_unlock(&fiq->waitq.lock);
 	return err;
 }
 
@@ -1359,16 +1349,17 @@ static ssize_t fuse_dev_read(struct kiocb *iocb, struct iov_iter *to)
 {
 	struct fuse_copy_state cs;
 	struct file *file = iocb->ki_filp;
-	struct fuse_conn *fc = fuse_get_conn(file);
-	if (!fc)
+	struct fuse_dev *fud = fuse_get_dev(file);
+
+	if (!fud)
 		return -EPERM;
 
 	if (!iter_is_iovec(to))
 		return -EINVAL;
 
-	fuse_copy_init(&cs, fc, 1, to);
+	fuse_copy_init(&cs, 1, to);
 
-	return fuse_dev_do_read(fc, file, &cs, iov_iter_count(to));
+	return fuse_dev_do_read(fud, file, &cs, iov_iter_count(to));
 }
 
 static ssize_t fuse_dev_splice_read(struct file *in, loff_t *ppos,
@@ -1380,18 +1371,19 @@ static ssize_t fuse_dev_splice_read(struct file *in, loff_t *ppos,
 	int do_wakeup = 0;
 	struct pipe_buffer *bufs;
 	struct fuse_copy_state cs;
-	struct fuse_conn *fc = fuse_get_conn(in);
-	if (!fc)
+	struct fuse_dev *fud = fuse_get_dev(in);
+
+	if (!fud)
 		return -EPERM;
 
 	bufs = kmalloc(pipe->buffers * sizeof(struct pipe_buffer), GFP_KERNEL);
 	if (!bufs)
 		return -ENOMEM;
 
-	fuse_copy_init(&cs, fc, 1, NULL);
+	fuse_copy_init(&cs, 1, NULL);
 	cs.pipebufs = bufs;
 	cs.pipe = pipe;
-	ret = fuse_dev_do_read(fc, in, &cs, len);
+	ret = fuse_dev_do_read(fud, in, &cs, len);
 	if (ret < 0)
 		goto out;
 
@@ -1830,11 +1822,11 @@ static int fuse_notify(struct fuse_conn *fc, enum fuse_notify_code code,
 }
 
 /* Look up request on processing list by unique ID */
-static struct fuse_req *request_find(struct fuse_conn *fc, u64 unique)
+static struct fuse_req *request_find(struct fuse_pqueue *fpq, u64 unique)
 {
 	struct fuse_req *req;
 
-	list_for_each_entry(req, &fc->processing, list) {
+	list_for_each_entry(req, &fpq->processing, list) {
 		if (req->in.h.unique == unique || req->intr_unique == unique)
 			return req;
 	}
@@ -1871,10 +1863,12 @@ static int copy_out_args(struct fuse_copy_state *cs, struct fuse_out *out,
  * it from the list and copy the rest of the buffer to the request.
  * The request is finished by calling request_end()
  */
-static ssize_t fuse_dev_do_write(struct fuse_conn *fc,
+static ssize_t fuse_dev_do_write(struct fuse_dev *fud,
 				 struct fuse_copy_state *cs, size_t nbytes)
 {
 	int err;
+	struct fuse_conn *fc = fud->fc;
+	struct fuse_pqueue *fpq = &fud->pq;
 	struct fuse_req *req;
 	struct fuse_out_header oh;
 
@@ -1902,63 +1896,60 @@ static ssize_t fuse_dev_do_write(struct fuse_conn *fc,
 	if (oh.error <= -1000 || oh.error > 0)
 		goto err_finish;
 
-	spin_lock(&fc->lock);
+	spin_lock(&fpq->lock);
 	err = -ENOENT;
-	if (!fc->connected)
-		goto err_unlock;
+	if (!fpq->connected)
+		goto err_unlock_pq;
 
-	req = request_find(fc, oh.unique);
+	req = request_find(fpq, oh.unique);
 	if (!req)
-		goto err_unlock;
+		goto err_unlock_pq;
 
-	if (req->aborted) {
-		spin_unlock(&fc->lock);
-		fuse_copy_finish(cs);
-		spin_lock(&fc->lock);
-		request_end(fc, req);
-		return -ENOENT;
-	}
 	/* Is it an interrupt reply? */
 	if (req->intr_unique == oh.unique) {
+		spin_unlock(&fpq->lock);
+
 		err = -EINVAL;
 		if (nbytes != sizeof(struct fuse_out_header))
-			goto err_unlock;
+			goto err_finish;
 
 		if (oh.error == -ENOSYS)
 			fc->no_interrupt = 1;
 		else if (oh.error == -EAGAIN)
-			queue_interrupt(fc, req);
+			queue_interrupt(&fc->iq, req);
 
-		spin_unlock(&fc->lock);
 		fuse_copy_finish(cs);
 		return nbytes;
 	}
 
-	req->state = FUSE_REQ_WRITING;
-	list_move(&req->list, &fc->io);
+	clear_bit(FR_SENT, &req->flags);
+	list_move(&req->list, &fpq->io);
 	req->out.h = oh;
-	req->locked = 1;
+	set_bit(FR_LOCKED, &req->flags);
+	spin_unlock(&fpq->lock);
 	cs->req = req;
 	if (!req->out.page_replace)
 		cs->move_pages = 0;
-	spin_unlock(&fc->lock);
 
 	err = copy_out_args(cs, &req->out, nbytes);
 	fuse_copy_finish(cs);
 
-	spin_lock(&fc->lock);
-	req->locked = 0;
-	if (!err) {
-		if (req->aborted)
-			err = -ENOENT;
-	} else if (!req->aborted)
+	spin_lock(&fpq->lock);
+	clear_bit(FR_LOCKED, &req->flags);
+	if (!fpq->connected)
+		err = -ENOENT;
+	else if (err)
 		req->out.h.error = -EIO;
+	if (!test_bit(FR_PRIVATE, &req->flags))
+		list_del_init(&req->list);
+	spin_unlock(&fpq->lock);
+
 	request_end(fc, req);
 
 	return err ? err : nbytes;
 
- err_unlock:
-	spin_unlock(&fc->lock);
+ err_unlock_pq:
+	spin_unlock(&fpq->lock);
  err_finish:
 	fuse_copy_finish(cs);
 	return err;
@@ -1967,16 +1958,17 @@ static ssize_t fuse_dev_do_write(struct fuse_conn *fc,
 static ssize_t fuse_dev_write(struct kiocb *iocb, struct iov_iter *from)
 {
 	struct fuse_copy_state cs;
-	struct fuse_conn *fc = fuse_get_conn(iocb->ki_filp);
-	if (!fc)
+	struct fuse_dev *fud = fuse_get_dev(iocb->ki_filp);
+
+	if (!fud)
 		return -EPERM;
 
 	if (!iter_is_iovec(from))
 		return -EINVAL;
 
-	fuse_copy_init(&cs, fc, 0, from);
+	fuse_copy_init(&cs, 0, from);
 
-	return fuse_dev_do_write(fc, &cs, iov_iter_count(from));
+	return fuse_dev_do_write(fud, &cs, iov_iter_count(from));
 }
 
 static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
@@ -1987,12 +1979,12 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
 	unsigned idx;
 	struct pipe_buffer *bufs;
 	struct fuse_copy_state cs;
-	struct fuse_conn *fc;
+	struct fuse_dev *fud;
 	size_t rem;
 	ssize_t ret;
 
-	fc = fuse_get_conn(out);
-	if (!fc)
+	fud = fuse_get_dev(out);
+	if (!fud)
 		return -EPERM;
 
 	bufs = kmalloc(pipe->buffers * sizeof(struct pipe_buffer), GFP_KERNEL);
@@ -2039,7 +2031,7 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
 	}
 	pipe_unlock(pipe);
 
-	fuse_copy_init(&cs, fc, 0, NULL);
+	fuse_copy_init(&cs, 0, NULL);
 	cs.pipebufs = bufs;
 	cs.nr_segs = nbuf;
 	cs.pipe = pipe;
@@ -2047,7 +2039,7 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
 	if (flags & SPLICE_F_MOVE)
 		cs.move_pages = 1;
 
-	ret = fuse_dev_do_write(fc, &cs, len);
+	ret = fuse_dev_do_write(fud, &cs, len);
 
 	for (idx = 0; idx < nbuf; idx++) {
 		struct pipe_buffer *buf = &bufs[idx];
@@ -2061,18 +2053,21 @@ out:
 static unsigned fuse_dev_poll(struct file *file, poll_table *wait)
 {
 	unsigned mask = POLLOUT | POLLWRNORM;
-	struct fuse_conn *fc = fuse_get_conn(file);
-	if (!fc)
+	struct fuse_iqueue *fiq;
+	struct fuse_dev *fud = fuse_get_dev(file);
+
+	if (!fud)
 		return POLLERR;
 
-	poll_wait(file, &fc->waitq, wait);
+	fiq = &fud->fc->iq;
+	poll_wait(file, &fiq->waitq, wait);
 
-	spin_lock(&fc->lock);
-	if (!fc->connected)
+	spin_lock(&fiq->waitq.lock);
+	if (!fiq->connected)
 		mask = POLLERR;
-	else if (request_pending(fc))
+	else if (request_pending(fiq))
 		mask |= POLLIN | POLLRDNORM;
-	spin_unlock(&fc->lock);
+	spin_unlock(&fiq->waitq.lock);
 
 	return mask;
 }
@@ -2083,67 +2078,18 @@ static unsigned fuse_dev_poll(struct file *file, poll_table *wait)
  * This function releases and reacquires fc->lock
  */
 static void end_requests(struct fuse_conn *fc, struct list_head *head)
-__releases(fc->lock)
-__acquires(fc->lock)
 {
 	while (!list_empty(head)) {
 		struct fuse_req *req;
 		req = list_entry(head->next, struct fuse_req, list);
 		req->out.h.error = -ECONNABORTED;
-		request_end(fc, req);
-		spin_lock(&fc->lock);
-	}
-}
-
-/*
- * Abort requests under I/O
- *
- * The requests are set to aborted and finished, and the request
- * waiter is woken up.  This will make request_wait_answer() wait
- * until the request is unlocked and then return.
- *
- * If the request is asynchronous, then the end function needs to be
- * called after waiting for the request to be unlocked (if it was
- * locked).
- */
-static void end_io_requests(struct fuse_conn *fc)
-__releases(fc->lock)
-__acquires(fc->lock)
-{
-	while (!list_empty(&fc->io)) {
-		struct fuse_req *req =
-			list_entry(fc->io.next, struct fuse_req, list);
-		void (*end) (struct fuse_conn *, struct fuse_req *) = req->end;
-
-		req->aborted = 1;
-		req->out.h.error = -ECONNABORTED;
-		req->state = FUSE_REQ_FINISHED;
+		clear_bit(FR_PENDING, &req->flags);
+		clear_bit(FR_SENT, &req->flags);
 		list_del_init(&req->list);
-		wake_up(&req->waitq);
-		if (end) {
-			req->end = NULL;
-			__fuse_get_request(req);
-			spin_unlock(&fc->lock);
-			wait_event(req->waitq, !req->locked);
-			end(fc, req);
-			fuse_put_request(fc, req);
-			spin_lock(&fc->lock);
-		}
+		request_end(fc, req);
 	}
 }
 
-static void end_queued_requests(struct fuse_conn *fc)
-__releases(fc->lock)
-__acquires(fc->lock)
-{
-	fc->max_background = UINT_MAX;
-	flush_bg_queue(fc);
-	end_requests(fc, &fc->pending);
-	end_requests(fc, &fc->processing);
-	while (forget_pending(fc))
-		kfree(dequeue_forget(fc, 1, NULL));
-}
-
 static void end_polls(struct fuse_conn *fc)
 {
 	struct rb_node *p;
@@ -2162,67 +2108,156 @@ static void end_polls(struct fuse_conn *fc)
 /*
  * Abort all requests.
  *
- * Emergency exit in case of a malicious or accidental deadlock, or
- * just a hung filesystem.
+ * Emergency exit in case of a malicious or accidental deadlock, or just a hung
+ * filesystem.
  *
- * The same effect is usually achievable through killing the
- * filesystem daemon and all users of the filesystem.  The exception
- * is the combination of an asynchronous request and the tricky
- * deadlock (see Documentation/filesystems/fuse.txt).
+ * The same effect is usually achievable through killing the filesystem daemon
+ * and all users of the filesystem.  The exception is the combination of an
+ * asynchronous request and the tricky deadlock (see
+ * Documentation/filesystems/fuse.txt).
  *
- * During the aborting, progression of requests from the pending and
- * processing lists onto the io list, and progression of new requests
- * onto the pending list is prevented by req->connected being false.
- *
- * Progression of requests under I/O to the processing list is
- * prevented by the req->aborted flag being true for these requests.
- * For this reason requests on the io list must be aborted first.
+ * Aborting requests under I/O goes as follows: 1: Separate out unlocked
+ * requests, they should be finished off immediately.  Locked requests will be
+ * finished after unlock; see unlock_request(). 2: Finish off the unlocked
+ * requests.  It is possible that some request will finish before we can.  This
+ * is OK, the request will in that case be removed from the list before we touch
+ * it.
  */
 void fuse_abort_conn(struct fuse_conn *fc)
 {
+	struct fuse_iqueue *fiq = &fc->iq;
+
 	spin_lock(&fc->lock);
 	if (fc->connected) {
+		struct fuse_dev *fud;
+		struct fuse_req *req, *next;
+		LIST_HEAD(to_end1);
+		LIST_HEAD(to_end2);
+
 		fc->connected = 0;
 		fc->blocked = 0;
 		fuse_set_initialized(fc);
-		end_io_requests(fc);
-		end_queued_requests(fc);
+		list_for_each_entry(fud, &fc->devices, entry) {
+			struct fuse_pqueue *fpq = &fud->pq;
+
+			spin_lock(&fpq->lock);
+			fpq->connected = 0;
+			list_for_each_entry_safe(req, next, &fpq->io, list) {
+				req->out.h.error = -ECONNABORTED;
+				spin_lock(&req->waitq.lock);
+				set_bit(FR_ABORTED, &req->flags);
+				if (!test_bit(FR_LOCKED, &req->flags)) {
+					set_bit(FR_PRIVATE, &req->flags);
+					list_move(&req->list, &to_end1);
+				}
+				spin_unlock(&req->waitq.lock);
+			}
+			list_splice_init(&fpq->processing, &to_end2);
+			spin_unlock(&fpq->lock);
+		}
+		fc->max_background = UINT_MAX;
+		flush_bg_queue(fc);
+
+		spin_lock(&fiq->waitq.lock);
+		fiq->connected = 0;
+		list_splice_init(&fiq->pending, &to_end2);
+		while (forget_pending(fiq))
+			kfree(dequeue_forget(fiq, 1, NULL));
+		wake_up_all_locked(&fiq->waitq);
+		spin_unlock(&fiq->waitq.lock);
+		kill_fasync(&fiq->fasync, SIGIO, POLL_IN);
 		end_polls(fc);
-		wake_up_all(&fc->waitq);
 		wake_up_all(&fc->blocked_waitq);
-		kill_fasync(&fc->fasync, SIGIO, POLL_IN);
+		spin_unlock(&fc->lock);
+
+		while (!list_empty(&to_end1)) {
+			req = list_first_entry(&to_end1, struct fuse_req, list);
+			__fuse_get_request(req);
+			list_del_init(&req->list);
+			request_end(fc, req);
+		}
+		end_requests(fc, &to_end2);
+	} else {
+		spin_unlock(&fc->lock);
 	}
-	spin_unlock(&fc->lock);
 }
 EXPORT_SYMBOL_GPL(fuse_abort_conn);
 
 int fuse_dev_release(struct inode *inode, struct file *file)
 {
-	struct fuse_conn *fc = fuse_get_conn(file);
-	if (fc) {
-		spin_lock(&fc->lock);
-		fc->connected = 0;
-		fc->blocked = 0;
-		fuse_set_initialized(fc);
-		end_queued_requests(fc);
-		end_polls(fc);
-		wake_up_all(&fc->blocked_waitq);
-		spin_unlock(&fc->lock);
-		fuse_conn_put(fc);
-	}
+	struct fuse_dev *fud = fuse_get_dev(file);
 
+	if (fud) {
+		struct fuse_conn *fc = fud->fc;
+		struct fuse_pqueue *fpq = &fud->pq;
+
+		WARN_ON(!list_empty(&fpq->io));
+		end_requests(fc, &fpq->processing);
+		/* Are we the last open device? */
+		if (atomic_dec_and_test(&fc->dev_count)) {
+			WARN_ON(fc->iq.fasync != NULL);
+			fuse_abort_conn(fc);
+		}
+		fuse_dev_free(fud);
+	}
 	return 0;
 }
 EXPORT_SYMBOL_GPL(fuse_dev_release);
 
 static int fuse_dev_fasync(int fd, struct file *file, int on)
 {
-	struct fuse_conn *fc = fuse_get_conn(file);
-	if (!fc)
+	struct fuse_dev *fud = fuse_get_dev(file);
+
+	if (!fud)
 		return -EPERM;
 
 	/* No locking - fasync_helper does its own locking */
-	return fasync_helper(fd, file, on, &fc->fasync);
+	return fasync_helper(fd, file, on, &fud->fc->iq.fasync);
+}
+
+static int fuse_device_clone(struct fuse_conn *fc, struct file *new)
+{
+	struct fuse_dev *fud;
+
+	if (new->private_data)
+		return -EINVAL;
+
+	fud = fuse_dev_alloc(fc);
+	if (!fud)
+		return -ENOMEM;
+
+	new->private_data = fud;
+	atomic_inc(&fc->dev_count);
+
+	return 0;
+}
+
+static long fuse_dev_ioctl(struct file *file, unsigned int cmd,
+			   unsigned long arg)
+{
+	int err = -ENOTTY;
+
+	if (cmd == FUSE_DEV_IOC_CLONE) {
+		int oldfd;
+
+		err = -EFAULT;
+		if (!get_user(oldfd, (__u32 __user *) arg)) {
+			struct file *old = fget(oldfd);
+
+			err = -EINVAL;
+			if (old) {
+				struct fuse_dev *fud = fuse_get_dev(old);
+
+				if (fud) {
+					mutex_lock(&fuse_mutex);
+					err = fuse_device_clone(fud->fc, file);
+					mutex_unlock(&fuse_mutex);
+				}
+				fput(old);
+			}
+		}
+	}
+	return err;
 }
 
 const struct file_operations fuse_dev_operations = {
@@ -2236,6 +2271,8 @@ const struct file_operations fuse_dev_operations = {
 	.poll		= fuse_dev_poll,
 	.release	= fuse_dev_release,
 	.fasync		= fuse_dev_fasync,
+	.unlocked_ioctl = fuse_dev_ioctl,
+	.compat_ioctl   = fuse_dev_ioctl,
 };
 EXPORT_SYMBOL_GPL(fuse_dev_operations);
 
diff --git a/fs/fuse/file.c b/fs/fuse/file.c
index 8c5e2fa68835..f523f2f04c19 100644
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@ -96,17 +96,17 @@ static void fuse_file_put(struct fuse_file *ff, bool sync)
 			 * Drop the release request when client does not
 			 * implement 'open'
 			 */
-			req->background = 0;
+			__clear_bit(FR_BACKGROUND, &req->flags);
 			iput(req->misc.release.inode);
 			fuse_put_request(ff->fc, req);
 		} else if (sync) {
-			req->background = 0;
+			__clear_bit(FR_BACKGROUND, &req->flags);
 			fuse_request_send(ff->fc, req);
 			iput(req->misc.release.inode);
 			fuse_put_request(ff->fc, req);
 		} else {
 			req->end = fuse_release_end;
-			req->background = 1;
+			__set_bit(FR_BACKGROUND, &req->flags);
 			fuse_request_send_background(ff->fc, req);
 		}
 		kfree(ff);
@@ -299,8 +299,8 @@ void fuse_sync_release(struct fuse_file *ff, int flags)
 {
 	WARN_ON(atomic_read(&ff->count) > 1);
 	fuse_prepare_release(ff, flags, FUSE_RELEASE);
-	ff->reserved_req->force = 1;
-	ff->reserved_req->background = 0;
+	__set_bit(FR_FORCE, &ff->reserved_req->flags);
+	__clear_bit(FR_BACKGROUND, &ff->reserved_req->flags);
 	fuse_request_send(ff->fc, ff->reserved_req);
 	fuse_put_request(ff->fc, ff->reserved_req);
 	kfree(ff);
@@ -426,7 +426,7 @@ static int fuse_flush(struct file *file, fl_owner_t id)
 	req->in.numargs = 1;
 	req->in.args[0].size = sizeof(inarg);
 	req->in.args[0].value = &inarg;
-	req->force = 1;
+	__set_bit(FR_FORCE, &req->flags);
 	fuse_request_send(fc, req);
 	err = req->out.h.error;
 	fuse_put_request(fc, req);
@@ -1169,7 +1169,7 @@ static ssize_t fuse_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	if (err <= 0)
 		goto out;
 
-	err = file_remove_suid(file);
+	err = file_remove_privs(file);
 	if (err)
 		goto out;
 
@@ -1611,7 +1611,8 @@ static int fuse_writepage_locked(struct page *page)
 	if (!req)
 		goto err;
 
-	req->background = 1; /* writeback always goes to bg_queue */
+	/* writeback always goes to bg_queue */
+	__set_bit(FR_BACKGROUND, &req->flags);
 	tmp_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
 	if (!tmp_page)
 		goto err_free;
@@ -1742,8 +1743,7 @@ static bool fuse_writepage_in_flight(struct fuse_req *new_req,
 		}
 	}
 
-	if (old_req->num_pages == 1 && (old_req->state == FUSE_REQ_INIT ||
-					old_req->state == FUSE_REQ_PENDING)) {
+	if (old_req->num_pages == 1 && test_bit(FR_PENDING, &old_req->flags)) {
 		struct backing_dev_info *bdi = inode_to_bdi(page->mapping->host);
 
 		copy_highpage(old_req->pages[0], page);
@@ -1830,7 +1830,7 @@ static int fuse_writepages_fill(struct page *page,
 		req->misc.write.in.write_flags |= FUSE_WRITE_CACHE;
 		req->misc.write.next = NULL;
 		req->in.argpages = 1;
-		req->background = 1;
+		__set_bit(FR_BACKGROUND, &req->flags);
 		req->num_pages = 0;
 		req->end = fuse_writepage_end;
 		req->inode = inode;
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h
index 7354dc142a50..405113101db8 100644
--- a/fs/fuse/fuse_i.h
+++ b/fs/fuse/fuse_i.h
@@ -241,16 +241,6 @@ struct fuse_args {
 
 #define FUSE_ARGS(args) struct fuse_args args = {}
 
-/** The request state */
-enum fuse_req_state {
-	FUSE_REQ_INIT = 0,
-	FUSE_REQ_PENDING,
-	FUSE_REQ_READING,
-	FUSE_REQ_SENT,
-	FUSE_REQ_WRITING,
-	FUSE_REQ_FINISHED
-};
-
 /** The request IO state (for asynchronous processing) */
 struct fuse_io_priv {
 	int async;
@@ -267,7 +257,40 @@ struct fuse_io_priv {
 };
 
 /**
+ * Request flags
+ *
+ * FR_ISREPLY:		set if the request has reply
+ * FR_FORCE:		force sending of the request even if interrupted
+ * FR_BACKGROUND:	request is sent in the background
+ * FR_WAITING:		request is counted as "waiting"
+ * FR_ABORTED:		the request was aborted
+ * FR_INTERRUPTED:	the request has been interrupted
+ * FR_LOCKED:		data is being copied to/from the request
+ * FR_PENDING:		request is not yet in userspace
+ * FR_SENT:		request is in userspace, waiting for an answer
+ * FR_FINISHED:		request is finished
+ * FR_PRIVATE:		request is on private list
+ */
+enum fuse_req_flag {
+	FR_ISREPLY,
+	FR_FORCE,
+	FR_BACKGROUND,
+	FR_WAITING,
+	FR_ABORTED,
+	FR_INTERRUPTED,
+	FR_LOCKED,
+	FR_PENDING,
+	FR_SENT,
+	FR_FINISHED,
+	FR_PRIVATE,
+};
+
+/**
  * A request to the client
+ *
+ * .waitq.lock protects the following fields:
+ *   - FR_ABORTED
+ *   - FR_LOCKED (may also be modified under fc->lock, tested under both)
  */
 struct fuse_req {
 	/** This can be on either pending processing or io lists in
@@ -283,35 +306,8 @@ struct fuse_req {
 	/** Unique ID for the interrupt request */
 	u64 intr_unique;
 
-	/*
-	 * The following bitfields are either set once before the
-	 * request is queued or setting/clearing them is protected by
-	 * fuse_conn->lock
-	 */
-
-	/** True if the request has reply */
-	unsigned isreply:1;
-
-	/** Force sending of the request even if interrupted */
-	unsigned force:1;
-
-	/** The request was aborted */
-	unsigned aborted:1;
-
-	/** Request is sent in the background */
-	unsigned background:1;
-
-	/** The request has been interrupted */
-	unsigned interrupted:1;
-
-	/** Data is being copied to/from the request */
-	unsigned locked:1;
-
-	/** Request is counted as "waiting" */
-	unsigned waiting:1;
-
-	/** State of the request */
-	enum fuse_req_state state;
+	/* Request flags, updated with test/set/clear_bit() */
+	unsigned long flags;
 
 	/** The request input */
 	struct fuse_in in;
@@ -380,6 +376,61 @@ struct fuse_req {
 	struct file *stolen_file;
 };
 
+struct fuse_iqueue {
+	/** Connection established */
+	unsigned connected;
+
+	/** Readers of the connection are waiting on this */
+	wait_queue_head_t waitq;
+
+	/** The next unique request id */
+	u64 reqctr;
+
+	/** The list of pending requests */
+	struct list_head pending;
+
+	/** Pending interrupts */
+	struct list_head interrupts;
+
+	/** Queue of pending forgets */
+	struct fuse_forget_link forget_list_head;
+	struct fuse_forget_link *forget_list_tail;
+
+	/** Batching of FORGET requests (positive indicates FORGET batch) */
+	int forget_batch;
+
+	/** O_ASYNC requests */
+	struct fasync_struct *fasync;
+};
+
+struct fuse_pqueue {
+	/** Connection established */
+	unsigned connected;
+
+	/** Lock protecting accessess to  members of this structure */
+	spinlock_t lock;
+
+	/** The list of requests being processed */
+	struct list_head processing;
+
+	/** The list of requests under I/O */
+	struct list_head io;
+};
+
+/**
+ * Fuse device instance
+ */
+struct fuse_dev {
+	/** Fuse connection for this device */
+	struct fuse_conn *fc;
+
+	/** Processing queue */
+	struct fuse_pqueue pq;
+
+	/** list entry on fc->devices */
+	struct list_head entry;
+};
+
 /**
  * A Fuse connection.
  *
@@ -394,6 +445,9 @@ struct fuse_conn {
 	/** Refcount */
 	atomic_t count;
 
+	/** Number of fuse_dev's */
+	atomic_t dev_count;
+
 	struct rcu_head rcu;
 
 	/** The user id for this mount */
@@ -411,17 +465,8 @@ struct fuse_conn {
 	/** Maximum write size */
 	unsigned max_write;
 
-	/** Readers of the connection are waiting on this */
-	wait_queue_head_t waitq;
-
-	/** The list of pending requests */
-	struct list_head pending;
-
-	/** The list of requests being processed */
-	struct list_head processing;
-
-	/** The list of requests under I/O */
-	struct list_head io;
+	/** Input queue */
+	struct fuse_iqueue iq;
 
 	/** The next unique kernel file handle */
 	u64 khctr;
@@ -444,16 +489,6 @@ struct fuse_conn {
 	/** The list of background requests set aside for later queuing */
 	struct list_head bg_queue;
 
-	/** Pending interrupts */
-	struct list_head interrupts;
-
-	/** Queue of pending forgets */
-	struct fuse_forget_link forget_list_head;
-	struct fuse_forget_link *forget_list_tail;
-
-	/** Batching of FORGET requests (positive indicates FORGET batch) */
-	int forget_batch;
-
 	/** Flag indicating that INIT reply has been received. Allocating
 	 * any fuse request will be suspended until the flag is set */
 	int initialized;
@@ -469,9 +504,6 @@ struct fuse_conn {
 	/** waitq for reserved requests */
 	wait_queue_head_t reserved_req_waitq;
 
-	/** The next unique request id */
-	u64 reqctr;
-
 	/** Connection established, cleared on umount, connection
 	    abort and device release */
 	unsigned connected;
@@ -594,9 +626,6 @@ struct fuse_conn {
 	/** number of dentries used in the above array */
 	int ctl_ndents;
 
-	/** O_ASYNC requests */
-	struct fasync_struct *fasync;
-
 	/** Key for lock owner ID scrambling */
 	u32 scramble_key[4];
 
