Merge branch 'foreign/liubo/replace-lockup' into for-chris-4.6

15 files changed, 236 insertions, 155 deletions

diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
index b90cd3776f8e..f6dac40f87ff 100644
--- a/fs/btrfs/backref.c
+++ b/fs/btrfs/backref.c
@@ -1406,7 +1406,8 @@ char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 			read_extent_buffer(eb, dest + bytes_left,
 					   name_off, name_len);
 		if (eb != eb_in) {
-			btrfs_tree_read_unlock_blocking(eb);
+			if (!path->skip_locking)
+				btrfs_tree_read_unlock_blocking(eb);
 			free_extent_buffer(eb);
 		}
 		ret = btrfs_find_item(fs_root, path, parent, 0,
@@ -1426,9 +1427,10 @@ char *btrfs_ref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
 		eb = path->nodes[0];
 		/* make sure we can use eb after releasing the path */
 		if (eb != eb_in) {
-			atomic_inc(&eb->refs);
-			btrfs_tree_read_lock(eb);
-			btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
+			if (!path->skip_locking)
+				btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
+			path->nodes[0] = NULL;
+			path->locks[0] = 0;
 		}
 		btrfs_release_path(path);
 		iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index c473c42d7d6c..3346cd8f9910 100644
--- a/fs/btrfs/compression.c
+++ b/fs/btrfs/compression.c
@@ -637,11 +637,7 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 	faili = nr_pages - 1;
 	cb->nr_pages = nr_pages;
 
-	/* In the parent-locked case, we only locked the range we are
-	 * interested in.  In all other cases, we can opportunistically
-	 * cache decompressed data that goes beyond the requested range. */
-	if (!(bio_flags & EXTENT_BIO_PARENT_LOCKED))
-		add_ra_bio_pages(inode, em_start + em_len, cb);
+	add_ra_bio_pages(inode, em_start + em_len, cb);
 
 	/* include any pages we added in add_ra-bio_pages */
 	uncompressed_len = bio->bi_vcnt * PAGE_CACHE_SIZE;
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index e2a5cc0d4a14..38f34dadb86a 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -1005,8 +1005,10 @@ struct btrfs_dev_replace {
 	pid_t lock_owner;
 	atomic_t nesting_level;
 	struct mutex lock_finishing_cancel_unmount;
-	struct mutex lock_management_lock;
-	struct mutex lock;
+	rwlock_t lock;
+	atomic_t read_locks;
+	atomic_t blocking_readers;
+	wait_queue_head_t read_lock_wq;
 
 	struct btrfs_scrub_progress scrub_progress;
 };
diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c
index 0be47e4b8136..b57daa895cea 100644
--- a/fs/btrfs/delayed-inode.c
+++ b/fs/btrfs/delayed-inode.c
@@ -1689,7 +1689,7 @@ int btrfs_should_delete_dir_index(struct list_head *del_list,
  *
  */
 int btrfs_readdir_delayed_dir_index(struct dir_context *ctx,
-				    struct list_head *ins_list)
+				    struct list_head *ins_list, bool *emitted)
 {
 	struct btrfs_dir_item *di;
 	struct btrfs_delayed_item *curr, *next;
@@ -1733,6 +1733,7 @@ int btrfs_readdir_delayed_dir_index(struct dir_context *ctx,
 
 		if (over)
 			return 1;
+		*emitted = true;
 	}
 	return 0;
 }
diff --git a/fs/btrfs/delayed-inode.h b/fs/btrfs/delayed-inode.h
index f70119f25421..0167853c84ae 100644
--- a/fs/btrfs/delayed-inode.h
+++ b/fs/btrfs/delayed-inode.h
@@ -144,7 +144,7 @@ void btrfs_put_delayed_items(struct list_head *ins_list,
 int btrfs_should_delete_dir_index(struct list_head *del_list,
 				  u64 index);
 int btrfs_readdir_delayed_dir_index(struct dir_context *ctx,
-				    struct list_head *ins_list);
+				    struct list_head *ins_list, bool *emitted);
 
 /* for init */
 int __init btrfs_delayed_inode_init(void);
diff --git a/fs/btrfs/dev-replace.c b/fs/btrfs/dev-replace.c
index 01ce5fcecc5c..ff2db7a6c894 100644
--- a/fs/btrfs/dev-replace.c
+++ b/fs/btrfs/dev-replace.c
@@ -202,13 +202,13 @@ int btrfs_run_dev_replace(struct btrfs_trans_handle *trans,
 	struct btrfs_dev_replace_item *ptr;
 	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
 
-	btrfs_dev_replace_lock(dev_replace);
+	btrfs_dev_replace_lock(dev_replace, 0);
 	if (!dev_replace->is_valid ||
 	    !dev_replace->item_needs_writeback) {
-		btrfs_dev_replace_unlock(dev_replace);
+		btrfs_dev_replace_unlock(dev_replace, 0);
 		return 0;
 	}
-	btrfs_dev_replace_unlock(dev_replace);
+	btrfs_dev_replace_unlock(dev_replace, 0);
 
 	key.objectid = 0;
 	key.type = BTRFS_DEV_REPLACE_KEY;
@@ -264,7 +264,7 @@ int btrfs_run_dev_replace(struct btrfs_trans_handle *trans,
 	ptr = btrfs_item_ptr(eb, path->slots[0],
 			     struct btrfs_dev_replace_item);
 
