Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull btrfs fixes and features from Chris Mason:
 "We've merged in the error handling patches from SuSE.  These are
  already shipping in the sles kernel, and they give btrfs the ability
  to abort transactions and go readonly on errors.  It involves a lot of
  churn as they clarify BUG_ONs, and remove the ones we now properly
  deal with.

  Josef reworked the way our metadata interacts with the page cache.
  page->private now points to the btrfs extent_buffer object, which
  makes everything faster.  He changed it so we write an whole extent
  buffer at a time instead of allowing individual pages to go down,,
  which will be important for the raid5/6 code (for the 3.5 merge
  window ;)

  Josef also made us more aggressive about dropping pages for metadata
  blocks that were freed due to COW.  Overall, our metadata caching is
  much faster now.

  We've integrated my patch for metadata bigger than the page size.
  This allows metadata blocks up to 64KB in size.  In practice 16K and
  32K seem to work best.  For workloads with lots of metadata, this cuts
  down the size of the extent allocation tree dramatically and fragments
  much less.

  Scrub was updated to support the larger block sizes, which ended up
  being a fairly large change (thanks Stefan Behrens).

  We also have an assortment of fixes and updates, especially to the
  balancing code (Ilya Dryomov), the back ref walker (Jan Schmidt) and
  the defragging code (Liu Bo)."

Fixed up trivial conflicts in fs/btrfs/scrub.c that were just due to
removal of the second argument to k[un]map_atomic() in commit
7ac687d9e047.

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (75 commits)
  Btrfs: update the checks for mixed block groups with big metadata blocks
  Btrfs: update to the right index of defragment
  Btrfs: do not bother to defrag an extent if it is a big real extent
  Btrfs: add a check to decide if we should defrag the range
  Btrfs: fix recursive defragment with autodefrag option
  Btrfs: fix the mismatch of page->mapping
  Btrfs: fix race between direct io and autodefrag
  Btrfs: fix deadlock during allocating chunks
  Btrfs: show useful info in space reservation tracepoint
  Btrfs: don't use crc items bigger than 4KB
  Btrfs: flush out and clean up any block device pages during mount
  btrfs: disallow unequal data/metadata blocksize for mixed block groups
  Btrfs: enhance superblock sanity checks
  Btrfs: change scrub to support big blocks
  Btrfs: minor cleanup in scrub
  Btrfs: introduce common define for max number of mirrors
  Btrfs: fix infinite loop in btrfs_shrink_device()
  Btrfs: fix memory leak in resolver code
  Btrfs: allow dup for data chunks in mixed mode
  Btrfs: validate target profiles only if we are going to use them
  ...

41 files changed, 4362 insertions, 2225 deletions

diff --git a/fs/btrfs/async-thread.c b/fs/btrfs/async-thread.c
index 0cc20b35c1c..42704149b72 100644
--- a/fs/btrfs/async-thread.c
+++ b/fs/btrfs/async-thread.c
@@ -171,11 +171,11 @@ out:
 	spin_unlock_irqrestore(&workers->lock, flags);
 }
 
-static noinline int run_ordered_completions(struct btrfs_workers *workers,
+static noinline void run_ordered_completions(struct btrfs_workers *workers,
 					    struct btrfs_work *work)
 {
 	if (!workers->ordered)
-		return 0;
+		return;
 
 	set_bit(WORK_DONE_BIT, &work->flags);
 
@@ -213,7 +213,6 @@ static noinline int run_ordered_completions(struct btrfs_workers *workers,
 	}
 
 	spin_unlock(&workers->order_lock);
-	return 0;
 }
 
 static void put_worker(struct btrfs_worker_thread *worker)
@@ -399,7 +398,7 @@ again:
 /*
  * this will wait for all the worker threads to shutdown
  */
-int btrfs_stop_workers(struct btrfs_workers *workers)
+void btrfs_stop_workers(struct btrfs_workers *workers)
 {
 	struct list_head *cur;
 	struct btrfs_worker_thread *worker;
@@ -427,7 +426,6 @@ int btrfs_stop_workers(struct btrfs_workers *workers)
 		put_worker(worker);
 	}
 	spin_unlock_irq(&workers->lock);
-	return 0;
 }
 
 /*
@@ -615,14 +613,14 @@ found:
  * it was taken from.  It is intended for use with long running work functions
  * that make some progress and want to give the cpu up for others.
  */
-int btrfs_requeue_work(struct btrfs_work *work)
+void btrfs_requeue_work(struct btrfs_work *work)
 {
 	struct btrfs_worker_thread *worker = work->worker;
 	unsigned long flags;
 	int wake = 0;
 
 	if (test_and_set_bit(WORK_QUEUED_BIT, &work->flags))
-		goto out;
+		return;
 
 	spin_lock_irqsave(&worker->lock, flags);
 	if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags))
@@ -649,9 +647,6 @@ int btrfs_requeue_work(struct btrfs_work *work)
 	if (wake)
 		wake_up_process(worker->task);
 	spin_unlock_irqrestore(&worker->lock, flags);
-out:
-
-	return 0;
 }
 
 void btrfs_set_work_high_prio(struct btrfs_work *work)
diff --git a/fs/btrfs/async-thread.h b/fs/btrfs/async-thread.h
index f34cc31fa3c..063698b90ce 100644
--- a/fs/btrfs/async-thread.h
+++ b/fs/btrfs/async-thread.h
@@ -111,9 +111,9 @@ struct btrfs_workers {
 