@@ -614,6 +643,9 @@ struct fuse_conn {
 
 	/** Read/write semaphore to hold when accessing sb. */
 	struct rw_semaphore killsb;
+
+	/** List of device instances belonging to this connection */
+	struct list_head devices;
 };
 
 static inline struct fuse_conn *get_fuse_conn_super(struct super_block *sb)
@@ -826,6 +858,9 @@ void fuse_conn_init(struct fuse_conn *fc);
  */
 void fuse_conn_put(struct fuse_conn *fc);
 
+struct fuse_dev *fuse_dev_alloc(struct fuse_conn *fc);
+void fuse_dev_free(struct fuse_dev *fud);
+
 /**
  * Add connection to control filesystem
  */
diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c
index 082ac1c97f39..2913db2a5b99 100644
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@ -362,8 +362,8 @@ static void fuse_send_destroy(struct fuse_conn *fc)
 	if (req && fc->conn_init) {
 		fc->destroy_req = NULL;
 		req->in.h.opcode = FUSE_DESTROY;
-		req->force = 1;
-		req->background = 0;
+		__set_bit(FR_FORCE, &req->flags);
+		__clear_bit(FR_BACKGROUND, &req->flags);
 		fuse_request_send(fc, req);
 		fuse_put_request(fc, req);
 	}
@@ -567,30 +567,46 @@ static int fuse_show_options(struct seq_file *m, struct dentry *root)
 	return 0;
 }
 
+static void fuse_iqueue_init(struct fuse_iqueue *fiq)
+{
+	memset(fiq, 0, sizeof(struct fuse_iqueue));
+	init_waitqueue_head(&fiq->waitq);
+	INIT_LIST_HEAD(&fiq->pending);
+	INIT_LIST_HEAD(&fiq->interrupts);
+	fiq->forget_list_tail = &fiq->forget_list_head;
+	fiq->connected = 1;
+}
+
+static void fuse_pqueue_init(struct fuse_pqueue *fpq)
+{
+	memset(fpq, 0, sizeof(struct fuse_pqueue));
+	spin_lock_init(&fpq->lock);
+	INIT_LIST_HEAD(&fpq->processing);
+	INIT_LIST_HEAD(&fpq->io);
+	fpq->connected = 1;
+}
+
 void fuse_conn_init(struct fuse_conn *fc)
 {
 	memset(fc, 0, sizeof(*fc));
 	spin_lock_init(&fc->lock);
 	init_rwsem(&fc->killsb);
 	atomic_set(&fc->count, 1);
-	init_waitqueue_head(&fc->waitq);
+	atomic_set(&fc->dev_count, 1);
 	init_waitqueue_head(&fc->blocked_waitq);
 	init_waitqueue_head(&fc->reserved_req_waitq);
-	INIT_LIST_HEAD(&fc->pending);
-	INIT_LIST_HEAD(&fc->processing);
-	INIT_LIST_HEAD(&fc->io);
-	INIT_LIST_HEAD(&fc->interrupts);
+	fuse_iqueue_init(&fc->iq);
 	INIT_LIST_HEAD(&fc->bg_queue);
 	INIT_LIST_HEAD(&fc->entry);
-	fc->forget_list_tail = &fc->forget_list_head;
+	INIT_LIST_HEAD(&fc->devices);
 	atomic_set(&fc->num_waiting, 0);
 	fc->max_background = FUSE_DEFAULT_MAX_BACKGROUND;
 	fc->congestion_threshold = FUSE_DEFAULT_CONGESTION_THRESHOLD;
 	fc->khctr = 0;
 	fc->polled_files = RB_ROOT;
-	fc->reqctr = 0;
 	fc->blocked = 0;
 	fc->initialized = 0;
+	fc->connected = 1;
 	fc->attr_version = 1;
 	get_random_bytes(&fc->scramble_key, sizeof(fc->scramble_key));
 }
@@ -930,6 +946,7 @@ static void fuse_send_init(struct fuse_conn *fc, struct fuse_req *req)
 
 static void fuse_free_conn(struct fuse_conn *fc)
 {
+	WARN_ON(!list_empty(&fc->devices));
 	kfree_rcu(fc, rcu);
 }
 
@@ -975,8 +992,42 @@ static int fuse_bdi_init(struct fuse_conn *fc, struct super_block *sb)
 	return 0;
 }
 
+struct fuse_dev *fuse_dev_alloc(struct fuse_conn *fc)
+{
+	struct fuse_dev *fud;
+
+	fud = kzalloc(sizeof(struct fuse_dev), GFP_KERNEL);
+	if (fud) {
+		fud->fc = fuse_conn_get(fc);
+		fuse_pqueue_init(&fud->pq);
+
+		spin_lock(&fc->lock);
+		list_add_tail(&fud->entry, &fc->devices);
+		spin_unlock(&fc->lock);
+	}
+
+	return fud;
+}
+EXPORT_SYMBOL_GPL(fuse_dev_alloc);
+
+void fuse_dev_free(struct fuse_dev *fud)
+{
+	struct fuse_conn *fc = fud->fc;
+
+	if (fc) {
+		spin_lock(&fc->lock);
+		list_del(&fud->entry);
+		spin_unlock(&fc->lock);
+
+		fuse_conn_put(fc);
+	}
+	kfree(fud);
+}
+EXPORT_SYMBOL_GPL(fuse_dev_free);
+
 static int fuse_fill_super(struct super_block *sb, void *data, int silent)
 {
+	struct fuse_dev *fud;
 	struct fuse_conn *fc;
 	struct inode *root;
 	struct fuse_mount_data d;
@@ -1026,12 +1077,17 @@ static int fuse_fill_super(struct super_block *sb, void *data, int silent)
 		goto err_fput;
 
 	fuse_conn_init(fc);
+	fc->release = fuse_free_conn;
+
+	fud = fuse_dev_alloc(fc);
+	if (!fud)
+		goto err_put_conn;
 
 	fc->dev = sb->s_dev;
 	fc->sb = sb;
 	err = fuse_bdi_init(fc, sb);
 	if (err)
-		goto err_put_conn;
+		goto err_dev_free;
 
 	sb->s_bdi = &fc->bdi;
 
@@ -1040,7 +1096,6 @@ static int fuse_fill_super(struct super_block *sb, void *data, int silent)
 		fc->dont_mask = 1;
 	sb->s_flags |= MS_POSIXACL;
 
-	fc->release = fuse_free_conn;
 	fc->flags = d.flags;
 	fc->user_id = d.user_id;
 	fc->group_id = d.group_id;
@@ -1053,14 +1108,14 @@ static int fuse_fill_super(struct super_block *sb, void *data, int silent)
 	root = fuse_get_root_inode(sb, d.rootmode);
 	root_dentry = d_make_root(root);
 	if (!root_dentry)
-		goto err_put_conn;
+		goto err_dev_free;
 	/* only now - we want root dentry with NULL ->d_op */
 	sb->s_d_op = &fuse_dentry_operations;
 
 	init_req = fuse_request_alloc(0);
 	if (!init_req)
 		goto err_put_root;
-	init_req->background = 1;
+	__set_bit(FR_BACKGROUND, &init_req->flags);
 
 	if (is_bdev) {
 		fc->destroy_req = fuse_request_alloc(0);
@@ -1079,8 +1134,7 @@ static int fuse_fill_super(struct super_block *sb, void *data, int silent)
 
 	list_add_tail(&fc->entry, &fuse_conn_list);
 	sb->s_root = root_dentry;
-	fc->connected = 1;
-	file->private_data = fuse_conn_get(fc);
+	file->private_data = fud;
 	mutex_unlock(&fuse_mutex);
 	/*
 	 * atomic_dec_and_test() in fput() provides the necessary
@@ -1099,6 +1153,8 @@ static int fuse_fill_super(struct super_block *sb, void *data, int silent)
 	fuse_request_free(init_req);
  err_put_root:
 	dput(root_dentry);
+ err_dev_free:
+	fuse_dev_free(fud);
  err_put_conn:
 	fuse_bdi_destroy(fc);
 	fuse_conn_put(fc);
@@ -1238,7 +1294,6 @@ static void fuse_fs_cleanup(void)
 }
 
 static struct kobject *fuse_kobj;
-static struct kobject *connections_kobj;
 
 static int fuse_sysfs_init(void)
 {
@@ -1250,11 +1305,9 @@ static int fuse_sysfs_init(void)
 		goto out_err;
 	}
 
-	connections_kobj = kobject_create_and_add("connections", fuse_kobj);
-	if (!connections_kobj) {
-		err = -ENOMEM;
+	err = sysfs_create_mount_point(fuse_kobj, "connections");
+	if (err)
 		goto out_fuse_unregister;
-	}
 
 	return 0;
 
@@ -1266,7 +1319,7 @@ static int fuse_sysfs_init(void)
 
 static void fuse_sysfs_cleanup(void)
 {
-	kobject_put(connections_kobj);
+	sysfs_remove_mount_point(fuse_kobj, "connections");
 	kobject_put(fuse_kobj);
 }
 
diff --git a/fs/hfs/hfs_fs.h b/fs/hfs/hfs_fs.h
index 95d255219b1e..1f1c7dcbcc2f 100644
--- a/fs/hfs/hfs_fs.h
+++ b/fs/hfs/hfs_fs.h
@@ -252,7 +252,7 @@ extern void hfs_mark_mdb_dirty(struct super_block *sb);
 #define __hfs_u_to_mtime(sec)	cpu_to_be32(sec + 2082844800U - sys_tz.tz_minuteswest * 60)
 #define __hfs_m_to_utime(sec)	(be32_to_cpu(sec) - 2082844800U  + sys_tz.tz_minuteswest * 60)
 
-#define HFS_I(inode)	(list_entry(inode, struct hfs_inode_info, vfs_inode))
+#define HFS_I(inode)	(container_of(inode, struct hfs_inode_info, vfs_inode))
 #define HFS_SB(sb)	((struct hfs_sb_info *)(sb)->s_fs_info)
 
 #define hfs_m_to_utime(time)	(struct timespec){ .tv_sec = __hfs_m_to_utime(time) }
diff --git a/fs/hfsplus/hfsplus_fs.h b/fs/hfsplus/hfsplus_fs.h
index b0441d65fa54..f91a1faf819e 100644
--- a/fs/hfsplus/hfsplus_fs.h
+++ b/fs/hfsplus/hfsplus_fs.h
@@ -263,7 +263,7 @@ struct hfsplus_inode_info {
 
 static inline struct hfsplus_inode_info *HFSPLUS_I(struct inode *inode)
 {
-	return list_entry(inode, struct hfsplus_inode_info, vfs_inode);
+	return container_of(inode, struct hfsplus_inode_info, vfs_inode);
 }
 
 /*
diff --git a/fs/hpfs/alloc.c b/fs/hpfs/alloc.c
index f005046e1591..d6a4b55d2ab0 100644
--- a/fs/hpfs/alloc.c
+++ b/fs/hpfs/alloc.c
@@ -484,3 +484,98 @@ struct anode *hpfs_alloc_anode(struct super_block *s, secno near, anode_secno *a
 	a->btree.first_free = cpu_to_le16(8);
 	return a;
 }
+
+static unsigned find_run(__le32 *bmp, unsigned *idx)
+{
+	unsigned len;
+	while (tstbits(bmp, *idx, 1)) {
+		(*idx)++;
+		if (unlikely(*idx >= 0x4000))
+			return 0;
+	}
+	len = 1;
+	while (!tstbits(bmp, *idx + len, 1))
+		len++;
+	return len;
+}
+
+static int do_trim(struct super_block *s, secno start, unsigned len, secno limit_start, secno limit_end, unsigned minlen, unsigned *result)
+{
+	int err;
+	secno end;
+	if (fatal_signal_pending(current))
+		return -EINTR;
+	end = start + len;
+	if (start < limit_start)
+		start = limit_start;
+	if (end > limit_end)
+		end = limit_end;
+	if (start >= end)
+		return 0;
+	if (end - start < minlen)
+		return 0;
+	err = sb_issue_discard(s, start, end - start, GFP_NOFS, 0);
+	if (err)
+		return err;
+	*result += end - start;
+	return 0;
+}
+
+int hpfs_trim_fs(struct super_block *s, u64 start, u64 end, u64 minlen, unsigned *result)
+{
+	int err = 0;
+	struct hpfs_sb_info *sbi = hpfs_sb(s);
+	unsigned idx, len, start_bmp, end_bmp;
+	__le32 *bmp;
+	struct quad_buffer_head qbh;
+
+	*result = 0;
+	if (!end || end > sbi->sb_fs_size)
+		end = sbi->sb_fs_size;
+	if (start >= sbi->sb_fs_size)
+		return 0;
+	if (minlen > 0x4000)
+		return 0;
+	if (start < sbi->sb_dirband_start + sbi->sb_dirband_size && end > sbi->sb_dirband_start) {
+		hpfs_lock(s);
+		if (s->s_flags & MS_RDONLY) {
+			err = -EROFS;
+			goto unlock_1;
+		}
+		if (!(bmp = hpfs_map_dnode_bitmap(s, &qbh))) {
+			err = -EIO;
+			goto unlock_1;
+		}
+		idx = 0;
+		while ((len = find_run(bmp, &idx)) && !err) {
+			err = do_trim(s, sbi->sb_dirband_start + idx * 4, len * 4, start, end, minlen, result);
+			idx += len;
+		}
+		hpfs_brelse4(&qbh);
+unlock_1:
+		hpfs_unlock(s);
+	}
+	start_bmp = start >> 14;
+	end_bmp = (end + 0x3fff) >> 14;
+	while (start_bmp < end_bmp && !err) {
+		hpfs_lock(s);
+		if (s->s_flags & MS_RDONLY) {
+			err = -EROFS;
+			goto unlock_2;
+		}
+		if (!(bmp = hpfs_map_bitmap(s, start_bmp, &qbh, "trim"))) {
+			err = -EIO;
+			goto unlock_2;
+		}
+		idx = 0;
+		while ((len = find_run(bmp, &idx)) && !err) {
+			err = do_trim(s, (start_bmp << 14) + idx, len, start, end, minlen, result);
+			idx += len;
+		}
+		hpfs_brelse4(&qbh);
+unlock_2:
+		hpfs_unlock(s);
+		start_bmp++;
+	}
+	return err;
+}
diff --git a/fs/hpfs/dir.c b/fs/hpfs/dir.c
index 2a8e07425de0..dc540bfcee1d 100644
--- a/fs/hpfs/dir.c
+++ b/fs/hpfs/dir.c
@@ -327,4 +327,5 @@ const struct file_operations hpfs_dir_ops =
 	.iterate	= hpfs_readdir,
 	.release	= hpfs_dir_release,
 	.fsync		= hpfs_file_fsync,
+	.unlocked_ioctl	= hpfs_ioctl,
 };
diff --git a/fs/hpfs/file.c b/fs/hpfs/file.c
index 6d8cfe9b52d6..7ca28d604bf7 100644
--- a/fs/hpfs/file.c
+++ b/fs/hpfs/file.c
@@ -203,6 +203,7 @@ const struct file_operations hpfs_file_ops =
 	.release	= hpfs_file_release,
 	.fsync		= hpfs_file_fsync,
 	.splice_read	= generic_file_splice_read,
+	.unlocked_ioctl	= hpfs_ioctl,
 };
 
 const struct inode_operations hpfs_file_iops =
diff --git a/fs/hpfs/hpfs_fn.h b/fs/hpfs/hpfs_fn.h
index b63b75fa00e7..c4867b5116dd 100644
--- a/fs/hpfs/hpfs_fn.h
+++ b/fs/hpfs/hpfs_fn.h
@@ -18,6 +18,8 @@
 #include <linux/pagemap.h>
 #include <linux/buffer_head.h>
 #include <linux/slab.h>
+#include <linux/sched.h>
+#include <linux/blkdev.h>
 #include <asm/unaligned.h>
 
 #include "hpfs.h"
@@ -200,6 +202,7 @@ void hpfs_free_dnode(struct super_block *, secno);
 struct dnode *hpfs_alloc_dnode(struct super_block *, secno, dnode_secno *, struct quad_buffer_head *);
 struct fnode *hpfs_alloc_fnode(struct super_block *, secno, fnode_secno *, struct buffer_head **);
 struct anode *hpfs_alloc_anode(struct super_block *, secno, anode_secno *, struct buffer_head **);
+int hpfs_trim_fs(struct super_block *, u64, u64, u64, unsigned *);
 
 /* anode.c */
 
@@ -304,7 +307,7 @@ extern const struct address_space_operations hpfs_symlink_aops;
 
 static inline struct hpfs_inode_info *hpfs_i(struct inode *inode)
 {
-	return list_entry(inode, struct hpfs_inode_info, vfs_inode);
+	return container_of(inode, struct hpfs_inode_info, vfs_inode);
 }
 
 static inline struct hpfs_sb_info *hpfs_sb(struct super_block *sb)
@@ -318,6 +321,7 @@ __printf(2, 3)
 void hpfs_error(struct super_block *, const char *, ...);
 int hpfs_stop_cycles(struct super_block *, int, int *, int *, char *);
 unsigned hpfs_get_free_dnodes(struct super_block *);
+long hpfs_ioctl(struct file *file, unsigned cmd, unsigned long arg);
 
 /*
  * local time (HPFS) to GMT (Unix)
diff --git a/fs/hpfs/super.c b/fs/hpfs/super.c
index 7cd00d3a7c9b..68a9bed05628 100644
--- a/fs/hpfs/super.c
+++ b/fs/hpfs/super.c
@@ -52,17 +52,20 @@ static void unmark_dirty(struct super_block *s)
 }
 
 /* Filesystem error... */
-static char err_buf[1024];
-
 void hpfs_error(struct super_block *s, const char *fmt, ...)
 {
+	struct va_format vaf;
 	va_list args;
 
 	va_start(args, fmt);
-	vsnprintf(err_buf, sizeof(err_buf), fmt, args);
+
+	vaf.fmt = fmt;
+	vaf.va = &args;
+
+	pr_err("filesystem error: %pV", &vaf);
+
 	va_end(args);
 
-	pr_err("filesystem error: %s", err_buf);
 	if (!hpfs_sb(s)->sb_was_error) {
 		if (hpfs_sb(s)->sb_err == 2) {
 			pr_cont("; crashing the system because you wanted it\n");
@@ -196,12 +199,39 @@ static int hpfs_statfs(struct dentry *dentry, struct kstatfs *buf)
 	return 0;
 }
 
+
+long hpfs_ioctl(struct file *file, unsigned cmd, unsigned long arg)
+{
+	switch (cmd) {
+		case FITRIM: {
+			struct fstrim_range range;
+			secno n_trimmed;
+			int r;
+			if (!capable(CAP_SYS_ADMIN))
+				return -EPERM;
+			if (copy_from_user(&range, (struct fstrim_range __user *)arg, sizeof(range)))
+				return -EFAULT;
+			r = hpfs_trim_fs(file_inode(file)->i_sb, range.start >> 9, (range.start + range.len) >> 9, (range.minlen + 511) >> 9, &n_trimmed);
+			if (r)
+				return r;
+			range.len = (u64)n_trimmed << 9;
+			if (copy_to_user((struct fstrim_range __user *)arg, &range, sizeof(range)))
+				return -EFAULT;
+			return 0;
+		}
+		default: {
+			return -ENOIOCTLCMD;
+		}
+	}
+}
+
+
 static struct kmem_cache * hpfs_inode_cachep;
 
 static struct inode *hpfs_alloc_inode(struct super_block *sb)
 {
 	struct hpfs_inode_info *ei;
-	ei = (struct hpfs_inode_info *)kmem_cache_alloc(hpfs_inode_cachep, GFP_NOFS);
+	ei = kmem_cache_alloc(hpfs_inode_cachep, GFP_NOFS);
 	if (!ei)
 		return NULL;
 	ei->vfs_inode.i_version = 1;
@@ -424,11 +454,14 @@ static int hpfs_remount_fs(struct super_block *s, int *flags, char *data)
 	int o;
 	struct hpfs_sb_info *sbi = hpfs_sb(s);
 	char *new_opts = kstrdup(data, GFP_KERNEL);
-	
+
+	if (!new_opts)
+		return -ENOMEM;
+
 	sync_filesystem(s);
 
 	*flags |= MS_NOATIME;
-	
+
 	hpfs_lock(s);
 	uid = sbi->sb_uid; gid = sbi->sb_gid;
 	umask = 0777 & ~sbi->sb_mode;
diff --git a/fs/inode.c b/fs/inode.c
index 069721f0cc0e..d30640f7a193 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -841,7 +841,11 @@ unsigned int get_next_ino(void)
 	}
 #endif
 
-	*p = ++res;
+	res++;
+	/* get_next_ino should not provide a 0 inode number */
+	if (unlikely(!res))
+		res++;
+	*p = res;
 	put_cpu_var(last_ino);
 	return res;
 }
@@ -1674,7 +1678,31 @@ int should_remove_suid(struct dentry *dentry)
 }
 EXPORT_SYMBOL(should_remove_suid);
 
-static int __remove_suid(struct dentry *dentry, int kill)
+/*
+ * Return mask of changes for notify_change() that need to be done as a
+ * response to write or truncate. Return 0 if nothing has to be changed.
+ * Negative value on error (change should be denied).
+ */
+int dentry_needs_remove_privs(struct dentry *dentry)
+{
+	struct inode *inode = d_inode(dentry);
+	int mask = 0;
+	int ret;
+
+	if (IS_NOSEC(inode))
+		return 0;
+
+	mask = should_remove_suid(dentry);
+	ret = security_inode_need_killpriv(dentry);
+	if (ret < 0)
+		return ret;
+	if (ret)
+		mask |= ATTR_KILL_PRIV;
+	return mask;
+}
+EXPORT_SYMBOL(dentry_needs_remove_privs);
+
+static int __remove_privs(struct dentry *dentry, int kill)
 {
 	struct iattr newattrs;
 
@@ -1686,33 +1714,32 @@ static int __remove_suid(struct dentry *dentry, int kill)
 	return notify_change(dentry, &newattrs, NULL);
 }
 
-int file_remove_suid(struct file *file)
+/*
+ * Remove special file priviledges (suid, capabilities) when file is written
+ * to or truncated.
+ */
+int file_remove_privs(struct file *file)
 {
 	struct dentry *dentry = file->f_path.dentry;
 	struct inode *inode = d_inode(dentry);
-	int killsuid;
-	int killpriv;
+	int kill;
 	int error = 0;
 
 	/* Fast path for nothing security related */
 	if (IS_NOSEC(inode))
 		return 0;
 
-	killsuid = should_remove_suid(dentry);
-	killpriv = security_inode_need_killpriv(dentry);
-
-	if (killpriv < 0)
-		return killpriv;
-	if (killpriv)
-		error = security_inode_killpriv(dentry);
-	if (!error && killsuid)
-		error = __remove_suid(dentry, killsuid);
-	if (!error && (inode->i_sb->s_flags & MS_NOSEC))
-		inode->i_flags |= S_NOSEC;
+	kill = file_needs_remove_privs(file);
+	if (kill < 0)
+		return kill;
+	if (kill)
+		error = __remove_privs(dentry, kill);
+	if (!error)
+		inode_has_no_xattr(inode);
 
 	return error;
 }
-EXPORT_SYMBOL(file_remove_suid);
+EXPORT_SYMBOL(file_remove_privs);
 
 /**
  *	file_update_time	-	update mtime and ctime time
@@ -1967,9 +1994,8 @@ EXPORT_SYMBOL(inode_dio_wait);
  * inode is being instantiated).  The reason for the cmpxchg() loop
  * --- which wouldn't be necessary if all code paths which modify
  * i_flags actually followed this rule, is that there is at least one
- * code path which doesn't today --- for example,
- * __generic_file_aio_write() calls file_remove_suid() without holding
- * i_mutex --- so we use cmpxchg() out of an abundance of caution.
+ * code path which doesn't today so we use cmpxchg() out of an abundance
+ * of caution.
  *
  * In the long run, i_mutex is overkill, and we should probably look
  * at using the i_lock spinlock to protect i_flags, and then make sure
diff --git a/fs/internal.h b/fs/internal.h
index 01dce1d1476b..4d5af583ab03 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -107,6 +107,7 @@ extern struct file *do_file_open_root(struct dentry *, struct vfsmount *,
 extern long do_handle_open(int mountdirfd,
 			   struct file_handle __user *ufh, int open_flag);
 extern int open_check_o_direct(struct file *f);
+extern int vfs_open(const struct path *, struct file *, const struct cred *);
 
 /*
  * inode.c
diff --git a/fs/jffs2/os-linux.h b/fs/jffs2/os-linux.h
index d200a9b8fd5e..824e61ede465 100644
--- a/fs/jffs2/os-linux.h
+++ b/fs/jffs2/os-linux.h
@@ -19,7 +19,7 @@
 struct kstatfs;
 struct kvec;
 
-#define JFFS2_INODE_INFO(i) (list_entry(i, struct jffs2_inode_info, vfs_inode))
+#define JFFS2_INODE_INFO(i) (container_of(i, struct jffs2_inode_info, vfs_inode))
 #define OFNI_EDONI_2SFFJ(f)  (&(f)->vfs_inode)
 #define JFFS2_SB_INFO(sb) (sb->s_fs_info)
 #define OFNI_BS_2SFFJ(c)  ((struct super_block *)c->os_priv)
diff --git a/fs/jfs/ioctl.c b/fs/jfs/ioctl.c
index 93a1232894f6..8db8b7d61e40 100644
--- a/fs/jfs/ioctl.c
+++ b/fs/jfs/ioctl.c
@@ -180,9 +180,6 @@ long jfs_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 	case JFS_IOC_SETFLAGS32:
 		cmd = JFS_IOC_SETFLAGS;
 		break;
-	case FITRIM:
-		cmd = FITRIM;
-		break;
 	}
 	return jfs_ioctl(filp, cmd, arg);
 }
diff --git a/fs/jfs/jfs_incore.h b/fs/jfs/jfs_incore.h
index fa7e795bd8ae..1f26d1910409 100644
--- a/fs/jfs/jfs_incore.h
+++ b/fs/jfs/jfs_incore.h
@@ -206,7 +206,7 @@ struct jfs_sb_info {
 
 static inline struct jfs_inode_info *JFS_IP(struct inode *inode)
 {
-	return list_entry(inode, struct jfs_inode_info, vfs_inode);
+	return container_of(inode, struct jfs_inode_info, vfs_inode);
 }
 
 static inline int jfs_dirtable_inline(struct inode *inode)
diff --git a/fs/kernfs/dir.c b/fs/kernfs/dir.c
index fffca9517321..2d48d28e1640 100644
--- a/fs/kernfs/dir.c
+++ b/fs/kernfs/dir.c
@@ -592,6 +592,9 @@ int kernfs_add_one(struct kernfs_node *kn)
 		goto out_unlock;
 
 	ret = -ENOENT;
+	if (parent->flags & KERNFS_EMPTY_DIR)
+		goto out_unlock;
+
 	if ((parent->flags & KERNFS_ACTIVATED) && !kernfs_active(parent))
 		goto out_unlock;
 
@@ -783,6 +786,38 @@ struct kernfs_node *kernfs_create_dir_ns(struct kernfs_node *parent,
 	return ERR_PTR(rc);
 }
 
+/**
+ * kernfs_create_empty_dir - create an always empty directory
+ * @parent: parent in which to create a new directory
+ * @name: name of the new directory
+ *
+ * Returns the created node on success, ERR_PTR() value on failure.
+ */
+struct kernfs_node *kernfs_create_empty_dir(struct kernfs_node *parent,
+					    const char *name)
+{
+	struct kernfs_node *kn;
+	int rc;
+
+	/* allocate */
+	kn = kernfs_new_node(parent, name, S_IRUGO|S_IXUGO|S_IFDIR, KERNFS_DIR);
+	if (!kn)
+		return ERR_PTR(-ENOMEM);
+
+	kn->flags |= KERNFS_EMPTY_DIR;
+	kn->dir.root = parent->dir.root;
+	kn->ns = NULL;
+	kn->priv = NULL;
+
+	/* link in */
+	rc = kernfs_add_one(kn);
+	if (!rc)
+		return kn;
+
+	kernfs_put(kn);
+	return ERR_PTR(rc);
+}
+
 static struct dentry *kernfs_iop_lookup(struct inode *dir,
 					struct dentry *dentry,
 					unsigned int flags)
@@ -1254,7 +1289,8 @@ int kernfs_rename_ns(struct kernfs_node *kn, struct kernfs_node *new_parent,
 	mutex_lock(&kernfs_mutex);
 
 	error = -ENOENT;
-	if (!kernfs_active(kn) || !kernfs_active(new_parent))
+	if (!kernfs_active(kn) || !kernfs_active(new_parent) ||
+	    (new_parent->flags & KERNFS_EMPTY_DIR))
 		goto out;
 
 	error = 0;
diff --git a/fs/kernfs/inode.c b/fs/kernfs/inode.c
index 2da8493a380b..756dd56aaf60 100644
--- a/fs/kernfs/inode.c
+++ b/fs/kernfs/inode.c
@@ -296,6 +296,8 @@ static void kernfs_init_inode(struct kernfs_node *kn, struct inode *inode)
 	case KERNFS_DIR:
 		inode->i_op = &kernfs_dir_iops;
 		inode->i_fop = &kernfs_dir_fops;
+		if (kn->flags & KERNFS_EMPTY_DIR)
+			make_empty_dir_inode(inode);
 		break;
 	case KERNFS_FILE:
 		inode->i_size = kn->attr.size;
diff --git a/fs/libfs.c b/fs/libfs.c
index 65e1feca8b98..102edfd39000 100644
--- a/fs/libfs.c
+++ b/fs/libfs.c
@@ -20,11 +20,6 @@
 