-	btrfs_dev_replace_lock(dev_replace);
+	btrfs_dev_replace_lock(dev_replace, 1);
 	if (dev_replace->srcdev)
 		btrfs_set_dev_replace_src_devid(eb, ptr,
 			dev_replace->srcdev->devid);
@@ -287,7 +287,7 @@ int btrfs_run_dev_replace(struct btrfs_trans_handle *trans,
 	btrfs_set_dev_replace_cursor_right(eb, ptr,
 		dev_replace->cursor_right);
 	dev_replace->item_needs_writeback = 0;
-	btrfs_dev_replace_unlock(dev_replace);
+	btrfs_dev_replace_unlock(dev_replace, 1);
 
 	btrfs_mark_buffer_dirty(eb);
 
@@ -356,7 +356,7 @@ int btrfs_dev_replace_start(struct btrfs_root *root,
 		return PTR_ERR(trans);
 	}
 
-	btrfs_dev_replace_lock(dev_replace);
+	btrfs_dev_replace_lock(dev_replace, 1);
 	switch (dev_replace->replace_state) {
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_NEVER_STARTED:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_FINISHED:
@@ -395,7 +395,7 @@ int btrfs_dev_replace_start(struct btrfs_root *root,
 	dev_replace->is_valid = 1;
 	dev_replace->item_needs_writeback = 1;
 	args->result = BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_ERROR;
-	btrfs_dev_replace_unlock(dev_replace);
+	btrfs_dev_replace_unlock(dev_replace, 1);
 
 	ret = btrfs_sysfs_add_device_link(tgt_device->fs_devices, tgt_device);
 	if (ret)
@@ -407,7 +407,7 @@ int btrfs_dev_replace_start(struct btrfs_root *root,
 	trans = btrfs_start_transaction(root, 0);
 	if (IS_ERR(trans)) {
 		ret = PTR_ERR(trans);
-		btrfs_dev_replace_lock(dev_replace);
+		btrfs_dev_replace_lock(dev_replace, 1);
 		goto leave;
 	}
 
@@ -433,7 +433,7 @@ int btrfs_dev_replace_start(struct btrfs_root *root,
 leave:
 	dev_replace->srcdev = NULL;
 	dev_replace->tgtdev = NULL;
-	btrfs_dev_replace_unlock(dev_replace);
+	btrfs_dev_replace_unlock(dev_replace, 1);
 	btrfs_destroy_dev_replace_tgtdev(fs_info, tgt_device);
 	return ret;
 }
@@ -471,18 +471,18 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 	/* don't allow cancel or unmount to disturb the finishing procedure */
 	mutex_lock(&dev_replace->lock_finishing_cancel_unmount);
 
-	btrfs_dev_replace_lock(dev_replace);
+	btrfs_dev_replace_lock(dev_replace, 0);
 	/* was the operation canceled, or is it finished? */
 	if (dev_replace->replace_state !=
 	    BTRFS_IOCTL_DEV_REPLACE_STATE_STARTED) {
-		btrfs_dev_replace_unlock(dev_replace);
+		btrfs_dev_replace_unlock(dev_replace, 0);
 		mutex_unlock(&dev_replace->lock_finishing_cancel_unmount);
 		return 0;
 	}
 
 	tgt_device = dev_replace->tgtdev;
 	src_device = dev_replace->srcdev;
-	btrfs_dev_replace_unlock(dev_replace);
+	btrfs_dev_replace_unlock(dev_replace, 0);
 
 	/*
 	 * flush all outstanding I/O and inode extent mappings before the
@@ -507,7 +507,7 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 	/* keep away write_all_supers() during the finishing procedure */
 	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
 	mutex_lock(&root->fs_info->chunk_mutex);
-	btrfs_dev_replace_lock(dev_replace);
+	btrfs_dev_replace_lock(dev_replace, 1);
 	dev_replace->replace_state =
 		scrub_ret ? BTRFS_IOCTL_DEV_REPLACE_STATE_CANCELED
 			  : BTRFS_IOCTL_DEV_REPLACE_STATE_FINISHED;
@@ -528,7 +528,7 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 			        rcu_str_deref(src_device->name),
 			      src_device->devid,
 			      rcu_str_deref(tgt_device->name), scrub_ret);
-		btrfs_dev_replace_unlock(dev_replace);
+		btrfs_dev_replace_unlock(dev_replace, 1);
 		mutex_unlock(&root->fs_info->chunk_mutex);
 		mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
 		mutex_unlock(&uuid_mutex);
@@ -565,7 +565,7 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 	list_add(&tgt_device->dev_alloc_list, &fs_info->fs_devices->alloc_list);
 	fs_info->fs_devices->rw_devices++;
 
-	btrfs_dev_replace_unlock(dev_replace);
+	btrfs_dev_replace_unlock(dev_replace, 1);
 
 	btrfs_rm_dev_replace_blocked(fs_info);
 
@@ -649,7 +649,7 @@ void btrfs_dev_replace_status(struct btrfs_fs_info *fs_info,
 	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
 	struct btrfs_device *srcdev;
 