 void btrfs_queue_worker(struct btrfs_workers *workers, struct btrfs_work *work);
 int btrfs_start_workers(struct btrfs_workers *workers);
-int btrfs_stop_workers(struct btrfs_workers *workers);
+void btrfs_stop_workers(struct btrfs_workers *workers);
 void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max,
 			struct btrfs_workers *async_starter);
-int btrfs_requeue_work(struct btrfs_work *work);
+void btrfs_requeue_work(struct btrfs_work *work);
 void btrfs_set_work_high_prio(struct btrfs_work *work);
 #endif
diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
index 0436c12da8c..f4e90748940 100644
--- a/fs/btrfs/backref.c
+++ b/fs/btrfs/backref.c
@@ -116,6 +116,7 @@ add_parent:
  * to a logical address
  */
 static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
+					int search_commit_root,
 					struct __prelim_ref *ref,
 					struct ulist *parents)
 {
@@ -131,6 +132,7 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
+	path->search_commit_root = !!search_commit_root;
 
 	root_key.objectid = ref->root_id;
 	root_key.type = BTRFS_ROOT_ITEM_KEY;
@@ -188,6 +190,7 @@ out:
  * resolve all indirect backrefs from the list
  */
 static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
+				   int search_commit_root,
 				   struct list_head *head)
 {
 	int err;
@@ -212,7 +215,8 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
 			continue;
 		if (ref->count == 0)
 			continue;
-		err = __resolve_indirect_ref(fs_info, ref, parents);
+		err = __resolve_indirect_ref(fs_info, search_commit_root,
+					     ref, parents);
 		if (err) {
 			if (ret == 0)
 				ret = err;
@@ -586,6 +590,7 @@ static int find_parent_nodes(struct btrfs_trans_handle *trans,
 	struct btrfs_delayed_ref_head *head;
 	int info_level = 0;
 	int ret;
+	int search_commit_root = (trans == BTRFS_BACKREF_SEARCH_COMMIT_ROOT);
 	struct list_head prefs_delayed;
 	struct list_head prefs;
 	struct __prelim_ref *ref;
@@ -600,6 +605,7 @@ static int find_parent_nodes(struct btrfs_trans_handle *trans,
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
+	path->search_commit_root = !!search_commit_root;
 
 	/*
 	 * grab both a lock on the path and a lock on the delayed ref head.
@@ -614,35 +620,39 @@ again:
 		goto out;
 	BUG_ON(ret == 0);
 
-	/*
-	 * look if there are updates for this ref queued and lock the head
-	 */
-	delayed_refs = &trans->transaction->delayed_refs;
-	spin_lock(&delayed_refs->lock);
-	head = btrfs_find_delayed_ref_head(trans, bytenr);
-	if (head) {
-		if (!mutex_trylock(&head->mutex)) {
-			atomic_inc(&head->node.refs);
-			spin_unlock(&delayed_refs->lock);
-
-			btrfs_release_path(path);
-
-			/*
-			 * Mutex was contended, block until it's
-			 * released and try again
-			 */
-			mutex_lock(&head->mutex);
-			mutex_unlock(&head->mutex);
-			btrfs_put_delayed_ref(&head->node);
-			goto again;
-		}
-		ret = __add_delayed_refs(head, seq, &info_key, &prefs_delayed);
-		if (ret) {
-			spin_unlock(&delayed_refs->lock);
-			goto out;
+	if (trans != BTRFS_BACKREF_SEARCH_COMMIT_ROOT) {
+		/*
+		 * look if there are updates for this ref queued and lock the
+		 * head
+		 */
+		delayed_refs = &trans->transaction->delayed_refs;
+		spin_lock(&delayed_refs->lock);
+		head = btrfs_find_delayed_ref_head(trans, bytenr);
+		if (head) {
+			if (!mutex_trylock(&head->mutex)) {
+				atomic_inc(&head->node.refs);
+				spin_unlock(&delayed_refs->lock);
+
+				btrfs_release_path(path);
+
+				/*
+				 * Mutex was contended, block until it's
+				 * released and try again
+				 */
+				mutex_lock(&head->mutex);
+				mutex_unlock(&head->mutex);
+				btrfs_put_delayed_ref(&head->node);
+				goto again;
+			}
+			ret = __add_delayed_refs(head, seq, &info_key,
+						 &prefs_delayed);
+			if (ret) {
+				spin_unlock(&delayed_refs->lock);
+				goto out;
+			}
 		}
+		spin_unlock(&delayed_refs->lock);
 	}
-	spin_unlock(&delayed_refs->lock);
 
 	if (path->slots[0]) {
 		struct extent_buffer *leaf;
@@ -679,7 +689,7 @@ again:
 	if (ret)
 		goto out;
 
-	ret = __resolve_indirect_refs(fs_info, &prefs);
+	ret = __resolve_indirect_refs(fs_info, search_commit_root, &prefs);
 	if (ret)
 		goto out;
 
@@ -1074,8 +1084,7 @@ int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
 	return 0;
 }
 
-static int iterate_leaf_refs(struct btrfs_fs_info *fs_info,
-				struct btrfs_path *path, u64 logical,
+static int iterate_leaf_refs(struct btrfs_fs_info *fs_info, u64 logical,
 				u64 orig_extent_item_objectid,
 				u64 extent_item_pos, u64 root,
 				iterate_extent_inodes_t *iterate, void *ctx)
@@ -1143,35 +1152,38 @@ static int iterate_leaf_refs(struct btrfs_fs_info *fs_info,
  * calls iterate() for every inode that references the extent identified by
  * the given parameters.
  * when the iterator function returns a non-zero value, iteration stops.
- * path is guaranteed to be in released state when iterate() is called.
  */
 int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
-				struct btrfs_path *path,
 				u64 extent_item_objectid, u64 extent_item_pos,
+				int search_commit_root,
 				iterate_extent_inodes_t *iterate, void *ctx)
 {
 	int ret;
 	struct list_head data_refs = LIST_HEAD_INIT(data_refs);
 	struct list_head shared_refs = LIST_HEAD_INIT(shared_refs);
 	struct btrfs_trans_handle *trans;
-	struct ulist *refs;
-	struct ulist *roots;
+	struct ulist *refs = NULL;
+	struct ulist *roots = NULL;
 	struct ulist_node *ref_node = NULL;
 	struct ulist_node *root_node = NULL;
 	struct seq_list seq_elem;
-	struct btrfs_delayed_ref_root *delayed_refs;
-
-	trans = btrfs_join_transaction(fs_info->extent_root);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
+	struct btrfs_delayed_ref_root *delayed_refs = NULL;
 
 	pr_debug("resolving all inodes for extent %llu\n",
 			extent_item_objectid);
 