 #include "internal.h"
 
-static inline int simple_positive(struct dentry *dentry)
-{
-	return d_really_is_positive(dentry) && !d_unhashed(dentry);
-}
-
 int simple_getattr(struct vfsmount *mnt, struct dentry *dentry,
 		   struct kstat *stat)
 {
@@ -1108,3 +1103,98 @@ const struct inode_operations simple_symlink_inode_operations = {
 	.readlink = generic_readlink
 };
 EXPORT_SYMBOL(simple_symlink_inode_operations);
+
+/*
+ * Operations for a permanently empty directory.
+ */
+static struct dentry *empty_dir_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
+{
+	return ERR_PTR(-ENOENT);
+}
+
+static int empty_dir_getattr(struct vfsmount *mnt, struct dentry *dentry,
+				 struct kstat *stat)
+{
+	struct inode *inode = d_inode(dentry);
+	generic_fillattr(inode, stat);
+	return 0;
+}
+
+static int empty_dir_setattr(struct dentry *dentry, struct iattr *attr)
+{
+	return -EPERM;
+}
+
+static int empty_dir_setxattr(struct dentry *dentry, const char *name,
+			      const void *value, size_t size, int flags)
+{
+	return -EOPNOTSUPP;
+}
+
+static ssize_t empty_dir_getxattr(struct dentry *dentry, const char *name,
+				  void *value, size_t size)
+{
+	return -EOPNOTSUPP;
+}
+
+static int empty_dir_removexattr(struct dentry *dentry, const char *name)
+{
+	return -EOPNOTSUPP;
+}
+
+static ssize_t empty_dir_listxattr(struct dentry *dentry, char *list, size_t size)
+{
+	return -EOPNOTSUPP;
+}
+
+static const struct inode_operations empty_dir_inode_operations = {
+	.lookup		= empty_dir_lookup,
+	.permission	= generic_permission,
+	.setattr	= empty_dir_setattr,
+	.getattr	= empty_dir_getattr,
+	.setxattr	= empty_dir_setxattr,
+	.getxattr	= empty_dir_getxattr,
+	.removexattr	= empty_dir_removexattr,
+	.listxattr	= empty_dir_listxattr,
+};
+
+static loff_t empty_dir_llseek(struct file *file, loff_t offset, int whence)
+{
+	/* An empty directory has two entries . and .. at offsets 0 and 1 */
+	return generic_file_llseek_size(file, offset, whence, 2, 2);
+}
+
+static int empty_dir_readdir(struct file *file, struct dir_context *ctx)
+{
+	dir_emit_dots(file, ctx);
+	return 0;
+}
+
+static const struct file_operations empty_dir_operations = {
+	.llseek		= empty_dir_llseek,
+	.read		= generic_read_dir,
+	.iterate	= empty_dir_readdir,
+	.fsync		= noop_fsync,
+};
+
+
+void make_empty_dir_inode(struct inode *inode)
+{
+	set_nlink(inode, 2);
+	inode->i_mode = S_IFDIR | S_IRUGO | S_IXUGO;
+	inode->i_uid = GLOBAL_ROOT_UID;
+	inode->i_gid = GLOBAL_ROOT_GID;
+	inode->i_rdev = 0;
+	inode->i_size = 2;
+	inode->i_blkbits = PAGE_SHIFT;
+	inode->i_blocks = 0;
+
+	inode->i_op = &empty_dir_inode_operations;
+	inode->i_fop = &empty_dir_operations;
+}
+
+bool is_empty_dir_inode(struct inode *inode)
+{
+	return (inode->i_fop == &empty_dir_operations) &&
+		(inode->i_op == &empty_dir_inode_operations);
+}
diff --git a/fs/minix/dir.c b/fs/minix/dir.c
index 118e4e7bc935..d19ac258105a 100644
--- a/fs/minix/dir.c
+++ b/fs/minix/dir.c
@@ -45,11 +45,6 @@ minix_last_byte(struct inode *inode, unsigned long page_nr)
 	return last_byte;
 }
 
-static inline unsigned long dir_pages(struct inode *inode)
-{
-	return (inode->i_size+PAGE_CACHE_SIZE-1)>>PAGE_CACHE_SHIFT;
-}
-
 static int dir_commit_chunk(struct page *page, loff_t pos, unsigned len)
 {
 	struct address_space *mapping = page->mapping;
diff --git a/fs/minix/minix.h b/fs/minix/minix.h
index 1ebd11854622..01ad81dcacc5 100644
--- a/fs/minix/minix.h
+++ b/fs/minix/minix.h
@@ -84,7 +84,7 @@ static inline struct minix_sb_info *minix_sb(struct super_block *sb)
 
 static inline struct minix_inode_info *minix_i(struct inode *inode)
 {
-	return list_entry(inode, struct minix_inode_info, vfs_inode);
+	return container_of(inode, struct minix_inode_info, vfs_inode);
 }
 
 static inline unsigned minix_blocks_needed(unsigned bits, unsigned blocksize)
diff --git a/fs/namei.c b/fs/namei.c
index 2dad0eaf91d3..ae4e4c18b2ac 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -792,7 +792,7 @@ static void set_root(struct nameidata *nd)
 	get_fs_root(current->fs, &nd->root);
 }
 
-static unsigned set_root_rcu(struct nameidata *nd)
+static void set_root_rcu(struct nameidata *nd)
 {
 	struct fs_struct *fs = current->fs;
 	unsigned seq;
@@ -802,7 +802,6 @@ static unsigned set_root_rcu(struct nameidata *nd)
 		nd->root = fs->root;
 		nd->root_seq = __read_seqcount_begin(&nd->root.dentry->d_seq);
 	} while (read_seqcount_retry(&fs->seq, seq));
-	return nd->root_seq;
 }
 
 static void path_put_conditional(struct path *path, struct nameidata *nd)
@@ -1998,7 +1997,8 @@ static const char *path_init(struct nameidata *nd, unsigned flags)
 	if (*s == '/') {
 		if (flags & LOOKUP_RCU) {
 			rcu_read_lock();
-			nd->seq = set_root_rcu(nd);
+			set_root_rcu(nd);
+			nd->seq = nd->root_seq;
 		} else {
 			set_root(nd);
 			path_get(&nd->root);
diff --git a/fs/namespace.c b/fs/namespace.c
index e99f1f4e00cd..c7cb8a526c05 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -2343,6 +2343,8 @@ unlock:
 	return err;
 }
 
+static bool fs_fully_visible(struct file_system_type *fs_type, int *new_mnt_flags);
+
 /*
  * create a new mount for userspace and request it to be added into the
  * namespace's tree
@@ -2374,6 +2376,10 @@ static int do_new_mount(struct path *path, const char *fstype, int flags,
 			flags |= MS_NODEV;
 			mnt_flags |= MNT_NODEV | MNT_LOCK_NODEV;
 		}
+		if (type->fs_flags & FS_USERNS_VISIBLE) {
+			if (!fs_fully_visible(type, &mnt_flags))
+				return -EPERM;
+		}
 	}
 
 	mnt = vfs_kern_mount(type, flags, name, data);
@@ -3175,9 +3181,10 @@ bool current_chrooted(void)
 	return chrooted;
 }
 
-bool fs_fully_visible(struct file_system_type *type)
+static bool fs_fully_visible(struct file_system_type *type, int *new_mnt_flags)
 {
 	struct mnt_namespace *ns = current->nsproxy->mnt_ns;
+	int new_flags = *new_mnt_flags;
 	struct mount *mnt;
 	bool visible = false;
 
@@ -3196,16 +3203,36 @@ bool fs_fully_visible(struct file_system_type *type)
 		if (mnt->mnt.mnt_root != mnt->mnt.mnt_sb->s_root)
 			continue;
 
-		/* This mount is not fully visible if there are any child mounts
-		 * that cover anything except for empty directories.
+		/* Verify the mount flags are equal to or more permissive
+		 * than the proposed new mount.
+		 */
+		if ((mnt->mnt.mnt_flags & MNT_LOCK_READONLY) &&
+		    !(new_flags & MNT_READONLY))
+			continue;
+		if ((mnt->mnt.mnt_flags & MNT_LOCK_NODEV) &&
+		    !(new_flags & MNT_NODEV))
+			continue;
+		if ((mnt->mnt.mnt_flags & MNT_LOCK_ATIME) &&
+		    ((mnt->mnt.mnt_flags & MNT_ATIME_MASK) != (new_flags & MNT_ATIME_MASK)))
+			continue;
+
+		/* This mount is not fully visible if there are any
+		 * locked child mounts that cover anything except for
+		 * empty directories.
 		 */
 		list_for_each_entry(child, &mnt->mnt_mounts, mnt_child) {
 			struct inode *inode = child->mnt_mountpoint->d_inode;
-			if (!S_ISDIR(inode->i_mode))
-				goto next;
-			if (inode->i_nlink > 2)
+			/* Only worry about locked mounts */
+			if (!(mnt->mnt.mnt_flags & MNT_LOCKED))
+				continue;
+			/* Is the directory permanetly empty? */
+			if (!is_empty_dir_inode(inode))
 				goto next;
 		}
+		/* Preserve the locked attributes */
+		*new_mnt_flags |= mnt->mnt.mnt_flags & (MNT_LOCK_READONLY | \
+							MNT_LOCK_NODEV    | \
+							MNT_LOCK_ATIME);
 		visible = true;
 		goto found;
 	next:	;
diff --git a/fs/ncpfs/dir.c b/fs/ncpfs/dir.c
index 80021c709af9..93575e91a7aa 100644
--- a/fs/ncpfs/dir.c
+++ b/fs/ncpfs/dir.c
@@ -1145,6 +1145,8 @@ static int ncp_rename(struct inode *old_dir, struct dentry *old_dentry,
 		case 0x00:
 			ncp_dbg(1, "renamed %pd -> %pd\n",
 				old_dentry, new_dentry);
+			ncp_d_prune(old_dentry);
+			ncp_d_prune(new_dentry);
 			break;
 		case 0x9E:
 			error = -ENAMETOOLONG;
diff --git a/fs/nfs/callback.c b/fs/nfs/callback.c
index 8d129bb7355a..682529c00996 100644
--- a/fs/nfs/callback.c
+++ b/fs/nfs/callback.c
@@ -458,7 +458,7 @@ check_gss_callback_principal(struct nfs_client *clp, struct svc_rqst *rqstp)
  * pg_authenticate method for nfsv4 callback threads.
  *
  * The authflavor has been negotiated, so an incorrect flavor is a server
- * bug. Drop packets with incorrect authflavor.
+ * bug. Deny packets with incorrect authflavor.
  *
  * All other checking done after NFS decoding where the nfs_client can be
  * found in nfs4_callback_compound
@@ -468,12 +468,12 @@ static int nfs_callback_authenticate(struct svc_rqst *rqstp)
 	switch (rqstp->rq_authop->flavour) {
 	case RPC_AUTH_NULL:
 		if (rqstp->rq_proc != CB_NULL)
-			return SVC_DROP;
+			return SVC_DENIED;
 		break;
 	case RPC_AUTH_GSS:
 		/* No RPC_AUTH_GSS support yet in NFSv4.1 */
 		 if (svc_is_backchannel(rqstp))
-			return SVC_DROP;
+			return SVC_DENIED;
 	}
 	return SVC_OK;
 }
diff --git a/fs/nfs/callback_proc.c b/fs/nfs/callback_proc.c
index 197806fb87ff..29e3c1b011b7 100644
--- a/fs/nfs/callback_proc.c
+++ b/fs/nfs/callback_proc.c
@@ -327,10 +327,8 @@ validate_seqid(struct nfs4_slot_table *tbl, struct cb_sequenceargs * args)
 	dprintk("%s slot table seqid: %u\n", __func__, slot->seq_nr);
 
 	/* Normal */
-	if (likely(args->csa_sequenceid == slot->seq_nr + 1)) {
-		slot->seq_nr++;
+	if (likely(args->csa_sequenceid == slot->seq_nr + 1))
 		goto out_ok;
-	}
 
 	/* Replay */
 	if (args->csa_sequenceid == slot->seq_nr) {
@@ -418,6 +416,7 @@ __be32 nfs4_callback_sequence(struct cb_sequenceargs *args,
 			      struct cb_process_state *cps)
 {
 	struct nfs4_slot_table *tbl;
+	struct nfs4_slot *slot;
 	struct nfs_client *clp;
 	int i;
 	__be32 status = htonl(NFS4ERR_BADSESSION);
@@ -429,25 +428,32 @@ __be32 nfs4_callback_sequence(struct cb_sequenceargs *args,
 
 	if (!(clp->cl_session->flags & SESSION4_BACK_CHAN))
 		goto out;
+
 	tbl = &clp->cl_session->bc_slot_table;
+	slot = tbl->slots + args->csa_slotid;
 
 	spin_lock(&tbl->slot_tbl_lock);
 	/* state manager is resetting the session */
 	if (test_bit(NFS4_SLOT_TBL_DRAINING, &tbl->slot_tbl_state)) {
-		spin_unlock(&tbl->slot_tbl_lock);
 		status = htonl(NFS4ERR_DELAY);
 		/* Return NFS4ERR_BADSESSION if we're draining the session
 		 * in order to reset it.
 		 */
 		if (test_bit(NFS4CLNT_SESSION_RESET, &clp->cl_state))
 			status = htonl(NFS4ERR_BADSESSION);
-		goto out;
+		goto out_unlock;
 	}
 
-	status = validate_seqid(&clp->cl_session->bc_slot_table, args);
-	spin_unlock(&tbl->slot_tbl_lock);
+	memcpy(&res->csr_sessionid, &args->csa_sessionid,
+	       sizeof(res->csr_sessionid));
+	res->csr_sequenceid = args->csa_sequenceid;
+	res->csr_slotid = args->csa_slotid;
+	res->csr_highestslotid = NFS41_BC_MAX_CALLBACKS - 1;
+	res->csr_target_highestslotid = NFS41_BC_MAX_CALLBACKS - 1;
+
+	status = validate_seqid(tbl, args);
 	if (status)
-		goto out;
+		goto out_unlock;
 
 	cps->slotid = args->csa_slotid;
 
@@ -458,15 +464,17 @@ __be32 nfs4_callback_sequence(struct cb_sequenceargs *args,
 	 */
 	if (referring_call_exists(clp, args->csa_nrclists, args->csa_rclists)) {
 		status = htonl(NFS4ERR_DELAY);
-		goto out;
+		goto out_unlock;
 	}
 
-	memcpy(&res->csr_sessionid, &args->csa_sessionid,
-	       sizeof(res->csr_sessionid));
-	res->csr_sequenceid = args->csa_sequenceid;
-	res->csr_slotid = args->csa_slotid;
-	res->csr_highestslotid = NFS41_BC_MAX_CALLBACKS - 1;
-	res->csr_target_highestslotid = NFS41_BC_MAX_CALLBACKS - 1;
+	/*
+	 * RFC5661 20.9.3
+	 * If CB_SEQUENCE returns an error, then the state of the slot
+	 * (sequence ID, cached reply) MUST NOT change.
+	 */
+	slot->seq_nr++;
+out_unlock:
+	spin_unlock(&tbl->slot_tbl_lock);
 
 out:
 	cps->clp = clp; /* put in nfs4_callback_compound */
diff --git a/fs/nfs/callback_xdr.c b/fs/nfs/callback_xdr.c
index 19ca95cdfd9b..6b1697a01dde 100644
--- a/fs/nfs/callback_xdr.c
+++ b/fs/nfs/callback_xdr.c
@@ -909,7 +909,7 @@ static __be32 nfs4_callback_compound(struct svc_rqst *rqstp, void *argp, void *r
 	xdr_init_encode(&xdr_out, &rqstp->rq_res, p);
 
 	status = decode_compound_hdr_arg(&xdr_in, &hdr_arg);
-	if (status == __constant_htonl(NFS4ERR_RESOURCE))
+	if (status == htonl(NFS4ERR_RESOURCE))
 		return rpc_garbage_args;
 
 	if (hdr_arg.minorversion == 0) {
diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index 892aefff3630..ecebb406cc1a 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -825,7 +825,6 @@ error:
  * Load up the server record from information gained in an fsinfo record
  */
 static void nfs_server_set_fsinfo(struct nfs_server *server,
-				  struct nfs_fh *mntfh,
 				  struct nfs_fsinfo *fsinfo)
 {
 	unsigned long max_rpc_payload;
@@ -901,7 +900,7 @@ int nfs_probe_fsinfo(struct nfs_server *server, struct nfs_fh *mntfh, struct nfs
 	if (error < 0)
 		goto out_error;
 
-	nfs_server_set_fsinfo(server, mntfh, &fsinfo);
+	nfs_server_set_fsinfo(server, &fsinfo);
 
 	/* Get some general file system info */
 	if (server->namelen == 0) {
@@ -1193,8 +1192,6 @@ void nfs_clients_init(struct net *net)
 }
 
 #ifdef CONFIG_PROC_FS
-static struct proc_dir_entry *proc_fs_nfs;
-
 static int nfs_server_list_open(struct inode *inode, struct file *file);
 static void *nfs_server_list_start(struct seq_file *p, loff_t *pos);
 static void *nfs_server_list_next(struct seq_file *p, void *v, loff_t *pos);
@@ -1364,27 +1361,29 @@ static int nfs_volume_list_show(struct seq_file *m, void *v)
 {
 	struct nfs_server *server;
 	struct nfs_client *clp;
-	char dev[8], fsid[17];
+	char dev[13];	// 8 for 2^24, 1 for ':', 3 for 2^8, 1 for '\0'
+	char fsid[34];	// 2 * 16 for %llx, 1 for ':', 1 for '\0'
 	struct nfs_net *nn = net_generic(seq_file_net(m), nfs_net_id);
 
 	/* display header on line 1 */
 	if (v == &nn->nfs_volume_list) {
-		seq_puts(m, "NV SERVER   PORT DEV     FSID              FSC\n");
+		seq_puts(m, "NV SERVER   PORT DEV          FSID"
+			    "                              FSC\n");
 		return 0;
 	}
 	/* display one transport per line on subsequent lines */
 	server = list_entry(v, struct nfs_server, master_link);
 	clp = server->nfs_client;
 
-	snprintf(dev, 8, "%u:%u",
+	snprintf(dev, sizeof(dev), "%u:%u",
 		 MAJOR(server->s_dev), MINOR(server->s_dev));
 
-	snprintf(fsid, 17, "%llx:%llx",
+	snprintf(fsid, sizeof(fsid), "%llx:%llx",
 		 (unsigned long long) server->fsid.major,
 		 (unsigned long long) server->fsid.minor);
 
 	rcu_read_lock();
-	seq_printf(m, "v%u %s %s %-7s %-17s %s\n",
+	seq_printf(m, "v%u %s %s %-12s %-33s %s\n",
 		   clp->rpc_ops->version,
 		   rpc_peeraddr2str(clp->cl_rpcclient, RPC_DISPLAY_HEX_ADDR),
 		   rpc_peeraddr2str(clp->cl_rpcclient, RPC_DISPLAY_HEX_PORT),
@@ -1434,27 +1433,20 @@ void nfs_fs_proc_net_exit(struct net *net)
  */
 int __init nfs_fs_proc_init(void)
 {
-	struct proc_dir_entry *p;
-
-	proc_fs_nfs = proc_mkdir("fs/nfsfs", NULL);
-	if (!proc_fs_nfs)
+	if (!proc_mkdir("fs/nfsfs", NULL))
 		goto error_0;
 
 	/* a file of servers with which we're dealing */
-	p = proc_symlink("servers", proc_fs_nfs, "../../net/nfsfs/servers");
-	if (!p)
+	if (!proc_symlink("fs/nfsfs/servers", NULL, "../../net/nfsfs/servers"))
 		goto error_1;
 
 	/* a file of volumes that we have mounted */
-	p = proc_symlink("volumes", proc_fs_nfs, "../../net/nfsfs/volumes");
-	if (!p)
-		goto error_2;
-	return 0;
+	if (!proc_symlink("fs/nfsfs/volumes", NULL, "../../net/nfsfs/volumes"))
+		goto error_1;
 
-error_2:
-	remove_proc_entry("servers", proc_fs_nfs);
+	return 0;
 error_1:
-	remove_proc_entry("fs/nfsfs", NULL);
+	remove_proc_subtree("fs/nfsfs", NULL);
 error_0:
 	return -ENOMEM;
 }
@@ -1464,9 +1456,7 @@ error_0:
  */
 void nfs_fs_proc_exit(void)
 {
-	remove_proc_entry("volumes", proc_fs_nfs);
-	remove_proc_entry("servers", proc_fs_nfs);
-	remove_proc_entry("fs/nfsfs", NULL);
+	remove_proc_subtree("fs/nfsfs", NULL);
 }
 
 #endif /* CONFIG_PROC_FS */
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index b2c8b31b2be7..547308a5ec6f 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -1470,9 +1470,6 @@ static int nfs_finish_open(struct nfs_open_context *ctx,
 {
 	int err;
 
-	if ((open_flags & (O_CREAT | O_EXCL)) == (O_CREAT | O_EXCL))
-		*opened |= FILE_CREATED;
-
 	err = finish_open(file, dentry, do_open, opened);
 	if (err)
 		goto out;
@@ -1771,7 +1768,7 @@ EXPORT_SYMBOL_GPL(nfs_mkdir);
 
 static void nfs_dentry_handle_enoent(struct dentry *dentry)
 {
-	if (d_really_is_positive(dentry) && !d_unhashed(dentry))
+	if (simple_positive(dentry))
 		d_delete(dentry);
 }
 
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index 8b8d83a526ce..cc4fa1ed61fc 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -555,31 +555,22 @@ static int nfs_launder_page(struct page *page)
 	return nfs_wb_page(inode, page);
 }
 
-#ifdef CONFIG_NFS_SWAP
 static int nfs_swap_activate(struct swap_info_struct *sis, struct file *file,
 						sector_t *span)
 {
-	int ret;
 	struct rpc_clnt *clnt = NFS_CLIENT(file->f_mapping->host);
 
 	*span = sis->pages;
 
-	rcu_read_lock();
-	ret = xs_swapper(rcu_dereference(clnt->cl_xprt), 1);
-	rcu_read_unlock();
-
-	return ret;
+	return rpc_clnt_swap_activate(clnt);
 }
 
 static void nfs_swap_deactivate(struct file *file)
 {
 	struct rpc_clnt *clnt = NFS_CLIENT(file->f_mapping->host);
 
-	rcu_read_lock();
-	xs_swapper(rcu_dereference(clnt->cl_xprt), 0);
-	rcu_read_unlock();
+	rpc_clnt_swap_deactivate(clnt);
 }
-#endif
 
 const struct address_space_operations nfs_file_aops = {
 	.readpage = nfs_readpage,
@@ -596,10 +587,8 @@ const struct address_space_operations nfs_file_aops = {
 	.launder_page = nfs_launder_page,
 	.is_dirty_writeback = nfs_check_dirty_writeback,
 	.error_remove_page = generic_error_remove_page,
-#ifdef CONFIG_NFS_SWAP
 	.swap_activate = nfs_swap_activate,
 	.swap_deactivate = nfs_swap_deactivate,
-#endif
 };
 
 /*
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c
index 7d05089e52d6..c12951b9551e 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.c
+++ b/fs/nfs/flexfilelayout/flexfilelayout.c
@@ -20,6 +20,7 @@
 #include "../nfs4trace.h"
 #include "../iostat.h"
 #include "../nfs.h"
+#include "../nfs42.h"
 
 #define NFSDBG_FACILITY         NFSDBG_PNFS_LD
 
@@ -182,17 +183,14 @@ static void _ff_layout_free_lseg(struct nfs4_ff_layout_segment *fls)
 
 static void ff_layout_sort_mirrors(struct nfs4_ff_layout_segment *fls)
 {
-	struct nfs4_ff_layout_mirror *tmp;
 	int i, j;
 
 	for (i = 0; i < fls->mirror_array_cnt - 1; i++) {
 		for (j = i + 1; j < fls->mirror_array_cnt; j++)
 			if (fls->mirror_array[i]->efficiency <
-			    fls->mirror_array[j]->efficiency) {
-				tmp = fls->mirror_array[i];
-				fls->mirror_array[i] = fls->mirror_array[j];
-				fls->mirror_array[j] = tmp;
-			}
+			    fls->mirror_array[j]->efficiency)
+				swap(fls->mirror_array[i],
+				     fls->mirror_array[j]);
 	}
 }
 
@@ -274,6 +272,7 @@ ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
 
 		spin_lock_init(&fls->mirror_array[i]->lock);
 		fls->mirror_array[i]->ds_count = ds_count;
+		fls->mirror_array[i]->lseg = &fls->generic_hdr;
 
 		/* deviceid */
 		rc = decode_deviceid(&stream, &devid);
@@ -344,6 +343,10 @@ ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
 			fls->mirror_array[i]->gid);
 	}
 
+	p = xdr_inline_decode(&stream, 4);
+	if (p)
+		fls->flags = be32_to_cpup(p);
+
 	ff_layout_sort_mirrors(fls);
 	rc = ff_layout_check_layout(lgr);
 	if (rc)
@@ -415,6 +418,146 @@ ff_layout_get_lseg_count(struct nfs4_ff_layout_segment *fls)
 	return 1;
 }
 
+static void
+nfs4_ff_start_busy_timer(struct nfs4_ff_busy_timer *timer)
+{
+	/* first IO request? */
+	if (atomic_inc_return(&timer->n_ops) == 1) {
+		timer->start_time = ktime_get();
+	}
+}
+
+static ktime_t
+nfs4_ff_end_busy_timer(struct nfs4_ff_busy_timer *timer)
+{
+	ktime_t start, now;
+
+	if (atomic_dec_return(&timer->n_ops) < 0)
+		WARN_ON_ONCE(1);
+
+	now = ktime_get();
+	start = timer->start_time;
+	timer->start_time = now;
+	return ktime_sub(now, start);
+}
+
+static ktime_t
+nfs4_ff_layout_calc_completion_time(struct rpc_task *task)
+{
+	return ktime_sub(ktime_get(), task->tk_start);
+}
+
+static bool
+nfs4_ff_layoutstat_start_io(struct nfs4_ff_layout_mirror *mirror,
+			    struct nfs4_ff_layoutstat *layoutstat)
+{
+	static const ktime_t notime = {0};
+	ktime_t now = ktime_get();
+
+	nfs4_ff_start_busy_timer(&layoutstat->busy_timer);
+	if (ktime_equal(mirror->start_time, notime))
+		mirror->start_time = now;
+	if (ktime_equal(mirror->last_report_time, notime))
+		mirror->last_report_time = now;
+	if (ktime_to_ms(ktime_sub(now, mirror->last_report_time)) >=
+			FF_LAYOUTSTATS_REPORT_INTERVAL) {
+		mirror->last_report_time = now;
+		return true;
+	}
+
+	return false;
+}
+
+static void
+nfs4_ff_layout_stat_io_update_requested(struct nfs4_ff_layoutstat *layoutstat,
+		__u64 requested)
+{
+	struct nfs4_ff_io_stat *iostat = &layoutstat->io_stat;
+
+	iostat->ops_requested++;
+	iostat->bytes_requested += requested;
+}
+
+static void
+nfs4_ff_layout_stat_io_update_completed(struct nfs4_ff_layoutstat *layoutstat,
+		__u64 requested,
+		__u64 completed,
+		ktime_t time_completed)
+{
+	struct nfs4_ff_io_stat *iostat = &layoutstat->io_stat;
+	ktime_t timer;
+
+	iostat->ops_completed++;
+	iostat->bytes_completed += completed;
+	iostat->bytes_not_delivered += requested - completed;
+
+	timer = nfs4_ff_end_busy_timer(&layoutstat->busy_timer);
+	iostat->total_busy_time =
+			ktime_add(iostat->total_busy_time, timer);
+	iostat->aggregate_completion_time =
+			ktime_add(iostat->aggregate_completion_time, time_completed);
+}
+
+static void
+nfs4_ff_layout_stat_io_start_read(struct nfs4_ff_layout_mirror *mirror,
+		__u64 requested)
+{
+	bool report;
+
+	spin_lock(&mirror->lock);
+	report = nfs4_ff_layoutstat_start_io(mirror, &mirror->read_stat);
+	nfs4_ff_layout_stat_io_update_requested(&mirror->read_stat, requested);
+	spin_unlock(&mirror->lock);
+
+	if (report)
+		pnfs_report_layoutstat(mirror->lseg->pls_layout->plh_inode);
+}
+
+static void
+nfs4_ff_layout_stat_io_end_read(struct rpc_task *task,
+		struct nfs4_ff_layout_mirror *mirror,
+		__u64 requested,
+		__u64 completed)
+{
+	spin_lock(&mirror->lock);
+	nfs4_ff_layout_stat_io_update_completed(&mirror->read_stat,
+			requested, completed,
+			nfs4_ff_layout_calc_completion_time(task));
+	spin_unlock(&mirror->lock);
+}
+
+static void
+nfs4_ff_layout_stat_io_start_write(struct nfs4_ff_layout_mirror *mirror,
+		__u64 requested)
+{
+	bool report;
+
+	spin_lock(&mirror->lock);
+	report = nfs4_ff_layoutstat_start_io(mirror , &mirror->write_stat);
+	nfs4_ff_layout_stat_io_update_requested(&mirror->write_stat, requested);
+	spin_unlock(&mirror->lock);
+
+	if (report)
+		pnfs_report_layoutstat(mirror->lseg->pls_layout->plh_inode);
+}
+
+static void
+nfs4_ff_layout_stat_io_end_write(struct rpc_task *task,
+		struct nfs4_ff_layout_mirror *mirror,
+		__u64 requested,
+		__u64 completed,
+		enum nfs3_stable_how committed)
+{
+	if (committed == NFS_UNSTABLE)
+		requested = completed = 0;
+
+	spin_lock(&mirror->lock);
+	nfs4_ff_layout_stat_io_update_completed(&mirror->write_stat,
+			requested, completed,
+			nfs4_ff_layout_calc_completion_time(task));
+	spin_unlock(&mirror->lock);
+}
+
 static int
 ff_layout_alloc_commit_info(struct pnfs_layout_segment *lseg,
 			    struct nfs_commit_info *cinfo,
@@ -631,7 +774,7 @@ static void ff_layout_reset_write(struct nfs_pgio_header *hdr, bool retry_pnfs)
 			nfs_direct_set_resched_writes(hdr->dreq);
 			/* fake unstable write to let common nfs resend pages */
 			hdr->verf.committed = NFS_UNSTABLE;
-			hdr->good_bytes = 0;
+			hdr->good_bytes = hdr->args.count;
 		}
 		return;
 	}
@@ -879,6 +1022,12 @@ static int ff_layout_read_done_cb(struct rpc_task *task,
 	return 0;
 }
 