-	btrfs_dev_replace_lock(dev_replace);
+	btrfs_dev_replace_lock(dev_replace, 0);
 	/* even if !dev_replace_is_valid, the values are good enough for
 	 * the replace_status ioctl */
 	args->result = BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_ERROR;
@@ -675,7 +675,7 @@ void btrfs_dev_replace_status(struct btrfs_fs_info *fs_info,
 			div_u64(btrfs_device_get_total_bytes(srcdev), 1000));
 		break;
 	}
-	btrfs_dev_replace_unlock(dev_replace);
+	btrfs_dev_replace_unlock(dev_replace, 0);
 }
 
 int btrfs_dev_replace_cancel(struct btrfs_fs_info *fs_info,
@@ -698,13 +698,13 @@ static u64 __btrfs_dev_replace_cancel(struct btrfs_fs_info *fs_info)
 		return -EROFS;
 
 	mutex_lock(&dev_replace->lock_finishing_cancel_unmount);
-	btrfs_dev_replace_lock(dev_replace);
+	btrfs_dev_replace_lock(dev_replace, 1);
 	switch (dev_replace->replace_state) {
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_NEVER_STARTED:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_FINISHED:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_CANCELED:
 		result = BTRFS_IOCTL_DEV_REPLACE_RESULT_NOT_STARTED;
-		btrfs_dev_replace_unlock(dev_replace);
+		btrfs_dev_replace_unlock(dev_replace, 1);
 		goto leave;
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_STARTED:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_SUSPENDED:
@@ -717,7 +717,7 @@ static u64 __btrfs_dev_replace_cancel(struct btrfs_fs_info *fs_info)
 	dev_replace->replace_state = BTRFS_IOCTL_DEV_REPLACE_STATE_CANCELED;
 	dev_replace->time_stopped = get_seconds();
 	dev_replace->item_needs_writeback = 1;
-	btrfs_dev_replace_unlock(dev_replace);
+	btrfs_dev_replace_unlock(dev_replace, 1);
 	btrfs_scrub_cancel(fs_info);
 
 	trans = btrfs_start_transaction(root, 0);
@@ -740,7 +740,7 @@ void btrfs_dev_replace_suspend_for_unmount(struct btrfs_fs_info *fs_info)
 	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
 
 	mutex_lock(&dev_replace->lock_finishing_cancel_unmount);
-	btrfs_dev_replace_lock(dev_replace);
+	btrfs_dev_replace_lock(dev_replace, 1);
 	switch (dev_replace->replace_state) {
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_NEVER_STARTED:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_FINISHED:
@@ -756,7 +756,7 @@ void btrfs_dev_replace_suspend_for_unmount(struct btrfs_fs_info *fs_info)
 		break;
 	}
 
-	btrfs_dev_replace_unlock(dev_replace);
+	btrfs_dev_replace_unlock(dev_replace, 1);
 	mutex_unlock(&dev_replace->lock_finishing_cancel_unmount);
 }
 
@@ -766,12 +766,12 @@ int btrfs_resume_dev_replace_async(struct btrfs_fs_info *fs_info)
 	struct task_struct *task;
 	struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
 
-	btrfs_dev_replace_lock(dev_replace);
+	btrfs_dev_replace_lock(dev_replace, 1);
 	switch (dev_replace->replace_state) {
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_NEVER_STARTED:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_FINISHED:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_CANCELED:
-		btrfs_dev_replace_unlock(dev_replace);
+		btrfs_dev_replace_unlock(dev_replace, 1);
 		return 0;
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_STARTED:
 		break;
@@ -784,10 +784,10 @@ int btrfs_resume_dev_replace_async(struct btrfs_fs_info *fs_info)
 		btrfs_info(fs_info, "cannot continue dev_replace, tgtdev is missing");
 		btrfs_info(fs_info,
 			"you may cancel the operation after 'mount -o degraded'");
-		btrfs_dev_replace_unlock(dev_replace);
+		btrfs_dev_replace_unlock(dev_replace, 1);
 		return 0;
 	}
-	btrfs_dev_replace_unlock(dev_replace);
+	btrfs_dev_replace_unlock(dev_replace, 1);
 
 	WARN_ON(atomic_xchg(
 		&fs_info->mutually_exclusive_operation_running, 1));
@@ -865,48 +865,58 @@ int btrfs_dev_replace_is_ongoing(struct btrfs_dev_replace *dev_replace)
 	return 1;
 }
 
-void btrfs_dev_replace_lock(struct btrfs_dev_replace *dev_replace)
+void btrfs_dev_replace_lock(struct btrfs_dev_replace *dev_replace, int rw)
 {
-	/* the beginning is just an optimization for the typical case */
-	if (atomic_read(&dev_replace->nesting_level) == 0) {
-acquire_lock:
-		/* this is not a nested case where the same thread
-		 * is trying to acqurire the same lock twice */
-		mutex_lock(&dev_replace->lock);
-		mutex_lock(&dev_replace->lock_management_lock);
-		dev_replace->lock_owner = current->pid;
-		atomic_inc(&dev_replace->nesting_level);
-		mutex_unlock(&dev_replace->lock_management_lock);
-		return;
+	if (rw == 1) {
+		/* write */
+again:
+		wait_event(dev_replace->read_lock_wq,
+			   atomic_read(&dev_replace->blocking_readers) == 0);
+		write_lock(&dev_replace->lock);
+		if (atomic_read(&dev_replace->blocking_readers)) {
+			write_unlock(&dev_replace->lock);
+			goto again;
+		}
+	} else {
+		read_lock(&dev_replace->lock);
+		atomic_inc(&dev_replace->read_locks);
 	}
+}
 