-	delayed_refs = &trans->transaction->delayed_refs;
-	spin_lock(&delayed_refs->lock);
-	btrfs_get_delayed_seq(delayed_refs, &seq_elem);
-	spin_unlock(&delayed_refs->lock);
+	if (search_commit_root) {
+		trans = BTRFS_BACKREF_SEARCH_COMMIT_ROOT;
+	} else {
+		trans = btrfs_join_transaction(fs_info->extent_root);
+		if (IS_ERR(trans))
+			return PTR_ERR(trans);
+
+		delayed_refs = &trans->transaction->delayed_refs;
+		spin_lock(&delayed_refs->lock);
+		btrfs_get_delayed_seq(delayed_refs, &seq_elem);
+		spin_unlock(&delayed_refs->lock);
+	}
 
 	ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid,
 				   extent_item_pos, seq_elem.seq,
@@ -1188,7 +1200,7 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 		while (!ret && (root_node = ulist_next(roots, root_node))) {
 			pr_debug("root %llu references leaf %llu\n",
 					root_node->val, ref_node->val);
-			ret = iterate_leaf_refs(fs_info, path, ref_node->val,
+			ret = iterate_leaf_refs(fs_info, ref_node->val,
 						extent_item_objectid,
 						extent_item_pos, root_node->val,
 						iterate, ctx);
@@ -1198,8 +1210,11 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
 	ulist_free(refs);
 	ulist_free(roots);
 out:
-	btrfs_put_delayed_seq(delayed_refs, &seq_elem);
-	btrfs_end_transaction(trans, fs_info->extent_root);
+	if (!search_commit_root) {
+		btrfs_put_delayed_seq(delayed_refs, &seq_elem);
+		btrfs_end_transaction(trans, fs_info->extent_root);
+	}
+
 	return ret;
 }
 
@@ -1210,6 +1225,7 @@ int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
 	int ret;
 	u64 extent_item_pos;
 	struct btrfs_key found_key;
+	int search_commit_root = path->search_commit_root;
 
 	ret = extent_from_logical(fs_info, logical, path,
 					&found_key);
@@ -1220,8 +1236,9 @@ int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
 		return ret;
 
 	extent_item_pos = logical - found_key.objectid;
-	ret = iterate_extent_inodes(fs_info, path, found_key.objectid,
-					extent_item_pos, iterate, ctx);
+	ret = iterate_extent_inodes(fs_info, found_key.objectid,
+					extent_item_pos, search_commit_root,
+					iterate, ctx);
 
 	return ret;
 }
@@ -1342,12 +1359,6 @@ int paths_from_inode(u64 inum, struct inode_fs_paths *ipath)
 				inode_to_path, ipath);
 }
 
-/*
- * allocates space to return multiple file system paths for an inode.
- * total_bytes to allocate are passed, note that space usable for actual path
- * information will be total_bytes - sizeof(struct inode_fs_paths).
- * the returned pointer must be freed with free_ipath() in the end.
- */
 struct btrfs_data_container *init_data_container(u32 total_bytes)
 {
 	struct btrfs_data_container *data;
@@ -1403,5 +1414,6 @@ struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root,
 
 void free_ipath(struct inode_fs_paths *ipath)
 {
+	kfree(ipath->fspath);
 	kfree(ipath);
 }
diff --git a/fs/btrfs/backref.h b/fs/btrfs/backref.h
index d00dfa9ca93..57ea2e959e4 100644
--- a/fs/btrfs/backref.h
+++ b/fs/btrfs/backref.h
@@ -22,6 +22,8 @@
 #include "ioctl.h"
 #include "ulist.h"
 
+#define BTRFS_BACKREF_SEARCH_COMMIT_ROOT ((struct btrfs_trans_handle *)0)
+
 struct inode_fs_paths {
 	struct btrfs_path		*btrfs_path;
 	struct btrfs_root		*fs_root;
@@ -44,9 +46,8 @@ int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
 				u64 *out_root, u8 *out_level);
 
 int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
-				struct btrfs_path *path,
 				u64 extent_item_objectid,
-				u64 extent_offset,
+				u64 extent_offset, int search_commit_root,
 				iterate_extent_inodes_t *iterate, void *ctx);
 
 int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index b805afb37fa..d286b40a567 100644
--- a/fs/btrfs/compression.c
+++ b/fs/btrfs/compression.c
@@ -226,8 +226,8 @@ out:
  * Clear the writeback bits on all of the file
  * pages for a compressed write
  */
-static noinline int end_compressed_writeback(struct inode *inode, u64 start,
-					     unsigned long ram_size)
+static noinline void end_compressed_writeback(struct inode *inode, u64 start,
+					      unsigned long ram_size)
 {
 	unsigned long index = start >> PAGE_CACHE_SHIFT;
 	unsigned long end_index = (start + ram_size - 1) >> PAGE_CACHE_SHIFT;
@@ -253,7 +253,6 @@ static noinline int end_compressed_writeback(struct inode *inode, u64 start,
 		index += ret;
 	}
 	/* the inode may be gone now */
-	return 0;
 }
 