+static bool
+ff_layout_need_layoutcommit(struct pnfs_layout_segment *lseg)
+{
+	return !(FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_LAYOUTCOMMIT);
+}
+
 /*
  * We reference the rpc_cred of the first WRITE that triggers the need for
  * a LAYOUTCOMMIT, and use it to send the layoutcommit compound.
@@ -891,6 +1040,9 @@ static int ff_layout_read_done_cb(struct rpc_task *task,
 static void
 ff_layout_set_layoutcommit(struct nfs_pgio_header *hdr)
 {
+	if (!ff_layout_need_layoutcommit(hdr->lseg))
+		return;
+
 	pnfs_set_layoutcommit(hdr->inode, hdr->lseg,
 			hdr->mds_offset + hdr->res.count);
 	dprintk("%s inode %lu pls_end_pos %lu\n", __func__, hdr->inode->i_ino,
@@ -909,6 +1061,10 @@ ff_layout_reset_to_mds(struct pnfs_layout_segment *lseg, int idx)
 static int ff_layout_read_prepare_common(struct rpc_task *task,
 					 struct nfs_pgio_header *hdr)
 {
+	nfs4_ff_layout_stat_io_start_read(
+			FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
+			hdr->args.count);
+
 	if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags))) {
 		rpc_exit(task, -EIO);
 		return -EIO;
@@ -962,15 +1118,15 @@ static void ff_layout_read_prepare_v4(struct rpc_task *task, void *data)
 {
 	struct nfs_pgio_header *hdr = data;
 
-	if (ff_layout_read_prepare_common(task, hdr))
-		return;
-
 	if (ff_layout_setup_sequence(hdr->ds_clp,
 				     &hdr->args.seq_args,
 				     &hdr->res.seq_res,
 				     task))
 		return;
 
+	if (ff_layout_read_prepare_common(task, hdr))
+		return;
+
 	if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
 			hdr->args.lock_context, FMODE_READ) == -EIO)
 		rpc_exit(task, -EIO); /* lost lock, terminate I/O */
@@ -982,6 +1138,10 @@ static void ff_layout_read_call_done(struct rpc_task *task, void *data)
 
 	dprintk("--> %s task->tk_status %d\n", __func__, task->tk_status);
 
+	nfs4_ff_layout_stat_io_end_read(task,
+			FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
+			hdr->args.count, hdr->res.count);
+
 	if (test_bit(NFS_IOHDR_REDO, &hdr->flags) &&
 	    task->tk_status == 0) {
 		nfs4_sequence_done(task, &hdr->res.seq_res);
@@ -1074,7 +1234,8 @@ static int ff_layout_commit_done_cb(struct rpc_task *task,
 		return -EAGAIN;
 	}
 
-	if (data->verf.committed == NFS_UNSTABLE)
+	if (data->verf.committed == NFS_UNSTABLE
+	    && ff_layout_need_layoutcommit(data->lseg))
 		pnfs_set_layoutcommit(data->inode, data->lseg, data->lwb);
 
 	return 0;
@@ -1083,6 +1244,10 @@ static int ff_layout_commit_done_cb(struct rpc_task *task,
 static int ff_layout_write_prepare_common(struct rpc_task *task,
 					  struct nfs_pgio_header *hdr)
 {
+	nfs4_ff_layout_stat_io_start_write(
+			FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
+			hdr->args.count);
+
 	if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags))) {
 		rpc_exit(task, -EIO);
 		return -EIO;
@@ -1116,15 +1281,15 @@ static void ff_layout_write_prepare_v4(struct rpc_task *task, void *data)
 {
 	struct nfs_pgio_header *hdr = data;
 
-	if (ff_layout_write_prepare_common(task, hdr))
-		return;
-
 	if (ff_layout_setup_sequence(hdr->ds_clp,
 				     &hdr->args.seq_args,
 				     &hdr->res.seq_res,
 				     task))
 		return;
 
+	if (ff_layout_write_prepare_common(task, hdr))
+		return;
+
 	if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
 			hdr->args.lock_context, FMODE_WRITE) == -EIO)
 		rpc_exit(task, -EIO); /* lost lock, terminate I/O */
@@ -1134,6 +1299,11 @@ static void ff_layout_write_call_done(struct rpc_task *task, void *data)
 {
 	struct nfs_pgio_header *hdr = data;
 
+	nfs4_ff_layout_stat_io_end_write(task,
+			FF_LAYOUT_COMP(hdr->lseg, hdr->pgio_mirror_idx),
+			hdr->args.count, hdr->res.count,
+			hdr->res.verf->committed);
+
 	if (test_bit(NFS_IOHDR_REDO, &hdr->flags) &&
 	    task->tk_status == 0) {
 		nfs4_sequence_done(task, &hdr->res.seq_res);
@@ -1152,8 +1322,17 @@ static void ff_layout_write_count_stats(struct rpc_task *task, void *data)
 	    &NFS_CLIENT(hdr->inode)->cl_metrics[NFSPROC4_CLNT_WRITE]);
 }
 
+static void ff_layout_commit_prepare_common(struct rpc_task *task,
+		struct nfs_commit_data *cdata)
+{
+	nfs4_ff_layout_stat_io_start_write(
+			FF_LAYOUT_COMP(cdata->lseg, cdata->ds_commit_index),
+			0);
+}
+
 static void ff_layout_commit_prepare_v3(struct rpc_task *task, void *data)
 {
+	ff_layout_commit_prepare_common(task, data);
 	rpc_call_start(task);
 }
 
@@ -1161,10 +1340,30 @@ static void ff_layout_commit_prepare_v4(struct rpc_task *task, void *data)
 {
 	struct nfs_commit_data *wdata = data;
 
-	ff_layout_setup_sequence(wdata->ds_clp,
+	if (ff_layout_setup_sequence(wdata->ds_clp,
 				 &wdata->args.seq_args,
 				 &wdata->res.seq_res,
-				 task);
+				 task))
+		return;
+	ff_layout_commit_prepare_common(task, data);
+}
+
+static void ff_layout_commit_done(struct rpc_task *task, void *data)
+{
+	struct nfs_commit_data *cdata = data;
+	struct nfs_page *req;
+	__u64 count = 0;
+
+	if (task->tk_status == 0) {
+		list_for_each_entry(req, &cdata->pages, wb_list)
+			count += req->wb_bytes;
+	}
+
+	nfs4_ff_layout_stat_io_end_write(task,
+			FF_LAYOUT_COMP(cdata->lseg, cdata->ds_commit_index),
+			count, count, NFS_FILE_SYNC);
+
+	pnfs_generic_write_commit_done(task, data);
 }
 
 static void ff_layout_commit_count_stats(struct rpc_task *task, void *data)
@@ -1205,14 +1404,14 @@ static const struct rpc_call_ops ff_layout_write_call_ops_v4 = {
 
 static const struct rpc_call_ops ff_layout_commit_call_ops_v3 = {
 	.rpc_call_prepare = ff_layout_commit_prepare_v3,
-	.rpc_call_done = pnfs_generic_write_commit_done,
+	.rpc_call_done = ff_layout_commit_done,
 	.rpc_count_stats = ff_layout_commit_count_stats,
 	.rpc_release = pnfs_generic_commit_release,
 };
 
 static const struct rpc_call_ops ff_layout_commit_call_ops_v4 = {
 	.rpc_call_prepare = ff_layout_commit_prepare_v4,
-	.rpc_call_done = pnfs_generic_write_commit_done,
+	.rpc_call_done = ff_layout_commit_done,
 	.rpc_count_stats = ff_layout_commit_count_stats,
 	.rpc_release = pnfs_generic_commit_release,
 };
@@ -1256,7 +1455,6 @@ ff_layout_read_pagelist(struct nfs_pgio_header *hdr)
 	fh = nfs4_ff_layout_select_ds_fh(lseg, idx);
 	if (fh)
 		hdr->args.fh = fh;
-
 	/*
 	 * Note that if we ever decide to split across DSes,
 	 * then we may need to handle dense-like offsets.
@@ -1385,6 +1583,7 @@ static int ff_layout_initiate_commit(struct nfs_commit_data *data, int how)
 	fh = select_ds_fh_from_commit(lseg, data->ds_commit_index);
 	if (fh)
 		data->args.fh = fh;
+
 	return nfs_initiate_commit(ds_clnt, data, ds->ds_clp->rpc_ops,
 				   vers == 3 ? &ff_layout_commit_call_ops_v3 :
 					       &ff_layout_commit_call_ops_v4,
@@ -1488,6 +1687,247 @@ out:
 	dprintk("%s: Return\n", __func__);
 }
 
+static int
+ff_layout_ntop4(const struct sockaddr *sap, char *buf, const size_t buflen)
+{
+	const struct sockaddr_in *sin = (struct sockaddr_in *)sap;
+
+	return snprintf(buf, buflen, "%pI4", &sin->sin_addr);
+}
+
+static size_t
+ff_layout_ntop6_noscopeid(const struct sockaddr *sap, char *buf,
+			  const int buflen)
+{
+	const struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)sap;
+	const struct in6_addr *addr = &sin6->sin6_addr;
+
+	/*
+	 * RFC 4291, Section 2.2.2
+	 *
+	 * Shorthanded ANY address
+	 */
+	if (ipv6_addr_any(addr))
+		return snprintf(buf, buflen, "::");
+
+	/*
+	 * RFC 4291, Section 2.2.2
+	 *
+	 * Shorthanded loopback address
+	 */
+	if (ipv6_addr_loopback(addr))
+		return snprintf(buf, buflen, "::1");
+
+	/*
+	 * RFC 4291, Section 2.2.3
+	 *
+	 * Special presentation address format for mapped v4
+	 * addresses.
+	 */
+	if (ipv6_addr_v4mapped(addr))
+		return snprintf(buf, buflen, "::ffff:%pI4",
+					&addr->s6_addr32[3]);
+
+	/*
+	 * RFC 4291, Section 2.2.1
+	 */
+	return snprintf(buf, buflen, "%pI6c", addr);
+}
+
+/* Derived from rpc_sockaddr2uaddr */
+static void
+ff_layout_encode_netaddr(struct xdr_stream *xdr, struct nfs4_pnfs_ds_addr *da)
+{
+	struct sockaddr *sap = (struct sockaddr *)&da->da_addr;
+	char portbuf[RPCBIND_MAXUADDRPLEN];
+	char addrbuf[RPCBIND_MAXUADDRLEN];
+	char *netid;
+	unsigned short port;
+	int len, netid_len;
+	__be32 *p;
+
+	switch (sap->sa_family) {
+	case AF_INET:
+		if (ff_layout_ntop4(sap, addrbuf, sizeof(addrbuf)) == 0)
+			return;
+		port = ntohs(((struct sockaddr_in *)sap)->sin_port);
+		netid = "tcp";
+		netid_len = 3;
+		break;
+	case AF_INET6:
+		if (ff_layout_ntop6_noscopeid(sap, addrbuf, sizeof(addrbuf)) == 0)
+			return;
+		port = ntohs(((struct sockaddr_in6 *)sap)->sin6_port);
+		netid = "tcp6";
+		netid_len = 4;
+		break;
+	default:
+		/* we only support tcp and tcp6 */
+		WARN_ON_ONCE(1);
+		return;
+	}
+
+	snprintf(portbuf, sizeof(portbuf), ".%u.%u", port >> 8, port & 0xff);
+	len = strlcat(addrbuf, portbuf, sizeof(addrbuf));
+
+	p = xdr_reserve_space(xdr, 4 + netid_len);
+	xdr_encode_opaque(p, netid, netid_len);
+
+	p = xdr_reserve_space(xdr, 4 + len);
+	xdr_encode_opaque(p, addrbuf, len);
+}
+
+static void
+ff_layout_encode_nfstime(struct xdr_stream *xdr,
+			 ktime_t t)
+{
+	struct timespec64 ts;
+	__be32 *p;
+
+	p = xdr_reserve_space(xdr, 12);
+	ts = ktime_to_timespec64(t);
+	p = xdr_encode_hyper(p, ts.tv_sec);
+	*p++ = cpu_to_be32(ts.tv_nsec);
+}
+
+static void
+ff_layout_encode_io_latency(struct xdr_stream *xdr,
+			    struct nfs4_ff_io_stat *stat)
+{
+	__be32 *p;
+
+	p = xdr_reserve_space(xdr, 5 * 8);
+	p = xdr_encode_hyper(p, stat->ops_requested);
+	p = xdr_encode_hyper(p, stat->bytes_requested);
+	p = xdr_encode_hyper(p, stat->ops_completed);
+	p = xdr_encode_hyper(p, stat->bytes_completed);
+	p = xdr_encode_hyper(p, stat->bytes_not_delivered);
+	ff_layout_encode_nfstime(xdr, stat->total_busy_time);
+	ff_layout_encode_nfstime(xdr, stat->aggregate_completion_time);
+}
+
+static void
+ff_layout_encode_layoutstats(struct xdr_stream *xdr,
+			     struct nfs42_layoutstat_args *args,
+			     struct nfs42_layoutstat_devinfo *devinfo)
+{
+	struct nfs4_ff_layout_mirror *mirror = devinfo->layout_private;
+	struct nfs4_pnfs_ds_addr *da;
+	struct nfs4_pnfs_ds *ds = mirror->mirror_ds->ds;
+	struct nfs_fh *fh = &mirror->fh_versions[0];
+	__be32 *p, *start;
+
+	da = list_first_entry(&ds->ds_addrs, struct nfs4_pnfs_ds_addr, da_node);
+	dprintk("%s: DS %s: encoding address %s\n",
+		__func__, ds->ds_remotestr, da->da_remotestr);
+	/* layoutupdate length */
+	start = xdr_reserve_space(xdr, 4);
+	/* netaddr4 */
+	ff_layout_encode_netaddr(xdr, da);
+	/* nfs_fh4 */
+	p = xdr_reserve_space(xdr, 4 + fh->size);
+	xdr_encode_opaque(p, fh->data, fh->size);
+	/* ff_io_latency4 read */
+	spin_lock(&mirror->lock);
+	ff_layout_encode_io_latency(xdr, &mirror->read_stat.io_stat);
+	/* ff_io_latency4 write */
+	ff_layout_encode_io_latency(xdr, &mirror->write_stat.io_stat);
+	spin_unlock(&mirror->lock);
+	/* nfstime4 */
+	ff_layout_encode_nfstime(xdr, ktime_sub(ktime_get(), mirror->start_time));
+	/* bool */
+	p = xdr_reserve_space(xdr, 4);
+	*p = cpu_to_be32(false);
+
+	*start = cpu_to_be32((xdr->p - start - 1) * 4);
+}
+
+static bool
+ff_layout_mirror_prepare_stats(struct nfs42_layoutstat_args *args,
+			       struct pnfs_layout_segment *pls,
+			       int *dev_count, int dev_limit)
+{
+	struct nfs4_ff_layout_mirror *mirror;
+	struct nfs4_deviceid_node *dev;
+	struct nfs42_layoutstat_devinfo *devinfo;
+	int i;
+
+	for (i = 0; i <= FF_LAYOUT_MIRROR_COUNT(pls); i++) {
+		if (*dev_count >= dev_limit)
+			break;
+		mirror = FF_LAYOUT_COMP(pls, i);
+		if (!mirror || !mirror->mirror_ds)
+			continue;
+		dev = FF_LAYOUT_DEVID_NODE(pls, i);
+		devinfo = &args->devinfo[*dev_count];
+		memcpy(&devinfo->dev_id, &dev->deviceid, NFS4_DEVICEID4_SIZE);
+		devinfo->offset = pls->pls_range.offset;
+		devinfo->length = pls->pls_range.length;
+		/* well, we don't really know if IO is continuous or not! */
+		devinfo->read_count = mirror->read_stat.io_stat.bytes_completed;
+		devinfo->read_bytes = mirror->read_stat.io_stat.bytes_completed;
+		devinfo->write_count = mirror->write_stat.io_stat.bytes_completed;
+		devinfo->write_bytes = mirror->write_stat.io_stat.bytes_completed;
+		devinfo->layout_type = LAYOUT_FLEX_FILES;
+		devinfo->layoutstats_encode = ff_layout_encode_layoutstats;
+		devinfo->layout_private = mirror;
+		/* lseg refcount put in cleanup_layoutstats */
+		pnfs_get_lseg(pls);
+
+		++(*dev_count);
+	}
+
+	return *dev_count < dev_limit;
+}
+
+static int
+ff_layout_prepare_layoutstats(struct nfs42_layoutstat_args *args)
+{
+	struct pnfs_layout_segment *pls;
+	int dev_count = 0;
+
+	spin_lock(&args->inode->i_lock);
+	list_for_each_entry(pls, &NFS_I(args->inode)->layout->plh_segs, pls_list) {
+		dev_count += FF_LAYOUT_MIRROR_COUNT(pls);
+	}
+	spin_unlock(&args->inode->i_lock);
+	/* For now, send at most PNFS_LAYOUTSTATS_MAXDEV statistics */
+	if (dev_count > PNFS_LAYOUTSTATS_MAXDEV) {
+		dprintk("%s: truncating devinfo to limit (%d:%d)\n",
+			__func__, dev_count, PNFS_LAYOUTSTATS_MAXDEV);
+		dev_count = PNFS_LAYOUTSTATS_MAXDEV;
+	}
+	args->devinfo = kmalloc(dev_count * sizeof(*args->devinfo), GFP_KERNEL);
+	if (!args->devinfo)
+		return -ENOMEM;
+
+	dev_count = 0;
+	spin_lock(&args->inode->i_lock);
+	list_for_each_entry(pls, &NFS_I(args->inode)->layout->plh_segs, pls_list) {
+		if (!ff_layout_mirror_prepare_stats(args, pls, &dev_count,
+						    PNFS_LAYOUTSTATS_MAXDEV)) {
+			break;
+		}
+	}
+	spin_unlock(&args->inode->i_lock);
+	args->num_dev = dev_count;
+
+	return 0;
+}
+
+static void
+ff_layout_cleanup_layoutstats(struct nfs42_layoutstat_data *data)
+{
+	struct nfs4_ff_layout_mirror *mirror;
+	int i;
+
+	for (i = 0; i < data->args.num_dev; i++) {
+		mirror = data->args.devinfo[i].layout_private;
+		data->args.devinfo[i].layout_private = NULL;
+		pnfs_put_lseg(mirror->lseg);
+	}
+}
+
 static struct pnfs_layoutdriver_type flexfilelayout_type = {
 	.id			= LAYOUT_FLEX_FILES,
 	.name			= "LAYOUT_FLEX_FILES",
@@ -1510,6 +1950,8 @@ static struct pnfs_layoutdriver_type flexfilelayout_type = {
 	.alloc_deviceid_node    = ff_layout_alloc_deviceid_node,
 	.encode_layoutreturn    = ff_layout_encode_layoutreturn,
 	.sync			= pnfs_nfs_generic_sync,
+	.prepare_layoutstats	= ff_layout_prepare_layoutstats,
+	.cleanup_layoutstats	= ff_layout_cleanup_layoutstats,
 };
 
 static int __init nfs4flexfilelayout_init(void)
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.h b/fs/nfs/flexfilelayout/flexfilelayout.h
index 070f20445b2d..f92f9a0a856b 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.h
+++ b/fs/nfs/flexfilelayout/flexfilelayout.h
@@ -9,12 +9,17 @@
 #ifndef FS_NFS_NFS4FLEXFILELAYOUT_H
 #define FS_NFS_NFS4FLEXFILELAYOUT_H
 
+#define FF_FLAGS_NO_LAYOUTCOMMIT 1
+
 #include "../pnfs.h"
 
 /* XXX: Let's filter out insanely large mirror count for now to avoid oom
  * due to network error etc. */
 #define NFS4_FLEXFILE_LAYOUT_MAX_MIRROR_CNT 4096
 
+/* LAYOUTSTATS report interval in ms */
+#define FF_LAYOUTSTATS_REPORT_INTERVAL (60000L)
+
 struct nfs4_ff_ds_version {
 	u32				version;
 	u32				minor_version;
@@ -41,24 +46,48 @@ struct nfs4_ff_layout_ds_err {
 	struct nfs4_deviceid		deviceid;
 };
 
+struct nfs4_ff_io_stat {
+	__u64				ops_requested;
+	__u64				bytes_requested;
+	__u64				ops_completed;
+	__u64				bytes_completed;
+	__u64				bytes_not_delivered;
+	ktime_t				total_busy_time;
+	ktime_t				aggregate_completion_time;
+};
+
+struct nfs4_ff_busy_timer {
+	ktime_t start_time;
+	atomic_t n_ops;
+};
+
+struct nfs4_ff_layoutstat {
+	struct nfs4_ff_io_stat io_stat;
+	struct nfs4_ff_busy_timer busy_timer;
+};
+
 struct nfs4_ff_layout_mirror {
+	struct pnfs_layout_segment	*lseg; /* back pointer */
 	u32				ds_count;
 	u32				efficiency;
 	struct nfs4_ff_layout_ds	*mirror_ds;
 	u32				fh_versions_cnt;
 	struct nfs_fh			*fh_versions;
 	nfs4_stateid			stateid;
-	struct nfs4_string		user_name;
-	struct nfs4_string		group_name;
 	u32				uid;
 	u32				gid;
 	struct rpc_cred			*cred;
 	spinlock_t			lock;
+	struct nfs4_ff_layoutstat	read_stat;
+	struct nfs4_ff_layoutstat	write_stat;
+	ktime_t				start_time;
+	ktime_t				last_report_time;
 };
 
 struct nfs4_ff_layout_segment {
 	struct pnfs_layout_segment	generic_hdr;
 	u64				stripe_unit;
+	u32				flags;
 	u32				mirror_array_cnt;
 	struct nfs4_ff_layout_mirror	**mirror_array;
 };
diff --git a/fs/nfs/flexfilelayout/flexfilelayoutdev.c b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
index 77a2d026aa12..f13e1969eedd 100644
--- a/fs/nfs/flexfilelayout/flexfilelayoutdev.c
+++ b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
@@ -324,7 +324,8 @@ static int ff_layout_update_mirror_cred(struct nfs4_ff_layout_mirror *mirror,
 				__func__, PTR_ERR(cred));
 			return PTR_ERR(cred);
 		} else {
-			mirror->cred = cred;
+			if (cmpxchg(&mirror->cred, NULL, cred))
+				put_rpccred(cred);
 		}
 	}
 	return 0;
@@ -386,7 +387,7 @@ nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
 	/* matching smp_wmb() in _nfs4_pnfs_v3/4_ds_connect */
 	smp_rmb();
 	if (ds->ds_clp)
-		goto out;
+		goto out_update_creds;
 
 	flavor = nfs4_ff_layout_choose_authflavor(mirror);
 
@@ -430,7 +431,7 @@ nfs4_ff_layout_prepare_ds(struct pnfs_layout_segment *lseg, u32 ds_idx,
 			}
 		}
 	}
-
+out_update_creds:
 	if (ff_layout_update_mirror_cred(mirror, ds))
 		ds = NULL;
 out:
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index f734562c6d24..b77b328a06d7 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -678,6 +678,8 @@ int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
 	if (!err) {
 		generic_fillattr(inode, stat);
 		stat->ino = nfs_compat_user_ino64(NFS_FILEID(inode));
+		if (S_ISDIR(inode->i_mode))
+			stat->blksize = NFS_SERVER(inode)->dtsize;
 	}
 out:
 	trace_nfs_getattr_exit(inode, err);
@@ -2008,17 +2010,15 @@ static int __init init_nfs_fs(void)
 	if (err)
 		goto out1;
 
-#ifdef CONFIG_PROC_FS
 	rpc_proc_register(&init_net, &nfs_rpcstat);
-#endif
-	if ((err = register_nfs_fs()) != 0)
+
+	err = register_nfs_fs();
+	if (err)
 		goto out0;
 
 	return 0;
 out0:
-#ifdef CONFIG_PROC_FS
 	rpc_proc_unregister(&init_net, "nfs");
-#endif
 	nfs_destroy_directcache();
 out1:
 	nfs_destroy_writepagecache();
@@ -2049,9 +2049,7 @@ static void __exit exit_nfs_fs(void)
 	nfs_destroy_nfspagecache();
 	nfs_fscache_unregister();
 	unregister_pernet_subsys(&nfs_net_ops);
-#ifdef CONFIG_PROC_FS
 	rpc_proc_unregister(&init_net, "nfs");
-#endif
 	unregister_nfs_fs();
 	nfs_fs_proc_exit();
 	nfsiod_stop();
diff --git a/fs/nfs/nfs3xdr.c b/fs/nfs/nfs3xdr.c
index 53852a4bd88b..9b04c2e6fffc 100644
--- a/fs/nfs/nfs3xdr.c
+++ b/fs/nfs/nfs3xdr.c
@@ -1342,7 +1342,7 @@ static void nfs3_xdr_enc_setacl3args(struct rpc_rqst *req,
 	if (args->npages != 0)
 		xdr_write_pages(xdr, args->pages, 0, args->len);
 	else
-		xdr_reserve_space(xdr, NFS_ACL_INLINE_BUFSIZE);
+		xdr_reserve_space(xdr, args->len);
 
 	error = nfsacl_encode(xdr->buf, base, args->inode,
 			    (args->mask & NFS_ACL) ?
diff --git a/fs/nfs/nfs42.h b/fs/nfs/nfs42.h
index 7afb8947dfdf..ff66ae700b89 100644
--- a/fs/nfs/nfs42.h
+++ b/fs/nfs/nfs42.h
@@ -5,11 +5,18 @@
 #ifndef __LINUX_FS_NFS_NFS4_2_H
 #define __LINUX_FS_NFS_NFS4_2_H
 