-	mutex_lock(&dev_replace->lock_management_lock);
-	if (atomic_read(&dev_replace->nesting_level) > 0 &&
-	    dev_replace->lock_owner == current->pid) {
-		WARN_ON(!mutex_is_locked(&dev_replace->lock));
-		atomic_inc(&dev_replace->nesting_level);
-		mutex_unlock(&dev_replace->lock_management_lock);
-		return;
+void btrfs_dev_replace_unlock(struct btrfs_dev_replace *dev_replace, int rw)
+{
+	if (rw == 1) {
+		/* write */
+		ASSERT(atomic_read(&dev_replace->blocking_readers) == 0);
+		write_unlock(&dev_replace->lock);
+	} else {
+		ASSERT(atomic_read(&dev_replace->read_locks) > 0);
+		atomic_dec(&dev_replace->read_locks);
+		read_unlock(&dev_replace->lock);
 	}
+}
 
-	mutex_unlock(&dev_replace->lock_management_lock);
-	goto acquire_lock;
+/* inc blocking cnt and release read lock */
+void btrfs_dev_replace_set_lock_blocking(
+					struct btrfs_dev_replace *dev_replace)
+{
+	/* only set blocking for read lock */
+	ASSERT(atomic_read(&dev_replace->read_locks) > 0);
+	atomic_inc(&dev_replace->blocking_readers);
+	read_unlock(&dev_replace->lock);
 }
 
-void btrfs_dev_replace_unlock(struct btrfs_dev_replace *dev_replace)
+/* acquire read lock and dec blocking cnt */
+void btrfs_dev_replace_clear_lock_blocking(
+					struct btrfs_dev_replace *dev_replace)
 {
-	WARN_ON(!mutex_is_locked(&dev_replace->lock));
-	mutex_lock(&dev_replace->lock_management_lock);
-	WARN_ON(atomic_read(&dev_replace->nesting_level) < 1);
-	WARN_ON(dev_replace->lock_owner != current->pid);
-	atomic_dec(&dev_replace->nesting_level);
-	if (atomic_read(&dev_replace->nesting_level) == 0) {
-		dev_replace->lock_owner = 0;
-		mutex_unlock(&dev_replace->lock_management_lock);
-		mutex_unlock(&dev_replace->lock);
-	} else {
-		mutex_unlock(&dev_replace->lock_management_lock);
-	}
+	/* only set blocking for read lock */
+	ASSERT(atomic_read(&dev_replace->read_locks) > 0);
+	ASSERT(atomic_read(&dev_replace->blocking_readers) > 0);
+	read_lock(&dev_replace->lock);
+	if (atomic_dec_and_test(&dev_replace->blocking_readers) &&
+	    waitqueue_active(&dev_replace->read_lock_wq))
+		wake_up(&dev_replace->read_lock_wq);
 }
 
 void btrfs_bio_counter_inc_noblocked(struct btrfs_fs_info *fs_info)
diff --git a/fs/btrfs/dev-replace.h b/fs/btrfs/dev-replace.h
index 20035cbbf021..29e3ef5f96bd 100644
--- a/fs/btrfs/dev-replace.h
+++ b/fs/btrfs/dev-replace.h
@@ -34,8 +34,11 @@ int btrfs_dev_replace_cancel(struct btrfs_fs_info *fs_info,
 void btrfs_dev_replace_suspend_for_unmount(struct btrfs_fs_info *fs_info);
 int btrfs_resume_dev_replace_async(struct btrfs_fs_info *fs_info);
 int btrfs_dev_replace_is_ongoing(struct btrfs_dev_replace *dev_replace);
-void btrfs_dev_replace_lock(struct btrfs_dev_replace *dev_replace);
-void btrfs_dev_replace_unlock(struct btrfs_dev_replace *dev_replace);
+void btrfs_dev_replace_lock(struct btrfs_dev_replace *dev_replace, int rw);
+void btrfs_dev_replace_unlock(struct btrfs_dev_replace *dev_replace, int rw);
+void btrfs_dev_replace_set_lock_blocking(struct btrfs_dev_replace *dev_replace);
+void btrfs_dev_replace_clear_lock_blocking(
+					struct btrfs_dev_replace *dev_replace);
 
 static inline void btrfs_dev_replace_stats_inc(atomic64_t *stat_value)
 {
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index de68b8b61fd2..1809475ace2f 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -2273,9 +2273,11 @@ static void btrfs_init_dev_replace_locks(struct btrfs_fs_info *fs_info)
 	fs_info->dev_replace.lock_owner = 0;
 	atomic_set(&fs_info->dev_replace.nesting_level, 0);
 	mutex_init(&fs_info->dev_replace.lock_finishing_cancel_unmount);
-	mutex_init(&fs_info->dev_replace.lock_management_lock);
-	mutex_init(&fs_info->dev_replace.lock);
+	rwlock_init(&fs_info->dev_replace.lock);
+	atomic_set(&fs_info->dev_replace.read_locks, 0);
+	atomic_set(&fs_info->dev_replace.blocking_readers, 0);
 	init_waitqueue_head(&fs_info->replace_wait);
+	init_waitqueue_head(&fs_info->dev_replace.read_lock_wq);
 }
 