 /*
@@ -392,16 +391,16 @@ int btrfs_submit_compressed_write(struct inode *inode, u64 start,
 			 */
 			atomic_inc(&cb->pending_bios);
 			ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 
 			if (!skip_sum) {
 				ret = btrfs_csum_one_bio(root, inode, bio,
 							 start, 1);
-				BUG_ON(ret);
+				BUG_ON(ret); /* -ENOMEM */
 			}
 
 			ret = btrfs_map_bio(root, WRITE, bio, 0, 1);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 
 			bio_put(bio);
 
@@ -421,15 +420,15 @@ int btrfs_submit_compressed_write(struct inode *inode, u64 start,
 	bio_get(bio);
 
 	ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
-	BUG_ON(ret);
+	BUG_ON(ret); /* -ENOMEM */
 
 	if (!skip_sum) {
 		ret = btrfs_csum_one_bio(root, inode, bio, start, 1);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 	}
 
 	ret = btrfs_map_bio(root, WRITE, bio, 0, 1);
-	BUG_ON(ret);
+	BUG_ON(ret); /* -ENOMEM */
 
 	bio_put(bio);
 	return 0;
@@ -497,7 +496,7 @@ static noinline int add_ra_bio_pages(struct inode *inode,
 		 * sure they map to this compressed extent on disk.
 		 */
 		set_page_extent_mapped(page);
-		lock_extent(tree, last_offset, end, GFP_NOFS);
+		lock_extent(tree, last_offset, end);
 		read_lock(&em_tree->lock);
 		em = lookup_extent_mapping(em_tree, last_offset,
 					   PAGE_CACHE_SIZE);
@@ -507,7 +506,7 @@ static noinline int add_ra_bio_pages(struct inode *inode,
 		    (last_offset + PAGE_CACHE_SIZE > extent_map_end(em)) ||
 		    (em->block_start >> 9) != cb->orig_bio->bi_sector) {
 			free_extent_map(em);
-			unlock_extent(tree, last_offset, end, GFP_NOFS);
+			unlock_extent(tree, last_offset, end);
 			unlock_page(page);
 			page_cache_release(page);
 			break;
@@ -535,7 +534,7 @@ static noinline int add_ra_bio_pages(struct inode *inode,
 			nr_pages++;
 			page_cache_release(page);
 		} else {
-			unlock_extent(tree, last_offset, end, GFP_NOFS);
+			unlock_extent(tree, last_offset, end);
 			unlock_page(page);
 			page_cache_release(page);
 			break;
@@ -662,7 +661,7 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 			bio_get(comp_bio);
 
 			ret = btrfs_bio_wq_end_io(root->fs_info, comp_bio, 0);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 
 			/*
 			 * inc the count before we submit the bio so
@@ -675,14 +674,14 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 			if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) {
 				ret = btrfs_lookup_bio_sums(root, inode,
 							comp_bio, sums);
-				BUG_ON(ret);
+				BUG_ON(ret); /* -ENOMEM */
 			}
 			sums += (comp_bio->bi_size + root->sectorsize - 1) /
 				root->sectorsize;
 
 			ret = btrfs_map_bio(root, READ, comp_bio,
 					    mirror_num, 0);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 
 			bio_put(comp_bio);
 
@@ -698,15 +697,15 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 	bio_get(comp_bio);
 
 	ret = btrfs_bio_wq_end_io(root->fs_info, comp_bio, 0);
-	BUG_ON(ret);
+	BUG_ON(ret); /* -ENOMEM */
 
 	if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) {
 		ret = btrfs_lookup_bio_sums(root, inode, comp_bio, sums);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 	}
 
 	ret = btrfs_map_bio(root, READ, comp_bio, mirror_num, 0);
-	BUG_ON(ret);
+	BUG_ON(ret); /* -ENOMEM */
 
 	bio_put(comp_bio);
 	return 0;
@@ -734,7 +733,7 @@ struct btrfs_compress_op *btrfs_compress_op[] = {
 	&btrfs_lzo_compress,
 };
 
-int __init btrfs_init_compress(void)
+void __init btrfs_init_compress(void)
 {
 	int i;
 
@@ -744,7 +743,6 @@ int __init btrfs_init_compress(void)
 		atomic_set(&comp_alloc_workspace[i], 0);
 		init_waitqueue_head(&comp_workspace_wait[i]);
 	}
-	return 0;
 }
 
 /*
diff --git a/fs/btrfs/compression.h b/fs/btrfs/compression.h
index a12059f4f0f..9afb0a62ae8 100644
--- a/fs/btrfs/compression.h
+++ b/fs/btrfs/compression.h
@@ -19,7 +19,7 @@
 #ifndef __BTRFS_COMPRESSION_
 #define __BTRFS_COMPRESSION_
 
-int btrfs_init_compress(void);
+void btrfs_init_compress(void);
 void btrfs_exit_compress(void);
 
 int btrfs_compress_pages(int type, struct address_space *mapping,
diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index 0639a555e16..e801f226d7e 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -36,7 +36,7 @@ static int balance_node_right(struct btrfs_trans_handle *trans,
 			      struct btrfs_root *root,
 			      struct extent_buffer *dst_buf,
 			      struct extent_buffer *src_buf);
-static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
+static void del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		   struct btrfs_path *path, int level, int slot);
 
 struct btrfs_path *btrfs_alloc_path(void)
@@ -156,10 +156,23 @@ struct extent_buffer *btrfs_root_node(struct btrfs_root *root)
 {
 	struct extent_buffer *eb;
 