+/*
+ * FIXME:  four LAYOUTSTATS calls per compound at most! Do we need to support
+ * more? Need to consider not to pre-alloc too much for a compound.
+ */
+#define PNFS_LAYOUTSTATS_MAXDEV (4)
+
 /* nfs4.2proc.c */
 int nfs42_proc_allocate(struct file *, loff_t, loff_t);
 int nfs42_proc_deallocate(struct file *, loff_t, loff_t);
 loff_t nfs42_proc_llseek(struct file *, loff_t, int);
-
+int nfs42_proc_layoutstats_generic(struct nfs_server *,
+				   struct nfs42_layoutstat_data *);
 /* nfs4.2xdr.h */
 extern struct rpc_procinfo nfs4_2_procedures[];
 
diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index 3a9e75235f30..f486b80f927a 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -10,6 +10,11 @@
 #include <linux/nfs_fs.h>
 #include "nfs4_fs.h"
 #include "nfs42.h"
+#include "iostat.h"
+#include "pnfs.h"
+#include "internal.h"
+
+#define NFSDBG_FACILITY NFSDBG_PNFS
 
 static int nfs42_set_rw_stateid(nfs4_stateid *dst, struct file *file,
 				fmode_t fmode)
@@ -165,3 +170,85 @@ loff_t nfs42_proc_llseek(struct file *filep, loff_t offset, int whence)
 
 	return vfs_setpos(filep, res.sr_offset, inode->i_sb->s_maxbytes);
 }
+
+static void
+nfs42_layoutstat_prepare(struct rpc_task *task, void *calldata)
+{
+	struct nfs42_layoutstat_data *data = calldata;
+	struct nfs_server *server = NFS_SERVER(data->args.inode);
+
+	nfs41_setup_sequence(nfs4_get_session(server), &data->args.seq_args,
+			     &data->res.seq_res, task);
+}
+
+static void
+nfs42_layoutstat_done(struct rpc_task *task, void *calldata)
+{
+	struct nfs42_layoutstat_data *data = calldata;
+
+	if (!nfs4_sequence_done(task, &data->res.seq_res))
+		return;
+
+	switch (task->tk_status) {
+	case 0:
+		break;
+	case -ENOTSUPP:
+	case -EOPNOTSUPP:
+		NFS_SERVER(data->inode)->caps &= ~NFS_CAP_LAYOUTSTATS;
+	default:
+		dprintk("%s server returns %d\n", __func__, task->tk_status);
+	}
+}
+
+static void
+nfs42_layoutstat_release(void *calldata)
+{
+	struct nfs42_layoutstat_data *data = calldata;
+	struct nfs_server *nfss = NFS_SERVER(data->args.inode);
+
+	if (nfss->pnfs_curr_ld->cleanup_layoutstats)
+		nfss->pnfs_curr_ld->cleanup_layoutstats(data);
+
+	pnfs_put_layout_hdr(NFS_I(data->args.inode)->layout);
+	smp_mb__before_atomic();
+	clear_bit(NFS_INO_LAYOUTSTATS, &NFS_I(data->args.inode)->flags);
+	smp_mb__after_atomic();
+	nfs_iput_and_deactive(data->inode);
+	kfree(data->args.devinfo);
+	kfree(data);
+}
+
+static const struct rpc_call_ops nfs42_layoutstat_ops = {
+	.rpc_call_prepare = nfs42_layoutstat_prepare,
+	.rpc_call_done = nfs42_layoutstat_done,
+	.rpc_release = nfs42_layoutstat_release,
+};
+
+int nfs42_proc_layoutstats_generic(struct nfs_server *server,
+				   struct nfs42_layoutstat_data *data)
+{
+	struct rpc_message msg = {
+		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTSTATS],
+		.rpc_argp = &data->args,
+		.rpc_resp = &data->res,
+	};
+	struct rpc_task_setup task_setup = {
+		.rpc_client = server->client,
+		.rpc_message = &msg,
+		.callback_ops = &nfs42_layoutstat_ops,
+		.callback_data = data,
+		.flags = RPC_TASK_ASYNC,
+	};
+	struct rpc_task *task;
+
+	data->inode = nfs_igrab_and_active(data->args.inode);
+	if (!data->inode) {
+		nfs42_layoutstat_release(data);
+		return -EAGAIN;
+	}
+	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
+	task = rpc_run_task(&task_setup);
+	if (IS_ERR(task))
+		return PTR_ERR(task);
+	return 0;
+}
diff --git a/fs/nfs/nfs42xdr.c b/fs/nfs/nfs42xdr.c
index 1a25b27248f2..a6bd27da6286 100644
--- a/fs/nfs/nfs42xdr.c
+++ b/fs/nfs/nfs42xdr.c
@@ -4,6 +4,8 @@
 #ifndef __LINUX_FS_NFS_NFS4_2XDR_H
 #define __LINUX_FS_NFS_NFS4_2XDR_H
 
+#include "nfs42.h"
+
 #define encode_fallocate_maxsz		(encode_stateid_maxsz + \
 					 2 /* offset */ + \
 					 2 /* length */)
@@ -22,6 +24,16 @@
 					 1 /* whence */ + \
 					 2 /* offset */ + \
 					 2 /* length */)
+#define encode_io_info_maxsz		4
+#define encode_layoutstats_maxsz	(op_decode_hdr_maxsz + \
+					2 /* offset */ + \
+					2 /* length */ + \
+					encode_stateid_maxsz + \
+					encode_io_info_maxsz + \
+					encode_io_info_maxsz + \
+					1 /* opaque devaddr4 length */ + \
+					XDR_QUADLEN(PNFS_LAYOUTSTATS_MAXSIZE))
+#define decode_layoutstats_maxsz	(op_decode_hdr_maxsz)
 
 #define NFS4_enc_allocate_sz		(compound_encode_hdr_maxsz + \
 					 encode_putfh_maxsz + \
@@ -45,6 +57,14 @@
 #define NFS4_dec_seek_sz		(compound_decode_hdr_maxsz + \
 					 decode_putfh_maxsz + \
 					 decode_seek_maxsz)
+#define NFS4_enc_layoutstats_sz		(compound_encode_hdr_maxsz + \
+					 encode_sequence_maxsz + \
+					 encode_putfh_maxsz + \
+					 PNFS_LAYOUTSTATS_MAXDEV * encode_layoutstats_maxsz)
+#define NFS4_dec_layoutstats_sz		(compound_decode_hdr_maxsz + \
+					 decode_sequence_maxsz + \
+					 decode_putfh_maxsz + \
+					 PNFS_LAYOUTSTATS_MAXDEV * decode_layoutstats_maxsz)
 
 
 static void encode_fallocate(struct xdr_stream *xdr,
@@ -81,6 +101,33 @@ static void encode_seek(struct xdr_stream *xdr,
 	encode_uint32(xdr, args->sa_what);
 }
 
+static void encode_layoutstats(struct xdr_stream *xdr,
+			       struct nfs42_layoutstat_args *args,
+			       struct nfs42_layoutstat_devinfo *devinfo,
+			       struct compound_hdr *hdr)
+{
+	__be32 *p;
+
+	encode_op_hdr(xdr, OP_LAYOUTSTATS, decode_layoutstats_maxsz, hdr);
+	p = reserve_space(xdr, 8 + 8);
+	p = xdr_encode_hyper(p, devinfo->offset);
+	p = xdr_encode_hyper(p, devinfo->length);
+	encode_nfs4_stateid(xdr, &args->stateid);
+	p = reserve_space(xdr, 4*8 + NFS4_DEVICEID4_SIZE + 4);
+	p = xdr_encode_hyper(p, devinfo->read_count);
+	p = xdr_encode_hyper(p, devinfo->read_bytes);
+	p = xdr_encode_hyper(p, devinfo->write_count);
+	p = xdr_encode_hyper(p, devinfo->write_bytes);
+	p = xdr_encode_opaque_fixed(p, devinfo->dev_id.data,
+			NFS4_DEVICEID4_SIZE);
+	/* Encode layoutupdate4 */
+	*p++ = cpu_to_be32(devinfo->layout_type);
+	if (devinfo->layoutstats_encode != NULL)
+		devinfo->layoutstats_encode(xdr, args, devinfo);
+	else
+		encode_uint32(xdr, 0);
+}
+
 /*
  * Encode ALLOCATE request
  */
@@ -137,6 +184,28 @@ static void nfs4_xdr_enc_seek(struct rpc_rqst *req,
 	encode_nops(&hdr);
 }
 
+/*
+ * Encode LAYOUTSTATS request
+ */
+static void nfs4_xdr_enc_layoutstats(struct rpc_rqst *req,
+				     struct xdr_stream *xdr,
+				     struct nfs42_layoutstat_args *args)
+{
+	int i;
+
+	struct compound_hdr hdr = {
+		.minorversion = nfs4_xdr_minorversion(&args->seq_args),
+	};
+
+	encode_compound_hdr(xdr, req, &hdr);
+	encode_sequence(xdr, &args->seq_args, &hdr);
+	encode_putfh(xdr, args->fh, &hdr);
+	WARN_ON(args->num_dev > PNFS_LAYOUTSTATS_MAXDEV);
+	for (i = 0; i < args->num_dev; i++)
+		encode_layoutstats(xdr, args, &args->devinfo[i], &hdr);
+	encode_nops(&hdr);
+}
+
 static int decode_allocate(struct xdr_stream *xdr, struct nfs42_falloc_res *res)
 {
 	return decode_op_hdr(xdr, OP_ALLOCATE);
@@ -169,6 +238,12 @@ out_overflow:
 	return -EIO;
 }
 
+static int decode_layoutstats(struct xdr_stream *xdr,
+			      struct nfs42_layoutstat_res *res)
+{
+	return decode_op_hdr(xdr, OP_LAYOUTSTATS);
+}
+
 /*
  * Decode ALLOCATE request
  */
@@ -246,4 +321,35 @@ static int nfs4_xdr_dec_seek(struct rpc_rqst *rqstp,
 out:
 	return status;
 }
+
+/*
+ * Decode LAYOUTSTATS request
+ */
+static int nfs4_xdr_dec_layoutstats(struct rpc_rqst *rqstp,
+				    struct xdr_stream *xdr,
+				    struct nfs42_layoutstat_res *res)
+{
+	struct compound_hdr hdr;
+	int status, i;
+
+	status = decode_compound_hdr(xdr, &hdr);
+	if (status)
+		goto out;
+	status = decode_sequence(xdr, &res->seq_res, rqstp);
+	if (status)
+		goto out;
+	status = decode_putfh(xdr);
+	if (status)
+		goto out;
+	WARN_ON(res->num_dev > PNFS_LAYOUTSTATS_MAXDEV);
+	for (i = 0; i < res->num_dev; i++) {
+		status = decode_layoutstats(xdr, res);
+		if (status)
+			goto out;
+	}
+out:
+	res->rpc_status = status;
+	return status;
+}
+
 #endif /* __LINUX_FS_NFS_NFS4_2XDR_H */
diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index fdef424b0cd3..ea3bee919a76 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -233,6 +233,7 @@ extern int nfs4_handle_exception(struct nfs_server *, int, struct nfs4_exception
 extern int nfs4_call_sync(struct rpc_clnt *, struct nfs_server *,
 			  struct rpc_message *, struct nfs4_sequence_args *,
 			  struct nfs4_sequence_res *, int);
+extern void nfs4_init_sequence(struct nfs4_sequence_args *, struct nfs4_sequence_res *, int);
 extern int nfs4_proc_setclientid(struct nfs_client *, u32, unsigned short, struct rpc_cred *, struct nfs4_setclientid_res *);
 extern int nfs4_proc_setclientid_confirm(struct nfs_client *, struct nfs4_setclientid_res *arg, struct rpc_cred *);
 extern int nfs4_proc_get_rootfh(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *, bool);
diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c
index e42be52a8c18..3aa6a9ba5113 100644
--- a/fs/nfs/nfs4client.c
+++ b/fs/nfs/nfs4client.c
@@ -676,7 +676,6 @@ found:
 		break;
 	}
 
-	/* No matching nfs_client found. */
 	spin_unlock(&nn->nfs_client_lock);
 	dprintk("NFS: <-- %s status = %d\n", __func__, status);
 	nfs_put_client(prev);
diff --git a/fs/nfs/nfs4file.c b/fs/nfs/nfs4file.c
index f58c17b3b480..dcd39d4e2efe 100644
--- a/fs/nfs/nfs4file.c
+++ b/fs/nfs/nfs4file.c
@@ -41,6 +41,10 @@ nfs4_file_open(struct inode *inode, struct file *filp)
 
 	dprintk("NFS: open file(%pd2)\n", dentry);
 
+	err = nfs_check_flags(openflags);
+	if (err)
+		return err;
+
 	if ((openflags & O_ACCMODE) == 3)
 		openflags--;
 
diff --git a/fs/nfs/nfs4getroot.c b/fs/nfs/nfs4getroot.c
index c0b3a16b4a00..039b3eb6d834 100644
--- a/fs/nfs/nfs4getroot.c
+++ b/fs/nfs/nfs4getroot.c
@@ -35,13 +35,6 @@ int nfs4_get_rootfh(struct nfs_server *server, struct nfs_fh *mntfh, bool auth_p
 		goto out;
 	}
 
-	if (fsinfo.fattr->valid & NFS_ATTR_FATTR_V4_REFERRAL) {
-		printk(KERN_ERR "nfs4_get_rootfh:"
-		       " getroot obtained referral\n");
-		ret = -EREMOTE;
-		goto out;
-	}
-
 	memcpy(&server->fsid, &fsinfo.fattr->fsid, sizeof(server->fsid));
 out:
 	nfs_free_fattr(fsinfo.fattr);
diff --git a/fs/nfs/nfs4idmap.c b/fs/nfs/nfs4idmap.c
index 2e1737c40a29..535dfc69c628 100644
--- a/fs/nfs/nfs4idmap.c
+++ b/fs/nfs/nfs4idmap.c
@@ -494,12 +494,7 @@ nfs_idmap_delete(struct nfs_client *clp)
 
 int nfs_idmap_init(void)
 {
-	int ret;
-	ret = nfs_idmap_init_keyring();
-	if (ret != 0)
-		goto out;
-out:
-	return ret;
+	return nfs_idmap_init_keyring();
 }
 
 void nfs_idmap_quit(void)
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 55e1e3af23a3..6f228b5af819 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -356,6 +356,9 @@ int nfs4_handle_exception(struct nfs_server *server, int errorcode, struct nfs4_
 		case 0:
 			return 0;
 		case -NFS4ERR_OPENMODE:
+		case -NFS4ERR_DELEG_REVOKED:
+		case -NFS4ERR_ADMIN_REVOKED:
+		case -NFS4ERR_BAD_STATEID:
 			if (inode && nfs4_have_delegation(inode, FMODE_READ)) {
 				nfs4_inode_return_delegation(inode);
 				exception->retry = 1;
@@ -367,15 +370,6 @@ int nfs4_handle_exception(struct nfs_server *server, int errorcode, struct nfs4_
 			if (ret < 0)
 				break;
 			goto wait_on_recovery;
-		case -NFS4ERR_DELEG_REVOKED:
-		case -NFS4ERR_ADMIN_REVOKED:
-		case -NFS4ERR_BAD_STATEID:
-			if (state == NULL)
-				break;
-			ret = nfs4_schedule_stateid_recovery(server, state);
-			if (ret < 0)
-				break;
-			goto wait_on_recovery;
 		case -NFS4ERR_EXPIRED:
 			if (state != NULL) {
 				ret = nfs4_schedule_stateid_recovery(server, state);
@@ -482,8 +476,8 @@ struct nfs4_call_sync_data {
 	struct nfs4_sequence_res *seq_res;
 };
 
-static void nfs4_init_sequence(struct nfs4_sequence_args *args,
-			       struct nfs4_sequence_res *res, int cache_reply)
+void nfs4_init_sequence(struct nfs4_sequence_args *args,
+			struct nfs4_sequence_res *res, int cache_reply)
 {
 	args->sa_slot = NULL;
 	args->sa_cache_this = cache_reply;
@@ -1553,6 +1547,13 @@ static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, fmode_t fmod
 	struct nfs4_state *newstate;
 	int ret;
 
+	if ((opendata->o_arg.claim == NFS4_OPEN_CLAIM_DELEGATE_CUR ||
+	     opendata->o_arg.claim == NFS4_OPEN_CLAIM_DELEG_CUR_FH) &&
+	    (opendata->o_arg.u.delegation_type & fmode) != fmode)
+		/* This mode can't have been delegated, so we must have
+		 * a valid open_stateid to cover it - not need to reclaim.
+		 */
+		return 0;
 	opendata->o_arg.open_flags = 0;
 	opendata->o_arg.fmode = fmode;
 	opendata->o_arg.share_access = nfs4_map_atomic_open_share(
@@ -1684,6 +1685,7 @@ static int nfs4_handle_delegation_recall_error(struct nfs_server *server, struct
 					"%d.\n", __func__, err);
 		case 0:
 		case -ENOENT:
+		case -EAGAIN:
 		case -ESTALE:
 			break;
 		case -NFS4ERR_BADSESSION:
@@ -3355,6 +3357,8 @@ static int nfs4_proc_lookup_common(struct rpc_clnt **clnt, struct inode *dir,
 			goto out;
 		case -NFS4ERR_MOVED:
 			err = nfs4_get_referral(client, dir, name, fattr, fhandle);
+			if (err == -NFS4ERR_MOVED)
+				err = nfs4_handle_exception(NFS_SERVER(dir), err, &exception);
 			goto out;
 		case -NFS4ERR_WRONGSEC:
 			err = -EPERM;
@@ -4955,49 +4959,128 @@ static void nfs4_init_boot_verifier(const struct nfs_client *clp,
 	memcpy(bootverf->data, verf, sizeof(bootverf->data));
 }
 
-static unsigned int
-nfs4_init_nonuniform_client_string(struct nfs_client *clp,
-				   char *buf, size_t len)
+static int
+nfs4_init_nonuniform_client_string(struct nfs_client *clp)
 {
-	unsigned int result;
+	int result;
+	size_t len;
+	char *str;
+	bool retried = false;
 
 	if (clp->cl_owner_id != NULL)
-		return strlcpy(buf, clp->cl_owner_id, len);
+		return 0;
+retry:
+	rcu_read_lock();
+	len = 10 + strlen(clp->cl_ipaddr) + 1 +
+		strlen(rpc_peeraddr2str(clp->cl_rpcclient, RPC_DISPLAY_ADDR)) +
+		1 +
+		strlen(rpc_peeraddr2str(clp->cl_rpcclient, RPC_DISPLAY_PROTO)) +
+		1;
+	rcu_read_unlock();
+
+	if (len > NFS4_OPAQUE_LIMIT + 1)
+		return -EINVAL;
+
+	/*
+	 * Since this string is allocated at mount time, and held until the
+	 * nfs_client is destroyed, we can use GFP_KERNEL here w/o worrying
+	 * about a memory-reclaim deadlock.
+	 */
+	str = kmalloc(len, GFP_KERNEL);
+	if (!str)
+		return -ENOMEM;
 
 	rcu_read_lock();
-	result = scnprintf(buf, len, "Linux NFSv4.0 %s/%s %s",
-				clp->cl_ipaddr,
-				rpc_peeraddr2str(clp->cl_rpcclient,
-							RPC_DISPLAY_ADDR),
-				rpc_peeraddr2str(clp->cl_rpcclient,
-							RPC_DISPLAY_PROTO));
+	result = scnprintf(str, len, "Linux NFSv4.0 %s/%s %s",
+			clp->cl_ipaddr,
+			rpc_peeraddr2str(clp->cl_rpcclient, RPC_DISPLAY_ADDR),
+			rpc_peeraddr2str(clp->cl_rpcclient, RPC_DISPLAY_PROTO));
 	rcu_read_unlock();
-	clp->cl_owner_id = kstrdup(buf, GFP_KERNEL);
-	return result;
+
+	/* Did something change? */
+	if (result >= len) {
+		kfree(str);
+		if (retried)
+			return -EINVAL;
+		retried = true;
+		goto retry;
+	}
+	clp->cl_owner_id = str;
+	return 0;
 }
 
-static unsigned int
-nfs4_init_uniform_client_string(struct nfs_client *clp,
-				char *buf, size_t len)
+static int
+nfs4_init_uniquifier_client_string(struct nfs_client *clp)
+{
+	int result;
+	size_t len;
+	char *str;
+
+	len = 10 + 10 + 1 + 10 + 1 +
+		strlen(nfs4_client_id_uniquifier) + 1 +
+		strlen(clp->cl_rpcclient->cl_nodename) + 1;
+
+	if (len > NFS4_OPAQUE_LIMIT + 1)
+		return -EINVAL;
+
+	/*
+	 * Since this string is allocated at mount time, and held until the
+	 * nfs_client is destroyed, we can use GFP_KERNEL here w/o worrying
+	 * about a memory-reclaim deadlock.
+	 */
+	str = kmalloc(len, GFP_KERNEL);
+	if (!str)
+		return -ENOMEM;
+
+	result = scnprintf(str, len, "Linux NFSv%u.%u %s/%s",
+			clp->rpc_ops->version, clp->cl_minorversion,
+			nfs4_client_id_uniquifier,
+			clp->cl_rpcclient->cl_nodename);
+	if (result >= len) {
+		kfree(str);
+		return -EINVAL;
+	}
+	clp->cl_owner_id = str;
+	return 0;
+}
+
+static int
+nfs4_init_uniform_client_string(struct nfs_client *clp)
 {
-	const char *nodename = clp->cl_rpcclient->cl_nodename;
-	unsigned int result;
+	int result;
+	size_t len;
+	char *str;
 
 	if (clp->cl_owner_id != NULL)
-		return strlcpy(buf, clp->cl_owner_id, len);
+		return 0;
 
 	if (nfs4_client_id_uniquifier[0] != '\0')
-		result = scnprintf(buf, len, "Linux NFSv%u.%u %s/%s",
-				clp->rpc_ops->version,
-				clp->cl_minorversion,
-				nfs4_client_id_uniquifier,
-				nodename);
-	else
-		result = scnprintf(buf, len, "Linux NFSv%u.%u %s",
-				clp->rpc_ops->version, clp->cl_minorversion,
-				nodename);
-	clp->cl_owner_id = kstrdup(buf, GFP_KERNEL);
-	return result;
+		return nfs4_init_uniquifier_client_string(clp);
+
+	len = 10 + 10 + 1 + 10 + 1 +
+		strlen(clp->cl_rpcclient->cl_nodename) + 1;
+
+	if (len > NFS4_OPAQUE_LIMIT + 1)
+		return -EINVAL;
+
+	/*
+	 * Since this string is allocated at mount time, and held until the
+	 * nfs_client is destroyed, we can use GFP_KERNEL here w/o worrying
+	 * about a memory-reclaim deadlock.
+	 */
+	str = kmalloc(len, GFP_KERNEL);
+	if (!str)
+		return -ENOMEM;
+
+	result = scnprintf(str, len, "Linux NFSv%u.%u %s",
+			clp->rpc_ops->version, clp->cl_minorversion,
+			clp->cl_rpcclient->cl_nodename);
+	if (result >= len) {
+		kfree(str);
+		return -EINVAL;
+	}
+	clp->cl_owner_id = str;
+	return 0;
 }
 
 /*
@@ -5044,7 +5127,7 @@ int nfs4_proc_setclientid(struct nfs_client *clp, u32 program,
 	struct nfs4_setclientid setclientid = {
 		.sc_verifier = &sc_verifier,
 		.sc_prog = program,
-		.sc_cb_ident = clp->cl_cb_ident,
+		.sc_clnt = clp,
 	};
 	struct rpc_message msg = {
 		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID],
@@ -5064,16 +5147,15 @@ int nfs4_proc_setclientid(struct nfs_client *clp, u32 program,
 
 	/* nfs_client_id4 */
 	nfs4_init_boot_verifier(clp, &sc_verifier);
+
 	if (test_bit(NFS_CS_MIGRATION, &clp->cl_flags))
-		setclientid.sc_name_len =
-				nfs4_init_uniform_client_string(clp,
-						setclientid.sc_name,
-						sizeof(setclientid.sc_name));
+		status = nfs4_init_uniform_client_string(clp);
 	else
-		setclientid.sc_name_len =
-				nfs4_init_nonuniform_client_string(clp,
-						setclientid.sc_name,
-						sizeof(setclientid.sc_name));
+		status = nfs4_init_nonuniform_client_string(clp);
+
+	if (status)
+		goto out;
+
 	/* cb_client4 */
 	setclientid.sc_netid_len =
 				nfs4_init_callback_netid(clp,
@@ -5083,9 +5165,9 @@ int nfs4_proc_setclientid(struct nfs_client *clp, u32 program,
 				sizeof(setclientid.sc_uaddr), "%s.%u.%u",
 				clp->cl_ipaddr, port >> 8, port & 255);
 
-	dprintk("NFS call  setclientid auth=%s, '%.*s'\n",
+	dprintk("NFS call  setclientid auth=%s, '%s'\n",
 		clp->cl_rpcclient->cl_auth->au_ops->au_name,
-		setclientid.sc_name_len, setclientid.sc_name);
+		clp->cl_owner_id);
 	task = rpc_run_task(&task_setup_data);
 	if (IS_ERR(task)) {
 		status = PTR_ERR(task);
@@ -5402,6 +5484,7 @@ static struct nfs4_unlockdata *nfs4_alloc_unlockdata(struct file_lock *fl,
 	atomic_inc(&lsp->ls_count);
 	/* Ensure we don't close file until we're done freeing locks! */
 	p->ctx = get_nfs_open_context(ctx);
+	get_file(fl->fl_file);
 	memcpy(&p->fl, fl, sizeof(p->fl));
 	p->server = NFS_SERVER(inode);
 	return p;
@@ -5413,6 +5496,7 @@ static void nfs4_locku_release_calldata(void *data)
 	nfs_free_seqid(calldata->arg.seqid);
 	nfs4_put_lock_state(calldata->lsp);
 	put_nfs_open_context(calldata->ctx);
+	fput(calldata->fl.fl_file);
 	kfree(calldata);
 }
 
@@ -6846,11 +6930,14 @@ static int _nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred,
 	};
 
 	nfs4_init_boot_verifier(clp, &verifier);
-	args.id_len = nfs4_init_uniform_client_string(clp, args.id,
-							sizeof(args.id));
-	dprintk("NFS call  exchange_id auth=%s, '%.*s'\n",
+
+	status = nfs4_init_uniform_client_string(clp);
+	if (status)
+		goto out;
+
+	dprintk("NFS call  exchange_id auth=%s, '%s'\n",
 		clp->cl_rpcclient->cl_auth->au_ops->au_name,
-		args.id_len, args.id);
+		clp->cl_owner_id);
 
 	res.server_owner = kzalloc(sizeof(struct nfs41_server_owner),
 					GFP_NOFS);
@@ -6885,7 +6972,7 @@ static int _nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred,
 		/* unsupported! */
 		WARN_ON_ONCE(1);
 		status = -EINVAL;
-		goto out_server_scope;
+		goto out_impl_id;
 	}
 
 	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
@@ -6913,6 +7000,7 @@ static int _nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred,
 		/* use the most recent implementation id */
 		kfree(clp->cl_implid);
 		clp->cl_implid = res.impl_id;
+		res.impl_id = NULL;
 
 		if (clp->cl_serverscope != NULL &&
 		    !nfs41_same_server_scope(clp->cl_serverscope,
@@ -6926,15 +7014,16 @@ static int _nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred,
 
 		if (clp->cl_serverscope == NULL) {
 			clp->cl_serverscope = res.server_scope;
-			goto out;
+			res.server_scope = NULL;
 		}
-	} else
-		kfree(res.impl_id);
+	}
 
-out_server_owner:
-	kfree(res.server_owner);
+out_impl_id:
+	kfree(res.impl_id);
 out_server_scope:
 	kfree(res.server_scope);
+out_server_owner:
+	kfree(res.server_owner);
 out:
 	if (clp->cl_implid != NULL)
 		dprintk("NFS reply exchange_id: Server Implementation ID: "
@@ -8061,9 +8150,8 @@ nfs4_proc_layoutcommit(struct nfs4_layoutcommit_data *data, bool sync)
 	struct rpc_task *task;
 	int status = 0;
 
-	dprintk("NFS: %4d initiating layoutcommit call. sync %d "
-		"lbw: %llu inode %lu\n",
-		data->task.tk_pid, sync,
+	dprintk("NFS: initiating layoutcommit call. sync %d "
+		"lbw: %llu inode %lu\n", sync,
 		data->args.lastbytewritten,
 		data->args.inode->i_ino);
 
@@ -8557,7 +8645,8 @@ static const struct nfs4_minor_version_ops nfs_v4_2_minor_ops = {
 		| NFS_CAP_ATOMIC_OPEN_V1
 		| NFS_CAP_ALLOCATE
 		| NFS_CAP_DEALLOCATE
-		| NFS_CAP_SEEK,
+		| NFS_CAP_SEEK
+		| NFS_CAP_LAYOUTSTATS,
 	.init_client = nfs41_init_client,
 	.shutdown_client = nfs41_shutdown_client,
 	.match_stateid = nfs41_match_stateid,
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index 2782cfca2265..605840dc89cf 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -309,7 +309,6 @@ int nfs41_init_clientid(struct nfs_client *clp, struct rpc_cred *cred)
 
 	if (test_bit(NFS4CLNT_LEASE_CONFIRM, &clp->cl_state))
 		goto do_confirm;
-	nfs4_begin_drain_session(clp);
 	status = nfs4_proc_exchange_id(clp, cred);
 	if (status != 0)
 		goto out;
@@ -1482,6 +1481,8 @@ restart:
 					spin_unlock(&state->state_lock);
 				}
 				nfs4_put_open_state(state);
+				clear_bit(NFS4CLNT_RECLAIM_NOGRACE,
+					&state->flags);
 				spin_lock(&sp->so_lock);
 				goto restart;
 			}
@@ -1830,6 +1831,7 @@ static int nfs4_establish_lease(struct nfs_client *clp)
 		clp->cl_mvops->reboot_recovery_ops;
 	int status;
 