 static void btrfs_init_qgroup(struct btrfs_fs_info *fs_info)
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 1b2073389dc2..ed2f32dd0d29 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -2897,12 +2897,11 @@ static int __do_readpage(struct extent_io_tree *tree,
 	struct block_device *bdev;
 	int ret;
 	int nr = 0;
-	int parent_locked = *bio_flags & EXTENT_BIO_PARENT_LOCKED;
 	size_t pg_offset = 0;
 	size_t iosize;
 	size_t disk_io_size;
 	size_t blocksize = inode->i_sb->s_blocksize;
-	unsigned long this_bio_flag = *bio_flags & EXTENT_BIO_PARENT_LOCKED;
+	unsigned long this_bio_flag = 0;
 
 	set_page_extent_mapped(page);
 
@@ -2942,18 +2941,16 @@ static int __do_readpage(struct extent_io_tree *tree,
 			kunmap_atomic(userpage);
 			set_extent_uptodate(tree, cur, cur + iosize - 1,
 					    &cached, GFP_NOFS);
-			if (!parent_locked)
-				unlock_extent_cached(tree, cur,
-						     cur + iosize - 1,
-						     &cached, GFP_NOFS);
+			unlock_extent_cached(tree, cur,
+					     cur + iosize - 1,
+					     &cached, GFP_NOFS);
 			break;
 		}
 		em = __get_extent_map(inode, page, pg_offset, cur,
 				      end - cur + 1, get_extent, em_cached);
 		if (IS_ERR_OR_NULL(em)) {
 			SetPageError(page);
-			if (!parent_locked)
-				unlock_extent(tree, cur, end);
+			unlock_extent(tree, cur, end);
 			break;
 		}
 		extent_offset = cur - em->start;
@@ -3038,12 +3035,9 @@ static int __do_readpage(struct extent_io_tree *tree,
 
 			set_extent_uptodate(tree, cur, cur + iosize - 1,
 					    &cached, GFP_NOFS);
-			if (parent_locked)
-				free_extent_state(cached);
-			else
-				unlock_extent_cached(tree, cur,
-						     cur + iosize - 1,
-						     &cached, GFP_NOFS);
+			unlock_extent_cached(tree, cur,
+					     cur + iosize - 1,
+					     &cached, GFP_NOFS);
 			cur = cur + iosize;
 			pg_offset += iosize;
 			continue;
@@ -3052,8 +3046,7 @@ static int __do_readpage(struct extent_io_tree *tree,
 		if (test_range_bit(tree, cur, cur_end,
 				   EXTENT_UPTODATE, 1, NULL)) {
 			check_page_uptodate(tree, page);
-			if (!parent_locked)
-				unlock_extent(tree, cur, cur + iosize - 1);
+			unlock_extent(tree, cur, cur + iosize - 1);
 			cur = cur + iosize;
 			pg_offset += iosize;
 			continue;
@@ -3063,8 +3056,7 @@ static int __do_readpage(struct extent_io_tree *tree,
 		 */
 		if (block_start == EXTENT_MAP_INLINE) {
 			SetPageError(page);
-			if (!parent_locked)
-				unlock_extent(tree, cur, cur + iosize - 1);
+			unlock_extent(tree, cur, cur + iosize - 1);
 			cur = cur + iosize;
 			pg_offset += iosize;
 			continue;
@@ -3083,8 +3075,7 @@ static int __do_readpage(struct extent_io_tree *tree,
 			*bio_flags = this_bio_flag;
 		} else {
 			SetPageError(page);
-			if (!parent_locked)
-				unlock_extent(tree, cur, cur + iosize - 1);
+			unlock_extent(tree, cur, cur + iosize - 1);
 		}
 		cur = cur + iosize;
 		pg_offset += iosize;
@@ -3214,20 +3205,6 @@ int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
 	return ret;
 }
 
-int extent_read_full_page_nolock(struct extent_io_tree *tree, struct page *page,
-				 get_extent_t *get_extent, int mirror_num)
-{
-	struct bio *bio = NULL;
-	unsigned long bio_flags = EXTENT_BIO_PARENT_LOCKED;
-	int ret;
-
-	ret = __do_readpage(tree, page, get_extent, NULL, &bio, mirror_num,
-			    &bio_flags, READ, NULL);
-	if (bio)
-		ret = submit_one_bio(READ, bio, mirror_num, bio_flags);
-	return ret;
-}
-
 static noinline void update_nr_written(struct page *page,
 				      struct writeback_control *wbc,
 				      unsigned long nr_written)
diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index 0377413bd4b9..880d5292e972 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -29,7 +29,6 @@
  */
 #define EXTENT_BIO_COMPRESSED 1
 #define EXTENT_BIO_TREE_LOG 2
-#define EXTENT_BIO_PARENT_LOCKED 4
 #define EXTENT_BIO_FLAG_SHIFT 16
 
 /* these are bit numbers for test/set bit */
@@ -210,8 +209,6 @@ static inline int lock_extent(struct extent_io_tree *tree, u64 start, u64 end)
 int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end);
 int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
 			  get_extent_t *get_extent, int mirror_num);
-int extent_read_full_page_nolock(struct extent_io_tree *tree, struct page *page,
-				 get_extent_t *get_extent, int mirror_num);
 int __init extent_io_init(void);
 void extent_io_exit(void);
 