-	rcu_read_lock();
-	eb = rcu_dereference(root->node);
-	extent_buffer_get(eb);
-	rcu_read_unlock();
+	while (1) {
+		rcu_read_lock();
+		eb = rcu_dereference(root->node);
+
+		/*
+		 * RCU really hurts here, we could free up the root node because
+		 * it was cow'ed but we may not get the new root node yet so do
+		 * the inc_not_zero dance and if it doesn't work then
+		 * synchronize_rcu and try again.
+		 */
+		if (atomic_inc_not_zero(&eb->refs)) {
+			rcu_read_unlock();
+			break;
+		}
+		rcu_read_unlock();
+		synchronize_rcu();
+	}
 	return eb;
 }
 
@@ -331,8 +344,13 @@ static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
 	if (btrfs_block_can_be_shared(root, buf)) {
 		ret = btrfs_lookup_extent_info(trans, root, buf->start,
 					       buf->len, &refs, &flags);
-		BUG_ON(ret);
-		BUG_ON(refs == 0);
+		if (ret)
+			return ret;
+		if (refs == 0) {
+			ret = -EROFS;
+			btrfs_std_error(root->fs_info, ret);
+			return ret;
+		}
 	} else {
 		refs = 1;
 		if (root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID ||
@@ -351,14 +369,14 @@ static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
 		     root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID) &&
 		    !(flags & BTRFS_BLOCK_FLAG_FULL_BACKREF)) {
 			ret = btrfs_inc_ref(trans, root, buf, 1, 1);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 
 			if (root->root_key.objectid ==
 			    BTRFS_TREE_RELOC_OBJECTID) {
 				ret = btrfs_dec_ref(trans, root, buf, 0, 1);
-				BUG_ON(ret);
+				BUG_ON(ret); /* -ENOMEM */
 				ret = btrfs_inc_ref(trans, root, cow, 1, 1);
-				BUG_ON(ret);
+				BUG_ON(ret); /* -ENOMEM */
 			}
 			new_flags |= BTRFS_BLOCK_FLAG_FULL_BACKREF;
 		} else {
@@ -368,14 +386,15 @@ static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
 				ret = btrfs_inc_ref(trans, root, cow, 1, 1);
 			else
 				ret = btrfs_inc_ref(trans, root, cow, 0, 1);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 		}
 		if (new_flags != 0) {
 			ret = btrfs_set_disk_extent_flags(trans, root,
 							  buf->start,
 							  buf->len,
 							  new_flags, 0);
-			BUG_ON(ret);
+			if (ret)
+				return ret;
 		}
 	} else {
 		if (flags & BTRFS_BLOCK_FLAG_FULL_BACKREF) {
@@ -384,9 +403,9 @@ static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
 				ret = btrfs_inc_ref(trans, root, cow, 1, 1);
 			else
 				ret = btrfs_inc_ref(trans, root, cow, 0, 1);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 			ret = btrfs_dec_ref(trans, root, buf, 1, 1);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 		}
 		clean_tree_block(trans, root, buf);
 		*last_ref = 1;
@@ -415,7 +434,7 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 {
 	struct btrfs_disk_key disk_key;
 	struct extent_buffer *cow;
-	int level;
+	int level, ret;
 	int last_ref = 0;
 	int unlock_orig = 0;
 	u64 parent_start;
@@ -467,7 +486,11 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 			    (unsigned long)btrfs_header_fsid(cow),
 			    BTRFS_FSID_SIZE);
 
-	update_ref_for_cow(trans, root, buf, cow, &last_ref);
+	ret = update_ref_for_cow(trans, root, buf, cow, &last_ref);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		return ret;
+	}
 
 	if (root->ref_cows)
 		btrfs_reloc_cow_block(trans, root, buf, cow);
@@ -504,7 +527,7 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 	}
 	if (unlock_orig)
 		btrfs_tree_unlock(buf);
-	free_extent_buffer(buf);
+	free_extent_buffer_stale(buf);
 	btrfs_mark_buffer_dirty(cow);
 	*cow_ret = cow;
 	return 0;
@@ -934,7 +957,12 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 
 		/* promote the child to a root */
 		child = read_node_slot(root, mid, 0);
-		BUG_ON(!child);
+		if (!child) {
+			ret = -EROFS;
+			btrfs_std_error(root->fs_info, ret);
+			goto enospc;
+		}
+
 		btrfs_tree_lock(child);
 		btrfs_set_lock_blocking(child);
 		ret = btrfs_cow_block(trans, root, child, mid, 0, &child);
@@ -959,7 +987,7 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 		root_sub_used(root, mid->len);
 		btrfs_free_tree_block(trans, root, mid, 0, 1, 0);
 		/* once for the root ptr */
-		free_extent_buffer(mid);
+		free_extent_buffer_stale(mid);
 		return 0;
 	}
 	if (btrfs_header_nritems(mid) >
@@ -1010,13 +1038,10 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 		if (btrfs_header_nritems(right) == 0) {
 			clean_tree_block(trans, root, right);
 			btrfs_tree_unlock(right);
-			wret = del_ptr(trans, root, path, level + 1, pslot +
-				       1);
-			if (wret)
-				ret = wret;
+			del_ptr(trans, root, path, level + 1, pslot + 1);
 			root_sub_used(root, right->len);
 			btrfs_free_tree_block(trans, root, right, 0, 1, 0);
-			free_extent_buffer(right);
+			free_extent_buffer_stale(right);
 			right = NULL;
 		} else {
 			struct btrfs_disk_key right_key;
@@ -1035,7 +1060,11 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 		 * otherwise we would have pulled some pointers from the
 		 * right
 		 */
-		BUG_ON(!left);
+		if (!left) {
+			ret = -EROFS;
+			btrfs_std_error(root->fs_info, ret);
+			goto enospc;
+		}
 		wret = balance_node_right(trans, root, mid, left);
 		if (wret < 0) {
 			ret = wret;
@@ -1051,12 +1080,10 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 	if (btrfs_header_nritems(mid) == 0) {
 		clean_tree_block(trans, root, mid);
 		btrfs_tree_unlock(mid);
-		wret = del_ptr(trans, root, path, level + 1, pslot);
-		if (wret)
-			ret = wret;
+		del_ptr(trans, root, path, level + 1, pslot);
 		root_sub_used(root, mid->len);
 		btrfs_free_tree_block(trans, root, mid, 0, 1, 0);
-		free_extent_buffer(mid);
+		free_extent_buffer_stale(mid);
 		mid = NULL;
 	} else {
 		/* update the parent key to reflect our changes */
@@ -1382,7 +1409,8 @@ static noinline int reada_for_balance(struct btrfs_root *root,
  * if lowest_unlock is 1, level 0 won't be unlocked
  */
 static noinline void unlock_up(struct btrfs_path *path, int level,
-			       int lowest_unlock)
+			       int lowest_unlock, int min_write_lock_level,
+			       int *write_lock_level)
 {
 	int i;
 	int skip_level = level;
@@ -1414,6 +1442,11 @@ static noinline void unlock_up(struct btrfs_path *path, int level,
 		if (i >= lowest_unlock && i > skip_level && path->locks[i]) {
 			btrfs_tree_unlock_rw(t, path->locks[i]);
 			path->locks[i] = 0;
+			if (write_lock_level &&
+			    i > min_write_lock_level &&
+			    i <= *write_lock_level) {
+				*write_lock_level = i - 1;
+			}
 		}
 	}
 }
@@ -1637,6 +1670,7 @@ int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root
 	/* everything at write_lock_level or lower must be write locked */
 	int write_lock_level = 0;
 	u8 lowest_level = 0;
+	int min_write_lock_level;
 