+	nfs4_begin_drain_session(clp);
 	cred = nfs4_get_clid_cred(clp);
 	if (cred == NULL)
 		return -ENOENT;
diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
index 0aea97841d30..558cd65dbdb7 100644
--- a/fs/nfs/nfs4xdr.c
+++ b/fs/nfs/nfs4xdr.c
@@ -139,7 +139,8 @@ static int nfs4_stat_to_errno(int);
 #define encode_setclientid_maxsz \
 				(op_encode_hdr_maxsz + \
 				XDR_QUADLEN(NFS4_VERIFIER_SIZE) + \
-				XDR_QUADLEN(NFS4_SETCLIENTID_NAMELEN) + \
+				/* client name */ \
+				1 + XDR_QUADLEN(NFS4_OPAQUE_LIMIT) + \
 				1 /* sc_prog */ + \
 				1 + XDR_QUADLEN(RPCBIND_MAXNETIDLEN) + \
 				1 + XDR_QUADLEN(RPCBIND_MAXUADDRLEN) + \
@@ -288,7 +289,8 @@ static int nfs4_stat_to_errno(int);
 #define encode_exchange_id_maxsz (op_encode_hdr_maxsz + \
 				encode_verifier_maxsz + \
 				1 /* co_ownerid.len */ + \
-				XDR_QUADLEN(NFS4_EXCHANGE_ID_LEN) + \
+				/* eia_clientowner */ \
+				1 + XDR_QUADLEN(NFS4_OPAQUE_LIMIT) + \
 				1 /* flags */ + \
 				1 /* spa_how */ + \
 				/* max is SP4_MACH_CRED (for now) */ + \
@@ -1667,13 +1669,14 @@ static void encode_setclientid(struct xdr_stream *xdr, const struct nfs4_setclie
 	encode_op_hdr(xdr, OP_SETCLIENTID, decode_setclientid_maxsz, hdr);
 	encode_nfs4_verifier(xdr, setclientid->sc_verifier);
 
-	encode_string(xdr, setclientid->sc_name_len, setclientid->sc_name);
+	encode_string(xdr, strlen(setclientid->sc_clnt->cl_owner_id),
+			setclientid->sc_clnt->cl_owner_id);
 	p = reserve_space(xdr, 4);
 	*p = cpu_to_be32(setclientid->sc_prog);
 	encode_string(xdr, setclientid->sc_netid_len, setclientid->sc_netid);
 	encode_string(xdr, setclientid->sc_uaddr_len, setclientid->sc_uaddr);
 	p = reserve_space(xdr, 4);
-	*p = cpu_to_be32(setclientid->sc_cb_ident);
+	*p = cpu_to_be32(setclientid->sc_clnt->cl_cb_ident);
 }
 
 static void encode_setclientid_confirm(struct xdr_stream *xdr, const struct nfs4_setclientid_res *arg, struct compound_hdr *hdr)
@@ -1747,7 +1750,8 @@ static void encode_exchange_id(struct xdr_stream *xdr,
 	encode_op_hdr(xdr, OP_EXCHANGE_ID, decode_exchange_id_maxsz, hdr);
 	encode_nfs4_verifier(xdr, args->verifier);
 
-	encode_string(xdr, args->id_len, args->id);
+	encode_string(xdr, strlen(args->client->cl_owner_id),
+			args->client->cl_owner_id);
 
 	encode_uint32(xdr, args->flags);
 	encode_uint32(xdr, args->state_protect.how);
@@ -7427,6 +7431,7 @@ struct rpc_procinfo	nfs4_procedures[] = {
 	PROC(SEEK,		enc_seek,		dec_seek),
 	PROC(ALLOCATE,		enc_allocate,		dec_allocate),
 	PROC(DEALLOCATE,	enc_deallocate,		dec_deallocate),
+	PROC(LAYOUTSTATS,	enc_layoutstats,	dec_layoutstats),
 #endif /* CONFIG_NFS_V4_2 */
 };
 
diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c
index 282b39369510..1da68d3b1eda 100644
--- a/fs/nfs/pagelist.c
+++ b/fs/nfs/pagelist.c
@@ -636,9 +636,8 @@ int nfs_initiate_pgio(struct rpc_clnt *clnt, struct nfs_pgio_header *hdr,
 
 	hdr->rw_ops->rw_initiate(hdr, &msg, rpc_ops, &task_setup_data, how);
 
-	dprintk("NFS: %5u initiated pgio call "
+	dprintk("NFS: initiated pgio call "
 		"(req %s/%llu, %u bytes @ offset %llu)\n",
-		hdr->task.tk_pid,
 		hdr->inode->i_sb->s_id,
 		(unsigned long long)NFS_FILEID(hdr->inode),
 		hdr->args.count,
@@ -690,8 +689,6 @@ static int nfs_pgio_error(struct nfs_pageio_descriptor *desc,
 static void nfs_pgio_release(void *calldata)
 {
 	struct nfs_pgio_header *hdr = calldata;
-	if (hdr->rw_ops->rw_release)
-		hdr->rw_ops->rw_release(hdr);
 	nfs_pgio_data_destroy(hdr);
 	hdr->completion_ops->completion(hdr);
 }
@@ -711,7 +708,9 @@ static void nfs_pageio_mirror_init(struct nfs_pgio_mirror *mirror,
  * nfs_pageio_init - initialise a page io descriptor
  * @desc: pointer to descriptor
  * @inode: pointer to inode
- * @doio: pointer to io function
+ * @pg_ops: pointer to pageio operations
+ * @compl_ops: pointer to pageio completion operations
+ * @rw_ops: pointer to nfs read/write operations
  * @bsize: io block size
  * @io_flags: extra parameters for the io function
  */
@@ -1186,6 +1185,7 @@ int nfs_pageio_add_request(struct nfs_pageio_descriptor *desc,
  * nfs_pageio_complete_mirror - Complete I/O on the current mirror of an
  *				nfs_pageio_descriptor
  * @desc: pointer to io descriptor
+ * @mirror_idx: pointer to mirror index
  */
 static void nfs_pageio_complete_mirror(struct nfs_pageio_descriptor *desc,
 				       u32 mirror_idx)
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 230606243be6..0ba9a02c9566 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -35,6 +35,7 @@
 #include "iostat.h"
 #include "nfs4trace.h"
 #include "delegation.h"
+#include "nfs42.h"
 
 #define NFSDBG_FACILITY		NFSDBG_PNFS
 #define PNFS_LAYOUTGET_RETRY_TIMEOUT (120*HZ)
@@ -1821,6 +1822,7 @@ int pnfs_write_done_resend_to_mds(struct nfs_pgio_header *hdr)
 	/* Resend all requests through the MDS */
 	nfs_pageio_init_write(&pgio, hdr->inode, FLUSH_STABLE, true,
 			      hdr->completion_ops);
+	set_bit(NFS_CONTEXT_RESEND_WRITES, &hdr->args.context->flags);
 	return nfs_pageio_resend(&pgio, hdr);
 }
 EXPORT_SYMBOL_GPL(pnfs_write_done_resend_to_mds);
@@ -1865,6 +1867,7 @@ pnfs_write_through_mds(struct nfs_pageio_descriptor *desc,
 		mirror->pg_recoalesce = 1;
 	}
 	nfs_pgio_data_destroy(hdr);
+	hdr->release(hdr);
 }
 
 static enum pnfs_try_status
@@ -1979,6 +1982,7 @@ pnfs_read_through_mds(struct nfs_pageio_descriptor *desc,
 		mirror->pg_recoalesce = 1;
 	}
 	nfs_pgio_data_destroy(hdr);
+	hdr->release(hdr);
 }
 
 /*
@@ -2247,3 +2251,63 @@ struct nfs4_threshold *pnfs_mdsthreshold_alloc(void)
 	}
 	return thp;
 }
+
+#if IS_ENABLED(CONFIG_NFS_V4_2)
+int
+pnfs_report_layoutstat(struct inode *inode)
+{
+	struct pnfs_layoutdriver_type *ld = NFS_SERVER(inode)->pnfs_curr_ld;
+	struct nfs_server *server = NFS_SERVER(inode);
+	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs42_layoutstat_data *data;
+	struct pnfs_layout_hdr *hdr;
+	int status = 0;
+
+	if (!pnfs_enabled_sb(server) || !ld->prepare_layoutstats)
+		goto out;
+
+	if (!nfs_server_capable(inode, NFS_CAP_LAYOUTSTATS))
+		goto out;
+
+	if (test_and_set_bit(NFS_INO_LAYOUTSTATS, &nfsi->flags))
+		goto out;
+
+	spin_lock(&inode->i_lock);
+	if (!NFS_I(inode)->layout) {
+		spin_unlock(&inode->i_lock);
+		goto out;
+	}
+	hdr = NFS_I(inode)->layout;
+	pnfs_get_layout_hdr(hdr);
+	spin_unlock(&inode->i_lock);
+
+	data = kzalloc(sizeof(*data), GFP_KERNEL);
+	if (!data) {
+		status = -ENOMEM;
+		goto out_put;
+	}
+
+	data->args.fh = NFS_FH(inode);
+	data->args.inode = inode;
+	nfs4_stateid_copy(&data->args.stateid, &hdr->plh_stateid);
+	status = ld->prepare_layoutstats(&data->args);
+	if (status)
+		goto out_free;
+
+	status = nfs42_proc_layoutstats_generic(NFS_SERVER(inode), data);
+
+out:
+	dprintk("%s returns %d\n", __func__, status);
+	return status;
+
+out_free:
+	kfree(data);
+out_put:
+	pnfs_put_layout_hdr(hdr);
+	smp_mb__before_atomic();
+	clear_bit(NFS_INO_LAYOUTSTATS, &nfsi->flags);
+	smp_mb__after_atomic();
+	goto out;
+}
+EXPORT_SYMBOL_GPL(pnfs_report_layoutstat);
+#endif
diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h
index 1e6308f82fc3..3e6ab7bfbabd 100644
--- a/fs/nfs/pnfs.h
+++ b/fs/nfs/pnfs.h
@@ -178,6 +178,8 @@ struct pnfs_layoutdriver_type {
 	void (*encode_layoutcommit) (struct pnfs_layout_hdr *lo,
 				     struct xdr_stream *xdr,
 				     const struct nfs4_layoutcommit_args *args);
+	int (*prepare_layoutstats) (struct nfs42_layoutstat_args *args);
+	void (*cleanup_layoutstats) (struct nfs42_layoutstat_data *data);
 };
 
 struct pnfs_layout_hdr {
@@ -290,7 +292,6 @@ int pnfs_write_done_resend_to_mds(struct nfs_pgio_header *);
 struct nfs4_threshold *pnfs_mdsthreshold_alloc(void);
 void pnfs_error_mark_layout_for_return(struct inode *inode,
 				       struct pnfs_layout_segment *lseg);
-
 /* nfs4_deviceid_flags */
 enum {
 	NFS_DEVICEID_INVALID = 0,       /* set when MDS clientid recalled */
@@ -689,4 +690,14 @@ static inline void nfs4_pnfs_v3_ds_connect_unload(void)
 
 #endif /* CONFIG_NFS_V4_1 */
 
+#if IS_ENABLED(CONFIG_NFS_V4_2)
+int pnfs_report_layoutstat(struct inode *inode);
+#else
+static inline int
+pnfs_report_layoutstat(struct inode *inode)
+{
+	return 0;
+}
+#endif
+
 #endif /* FS_NFS_PNFS_H */
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index e6c262555e08..65869ca9c851 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -1290,6 +1290,7 @@ static void nfs_initiate_write(struct nfs_pgio_header *hdr,
 static void nfs_redirty_request(struct nfs_page *req)
 {
 	nfs_mark_request_dirty(req);
+	set_bit(NFS_CONTEXT_RESEND_WRITES, &req->wb_context->flags);
 	nfs_unlock_request(req);
 	nfs_end_page_writeback(req);
 	nfs_release_request(req);
@@ -1348,11 +1349,6 @@ void nfs_commit_prepare(struct rpc_task *task, void *calldata)
 	NFS_PROTO(data->inode)->commit_rpc_prepare(task, data);
 }
 
-static void nfs_writeback_release_common(struct nfs_pgio_header *hdr)
-{
-	/* do nothing! */
-}
-
 /*
  * Special version of should_remove_suid() that ignores capabilities.
  */
@@ -1556,7 +1552,7 @@ int nfs_initiate_commit(struct rpc_clnt *clnt, struct nfs_commit_data *data,
 	/* Set up the initial task struct.  */
 	nfs_ops->commit_setup(data, &msg);
 
-	dprintk("NFS: %5u initiated commit call\n", data->task.tk_pid);
+	dprintk("NFS: initiated commit call\n");
 
 	nfs4_state_protect(NFS_SERVER(data->inode)->nfs_client,
 		NFS_SP4_MACH_CRED_COMMIT, &task_setup_data.rpc_client, &msg);
@@ -2013,7 +2009,6 @@ static const struct nfs_rw_ops nfs_rw_write_ops = {
 	.rw_mode		= FMODE_WRITE,
 	.rw_alloc_header	= nfs_writehdr_alloc,
 	.rw_free_header		= nfs_writehdr_free,
-	.rw_release		= nfs_writeback_release_common,
 	.rw_done		= nfs_writeback_done,
 	.rw_result		= nfs_writeback_result,
 	.rw_initiate		= nfs_initiate_write,
diff --git a/fs/nilfs2/dir.c b/fs/nilfs2/dir.c
index 0ee0bed3649b..6b8b92b19cec 100644
--- a/fs/nilfs2/dir.c
+++ b/fs/nilfs2/dir.c
@@ -61,11 +61,6 @@ static inline void nilfs_put_page(struct page *page)
 	page_cache_release(page);
 }
 
-static inline unsigned long dir_pages(struct inode *inode)
-{
-	return (inode->i_size+PAGE_CACHE_SIZE-1)>>PAGE_CACHE_SHIFT;
-}
-
 /*
  * Return the offset into page `page_nr' of the last valid
  * byte in that page, plus one.
diff --git a/fs/nilfs2/inode.c b/fs/nilfs2/inode.c
index 258d9fe2521a..4a73d6dffabf 100644
--- a/fs/nilfs2/inode.c
+++ b/fs/nilfs2/inode.c
@@ -307,31 +307,13 @@ static int nilfs_write_end(struct file *file, struct address_space *mapping,
 static ssize_t
 nilfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter, loff_t offset)
 {
-	struct file *file = iocb->ki_filp;
-	struct address_space *mapping = file->f_mapping;
-	struct inode *inode = file->f_mapping->host;
-	size_t count = iov_iter_count(iter);
-	ssize_t size;
+	struct inode *inode = file_inode(iocb->ki_filp);
 
 	if (iov_iter_rw(iter) == WRITE)
 		return 0;
 
 	/* Needs synchronization with the cleaner */
-	size = blockdev_direct_IO(iocb, inode, iter, offset, nilfs_get_block);
-
-	/*
-	 * In case of error extending write may have instantiated a few
-	 * blocks outside i_size. Trim these off again.
-	 */
-	if (unlikely(iov_iter_rw(iter) == WRITE && size < 0)) {
-		loff_t isize = i_size_read(inode);
-		loff_t end = offset + count;
-
-		if (end > isize)
-			nilfs_write_failed(mapping, end);
-	}
-
-	return size;
+	return blockdev_direct_IO(iocb, inode, iter, offset, nilfs_get_block);
 }
 
 const struct address_space_operations nilfs_aops = {
diff --git a/fs/nilfs2/ioctl.c b/fs/nilfs2/ioctl.c
index 9a20e513d7eb..aba43811d6ef 100644
--- a/fs/nilfs2/ioctl.c
+++ b/fs/nilfs2/ioctl.c
@@ -1369,7 +1369,6 @@ long nilfs_compat_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 	case NILFS_IOCTL_SYNC:
 	case NILFS_IOCTL_RESIZE:
 	case NILFS_IOCTL_SET_ALLOC_RANGE:
-	case FITRIM:
 		break;
 	default:
 		return -ENOIOCTLCMD;
diff --git a/fs/ntfs/file.c b/fs/ntfs/file.c
index 2cd653670764..262561fea923 100644
--- a/fs/ntfs/file.c
+++ b/fs/ntfs/file.c
@@ -382,7 +382,7 @@ static ssize_t ntfs_prepare_file_for_write(struct kiocb *iocb,
 	base_ni = ni;
 	if (NInoAttr(ni))
 		base_ni = ni->ext.base_ntfs_ino;
-	err = file_remove_suid(file);
+	err = file_remove_privs(file);
 	if (unlikely(err))
 		goto out;
 	/*
diff --git a/fs/ntfs/inode.h b/fs/ntfs/inode.h
index 76b6cfb579d7..b3c3469de6cb 100644
--- a/fs/ntfs/inode.h
+++ b/fs/ntfs/inode.h
@@ -239,7 +239,7 @@ typedef struct {
  */
 static inline ntfs_inode *NTFS_I(struct inode *inode)
 {
-	return (ntfs_inode *)list_entry(inode, big_ntfs_inode, vfs_inode);
+	return (ntfs_inode *)container_of(inode, big_ntfs_inode, vfs_inode);
 }
 
 static inline struct inode *VFS_I(ntfs_inode *ni)
diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c
index 53e6c40ed4c6..3cb097ccce60 100644
--- a/fs/ocfs2/ioctl.c
+++ b/fs/ocfs2/ioctl.c
@@ -980,7 +980,6 @@ long ocfs2_compat_ioctl(struct file *file, unsigned cmd, unsigned long arg)
 	case OCFS2_IOC_GROUP_EXTEND:
 	case OCFS2_IOC_GROUP_ADD:
 	case OCFS2_IOC_GROUP_ADD64:
-	case FITRIM:
 		break;
 	case OCFS2_IOC_REFLINK:
 		if (copy_from_user(&args, argp, sizeof(args)))
diff --git a/fs/open.c b/fs/open.c
index e0250bdcc440..e33dab287fa0 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -51,8 +51,10 @@ int do_truncate(struct dentry *dentry, loff_t length, unsigned int time_attrs,
 		newattrs.ia_valid |= ATTR_FILE;
 	}
 
-	/* Remove suid/sgid on truncate too */
-	ret = should_remove_suid(dentry);
+	/* Remove suid, sgid, and file capabilities on truncate too */
+	ret = dentry_needs_remove_privs(dentry);
+	if (ret < 0)
+		return ret;
 	if (ret)
 		newattrs.ia_valid |= ret | ATTR_FORCE;
 
@@ -678,18 +680,18 @@ int open_check_o_direct(struct file *f)
 }
 
 static int do_dentry_open(struct file *f,
+			  struct inode *inode,
 			  int (*open)(struct inode *, struct file *),
 			  const struct cred *cred)
 {
 	static const struct file_operations empty_fops = {};
-	struct inode *inode;
 	int error;
 
 	f->f_mode = OPEN_FMODE(f->f_flags) | FMODE_LSEEK |
 				FMODE_PREAD | FMODE_PWRITE;
 
 	path_get(&f->f_path);
-	inode = f->f_inode = f->f_path.dentry->d_inode;
+	f->f_inode = inode;
 	f->f_mapping = inode->i_mapping;
 
 	if (unlikely(f->f_flags & O_PATH)) {
@@ -793,7 +795,8 @@ int finish_open(struct file *file, struct dentry *dentry,
 	BUG_ON(*opened & FILE_OPENED); /* once it's opened, it's opened */
 
 	file->f_path.dentry = dentry;
-	error = do_dentry_open(file, open, current_cred());
+	error = do_dentry_open(file, d_backing_inode(dentry), open,
+			       current_cred());
 	if (!error)
 		*opened |= FILE_OPENED;
 
@@ -822,6 +825,34 @@ int finish_no_open(struct file *file, struct dentry *dentry)
 }
 EXPORT_SYMBOL(finish_no_open);
 
+char *file_path(struct file *filp, char *buf, int buflen)
+{
+	return d_path(&filp->f_path, buf, buflen);
+}
+EXPORT_SYMBOL(file_path);
+
+/**
+ * vfs_open - open the file at the given path
+ * @path: path to open
+ * @file: newly allocated file with f_flag initialized
+ * @cred: credentials to use
+ */
+int vfs_open(const struct path *path, struct file *file,
+	     const struct cred *cred)
+{
+	struct dentry *dentry = path->dentry;
+	struct inode *inode = dentry->d_inode;
+
+	file->f_path = *path;
+	if (dentry->d_flags & DCACHE_OP_SELECT_INODE) {
+		inode = dentry->d_op->d_select_inode(dentry, file->f_flags);
+		if (IS_ERR(inode))
+			return PTR_ERR(inode);
+	}
+
+	return do_dentry_open(file, inode, NULL, cred);
+}
+
 struct file *dentry_open(const struct path *path, int flags,
 			 const struct cred *cred)
 {
@@ -853,26 +884,6 @@ struct file *dentry_open(const struct path *path, int flags,
 }
 EXPORT_SYMBOL(dentry_open);
 
-/**
- * vfs_open - open the file at the given path
- * @path: path to open
- * @filp: newly allocated file with f_flag initialized
- * @cred: credentials to use
- */
-int vfs_open(const struct path *path, struct file *filp,
-	     const struct cred *cred)
-{
-	struct inode *inode = path->dentry->d_inode;
-
-	if (inode->i_op->dentry_open)
-		return inode->i_op->dentry_open(path->dentry, filp, cred);
-	else {
-		filp->f_path = *path;
-		return do_dentry_open(filp, NULL, cred);
-	}
-}
-EXPORT_SYMBOL(vfs_open);
-
 static inline int build_open_flags(int flags, umode_t mode, struct open_flags *op)
 {
 	int lookup_flags = 0;
diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c
index 308379b2d0b2..d9da5a4e9382 100644
--- a/fs/overlayfs/inode.c
+++ b/fs/overlayfs/inode.c
@@ -337,37 +337,33 @@ static bool ovl_open_need_copy_up(int flags, enum ovl_path_type type,
 	return true;
 }
 
-static int ovl_dentry_open(struct dentry *dentry, struct file *file,
-		    const struct cred *cred)
+struct inode *ovl_d_select_inode(struct dentry *dentry, unsigned file_flags)
 {
 	int err;
 	struct path realpath;
 	enum ovl_path_type type;
-	bool want_write = false;
+
+	if (d_is_dir(dentry))
+		return d_backing_inode(dentry);
 
 	type = ovl_path_real(dentry, &realpath);
-	if (ovl_open_need_copy_up(file->f_flags, type, realpath.dentry)) {
-		want_write = true;
+	if (ovl_open_need_copy_up(file_flags, type, realpath.dentry)) {
 		err = ovl_want_write(dentry);
 		if (err)
-			goto out;
+			return ERR_PTR(err);
 
-		if (file->f_flags & O_TRUNC)
+		if (file_flags & O_TRUNC)
 			err = ovl_copy_up_last(dentry, NULL, true);
 		else
 			err = ovl_copy_up(dentry);
+		ovl_drop_write(dentry);
 		if (err)
-			goto out_drop_write;
+			return ERR_PTR(err);
 
 		ovl_path_upper(dentry, &realpath);
 	}
 
-	err = vfs_open(&realpath, file, cred);
-out_drop_write:
-	if (want_write)
-		ovl_drop_write(dentry);
-out:
-	return err;
+	return d_backing_inode(realpath.dentry);
 }
 
 static const struct inode_operations ovl_file_inode_operations = {
@@ -378,7 +374,6 @@ static const struct inode_operations ovl_file_inode_operations = {
 	.getxattr	= ovl_getxattr,
 	.listxattr	= ovl_listxattr,
 	.removexattr	= ovl_removexattr,
-	.dentry_open	= ovl_dentry_open,
 };
 
 static const struct inode_operations ovl_symlink_inode_operations = {
diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h
index 17ac5afc9ffb..ea5a40b06e3a 100644
--- a/fs/overlayfs/overlayfs.h
+++ b/fs/overlayfs/overlayfs.h
@@ -173,6 +173,7 @@ ssize_t ovl_getxattr(struct dentry *dentry, const char *name,
 		     void *value, size_t size);
 ssize_t ovl_listxattr(struct dentry *dentry, char *list, size_t size);
 int ovl_removexattr(struct dentry *dentry, const char *name);
+struct inode *ovl_d_select_inode(struct dentry *dentry, unsigned file_flags);
 
 struct inode *ovl_new_inode(struct super_block *sb, umode_t mode,
 			    struct ovl_entry *oe);
diff --git a/fs/overlayfs/readdir.c b/fs/overlayfs/readdir.c
index 907870e81a72..70e9af551600 100644
--- a/fs/overlayfs/readdir.c
+++ b/fs/overlayfs/readdir.c
@@ -23,6 +23,7 @@ struct ovl_cache_entry {
 	u64 ino;
 	struct list_head l_node;
 	struct rb_node node;
+	struct ovl_cache_entry *next_maybe_whiteout;
 	bool is_whiteout;
 	char name[];
 };
@@ -39,7 +40,7 @@ struct ovl_readdir_data {
 	struct rb_root root;
 	struct list_head *list;
 	struct list_head middle;
-	struct dentry *dir;
+	struct ovl_cache_entry *first_maybe_whiteout;
 	int count;
 	int err;
 };
@@ -79,7 +80,7 @@ static struct ovl_cache_entry *ovl_cache_entry_find(struct rb_root *root,
 	return NULL;
 }
 
-static struct ovl_cache_entry *ovl_cache_entry_new(struct dentry *dir,
+static struct ovl_cache_entry *ovl_cache_entry_new(struct ovl_readdir_data *rdd,
 						   const char *name, int len,
 						   u64 ino, unsigned int d_type)
 {
@@ -98,29 +99,8 @@ static struct ovl_cache_entry *ovl_cache_entry_new(struct dentry *dir,
 	p->is_whiteout = false;
 
 	if (d_type == DT_CHR) {
-		struct dentry *dentry;
-		const struct cred *old_cred;
-		struct cred *override_cred;
-
-		override_cred = prepare_creds();
-		if (!override_cred) {
-			kfree(p);
-			return NULL;
-		}
-
-		/*
-		 * CAP_DAC_OVERRIDE for lookup
-		 */
-		cap_raise(override_cred->cap_effective, CAP_DAC_OVERRIDE);
-		old_cred = override_creds(override_cred);
-
-		dentry = lookup_one_len(name, dir, len);
-		if (!IS_ERR(dentry)) {
-			p->is_whiteout = ovl_is_whiteout(dentry);
-			dput(dentry);
-		}
-		revert_creds(old_cred);
-		put_cred(override_cred);
+		p->next_maybe_whiteout = rdd->first_maybe_whiteout;
+		rdd->first_maybe_whiteout = p;
 	}
 	return p;
 }
@@ -148,7 +128,7 @@ static int ovl_cache_entry_add_rb(struct ovl_readdir_data *rdd,
 			return 0;
 	}
 
-	p = ovl_cache_entry_new(rdd->dir, name, len, ino, d_type);
+	p = ovl_cache_entry_new(rdd, name, len, ino, d_type);
 	if (p == NULL)
 		return -ENOMEM;
 
@@ -169,7 +149,7 @@ static int ovl_fill_lower(struct ovl_readdir_data *rdd,
 	if (p) {
 		list_move_tail(&p->l_node, &rdd->middle);
 	} else {
-		p = ovl_cache_entry_new(rdd->dir, name, namelen, ino, d_type);
+		p = ovl_cache_entry_new(rdd, name, namelen, ino, d_type);
 		if (p == NULL)
 			rdd->err = -ENOMEM;
 		else
@@ -219,6 +199,43 @@ static int ovl_fill_merge(struct dir_context *ctx, const char *name,
 		return ovl_fill_lower(rdd, name, namelen, offset, ino, d_type);
 }
 
+static int ovl_check_whiteouts(struct dentry *dir, struct ovl_readdir_data *rdd)
+{
+	int err;
+	struct ovl_cache_entry *p;
+	struct dentry *dentry;
+	const struct cred *old_cred;
+	struct cred *override_cred;
+
+	override_cred = prepare_creds();
+	if (!override_cred)
+		return -ENOMEM;
+
+	/*
+	 * CAP_DAC_OVERRIDE for lookup
+	 */
+	cap_raise(override_cred->cap_effective, CAP_DAC_OVERRIDE);
+	old_cred = override_creds(override_cred);
+
+	err = mutex_lock_killable(&dir->d_inode->i_mutex);
+	if (!err) {
+		while (rdd->first_maybe_whiteout) {
+			p = rdd->first_maybe_whiteout;
+			rdd->first_maybe_whiteout = p->next_maybe_whiteout;
+			dentry = lookup_one_len(p->name, dir, p->len);
+			if (!IS_ERR(dentry)) {
+				p->is_whiteout = ovl_is_whiteout(dentry);
+				dput(dentry);
+			}
+		}
+		mutex_unlock(&dir->d_inode->i_mutex);
+	}
+	revert_creds(old_cred);
+	put_cred(override_cred);
+
+	return err;
+}
+
 static inline int ovl_dir_read(struct path *realpath,
 			       struct ovl_readdir_data *rdd)
 {
@@ -229,7 +246,7 @@ static inline int ovl_dir_read(struct path *realpath,
 	if (IS_ERR(realfile))
 		return PTR_ERR(realfile);
 