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 350b77220153..d4381f5ef52d 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -5731,6 +5731,7 @@ static int btrfs_real_readdir(struct file *file, struct dir_context *ctx)
 	char *name_ptr;
 	int name_len;
 	int is_curr = 0;	/* ctx->pos points to the current index? */
+	bool emitted;
 
 	/* FIXME, use a real flag for deciding about the key type */
 	if (root->fs_info->tree_root == root)
@@ -5759,6 +5760,7 @@ static int btrfs_real_readdir(struct file *file, struct dir_context *ctx)
 	if (ret < 0)
 		goto err;
 
+	emitted = false;
 	while (1) {
 		leaf = path->nodes[0];
 		slot = path->slots[0];
@@ -5838,6 +5840,7 @@ skip:
 
 			if (over)
 				goto nopos;
+			emitted = true;
 			di_len = btrfs_dir_name_len(leaf, di) +
 				 btrfs_dir_data_len(leaf, di) + sizeof(*di);
 			di_cur += di_len;
@@ -5850,11 +5853,20 @@ next:
 	if (key_type == BTRFS_DIR_INDEX_KEY) {
 		if (is_curr)
 			ctx->pos++;
-		ret = btrfs_readdir_delayed_dir_index(ctx, &ins_list);
+		ret = btrfs_readdir_delayed_dir_index(ctx, &ins_list, &emitted);
 		if (ret)
 			goto nopos;
 	}
 
+	/*
+	 * If we haven't emitted any dir entry, we must not touch ctx->pos as
+	 * it was was set to the termination value in previous call. We assume
+	 * that "." and ".." were emitted if we reach this point and set the
+	 * termination value as well for an empty directory.
+	 */
+	if (ctx->pos > 2 && !emitted)
+		goto nopos;
+
 	/* Reached end of directory/root. Bump pos past the last item. */
 	ctx->pos++;
 
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 9ba7b5be3fe5..a0be375702f9 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -2794,24 +2794,29 @@ out:
 static struct page *extent_same_get_page(struct inode *inode, pgoff_t index)
 {
 	struct page *page;
-	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
 
 	page = grab_cache_page(inode->i_mapping, index);
 	if (!page)
-		return NULL;
+		return ERR_PTR(-ENOMEM);
 
 	if (!PageUptodate(page)) {
-		if (extent_read_full_page_nolock(tree, page, btrfs_get_extent,
-						 0))
-			return NULL;
+		int ret;
+
+		ret = btrfs_readpage(NULL, page);
+		if (ret)
+			return ERR_PTR(ret);
 		lock_page(page);
 		if (!PageUptodate(page)) {
 			unlock_page(page);
 			page_cache_release(page);
-			return NULL;
+			return ERR_PTR(-EIO);
+		}
+		if (page->mapping != inode->i_mapping) {
+			unlock_page(page);
+			page_cache_release(page);
+			return ERR_PTR(-EAGAIN);
 		}
 	}
-	unlock_page(page);
 
 	return page;
 }
@@ -2823,17 +2828,31 @@ static int gather_extent_pages(struct inode *inode, struct page **pages,
 	pgoff_t index = off >> PAGE_CACHE_SHIFT;
 
 	for (i = 0; i < num_pages; i++) {
+again:
 		pages[i] = extent_same_get_page(inode, index + i);
-		if (!pages[i])
-			return -ENOMEM;
+		if (IS_ERR(pages[i])) {
+			int err = PTR_ERR(pages[i]);
+
+			if (err == -EAGAIN)
+				goto again;
+			pages[i] = NULL;
+			return err;
+		}
 	}
 	return 0;
 }
 
-static inline void lock_extent_range(struct inode *inode, u64 off, u64 len)
+static int lock_extent_range(struct inode *inode, u64 off, u64 len,
+			     bool retry_range_locking)
 {
-	/* do any pending delalloc/csum calc on src, one way or
-	   another, and lock file content */
+	/*
+	 * Do any pending delalloc/csum calculations on inode, one way or
+	 * another, and lock file content.
+	 * The locking order is:
+	 *
+	 *   1) pages
+	 *   2) range in the inode's io tree
+	 */
 	while (1) {
 		struct btrfs_ordered_extent *ordered;
 		lock_extent(&BTRFS_I(inode)->io_tree, off, off + len - 1);
@@ -2851,8 +2870,11 @@ static inline void lock_extent_range(struct inode *inode, u64 off, u64 len)
 		unlock_extent(&BTRFS_I(inode)->io_tree, off, off + len - 1);
 		if (ordered)
 			btrfs_put_ordered_extent(ordered);
+		if (!retry_range_locking)
+			return -EAGAIN;
 		btrfs_wait_ordered_range(inode, off, len);
 	}
+	return 0;
 }
 