 	lowest_level = p->lowest_level;
 	WARN_ON(lowest_level && ins_len > 0);
@@ -1664,6 +1698,8 @@ int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root
 	if (cow && (p->keep_locks || p->lowest_level))
 		write_lock_level = BTRFS_MAX_LEVEL;
 
+	min_write_lock_level = write_lock_level;
+
 again:
 	/*
 	 * we try very hard to do read locks on the root
@@ -1795,7 +1831,8 @@ cow_done:
 				goto again;
 			}
 
-			unlock_up(p, level, lowest_unlock);
+			unlock_up(p, level, lowest_unlock,
+				  min_write_lock_level, &write_lock_level);
 
 			if (level == lowest_level) {
 				if (dec)
@@ -1857,7 +1894,8 @@ cow_done:
 				}
 			}
 			if (!p->search_for_split)
-				unlock_up(p, level, lowest_unlock);
+				unlock_up(p, level, lowest_unlock,
+					  min_write_lock_level, &write_lock_level);
 			goto done;
 		}
 	}
@@ -1881,15 +1919,12 @@ done:
  * fixing up pointers when a given leaf/node is not in slot 0 of the
  * higher levels
  *
- * If this fails to write a tree block, it returns -1, but continues
- * fixing up the blocks in ram so the tree is consistent.
  */
-static int fixup_low_keys(struct btrfs_trans_handle *trans,
-			  struct btrfs_root *root, struct btrfs_path *path,
-			  struct btrfs_disk_key *key, int level)
+static void fixup_low_keys(struct btrfs_trans_handle *trans,
+			   struct btrfs_root *root, struct btrfs_path *path,
+			   struct btrfs_disk_key *key, int level)
 {
 	int i;
-	int ret = 0;
 	struct extent_buffer *t;
 
 	for (i = level; i < BTRFS_MAX_LEVEL; i++) {
@@ -1902,7 +1937,6 @@ static int fixup_low_keys(struct btrfs_trans_handle *trans,
 		if (tslot != 0)
 			break;
 	}
-	return ret;
 }
 
 /*
@@ -1911,9 +1945,9 @@ static int fixup_low_keys(struct btrfs_trans_handle *trans,
  * This function isn't completely safe. It's the caller's responsibility
  * that the new key won't break the order
  */
-int btrfs_set_item_key_safe(struct btrfs_trans_handle *trans,
-			    struct btrfs_root *root, struct btrfs_path *path,
-			    struct btrfs_key *new_key)
+void btrfs_set_item_key_safe(struct btrfs_trans_handle *trans,
+			     struct btrfs_root *root, struct btrfs_path *path,
+			     struct btrfs_key *new_key)
 {
 	struct btrfs_disk_key disk_key;
 	struct extent_buffer *eb;
@@ -1923,13 +1957,11 @@ int btrfs_set_item_key_safe(struct btrfs_trans_handle *trans,
 	slot = path->slots[0];
 	if (slot > 0) {
 		btrfs_item_key(eb, &disk_key, slot - 1);
-		if (comp_keys(&disk_key, new_key) >= 0)
-			return -1;
+		BUG_ON(comp_keys(&disk_key, new_key) >= 0);
 	}
 	if (slot < btrfs_header_nritems(eb) - 1) {
 		btrfs_item_key(eb, &disk_key, slot + 1);
-		if (comp_keys(&disk_key, new_key) <= 0)
-			return -1;
+		BUG_ON(comp_keys(&disk_key, new_key) <= 0);
 	}
 
 	btrfs_cpu_key_to_disk(&disk_key, new_key);
@@ -1937,7 +1969,6 @@ int btrfs_set_item_key_safe(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(eb);
 	if (slot == 0)
 		fixup_low_keys(trans, root, path, &disk_key, 1);
-	return 0;
 }
 