-	rdd->dir = realpath->dentry;
+	rdd->first_maybe_whiteout = NULL;
 	rdd->ctx.pos = 0;
 	do {
 		rdd->count = 0;
@@ -238,6 +255,10 @@ static inline int ovl_dir_read(struct path *realpath,
 		if (err >= 0)
 			err = rdd->err;
 	} while (!err && rdd->count);
+
+	if (!err && rdd->first_maybe_whiteout)
+		err = ovl_check_whiteouts(realpath->dentry, rdd);
+
 	fput(realfile);
 
 	return err;
diff --git a/fs/overlayfs/super.c b/fs/overlayfs/super.c
index bf8537c7f455..7466ff339c66 100644
--- a/fs/overlayfs/super.c
+++ b/fs/overlayfs/super.c
@@ -273,8 +273,56 @@ static void ovl_dentry_release(struct dentry *dentry)
 	}
 }
 
+static int ovl_dentry_revalidate(struct dentry *dentry, unsigned int flags)
+{
+	struct ovl_entry *oe = dentry->d_fsdata;
+	unsigned int i;
+	int ret = 1;
+
+	for (i = 0; i < oe->numlower; i++) {
+		struct dentry *d = oe->lowerstack[i].dentry;
+
+		if (d->d_flags & DCACHE_OP_REVALIDATE) {
+			ret = d->d_op->d_revalidate(d, flags);
+			if (ret < 0)
+				return ret;
+			if (!ret) {
+				if (!(flags & LOOKUP_RCU))
+					d_invalidate(d);
+				return -ESTALE;
+			}
+		}
+	}
+	return 1;
+}
+
+static int ovl_dentry_weak_revalidate(struct dentry *dentry, unsigned int flags)
+{
+	struct ovl_entry *oe = dentry->d_fsdata;
+	unsigned int i;
+	int ret = 1;
+
+	for (i = 0; i < oe->numlower; i++) {
+		struct dentry *d = oe->lowerstack[i].dentry;
+
+		if (d->d_flags & DCACHE_OP_WEAK_REVALIDATE) {
+			ret = d->d_op->d_weak_revalidate(d, flags);
+			if (ret <= 0)
+				break;
+		}
+	}
+	return ret;
+}
+
 static const struct dentry_operations ovl_dentry_operations = {
 	.d_release = ovl_dentry_release,
+	.d_select_inode = ovl_d_select_inode,
+};
+
+static const struct dentry_operations ovl_reval_dentry_operations = {
+	.d_release = ovl_dentry_release,
+	.d_revalidate = ovl_dentry_revalidate,
+	.d_weak_revalidate = ovl_dentry_weak_revalidate,
 };
 
 static struct ovl_entry *ovl_alloc_entry(unsigned int numlower)
@@ -288,6 +336,20 @@ static struct ovl_entry *ovl_alloc_entry(unsigned int numlower)
 	return oe;
 }
 
+static bool ovl_dentry_remote(struct dentry *dentry)
+{
+	return dentry->d_flags &
+		(DCACHE_OP_REVALIDATE | DCACHE_OP_WEAK_REVALIDATE);
+}
+
+static bool ovl_dentry_weird(struct dentry *dentry)
+{
+	return dentry->d_flags & (DCACHE_NEED_AUTOMOUNT |
+				  DCACHE_MANAGE_TRANSIT |
+				  DCACHE_OP_HASH |
+				  DCACHE_OP_COMPARE);
+}
+
 static inline struct dentry *ovl_lookup_real(struct dentry *dir,
 					     struct qstr *name)
 {
@@ -303,6 +365,10 @@ static inline struct dentry *ovl_lookup_real(struct dentry *dir,
 	} else if (!dentry->d_inode) {
 		dput(dentry);
 		dentry = NULL;
+	} else if (ovl_dentry_weird(dentry)) {
+		dput(dentry);
+		/* Don't support traversing automounts and other weirdness */
+		dentry = ERR_PTR(-EREMOTE);
 	}
 	return dentry;
 }
@@ -350,6 +416,11 @@ struct dentry *ovl_lookup(struct inode *dir, struct dentry *dentry,
 			goto out;
 
 		if (this) {
+			if (unlikely(ovl_dentry_remote(this))) {
+				dput(this);
+				err = -EREMOTE;
+				goto out;
+			}
 			if (ovl_is_whiteout(this)) {
 				dput(this);
 				this = NULL;
@@ -694,25 +765,6 @@ static void ovl_unescape(char *s)
 	}
 }
 
-static bool ovl_is_allowed_fs_type(struct dentry *root)
-{
-	const struct dentry_operations *dop = root->d_op;
-
-	/*
-	 * We don't support:
-	 *  - automount filesystems
-	 *  - filesystems with revalidate (FIXME for lower layer)
-	 *  - filesystems with case insensitive names
-	 */
-	if (dop &&
-	    (dop->d_manage || dop->d_automount ||
-	     dop->d_revalidate || dop->d_weak_revalidate ||
-	     dop->d_compare || dop->d_hash)) {
-		return false;
-	}
-	return true;
-}
-
 static int ovl_mount_dir_noesc(const char *name, struct path *path)
 {
 	int err = -EINVAL;
@@ -727,7 +779,7 @@ static int ovl_mount_dir_noesc(const char *name, struct path *path)
 		goto out;
 	}
 	err = -EINVAL;
-	if (!ovl_is_allowed_fs_type(path->dentry)) {
+	if (ovl_dentry_weird(path->dentry)) {
 		pr_err("overlayfs: filesystem on '%s' not supported\n", name);
 		goto out_put;
 	}
@@ -751,13 +803,21 @@ static int ovl_mount_dir(const char *name, struct path *path)
 	if (tmp) {
 		ovl_unescape(tmp);
 		err = ovl_mount_dir_noesc(tmp, path);
+
+		if (!err)
+			if (ovl_dentry_remote(path->dentry)) {
+				pr_err("overlayfs: filesystem on '%s' not supported as upperdir\n",
+				       tmp);
+				path_put(path);
+				err = -EINVAL;
+			}
 		kfree(tmp);
 	}
 	return err;
 }
 
 static int ovl_lower_dir(const char *name, struct path *path, long *namelen,
-			 int *stack_depth)
+			 int *stack_depth, bool *remote)
 {
 	int err;
 	struct kstatfs statfs;
@@ -774,6 +834,9 @@ static int ovl_lower_dir(const char *name, struct path *path, long *namelen,
 	*namelen = max(*namelen, statfs.f_namelen);
 	*stack_depth = max(*stack_depth, path->mnt->mnt_sb->s_stack_depth);
 
+	if (ovl_dentry_remote(path->dentry))
+		*remote = true;
+
 	return 0;
 
 out_put:
@@ -827,6 +890,7 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	unsigned int numlower;
 	unsigned int stacklen = 0;
 	unsigned int i;
+	bool remote = false;
 	int err;
 
 	err = -ENOMEM;
@@ -900,7 +964,8 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	lower = lowertmp;
 	for (numlower = 0; numlower < stacklen; numlower++) {
 		err = ovl_lower_dir(lower, &stack[numlower],
-				    &ufs->lower_namelen, &sb->s_stack_depth);
+				    &ufs->lower_namelen, &sb->s_stack_depth,
+				    &remote);
 		if (err)
 			goto out_put_lowerpath;
 
@@ -958,7 +1023,10 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent)
 	if (!ufs->upper_mnt)
 		sb->s_flags |= MS_RDONLY;
 
-	sb->s_d_op = &ovl_dentry_operations;
+	if (remote)
+		sb->s_d_op = &ovl_reval_dentry_operations;
+	else
+		sb->s_d_op = &ovl_dentry_operations;
 
 	err = -ENOMEM;
 	oe = ovl_alloc_entry(numlower);
diff --git a/fs/posix_acl.c b/fs/posix_acl.c
index 84bb65b83570..4fb17ded7d47 100644
--- a/fs/posix_acl.c
+++ b/fs/posix_acl.c
@@ -547,51 +547,45 @@ posix_acl_create(struct inode *dir, umode_t *mode,
 		struct posix_acl **default_acl, struct posix_acl **acl)
 {
 	struct posix_acl *p;
+	struct posix_acl *clone;
 	int ret;
 
+	*acl = NULL;
+	*default_acl = NULL;
+
 	if (S_ISLNK(*mode) || !IS_POSIXACL(dir))
-		goto no_acl;
+		return 0;
 
 	p = get_acl(dir, ACL_TYPE_DEFAULT);
-	if (IS_ERR(p)) {
-		if (p == ERR_PTR(-EOPNOTSUPP))
-			goto apply_umask;
-		return PTR_ERR(p);
+	if (!p || p == ERR_PTR(-EOPNOTSUPP)) {
+		*mode &= ~current_umask();
+		return 0;
 	}
+	if (IS_ERR(p))
+		return PTR_ERR(p);
 
-	if (!p)
-		goto apply_umask;
-
-	*acl = posix_acl_clone(p, GFP_NOFS);
-	if (!*acl)
+	clone = posix_acl_clone(p, GFP_NOFS);
+	if (!clone)
 		goto no_mem;
 
-	ret = posix_acl_create_masq(*acl, mode);
+	ret = posix_acl_create_masq(clone, mode);
 	if (ret < 0)
 		goto no_mem_clone;
 
-	if (ret == 0) {
-		posix_acl_release(*acl);
-		*acl = NULL;
-	}
+	if (ret == 0)
+		posix_acl_release(clone);
+	else
+		*acl = clone;
 
-	if (!S_ISDIR(*mode)) {
+	if (!S_ISDIR(*mode))
 		posix_acl_release(p);
-		*default_acl = NULL;
-	} else {
+	else
 		*default_acl = p;
-	}
-	return 0;
 
-apply_umask:
-	*mode &= ~current_umask();
-no_acl:
-	*default_acl = NULL;
-	*acl = NULL;
 	return 0;
 
 no_mem_clone:
-	posix_acl_release(*acl);
+	posix_acl_release(clone);
 no_mem:
 	posix_acl_release(p);
 	return -ENOMEM;
diff --git a/fs/proc/base.c b/fs/proc/base.c
index 1d540b3f226f..87782e874b6a 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -491,14 +491,17 @@ static int proc_pid_stack(struct seq_file *m, struct pid_namespace *ns,
 }
 #endif
 
-#ifdef CONFIG_SCHEDSTATS
+#ifdef CONFIG_SCHED_INFO
 /*
  * Provides /proc/PID/schedstat
  */
 static int proc_pid_schedstat(struct seq_file *m, struct pid_namespace *ns,
 			      struct pid *pid, struct task_struct *task)
 {
-	seq_printf(m, "%llu %llu %lu\n",
+	if (unlikely(!sched_info_on()))
+		seq_printf(m, "0 0 0\n");
+	else
+		seq_printf(m, "%llu %llu %lu\n",
 		   (unsigned long long)task->se.sum_exec_runtime,
 		   (unsigned long long)task->sched_info.run_delay,
 		   task->sched_info.pcount);
@@ -2787,7 +2790,7 @@ static const struct pid_entry tgid_base_stuff[] = {
 #ifdef CONFIG_STACKTRACE
 	ONE("stack",      S_IRUSR, proc_pid_stack),
 #endif
-#ifdef CONFIG_SCHEDSTATS
+#ifdef CONFIG_SCHED_INFO
 	ONE("schedstat",  S_IRUGO, proc_pid_schedstat),
 #endif
 #ifdef CONFIG_LATENCYTOP
@@ -3135,7 +3138,7 @@ static const struct pid_entry tid_base_stuff[] = {
 #ifdef CONFIG_STACKTRACE
 	ONE("stack",      S_IRUSR, proc_pid_stack),
 #endif
-#ifdef CONFIG_SCHEDSTATS
+#ifdef CONFIG_SCHED_INFO
 	ONE("schedstat", S_IRUGO, proc_pid_schedstat),
 #endif
 #ifdef CONFIG_LATENCYTOP
diff --git a/fs/proc/generic.c b/fs/proc/generic.c
index df6327a2b865..e5dee5c3188e 100644
--- a/fs/proc/generic.c
+++ b/fs/proc/generic.c
@@ -373,6 +373,10 @@ static struct proc_dir_entry *__proc_create(struct proc_dir_entry **parent,
 		WARN(1, "create '/proc/%s' by hand\n", qstr.name);
 		return NULL;
 	}
+	if (is_empty_pde(*parent)) {
+		WARN(1, "attempt to add to permanently empty directory");
+		return NULL;
+	}
 
 	ent = kzalloc(sizeof(struct proc_dir_entry) + qstr.len + 1, GFP_KERNEL);
 	if (!ent)
@@ -455,6 +459,25 @@ struct proc_dir_entry *proc_mkdir(const char *name,
 }
 EXPORT_SYMBOL(proc_mkdir);
 
+struct proc_dir_entry *proc_create_mount_point(const char *name)
+{
+	umode_t mode = S_IFDIR | S_IRUGO | S_IXUGO;
+	struct proc_dir_entry *ent, *parent = NULL;
+
+	ent = __proc_create(&parent, name, mode, 2);
+	if (ent) {
+		ent->data = NULL;
+		ent->proc_fops = NULL;
+		ent->proc_iops = NULL;
+		if (proc_register(parent, ent) < 0) {
+			kfree(ent);
+			parent->nlink--;
+			ent = NULL;
+		}
+	}
+	return ent;
+}
+
 struct proc_dir_entry *proc_create_data(const char *name, umode_t mode,
 					struct proc_dir_entry *parent,
 					const struct file_operations *proc_fops,
diff --git a/fs/proc/inode.c b/fs/proc/inode.c
index afe232b9df6e..bd95b9fdebb0 100644
--- a/fs/proc/inode.c
+++ b/fs/proc/inode.c
@@ -422,6 +422,10 @@ struct inode *proc_get_inode(struct super_block *sb, struct proc_dir_entry *de)
 		inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
 		PROC_I(inode)->pde = de;
 
+		if (is_empty_pde(de)) {
+			make_empty_dir_inode(inode);
+			return inode;
+		}
 		if (de->mode) {
 			inode->i_mode = de->mode;
 			inode->i_uid = de->uid;
diff --git a/fs/proc/internal.h b/fs/proc/internal.h
index c835b94c0cd3..aa2781095bd1 100644
--- a/fs/proc/internal.h
+++ b/fs/proc/internal.h
@@ -191,6 +191,12 @@ static inline struct proc_dir_entry *pde_get(struct proc_dir_entry *pde)
 }
 extern void pde_put(struct proc_dir_entry *);
 
+static inline bool is_empty_pde(const struct proc_dir_entry *pde)
+{
+	return S_ISDIR(pde->mode) && !pde->proc_iops;
+}
+struct proc_dir_entry *proc_create_mount_point(const char *name);
+
 /*
  * inode.c
  */
diff --git a/fs/proc/nommu.c b/fs/proc/nommu.c
index d4a35746cab9..f8595e8b5cd0 100644
--- a/fs/proc/nommu.c
+++ b/fs/proc/nommu.c
@@ -64,7 +64,7 @@ static int nommu_region_show(struct seq_file *m, struct vm_region *region)
 
 	if (file) {
 		seq_pad(m, ' ');
-		seq_path(m, &file->f_path, "");
+		seq_file_path(m, file, "");
 	}
 
 	seq_putc(m, '\n');
diff --git a/fs/proc/proc_sysctl.c b/fs/proc/proc_sysctl.c
index fea2561d773b..fdda62e6115e 100644
--- a/fs/proc/proc_sysctl.c
+++ b/fs/proc/proc_sysctl.c
@@ -19,6 +19,28 @@ static const struct inode_operations proc_sys_inode_operations;
 static const struct file_operations proc_sys_dir_file_operations;
 static const struct inode_operations proc_sys_dir_operations;
 
+/* Support for permanently empty directories */
+
+struct ctl_table sysctl_mount_point[] = {
+	{ }
+};
+
+static bool is_empty_dir(struct ctl_table_header *head)
+{
+	return head->ctl_table[0].child == sysctl_mount_point;
+}
+
+static void set_empty_dir(struct ctl_dir *dir)
+{
+	dir->header.ctl_table[0].child = sysctl_mount_point;
+}
+
+static void clear_empty_dir(struct ctl_dir *dir)
+
+{
+	dir->header.ctl_table[0].child = NULL;
+}
+
 void proc_sys_poll_notify(struct ctl_table_poll *poll)
 {
 	if (!poll)
@@ -187,6 +209,17 @@ static int insert_header(struct ctl_dir *dir, struct ctl_table_header *header)
 	struct ctl_table *entry;
 	int err;
 
+	/* Is this a permanently empty directory? */
+	if (is_empty_dir(&dir->header))
+		return -EROFS;
+
+	/* Am I creating a permanently empty directory? */
+	if (header->ctl_table == sysctl_mount_point) {
+		if (!RB_EMPTY_ROOT(&dir->root))
+			return -EINVAL;
+		set_empty_dir(dir);
+	}
+
 	dir->header.nreg++;
 	header->parent = dir;
 	err = insert_links(header);
@@ -202,6 +235,8 @@ fail:
 	erase_header(header);
 	put_links(header);
 fail_links:
+	if (header->ctl_table == sysctl_mount_point)
+		clear_empty_dir(dir);
 	header->parent = NULL;
 	drop_sysctl_table(&dir->header);
 	return err;
@@ -419,6 +454,8 @@ static struct inode *proc_sys_make_inode(struct super_block *sb,
 		inode->i_mode |= S_IFDIR;
 		inode->i_op = &proc_sys_dir_operations;
 		inode->i_fop = &proc_sys_dir_file_operations;
+		if (is_empty_dir(head))
+			make_empty_dir_inode(inode);
 	}
 out:
 	return inode;
diff --git a/fs/proc/root.c b/fs/proc/root.c
index b7fa4bfe896a..68feb0f70e63 100644
--- a/fs/proc/root.c
+++ b/fs/proc/root.c
@@ -112,9 +112,6 @@ static struct dentry *proc_mount(struct file_system_type *fs_type,
 		ns = task_active_pid_ns(current);
 		options = data;
 
-		if (!capable(CAP_SYS_ADMIN) && !fs_fully_visible(fs_type))
-			return ERR_PTR(-EPERM);
-
 		/* Does the mounter have privilege over the pid namespace? */
 		if (!ns_capable(ns->user_ns, CAP_SYS_ADMIN))
 			return ERR_PTR(-EPERM);
@@ -159,7 +156,7 @@ static struct file_system_type proc_fs_type = {
 	.name		= "proc",
 	.mount		= proc_mount,
 	.kill_sb	= proc_kill_sb,
-	.fs_flags	= FS_USERNS_MOUNT,
+	.fs_flags	= FS_USERNS_VISIBLE | FS_USERNS_MOUNT,
 };
 
 void __init proc_root_init(void)
@@ -182,10 +179,10 @@ void __init proc_root_init(void)
 #endif
 	proc_mkdir("fs", NULL);
 	proc_mkdir("driver", NULL);
-	proc_mkdir("fs/nfsd", NULL); /* somewhere for the nfsd filesystem to be mounted */
+	proc_create_mount_point("fs/nfsd"); /* somewhere for the nfsd filesystem to be mounted */
 #if defined(CONFIG_SUN_OPENPROMFS) || defined(CONFIG_SUN_OPENPROMFS_MODULE)
 	/* just give it a mountpoint */
-	proc_mkdir("openprom", NULL);
+	proc_create_mount_point("openprom");
 #endif
 	proc_tty_init();
 	proc_mkdir("bus", NULL);
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 6dee68d013ff..ca1e091881d4 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -310,7 +310,7 @@ show_map_vma(struct seq_file *m, struct vm_area_struct *vma, int is_pid)
 	 */
 	if (file) {
 		seq_pad(m, ' ');
-		seq_path(m, &file->f_path, "\n");
+		seq_file_path(m, file, "\n");
 		goto done;
 	}
 
@@ -1509,7 +1509,7 @@ static int show_numa_map(struct seq_file *m, void *v, int is_pid)
 
 	if (file) {
 		seq_puts(m, " file=");
-		seq_path(m, &file->f_path, "\n\t= ");
+		seq_file_path(m, file, "\n\t= ");
 	} else if (vma->vm_start <= mm->brk && vma->vm_end >= mm->start_brk) {
 		seq_puts(m, " heap");
 	} else {
diff --git a/fs/proc/task_nommu.c b/fs/proc/task_nommu.c
index 599ec2e20104..e0d64c92e4f6 100644
--- a/fs/proc/task_nommu.c
+++ b/fs/proc/task_nommu.c
@@ -180,7 +180,7 @@ static int nommu_vma_show(struct seq_file *m, struct vm_area_struct *vma,
 
 	if (file) {
 		seq_pad(m, ' ');
-		seq_path(m, &file->f_path, "");
+		seq_file_path(m, file, "");
 	} else if (mm) {
 		pid_t tid = pid_of_stack(priv, vma, is_pid);
 
diff --git a/fs/pstore/inode.c b/fs/pstore/inode.c
index dc43b5f29305..3adcc4669fac 100644
--- a/fs/pstore/inode.c
+++ b/fs/pstore/inode.c
@@ -461,22 +461,18 @@ static struct file_system_type pstore_fs_type = {
 	.kill_sb	= pstore_kill_sb,
 };
 
-static struct kobject *pstore_kobj;
-
 static int __init init_pstore_fs(void)
 {
-	int err = 0;
+	int err;
 
 	/* Create a convenient mount point for people to access pstore */
-	pstore_kobj = kobject_create_and_add("pstore", fs_kobj);
-	if (!pstore_kobj) {
-		err = -ENOMEM;
+	err = sysfs_create_mount_point(fs_kobj, "pstore");
+	if (err)
 		goto out;
-	}
 
 	err = register_filesystem(&pstore_fs_type);
 	if (err < 0)
-		kobject_put(pstore_kobj);
+		sysfs_remove_mount_point(fs_kobj, "pstore");
 
 out:
 	return err;
diff --git a/fs/qnx6/dir.c b/fs/qnx6/dir.c
index 8d64bb5366bf..e1f37278cf97 100644
--- a/fs/qnx6/dir.c
+++ b/fs/qnx6/dir.c
@@ -32,11 +32,6 @@ static struct page *qnx6_get_page(struct inode *dir, unsigned long n)
 	return page;
 }
 
-static inline unsigned long dir_pages(struct inode *inode)
-{
-	return (inode->i_size+PAGE_CACHE_SIZE-1)>>PAGE_CACHE_SHIFT;
-}
-
 static unsigned last_entry(struct inode *inode, unsigned long page_nr)
 {
 	unsigned long last_byte = inode->i_size;
diff --git a/fs/seq_file.c b/fs/seq_file.c
index 1d9c1cbd4d0b..ce9e39fd5daf 100644
--- a/fs/seq_file.c
+++ b/fs/seq_file.c
@@ -490,6 +490,20 @@ int seq_path(struct seq_file *m, const struct path *path, const char *esc)
 }
 EXPORT_SYMBOL(seq_path);
 
+/**
+ * seq_file_path - seq_file interface to print a pathname of a file
+ * @m: the seq_file handle
+ * @file: the struct file to print
+ * @esc: set of characters to escape in the output
+ *
+ * return the absolute path to the file.
+ */
+int seq_file_path(struct seq_file *m, struct file *file, const char *esc)
+{
+	return seq_path(m, &file->f_path, esc);
+}
+EXPORT_SYMBOL(seq_file_path);
+
 /*
  * Same as seq_path, but relative to supplied root.
  */
diff --git a/fs/squashfs/squashfs_fs_i.h b/fs/squashfs/squashfs_fs_i.h
index 73588e7700ed..d09fcd6fb85d 100644
--- a/fs/squashfs/squashfs_fs_i.h
+++ b/fs/squashfs/squashfs_fs_i.h
@@ -49,6 +49,6 @@ struct squashfs_inode_info {
 
 static inline struct squashfs_inode_info *squashfs_i(struct inode *inode)
 {
-	return list_entry(inode, struct squashfs_inode_info, vfs_inode);
+	return container_of(inode, struct squashfs_inode_info, vfs_inode);
 }
 #endif
diff --git a/fs/super.c b/fs/super.c
index 928c20f47af9..b61372354f2b 100644
--- a/fs/super.c
+++ b/fs/super.c
@@ -842,7 +842,7 @@ int get_anon_bdev(dev_t *p)
 	else if (error)
 		return -EAGAIN;
 
-	if (dev == (1 << MINORBITS)) {
+	if (dev >= (1 << MINORBITS)) {
 		spin_lock(&unnamed_dev_lock);
 		ida_remove(&unnamed_dev_ida, dev);
 		if (unnamed_dev_start > dev)
diff --git a/fs/sysfs/dir.c b/fs/sysfs/dir.c
index 0b45ff42f374..94374e435025 100644
--- a/fs/sysfs/dir.c
+++ b/fs/sysfs/dir.c
@@ -121,3 +121,37 @@ int sysfs_move_dir_ns(struct kobject *kobj, struct kobject *new_parent_kobj,
 
 	return kernfs_rename_ns(kn, new_parent, kn->name, new_ns);
 }
+
+/**
+ * sysfs_create_mount_point - create an always empty directory
+ * @parent_kobj:  kobject that will contain this always empty directory
+ * @name: The name of the always empty directory to add
+ */
+int sysfs_create_mount_point(struct kobject *parent_kobj, const char *name)
+{
+	struct kernfs_node *kn, *parent = parent_kobj->sd;
+
+	kn = kernfs_create_empty_dir(parent, name);
+	if (IS_ERR(kn)) {
+		if (PTR_ERR(kn) == -EEXIST)
+			sysfs_warn_dup(parent, name);
+		return PTR_ERR(kn);
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(sysfs_create_mount_point);
+
+/**
+ *	sysfs_remove_mount_point - remove an always empty directory.
+ *	@parent_kobj: kobject that will contain this always empty directory
+ *	@name: The name of the always empty directory to remove
+ *
+ */
+void sysfs_remove_mount_point(struct kobject *parent_kobj, const char *name)
+{
+	struct kernfs_node *parent = parent_kobj->sd;
+
+	kernfs_remove_by_name_ns(parent, name, NULL);
+}
+EXPORT_SYMBOL_GPL(sysfs_remove_mount_point);
diff --git a/fs/sysfs/mount.c b/fs/sysfs/mount.c
index 8a49486bf30c..1c6ac6fcee9f 100644
--- a/fs/sysfs/mount.c
+++ b/fs/sysfs/mount.c
@@ -31,9 +31,6 @@ static struct dentry *sysfs_mount(struct file_system_type *fs_type,
 	bool new_sb;
 
 	if (!(flags & MS_KERNMOUNT)) {
-		if (!capable(CAP_SYS_ADMIN) && !fs_fully_visible(fs_type))
-			return ERR_PTR(-EPERM);
-
 		if (!kobj_ns_current_may_mount(KOBJ_NS_TYPE_NET))
 			return ERR_PTR(-EPERM);
 	}
@@ -58,7 +55,7 @@ static struct file_system_type sysfs_fs_type = {
 	.name		= "sysfs",
 	.mount		= sysfs_mount,
 	.kill_sb	= sysfs_kill_sb,
-	.fs_flags	= FS_USERNS_MOUNT,
+	.fs_flags	= FS_USERNS_VISIBLE | FS_USERNS_MOUNT,
 };
 
 int __init sysfs_init(void)
diff --git a/fs/sysv/dir.c b/fs/sysv/dir.c
index 8f3555f00c54..63c1bcb224ee 100644
--- a/fs/sysv/dir.c
+++ b/fs/sysv/dir.c
@@ -33,11 +33,6 @@ static inline void dir_put_page(struct page *page)
 	page_cache_release(page);
 }
 
-static inline unsigned long dir_pages(struct inode *inode)
-{
-	return (inode->i_size+PAGE_CACHE_SIZE-1)>>PAGE_CACHE_SHIFT;
-}
-
 static int dir_commit_chunk(struct page *page, loff_t pos, unsigned len)
 {
 	struct address_space *mapping = page->mapping;
diff --git a/fs/sysv/sysv.h b/fs/sysv/sysv.h
index 2c13525131cd..6c212288adcb 100644
--- a/fs/sysv/sysv.h
+++ b/fs/sysv/sysv.h
@@ -73,7 +73,7 @@ struct sysv_inode_info {
 
 static inline struct sysv_inode_info *SYSV_I(struct inode *inode)
 {
-	return list_entry(inode, struct sysv_inode_info, vfs_inode);
+	return container_of(inode, struct sysv_inode_info, vfs_inode);
 }
 
 static inline struct sysv_sb_info *SYSV_SB(struct super_block *sb)
diff --git a/fs/tracefs/inode.c b/fs/tracefs/inode.c
index d92bdf3b079a..cbc8d5d2755a 100644
--- a/fs/tracefs/inode.c
+++ b/fs/tracefs/inode.c
@@ -496,16 +496,11 @@ struct dentry *tracefs_create_instance_dir(const char *name, struct dentry *pare
 	return dentry;
 }
 