 static void btrfs_double_inode_unlock(struct inode *inode1, struct inode *inode2)
@@ -2877,15 +2899,24 @@ static void btrfs_double_extent_unlock(struct inode *inode1, u64 loff1,
 	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)
+static int btrfs_double_extent_lock(struct inode *inode1, u64 loff1,
+				    struct inode *inode2, u64 loff2, u64 len,
+				    bool retry_range_locking)
 {
+	int ret;
+
 	if (inode1 < inode2) {
 		swap(inode1, inode2);
 		swap(loff1, loff2);
 	}
-	lock_extent_range(inode1, loff1, len);
-	lock_extent_range(inode2, loff2, len);
+	ret = lock_extent_range(inode1, loff1, len, retry_range_locking);
+	if (ret)
+		return ret;
+	ret = lock_extent_range(inode2, loff2, len, retry_range_locking);
+	if (ret)
+		unlock_extent(&BTRFS_I(inode1)->io_tree, loff1,
+			      loff1 + len - 1);
+	return ret;
 }
 
 struct cmp_pages {
@@ -2901,11 +2932,15 @@ static void btrfs_cmp_data_free(struct cmp_pages *cmp)
 
 	for (i = 0; i < cmp->num_pages; i++) {
 		pg = cmp->src_pages[i];
-		if (pg)
+		if (pg) {
+			unlock_page(pg);
 			page_cache_release(pg);
+		}
 		pg = cmp->dst_pages[i];
-		if (pg)
+		if (pg) {
+			unlock_page(pg);
 			page_cache_release(pg);
+		}
 	}
 	kfree(cmp->src_pages);
 	kfree(cmp->dst_pages);
@@ -2966,6 +3001,8 @@ static int btrfs_cmp_data(struct inode *src, u64 loff, struct inode *dst,
 
 		src_page = cmp->src_pages[i];
 		dst_page = cmp->dst_pages[i];
+		ASSERT(PageLocked(src_page));
+		ASSERT(PageLocked(dst_page));
 
 		addr = kmap_atomic(src_page);
 		dst_addr = kmap_atomic(dst_page);
@@ -3078,14 +3115,46 @@ static int btrfs_extent_same(struct inode *src, u64 loff, u64 olen,
 		goto out_unlock;
 	}
 
+again:
 	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);
+		ret = lock_extent_range(src, same_lock_start, same_lock_len,
+					false);
 	else
-		btrfs_double_extent_lock(src, loff, dst, dst_loff, len);
+		ret = btrfs_double_extent_lock(src, loff, dst, dst_loff, len,
+					       false);
+	/*
+	 * If one of the inodes has dirty pages in the respective range or
+	 * ordered extents, we need to flush dellaloc and wait for all ordered
+	 * extents in the range. We must unlock the pages and the ranges in the
+	 * io trees to avoid deadlocks when flushing delalloc (requires locking
+	 * pages) and when waiting for ordered extents to complete (they require
+	 * range locking).
+	 */
+	if (ret == -EAGAIN) {
+		/*
+		 * Ranges in the io trees already unlocked. Now unlock all
+		 * pages before waiting for all IO to complete.
+		 */
+		btrfs_cmp_data_free(&cmp);
+		if (same_inode) {
+			btrfs_wait_ordered_range(src, same_lock_start,
+						 same_lock_len);
+		} else {
+			btrfs_wait_ordered_range(src, loff, len);
+			btrfs_wait_ordered_range(dst, dst_loff, len);
+		}
+		goto again;
+	}
+	ASSERT(ret == 0);
+	if (WARN_ON(ret)) {
+		/* ranges in the io trees already unlocked */
+		btrfs_cmp_data_free(&cmp);
+		return ret;
+	}
 
 	/* pass original length for comparison so we stay within i_size */
 	ret = btrfs_cmp_data(src, loff, dst, dst_loff, olen, &cmp);
@@ -3795,9 +3864,15 @@ static noinline int btrfs_clone_files(struct file *file, struct file *file_src,
 		u64 lock_start = min_t(u64, off, destoff);
 		u64 lock_len = max_t(u64, off, destoff) + len - lock_start;
 
-		lock_extent_range(src, lock_start, lock_len);
+		ret = lock_extent_range(src, lock_start, lock_len, true);
 	} else {
-		btrfs_double_extent_lock(src, off, inode, destoff, len);
+		ret = btrfs_double_extent_lock(src, off, inode, destoff, len,
+					       true);
+	}
+	ASSERT(ret == 0);
+	if (WARN_ON(ret)) {
+		/* ranges in the io trees already unlocked */
+		goto out_unlock;
 	}
 
 	ret = btrfs_clone(src, inode, off, olen, len, destoff, 0);
diff --git a/fs/btrfs/reada.c b/fs/btrfs/reada.c
index bf69c008249c..b892914968c1 100644
--- a/fs/btrfs/reada.c
+++ b/fs/btrfs/reada.c
@@ -396,7 +396,7 @@ static struct reada_extent *reada_find_extent(struct btrfs_root *root,
 	}
 