 /*
@@ -2140,12 +2171,11 @@ static noinline int insert_new_root(struct btrfs_trans_handle *trans,
  *
  * slot and level indicate where you want the key to go, and
  * blocknr is the block the key points to.
- *
- * returns zero on success and < 0 on any error
  */
-static int insert_ptr(struct btrfs_trans_handle *trans, struct btrfs_root
-		      *root, struct btrfs_path *path, struct btrfs_disk_key
-		      *key, u64 bytenr, int slot, int level)
+static void insert_ptr(struct btrfs_trans_handle *trans,
+		       struct btrfs_root *root, struct btrfs_path *path,
+		       struct btrfs_disk_key *key, u64 bytenr,
+		       int slot, int level)
 {
 	struct extent_buffer *lower;
 	int nritems;
@@ -2155,8 +2185,7 @@ static int insert_ptr(struct btrfs_trans_handle *trans, struct btrfs_root
 	lower = path->nodes[level];
 	nritems = btrfs_header_nritems(lower);
 	BUG_ON(slot > nritems);
-	if (nritems == BTRFS_NODEPTRS_PER_BLOCK(root))
-		BUG();
+	BUG_ON(nritems == BTRFS_NODEPTRS_PER_BLOCK(root));
 	if (slot != nritems) {
 		memmove_extent_buffer(lower,
 			      btrfs_node_key_ptr_offset(slot + 1),
@@ -2169,7 +2198,6 @@ static int insert_ptr(struct btrfs_trans_handle *trans, struct btrfs_root
 	btrfs_set_node_ptr_generation(lower, slot, trans->transid);
 	btrfs_set_header_nritems(lower, nritems + 1);
 	btrfs_mark_buffer_dirty(lower);
-	return 0;
 }
 
 /*
@@ -2190,7 +2218,6 @@ static noinline int split_node(struct btrfs_trans_handle *trans,
 	struct btrfs_disk_key disk_key;
 	int mid;
 	int ret;
-	int wret;
 	u32 c_nritems;
 
 	c = path->nodes[level];
@@ -2247,11 +2274,8 @@ static noinline int split_node(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(c);
 	btrfs_mark_buffer_dirty(split);
 
-	wret = insert_ptr(trans, root, path, &disk_key, split->start,
-			  path->slots[level + 1] + 1,
-			  level + 1);
-	if (wret)
-		ret = wret;
+	insert_ptr(trans, root, path, &disk_key, split->start,
+		   path->slots[level + 1] + 1, level + 1);
 
 	if (path->slots[level] >= mid) {
 		path->slots[level] -= mid;
@@ -2320,6 +2344,7 @@ static noinline int __push_leaf_right(struct btrfs_trans_handle *trans,
 {
 	struct extent_buffer *left = path->nodes[0];
 	struct extent_buffer *upper = path->nodes[1];
+	struct btrfs_map_token token;
 	struct btrfs_disk_key disk_key;
 	int slot;
 	u32 i;
@@ -2331,6 +2356,8 @@ static noinline int __push_leaf_right(struct btrfs_trans_handle *trans,
 	u32 data_end;
 	u32 this_item_size;
 
+	btrfs_init_map_token(&token);
+
 	if (empty)
 		nr = 0;
 	else
@@ -2408,8 +2435,8 @@ static noinline int __push_leaf_right(struct btrfs_trans_handle *trans,
 	push_space = BTRFS_LEAF_DATA_SIZE(root);
 	for (i = 0; i < right_nritems; i++) {
 		item = btrfs_item_nr(right, i);
-		push_space -= btrfs_item_size(right, item);
-		btrfs_set_item_offset(right, item, push_space);
+		push_space -= btrfs_token_item_size(right, item, &token);
+		btrfs_set_token_item_offset(right, item, push_space, &token);
 	}
 
 	left_nritems -= push_items;
@@ -2537,9 +2564,11 @@ static noinline int __push_leaf_left(struct btrfs_trans_handle *trans,
 	u32 old_left_nritems;
 	u32 nr;
 	int ret = 0;
-	int wret;
 	u32 this_item_size;
 	u32 old_left_item_size;
+	struct btrfs_map_token token;
+
+	btrfs_init_map_token(&token);
 
 	if (empty)
 		nr = min(right_nritems, max_slot);
@@ -2600,9 +2629,10 @@ static noinline int __push_leaf_left(struct btrfs_trans_handle *trans,
 
 		item = btrfs_item_nr(left, i);
 
-		ioff = btrfs_item_offset(left, item);
-		btrfs_set_item_offset(left, item,
-		      ioff - (BTRFS_LEAF_DATA_SIZE(root) - old_left_item_size));
+		ioff = btrfs_token_item_offset(left, item, &token);
+		btrfs_set_token_item_offset(left, item,
+		      ioff - (BTRFS_LEAF_DATA_SIZE(root) - old_left_item_size),
+		      &token);
 	}
 	btrfs_set_header_nritems(left, old_left_nritems + push_items);
 
@@ -2632,8 +2662,9 @@ static noinline int __push_leaf_left(struct btrfs_trans_handle *trans,
 	for (i = 0; i < right_nritems; i++) {
 		item = btrfs_item_nr(right, i);
 
-		push_space = push_space - btrfs_item_size(right, item);
-		btrfs_set_item_offset(right, item, push_space);
+		push_space = push_space - btrfs_token_item_size(right,
+								item, &token);
+		btrfs_set_token_item_offset(right, item, push_space, &token);
 	}
 
 	btrfs_mark_buffer_dirty(left);
@@ -2643,9 +2674,7 @@ static noinline int __push_leaf_left(struct btrfs_trans_handle *trans,
 		clean_tree_block(trans, root, right);
 
 	btrfs_item_key(right, &disk_key, 0);
-	wret = fixup_low_keys(trans, root, path, &disk_key, 1);
-	if (wret)
-		ret = wret;
+	fixup_low_keys(trans, root, path, &disk_key, 1);
 