-static inline int tracefs_positive(struct dentry *dentry)
-{
-	return dentry->d_inode && !d_unhashed(dentry);
-}
-
 static int __tracefs_remove(struct dentry *dentry, struct dentry *parent)
 {
 	int ret = 0;
 
-	if (tracefs_positive(dentry)) {
+	if (simple_positive(dentry)) {
 		if (dentry->d_inode) {
 			dget(dentry);
 			switch (dentry->d_inode->i_mode & S_IFMT) {
@@ -582,7 +577,7 @@ void tracefs_remove_recursive(struct dentry *dentry)
 	 */
 	spin_lock(&parent->d_lock);
 	list_for_each_entry(child, &parent->d_subdirs, d_child) {
-		if (!tracefs_positive(child))
+		if (!simple_positive(child))
 			continue;
 
 		/* perhaps simple_empty(child) makes more sense */
@@ -603,7 +598,7 @@ void tracefs_remove_recursive(struct dentry *dentry)
 		 * from d_subdirs. When releasing the parent->d_lock we can
 		 * no longer trust that the next pointer is valid.
 		 * Restart the loop. We'll skip this one with the
-		 * tracefs_positive() check.
+		 * simple_positive() check.
 		 */
 		goto loop;
 	}
@@ -631,14 +626,12 @@ bool tracefs_initialized(void)
 	return tracefs_registered;
 }
 
-static struct kobject *trace_kobj;
-
 static int __init tracefs_init(void)
 {
 	int retval;
 
-	trace_kobj = kobject_create_and_add("tracing", kernel_kobj);
-	if (!trace_kobj)
+	retval = sysfs_create_mount_point(kernel_kobj, "tracing");
+	if (retval)
 		return -EINVAL;
 
 	retval = register_filesystem(&trace_fs_type);
diff --git a/fs/udf/udf_i.h b/fs/udf/udf_i.h
index b5cd8ed2aa12..b1b9a63d8cf3 100644
--- a/fs/udf/udf_i.h
+++ b/fs/udf/udf_i.h
@@ -56,7 +56,7 @@ struct udf_inode_info {
 
 static inline struct udf_inode_info *UDF_I(struct inode *inode)
 {
-	return list_entry(inode, struct udf_inode_info, vfs_inode);
+	return container_of(inode, struct udf_inode_info, vfs_inode);
 }
 
 #endif /* _UDF_I_H) */
diff --git a/fs/ufs/balloc.c b/fs/ufs/balloc.c
index 2c1036080d52..a7106eda5024 100644
--- a/fs/ufs/balloc.c
+++ b/fs/ufs/balloc.c
@@ -51,8 +51,8 @@ void ufs_free_fragments(struct inode *inode, u64 fragment, unsigned count)
 	
 	if (ufs_fragnum(fragment) + count > uspi->s_fpg)
 		ufs_error (sb, "ufs_free_fragments", "internal error");
-	
-	lock_ufs(sb);
+
+	mutex_lock(&UFS_SB(sb)->s_lock);
 	
 	cgno = ufs_dtog(uspi, fragment);
 	bit = ufs_dtogd(uspi, fragment);
@@ -115,13 +115,13 @@ void ufs_free_fragments(struct inode *inode, u64 fragment, unsigned count)
 	if (sb->s_flags & MS_SYNCHRONOUS)
 		ubh_sync_block(UCPI_UBH(ucpi));
 	ufs_mark_sb_dirty(sb);
-	
-	unlock_ufs(sb);
+
+	mutex_unlock(&UFS_SB(sb)->s_lock);
 	UFSD("EXIT\n");
 	return;
 
 failed:
-	unlock_ufs(sb);
+	mutex_unlock(&UFS_SB(sb)->s_lock);
 	UFSD("EXIT (FAILED)\n");
 	return;
 }
@@ -151,7 +151,7 @@ void ufs_free_blocks(struct inode *inode, u64 fragment, unsigned count)
 		goto failed;
 	}
 
-	lock_ufs(sb);
+	mutex_lock(&UFS_SB(sb)->s_lock);
 	
 do_more:
 	overflow = 0;
@@ -211,12 +211,12 @@ do_more:
 	}
 
 	ufs_mark_sb_dirty(sb);
-	unlock_ufs(sb);
+	mutex_unlock(&UFS_SB(sb)->s_lock);
 	UFSD("EXIT\n");
 	return;
 
 failed_unlock:
-	unlock_ufs(sb);
+	mutex_unlock(&UFS_SB(sb)->s_lock);
 failed:
 	UFSD("EXIT (FAILED)\n");
 	return;
@@ -357,7 +357,7 @@ u64 ufs_new_fragments(struct inode *inode, void *p, u64 fragment,
 	usb1 = ubh_get_usb_first(uspi);
 	*err = -ENOSPC;
 
-	lock_ufs(sb);
+	mutex_lock(&UFS_SB(sb)->s_lock);
 	tmp = ufs_data_ptr_to_cpu(sb, p);
 
 	if (count + ufs_fragnum(fragment) > uspi->s_fpb) {
@@ -378,19 +378,19 @@ u64 ufs_new_fragments(struct inode *inode, void *p, u64 fragment,
 				  "fragment %llu, tmp %llu\n",
 				  (unsigned long long)fragment,
 				  (unsigned long long)tmp);
-			unlock_ufs(sb);
+			mutex_unlock(&UFS_SB(sb)->s_lock);
 			return INVBLOCK;
 		}
 		if (fragment < UFS_I(inode)->i_lastfrag) {
 			UFSD("EXIT (ALREADY ALLOCATED)\n");
-			unlock_ufs(sb);
+			mutex_unlock(&UFS_SB(sb)->s_lock);
 			return 0;
 		}
 	}
 	else {
 		if (tmp) {
 			UFSD("EXIT (ALREADY ALLOCATED)\n");
-			unlock_ufs(sb);
+			mutex_unlock(&UFS_SB(sb)->s_lock);
 			return 0;
 		}
 	}
@@ -399,7 +399,7 @@ u64 ufs_new_fragments(struct inode *inode, void *p, u64 fragment,
 	 * There is not enough space for user on the device
 	 */
 	if (!capable(CAP_SYS_RESOURCE) && ufs_freespace(uspi, UFS_MINFREE) <= 0) {
-		unlock_ufs(sb);
+		mutex_unlock(&UFS_SB(sb)->s_lock);
 		UFSD("EXIT (FAILED)\n");
 		return 0;
 	}
@@ -424,7 +424,7 @@ u64 ufs_new_fragments(struct inode *inode, void *p, u64 fragment,
 			ufs_clear_frags(inode, result + oldcount,
 					newcount - oldcount, locked_page != NULL);
 		}
-		unlock_ufs(sb);
+		mutex_unlock(&UFS_SB(sb)->s_lock);
 		UFSD("EXIT, result %llu\n", (unsigned long long)result);
 		return result;
 	}
@@ -439,7 +439,7 @@ u64 ufs_new_fragments(struct inode *inode, void *p, u64 fragment,
 						fragment + count);
 		ufs_clear_frags(inode, result + oldcount, newcount - oldcount,
 				locked_page != NULL);
-		unlock_ufs(sb);
+		mutex_unlock(&UFS_SB(sb)->s_lock);
 		UFSD("EXIT, result %llu\n", (unsigned long long)result);
 		return result;
 	}
@@ -477,7 +477,7 @@ u64 ufs_new_fragments(struct inode *inode, void *p, u64 fragment,
 		*err = 0;
 		UFS_I(inode)->i_lastfrag = max(UFS_I(inode)->i_lastfrag,
 						fragment + count);
-		unlock_ufs(sb);
+		mutex_unlock(&UFS_SB(sb)->s_lock);
 		if (newcount < request)
 			ufs_free_fragments (inode, result + newcount, request - newcount);
 		ufs_free_fragments (inode, tmp, oldcount);
@@ -485,7 +485,7 @@ u64 ufs_new_fragments(struct inode *inode, void *p, u64 fragment,
 		return result;
 	}
 
-	unlock_ufs(sb);
+	mutex_unlock(&UFS_SB(sb)->s_lock);
 	UFSD("EXIT (FAILED)\n");
 	return 0;
 }		
diff --git a/fs/ufs/dir.c b/fs/ufs/dir.c
index 1bfe8cabff0f..74f2e80288bf 100644
--- a/fs/ufs/dir.c
+++ b/fs/ufs/dir.c
@@ -65,11 +65,6 @@ static inline void ufs_put_page(struct page *page)
 	page_cache_release(page);
 }
 
-static inline unsigned long ufs_dir_pages(struct inode *inode)
-{
-	return (inode->i_size+PAGE_CACHE_SIZE-1)>>PAGE_CACHE_SHIFT;
-}
-
 ino_t ufs_inode_by_name(struct inode *dir, const struct qstr *qstr)
 {
 	ino_t res = 0;
@@ -87,7 +82,8 @@ ino_t ufs_inode_by_name(struct inode *dir, const struct qstr *qstr)
 
 /* Releases the page */
 void ufs_set_link(struct inode *dir, struct ufs_dir_entry *de,
-		  struct page *page, struct inode *inode)
+		  struct page *page, struct inode *inode,
+		  bool update_times)
 {
 	loff_t pos = page_offset(page) +
 			(char *) de - (char *) page_address(page);
@@ -103,7 +99,8 @@ void ufs_set_link(struct inode *dir, struct ufs_dir_entry *de,
 
 	err = ufs_commit_chunk(page, pos, len);
 	ufs_put_page(page);
-	dir->i_mtime = dir->i_ctime = CURRENT_TIME_SEC;
+	if (update_times)
+		dir->i_mtime = dir->i_ctime = CURRENT_TIME_SEC;
 	mark_inode_dirty(dir);
 }
 
@@ -256,7 +253,7 @@ struct ufs_dir_entry *ufs_find_entry(struct inode *dir, const struct qstr *qstr,
 	int namelen = qstr->len;
 	unsigned reclen = UFS_DIR_REC_LEN(namelen);
 	unsigned long start, n;
-	unsigned long npages = ufs_dir_pages(dir);
+	unsigned long npages = dir_pages(dir);
 	struct page *page = NULL;
 	struct ufs_inode_info *ui = UFS_I(dir);
 	struct ufs_dir_entry *de;
@@ -320,7 +317,7 @@ int ufs_add_link(struct dentry *dentry, struct inode *inode)
 	unsigned short rec_len, name_len;
 	struct page *page = NULL;
 	struct ufs_dir_entry *de;
-	unsigned long npages = ufs_dir_pages(dir);
+	unsigned long npages = dir_pages(dir);
 	unsigned long n;
 	char *kaddr;
 	loff_t pos;
@@ -437,7 +434,7 @@ ufs_readdir(struct file *file, struct dir_context *ctx)
 	struct super_block *sb = inode->i_sb;
 	unsigned int offset = pos & ~PAGE_CACHE_MASK;
 	unsigned long n = pos >> PAGE_CACHE_SHIFT;
-	unsigned long npages = ufs_dir_pages(inode);
+	unsigned long npages = dir_pages(inode);
 	unsigned chunk_mask = ~(UFS_SB(sb)->s_uspi->s_dirblksize - 1);
 	int need_revalidate = file->f_version != inode->i_version;
 	unsigned flags = UFS_SB(sb)->s_flags;
@@ -608,7 +605,7 @@ int ufs_empty_dir(struct inode * inode)
 {
 	struct super_block *sb = inode->i_sb;
 	struct page *page = NULL;
-	unsigned long i, npages = ufs_dir_pages(inode);
+	unsigned long i, npages = dir_pages(inode);
 
 	for (i = 0; i < npages; i++) {
 		char *kaddr;
diff --git a/fs/ufs/ialloc.c b/fs/ufs/ialloc.c
index 7caa01652888..fd0203ce1f7f 100644
--- a/fs/ufs/ialloc.c
+++ b/fs/ufs/ialloc.c
@@ -69,11 +69,11 @@ void ufs_free_inode (struct inode * inode)
 	
 	ino = inode->i_ino;
 
-	lock_ufs(sb);
+	mutex_lock(&UFS_SB(sb)->s_lock);
 
 	if (!((ino > 1) && (ino < (uspi->s_ncg * uspi->s_ipg )))) {
 		ufs_warning(sb, "ufs_free_inode", "reserved inode or nonexistent inode %u\n", ino);
-		unlock_ufs(sb);
+		mutex_unlock(&UFS_SB(sb)->s_lock);
 		return;
 	}
 	
@@ -81,7 +81,7 @@ void ufs_free_inode (struct inode * inode)
 	bit = ufs_inotocgoff (ino);
 	ucpi = ufs_load_cylinder (sb, cg);
 	if (!ucpi) {
-		unlock_ufs(sb);
+		mutex_unlock(&UFS_SB(sb)->s_lock);
 		return;
 	}
 	ucg = ubh_get_ucg(UCPI_UBH(ucpi));
@@ -115,7 +115,7 @@ void ufs_free_inode (struct inode * inode)
 		ubh_sync_block(UCPI_UBH(ucpi));
 	
 	ufs_mark_sb_dirty(sb);
-	unlock_ufs(sb);
+	mutex_unlock(&UFS_SB(sb)->s_lock);
 	UFSD("EXIT\n");
 }
 
@@ -193,7 +193,7 @@ struct inode *ufs_new_inode(struct inode *dir, umode_t mode)
 	sbi = UFS_SB(sb);
 	uspi = sbi->s_uspi;
 
-	lock_ufs(sb);
+	mutex_lock(&sbi->s_lock);
 
 	/*
 	 * Try to place the inode in its parent directory
@@ -331,21 +331,21 @@ cg_found:
 			sync_dirty_buffer(bh);
 		brelse(bh);
 	}
-	unlock_ufs(sb);
+	mutex_unlock(&sbi->s_lock);
 
 	UFSD("allocating inode %lu\n", inode->i_ino);
 	UFSD("EXIT\n");
 	return inode;
 
 fail_remove_inode:
-	unlock_ufs(sb);
+	mutex_unlock(&sbi->s_lock);
 	clear_nlink(inode);
 	unlock_new_inode(inode);
 	iput(inode);
 	UFSD("EXIT (FAILED): err %d\n", err);
 	return ERR_PTR(err);
 failed:
-	unlock_ufs(sb);
+	mutex_unlock(&sbi->s_lock);
 	make_bad_inode(inode);
 	iput (inode);
 	UFSD("EXIT (FAILED): err %d\n", err);
diff --git a/fs/ufs/inode.c b/fs/ufs/inode.c
index 99aaf5c9bf4d..f913a6924b23 100644
--- a/fs/ufs/inode.c
+++ b/fs/ufs/inode.c
@@ -903,6 +903,9 @@ void ufs_evict_inode(struct inode * inode)
 	invalidate_inode_buffers(inode);
 	clear_inode(inode);
 
-	if (want_delete)
+	if (want_delete) {
+		lock_ufs(inode->i_sb);
 		ufs_free_inode(inode);
+		unlock_ufs(inode->i_sb);
+	}
 }
diff --git a/fs/ufs/namei.c b/fs/ufs/namei.c
index f773deb1d2e3..47966554317c 100644
--- a/fs/ufs/namei.c
+++ b/fs/ufs/namei.c
@@ -56,11 +56,9 @@ static struct dentry *ufs_lookup(struct inode * dir, struct dentry *dentry, unsi
 	if (dentry->d_name.len > UFS_MAXNAMLEN)
 		return ERR_PTR(-ENAMETOOLONG);
 
-	lock_ufs(dir->i_sb);
 	ino = ufs_inode_by_name(dir, &dentry->d_name);
 	if (ino)
 		inode = ufs_iget(dir->i_sb, ino);
-	unlock_ufs(dir->i_sb);
 	return d_splice_alias(inode, dentry);
 }
 
@@ -76,24 +74,16 @@ static int ufs_create (struct inode * dir, struct dentry * dentry, umode_t mode,
 		bool excl)
 {
 	struct inode *inode;
-	int err;
-
-	UFSD("BEGIN\n");
 
 	inode = ufs_new_inode(dir, mode);
-	err = PTR_ERR(inode);
+	if (IS_ERR(inode))
+		return PTR_ERR(inode);
 
-	if (!IS_ERR(inode)) {
-		inode->i_op = &ufs_file_inode_operations;
-		inode->i_fop = &ufs_file_operations;
-		inode->i_mapping->a_ops = &ufs_aops;
-		mark_inode_dirty(inode);
-		lock_ufs(dir->i_sb);
-		err = ufs_add_nondir(dentry, inode);
-		unlock_ufs(dir->i_sb);
-	}
-	UFSD("END: err=%d\n", err);
-	return err;
+	inode->i_op = &ufs_file_inode_operations;
+	inode->i_fop = &ufs_file_operations;
+	inode->i_mapping->a_ops = &ufs_aops;
+	mark_inode_dirty(inode);
+	return ufs_add_nondir(dentry, inode);
 }
 
 static int ufs_mknod(struct inode *dir, struct dentry *dentry, umode_t mode, dev_t rdev)
@@ -110,9 +100,7 @@ static int ufs_mknod(struct inode *dir, struct dentry *dentry, umode_t mode, dev
 		init_special_inode(inode, mode, rdev);
 		ufs_set_inode_dev(inode->i_sb, UFS_I(inode), rdev);
 		mark_inode_dirty(inode);
-		lock_ufs(dir->i_sb);
 		err = ufs_add_nondir(dentry, inode);
-		unlock_ufs(dir->i_sb);
 	}
 	return err;
 }
@@ -121,19 +109,18 @@ static int ufs_symlink (struct inode * dir, struct dentry * dentry,
 	const char * symname)
 {
 	struct super_block * sb = dir->i_sb;
-	int err = -ENAMETOOLONG;
+	int err;
 	unsigned l = strlen(symname)+1;
 	struct inode * inode;
 
 	if (l > sb->s_blocksize)
-		goto out_notlocked;
+		return -ENAMETOOLONG;
 
 	inode = ufs_new_inode(dir, S_IFLNK | S_IRWXUGO);
 	err = PTR_ERR(inode);
 	if (IS_ERR(inode))
-		goto out_notlocked;
+		return err;
 
-	lock_ufs(dir->i_sb);
 	if (l > UFS_SB(sb)->s_uspi->s_maxsymlinklen) {
 		/* slow symlink */
 		inode->i_op = &ufs_symlink_inode_operations;
@@ -150,17 +137,13 @@ static int ufs_symlink (struct inode * dir, struct dentry * dentry,
 	}
 	mark_inode_dirty(inode);
 
-	err = ufs_add_nondir(dentry, inode);
-out:
-	unlock_ufs(dir->i_sb);
-out_notlocked:
-	return err;
+	return ufs_add_nondir(dentry, inode);
 
 out_fail:
 	inode_dec_link_count(inode);
 	unlock_new_inode(inode);
 	iput(inode);
-	goto out;
+	return err;
 }
 
 static int ufs_link (struct dentry * old_dentry, struct inode * dir,
@@ -169,14 +152,16 @@ static int ufs_link (struct dentry * old_dentry, struct inode * dir,
 	struct inode *inode = d_inode(old_dentry);
 	int error;
 
-	lock_ufs(dir->i_sb);
-
 	inode->i_ctime = CURRENT_TIME_SEC;
 	inode_inc_link_count(inode);
 	ihold(inode);
 
-	error = ufs_add_nondir(dentry, inode);
-	unlock_ufs(dir->i_sb);
+	error = ufs_add_link(dentry, inode);
+	if (error) {
+		inode_dec_link_count(inode);
+		iput(inode);
+	} else
+		d_instantiate(dentry, inode);
 	return error;
 }
 
@@ -185,9 +170,12 @@ static int ufs_mkdir(struct inode * dir, struct dentry * dentry, umode_t mode)
 	struct inode * inode;
 	int err;
 
+	inode_inc_link_count(dir);
+
 	inode = ufs_new_inode(dir, S_IFDIR|mode);
+	err = PTR_ERR(inode);
 	if (IS_ERR(inode))
-		return PTR_ERR(inode);
+		goto out_dir;
 
 	inode->i_op = &ufs_dir_inode_operations;
 	inode->i_fop = &ufs_dir_operations;
@@ -195,9 +183,6 @@ static int ufs_mkdir(struct inode * dir, struct dentry * dentry, umode_t mode)
 
 	inode_inc_link_count(inode);
 
-	lock_ufs(dir->i_sb);
-	inode_inc_link_count(dir);
-
 	err = ufs_make_empty(inode, dir);
 	if (err)
 		goto out_fail;
@@ -205,20 +190,19 @@ static int ufs_mkdir(struct inode * dir, struct dentry * dentry, umode_t mode)
 	err = ufs_add_link(dentry, inode);
 	if (err)
 		goto out_fail;
-	unlock_ufs(dir->i_sb);
 
+	unlock_new_inode(inode);
 	d_instantiate(dentry, inode);
-out:
-	return err;
+	return 0;
 
 out_fail:
 	inode_dec_link_count(inode);
 	inode_dec_link_count(inode);
 	unlock_new_inode(inode);
 	iput (inode);
+out_dir:
 	inode_dec_link_count(dir);
-	unlock_ufs(dir->i_sb);
-	goto out;
+	return err;
 }
 
 static int ufs_unlink(struct inode *dir, struct dentry *dentry)
@@ -248,7 +232,6 @@ static int ufs_rmdir (struct inode * dir, struct dentry *dentry)
 	struct inode * inode = d_inode(dentry);
 	int err= -ENOTEMPTY;
 
-	lock_ufs(dir->i_sb);
 	if (ufs_empty_dir (inode)) {
 		err = ufs_unlink(dir, dentry);
 		if (!err) {
@@ -257,7 +240,6 @@ static int ufs_rmdir (struct inode * dir, struct dentry *dentry)
 			inode_dec_link_count(dir);
 		}
 	}
-	unlock_ufs(dir->i_sb);
 	return err;
 }
 
@@ -295,7 +277,7 @@ static int ufs_rename(struct inode *old_dir, struct dentry *old_dentry,
 		new_de = ufs_find_entry(new_dir, &new_dentry->d_name, &new_page);
 		if (!new_de)
 			goto out_dir;
-		ufs_set_link(new_dir, new_de, new_page, old_inode);
+		ufs_set_link(new_dir, new_de, new_page, old_inode, 1);
 		new_inode->i_ctime = CURRENT_TIME_SEC;
 		if (dir_de)
 			drop_nlink(new_inode);
@@ -318,7 +300,12 @@ static int ufs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	mark_inode_dirty(old_inode);
 
 	if (dir_de) {
-		ufs_set_link(old_inode, dir_de, dir_page, new_dir);
+		if (old_dir != new_dir)
+			ufs_set_link(old_inode, dir_de, dir_page, new_dir, 0);
+		else {
+			kunmap(dir_page);
+			page_cache_release(dir_page);
+		}
 		inode_dec_link_count(old_dir);
 	}
 	return 0;
diff --git a/fs/ufs/super.c b/fs/ufs/super.c
index 098508a93c7b..250579a80d90 100644
--- a/fs/ufs/super.c
+++ b/fs/ufs/super.c
@@ -695,6 +695,7 @@ static int ufs_sync_fs(struct super_block *sb, int wait)
 	unsigned flags;
 
 	lock_ufs(sb);
+	mutex_lock(&UFS_SB(sb)->s_lock);
 
 	UFSD("ENTER\n");
 
@@ -712,6 +713,7 @@ static int ufs_sync_fs(struct super_block *sb, int wait)
 	ufs_put_cstotal(sb);
 
 	UFSD("EXIT\n");
+	mutex_unlock(&UFS_SB(sb)->s_lock);
 	unlock_ufs(sb);
 
 	return 0;
@@ -800,6 +802,7 @@ static int ufs_fill_super(struct super_block *sb, void *data, int silent)
 	UFSD("flag %u\n", (int)(sb->s_flags & MS_RDONLY));
 	
 	mutex_init(&sbi->mutex);
+	mutex_init(&sbi->s_lock);
 	spin_lock_init(&sbi->work_lock);
 	INIT_DELAYED_WORK(&sbi->sync_work, delayed_sync_fs);
 	/*
@@ -1278,6 +1281,7 @@ static int ufs_remount (struct super_block *sb, int *mount_flags, char *data)
 
 	sync_filesystem(sb);
 	lock_ufs(sb);
+	mutex_lock(&UFS_SB(sb)->s_lock);
 	uspi = UFS_SB(sb)->s_uspi;
 	flags = UFS_SB(sb)->s_flags;
 	usb1 = ubh_get_usb_first(uspi);
@@ -1291,6 +1295,7 @@ static int ufs_remount (struct super_block *sb, int *mount_flags, char *data)
 	new_mount_opt = 0;
 	ufs_set_opt (new_mount_opt, ONERROR_LOCK);
 	if (!ufs_parse_options (data, &new_mount_opt)) {
+		mutex_unlock(&UFS_SB(sb)->s_lock);
 		unlock_ufs(sb);
 		return -EINVAL;
 	}
@@ -1298,12 +1303,14 @@ static int ufs_remount (struct super_block *sb, int *mount_flags, char *data)
 		new_mount_opt |= ufstype;
 	} else if ((new_mount_opt & UFS_MOUNT_UFSTYPE) != ufstype) {
 		pr_err("ufstype can't be changed during remount\n");
+		mutex_unlock(&UFS_SB(sb)->s_lock);
 		unlock_ufs(sb);
 		return -EINVAL;
 	}
 
 	if ((*mount_flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY)) {
 		UFS_SB(sb)->s_mount_opt = new_mount_opt;
+		mutex_unlock(&UFS_SB(sb)->s_lock);
 		unlock_ufs(sb);
 		return 0;
 	}
@@ -1327,6 +1334,7 @@ static int ufs_remount (struct super_block *sb, int *mount_flags, char *data)
 	 */
 #ifndef CONFIG_UFS_FS_WRITE
 		pr_err("ufs was compiled with read-only support, can't be mounted as read-write\n");
+		mutex_unlock(&UFS_SB(sb)->s_lock);
 		unlock_ufs(sb);
 		return -EINVAL;
 #else
@@ -1336,11 +1344,13 @@ static int ufs_remount (struct super_block *sb, int *mount_flags, char *data)
 		    ufstype != UFS_MOUNT_UFSTYPE_SUNx86 &&
 		    ufstype != UFS_MOUNT_UFSTYPE_UFS2) {
 			pr_err("this ufstype is read-only supported\n");
+			mutex_unlock(&UFS_SB(sb)->s_lock);
 			unlock_ufs(sb);
 			return -EINVAL;
 		}
 		if (!ufs_read_cylinder_structures(sb)) {
 			pr_err("failed during remounting\n");
+			mutex_unlock(&UFS_SB(sb)->s_lock);
 			unlock_ufs(sb);
 			return -EPERM;
 		}
@@ -1348,6 +1358,7 @@ static int ufs_remount (struct super_block *sb, int *mount_flags, char *data)
 #endif
 	}
 	UFS_SB(sb)->s_mount_opt = new_mount_opt;
+	mutex_unlock(&UFS_SB(sb)->s_lock);
 	unlock_ufs(sb);
 	return 0;
 }
diff --git a/fs/ufs/ufs.h b/fs/ufs/ufs.h
index 2a07396d5f9e..2e31ea2e35a3 100644
--- a/fs/ufs/ufs.h
+++ b/fs/ufs/ufs.h
@@ -30,6 +30,7 @@ struct ufs_sb_info {
 	int work_queued; /* non-zero if the delayed work is queued */
 	struct delayed_work sync_work; /* FS sync delayed work */
 	spinlock_t work_lock; /* protects sync_work and work_queued */
+	struct mutex s_lock;
 };
 
 struct ufs_inode_info {
@@ -105,7 +106,7 @@ extern int ufs_delete_entry(struct inode *, struct ufs_dir_entry *, struct page
 extern int ufs_empty_dir (struct inode *);
 extern struct ufs_dir_entry *ufs_dotdot(struct inode *, struct page **);
 extern void ufs_set_link(struct inode *dir, struct ufs_dir_entry *de,
-			 struct page *page, struct inode *inode);
+			 struct page *page, struct inode *inode, bool update_times);
 
 /* file.c */
 extern const struct inode_operations ufs_file_inode_operations;
diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
index 874507de3485..f0e8249722d4 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -577,6 +577,13 @@ restart:
 	if (error)
 		return error;
 
+	/* For changing security info in file_remove_privs() we need i_mutex */
+	if (*iolock == XFS_IOLOCK_SHARED && !IS_NOSEC(inode)) {
+		xfs_rw_iunlock(ip, *iolock);
+		*iolock = XFS_IOLOCK_EXCL;
+		xfs_rw_ilock(ip, *iolock);
+		goto restart;
+	}
 	/*
 	 * If the offset is beyond the size of the file, we need to zero any
 	 * blocks that fall between the existing EOF and the start of this
@@ -637,7 +644,9 @@ restart:
 	 * setgid bits if the process is not being run by root.  This keeps
 	 * people from modifying setuid and setgid binaries.
 	 */
-	return file_remove_suid(file);
+	if (!IS_NOSEC(inode))
+		return file_remove_privs(file);
+	return 0;
 }
 
 /*