 	/* insert extent in reada_tree + all per-device trees, all or nothing */
-	btrfs_dev_replace_lock(&fs_info->dev_replace);
+	btrfs_dev_replace_lock(&fs_info->dev_replace, 0);
 	spin_lock(&fs_info->reada_lock);
 	ret = radix_tree_insert(&fs_info->reada_tree, index, re);
 	if (ret == -EEXIST) {
@@ -404,12 +404,12 @@ static struct reada_extent *reada_find_extent(struct btrfs_root *root,
 		BUG_ON(!re_exist);
 		re_exist->refcnt++;
 		spin_unlock(&fs_info->reada_lock);
-		btrfs_dev_replace_unlock(&fs_info->dev_replace);
+		btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
 		goto error;
 	}
 	if (ret) {
 		spin_unlock(&fs_info->reada_lock);
-		btrfs_dev_replace_unlock(&fs_info->dev_replace);
+		btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
 		goto error;
 	}
 	prev_dev = NULL;
@@ -451,13 +451,13 @@ static struct reada_extent *reada_find_extent(struct btrfs_root *root,
 			BUG_ON(fs_info == NULL);
 			radix_tree_delete(&fs_info->reada_tree, index);
 			spin_unlock(&fs_info->reada_lock);
-			btrfs_dev_replace_unlock(&fs_info->dev_replace);
+			btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
 			goto error;
 		}
 		have_zone = 1;
 	}
 	spin_unlock(&fs_info->reada_lock);
-	btrfs_dev_replace_unlock(&fs_info->dev_replace);
+	btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
 
 	if (!have_zone)
 		goto error;
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 2de7817d0e1b..e42aa27c96e9 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -3859,16 +3859,16 @@ int btrfs_scrub_dev(struct btrfs_fs_info *fs_info, u64 devid, u64 start,
 		return -EIO;
 	}
 
-	btrfs_dev_replace_lock(&fs_info->dev_replace);
+	btrfs_dev_replace_lock(&fs_info->dev_replace, 0);
 	if (dev->scrub_device ||
 	    (!is_dev_replace &&
 	     btrfs_dev_replace_is_ongoing(&fs_info->dev_replace))) {
-		btrfs_dev_replace_unlock(&fs_info->dev_replace);
+		btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
 		mutex_unlock(&fs_info->scrub_lock);
 		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
 		return -EINPROGRESS;
 	}
-	btrfs_dev_replace_unlock(&fs_info->dev_replace);
+	btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
 
 	ret = scrub_workers_get(fs_info, is_dev_replace);
 	if (ret) {
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 6e0e4396d3a4..80857b4646c0 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -1715,12 +1715,12 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
 	} while (read_seqretry(&root->fs_info->profiles_lock, seq));
 
 	num_devices = root->fs_info->fs_devices->num_devices;
-	btrfs_dev_replace_lock(&root->fs_info->dev_replace);
+	btrfs_dev_replace_lock(&root->fs_info->dev_replace, 0);
 	if (btrfs_dev_replace_is_ongoing(&root->fs_info->dev_replace)) {
 		WARN_ON(num_devices < 1);
 		num_devices--;
 	}
-	btrfs_dev_replace_unlock(&root->fs_info->dev_replace);
+	btrfs_dev_replace_unlock(&root->fs_info->dev_replace, 0);
 
 	if ((all_avail & BTRFS_BLOCK_GROUP_RAID10) && num_devices <= 4) {
 		ret = BTRFS_ERROR_DEV_RAID10_MIN_NOT_MET;
@@ -3687,12 +3687,12 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 	}
 
 	num_devices = fs_info->fs_devices->num_devices;
-	btrfs_dev_replace_lock(&fs_info->dev_replace);
+	btrfs_dev_replace_lock(&fs_info->dev_replace, 0);
 	if (btrfs_dev_replace_is_ongoing(&fs_info->dev_replace)) {
 		BUG_ON(num_devices < 1);
 		num_devices--;
 	}
-	btrfs_dev_replace_unlock(&fs_info->dev_replace);
+	btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
 	allowed = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
 	if (num_devices == 1)
 		allowed |= BTRFS_BLOCK_GROUP_DUP;
@@ -5063,10 +5063,10 @@ int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len)
 		ret = 1;
 	free_extent_map(em);
 
-	btrfs_dev_replace_lock(&fs_info->dev_replace);
+	btrfs_dev_replace_lock(&fs_info->dev_replace, 0);
 	if (btrfs_dev_replace_is_ongoing(&fs_info->dev_replace))
 		ret++;
-	btrfs_dev_replace_unlock(&fs_info->dev_replace);
+	btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
 
 	return ret;
 }
@@ -5326,10 +5326,12 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
 	if (!bbio_ret)
 		goto out;
 
-	btrfs_dev_replace_lock(dev_replace);
+	btrfs_dev_replace_lock(dev_replace, 0);
 	dev_replace_is_ongoing = btrfs_dev_replace_is_ongoing(dev_replace);
 	if (!dev_replace_is_ongoing)
-		btrfs_dev_replace_unlock(dev_replace);
+		btrfs_dev_replace_unlock(dev_replace, 0);
+	else
+		btrfs_dev_replace_set_lock_blocking(dev_replace);
 
 	if (dev_replace_is_ongoing && mirror_num == map->num_stripes + 1 &&
 	    !(rw & (REQ_WRITE | REQ_DISCARD | REQ_GET_READ_MIRRORS)) &&
@@ -5752,8 +5754,10 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
 		bbio->mirror_num = map->num_stripes + 1;
 	}
 out:
-	if (dev_replace_is_ongoing)
-		btrfs_dev_replace_unlock(dev_replace);
+	if (dev_replace_is_ongoing) {
+		btrfs_dev_replace_clear_lock_blocking(dev_replace);
+		btrfs_dev_replace_unlock(dev_replace, 0);
+	}
 	free_extent_map(em);
 	return ret;
 }