 	/* then fixup the leaf pointer in the path */
 	if (path->slots[0] < push_items) {
@@ -2716,7 +2745,8 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root
 			      path->nodes[1], slot - 1, &left);
 	if (ret) {
 		/* we hit -ENOSPC, but it isn't fatal here */
-		ret = 1;
+		if (ret == -ENOSPC)
+			ret = 1;
 		goto out;
 	}
 
@@ -2738,22 +2768,21 @@ out:
 /*
  * split the path's leaf in two, making sure there is at least data_size
  * available for the resulting leaf level of the path.
- *
- * returns 0 if all went well and < 0 on failure.
  */
-static noinline int copy_for_split(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root,
-			       struct btrfs_path *path,
-			       struct extent_buffer *l,
-			       struct extent_buffer *right,
-			       int slot, int mid, int nritems)
+static noinline void copy_for_split(struct btrfs_trans_handle *trans,
+				    struct btrfs_root *root,
+				    struct btrfs_path *path,
+				    struct extent_buffer *l,
+				    struct extent_buffer *right,
+				    int slot, int mid, int nritems)
 {
 	int data_copy_size;
 	int rt_data_off;
 	int i;
-	int ret = 0;
-	int wret;
 	struct btrfs_disk_key disk_key;
+	struct btrfs_map_token token;
+
+	btrfs_init_map_token(&token);
 
 	nritems = nritems - mid;
 	btrfs_set_header_nritems(right, nritems);
@@ -2775,17 +2804,15 @@ static noinline int copy_for_split(struct btrfs_trans_handle *trans,
 		struct btrfs_item *item = btrfs_item_nr(right, i);
 		u32 ioff;
 
-		ioff = btrfs_item_offset(right, item);
-		btrfs_set_item_offset(right, item, ioff + rt_data_off);
+		ioff = btrfs_token_item_offset(right, item, &token);
+		btrfs_set_token_item_offset(right, item,
+					    ioff + rt_data_off, &token);
 	}
 
 	btrfs_set_header_nritems(l, mid);
-	ret = 0;
 	btrfs_item_key(right, &disk_key, 0);
-	wret = insert_ptr(trans, root, path, &disk_key, right->start,
-			  path->slots[1] + 1, 1);
-	if (wret)
-		ret = wret;
+	insert_ptr(trans, root, path, &disk_key, right->start,
+		   path->slots[1] + 1, 1);
 
 	btrfs_mark_buffer_dirty(right);
 	btrfs_mark_buffer_dirty(l);
@@ -2803,8 +2830,6 @@ static noinline int copy_for_split(struct btrfs_trans_handle *trans,
 	}
 
 	BUG_ON(path->slots[0] < 0);
-
-	return ret;
 }
 
 /*
@@ -2993,12 +3018,8 @@ again:
 	if (split == 0) {
 		if (mid <= slot) {
 			btrfs_set_header_nritems(right, 0);
-			wret = insert_ptr(trans, root, path,
-					  &disk_key, right->start,
-					  path->slots[1] + 1, 1);
-			if (wret)
-				ret = wret;
-
+			insert_ptr(trans, root, path, &disk_key, right->start,
+				   path->slots[1] + 1, 1);
 			btrfs_tree_unlock(path->nodes[0]);
 			free_extent_buffer(path->nodes[0]);
 			path->nodes[0] = right;
@@ -3006,29 +3027,21 @@ again:
 			path->slots[1] += 1;
 		} else {
 			btrfs_set_header_nritems(right, 0);
-			wret = insert_ptr(trans, root, path,
-					  &disk_key,
-					  right->start,
+			insert_ptr(trans, root, path, &disk_key, right->start,
 					  path->slots[1], 1);
-			if (wret)
-				ret = wret;
 			btrfs_tree_unlock(path->nodes[0]);
 			free_extent_buffer(path->nodes[0]);
 			path->nodes[0] = right;
 			path->slots[0] = 0;
-			if (path->slots[1] == 0) {
-				wret = fixup_low_keys(trans, root,
-						path, &disk_key, 1);
-				if (wret)
-					ret = wret;
-			}
+			if (path->slots[1] == 0)
+				fixup_low_keys(trans, root, path,
+					       &disk_key, 1);
 		}
 		btrfs_mark_buffer_dirty(right);
 		return ret;
 	}
 
-	ret = copy_for_split(trans, root, path, l, right, slot, mid, nritems);
-	BUG_ON(ret);
+	copy_for_split(trans, root, path, l, right, slot, mid, nritems);
 
 	if (split == 2) {
 		BUG_ON(num_doubles != 0);
@@ -3036,7 +3049,7 @@ again:
 		goto again;
 	}
 
-	return ret;
+	return 0;
 
 push_for_double:
 	push_for_double_split(trans, root, path, data_size);
@@ -3238,11 +3251,9 @@ int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
 		return ret;
 
 	path->slots[0]++;
-	ret = setup_items_for_insert(trans, root, path, new_key, &item_size,
-				     item_size, item_size +
-				     sizeof(struct btrfs_item), 1);
-	BUG_ON(ret);
-
+	setup_items_for_insert(trans, root, path, new_key, &item_size,
+			       item_size, item_size +
+			       sizeof(struct btrfs_item), 1);
 	leaf = path->nodes[0];
 	memcpy_extent_buffer(leaf,
 			     btrfs_item_ptr_offset(leaf, path->slots[0]),
@@ -3257,10 +3268,10 @@ int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
  * off the end of the item or if we shift the item to chop bytes off
  * the front.
  */
-int btrfs_truncate_item(struct btrfs_trans_handle *trans,
-			struct btrfs_root *root,
-			struct btrfs_path *path,
-			u32 new_size, int from_end)
+void btrfs_truncate_item(struct btrfs_trans_handle *trans,
+			 struct btrfs_root *root,
+			 struct btrfs_path *path,
+			 u32 new_size, int from_end)
 {
 	int slot;
 	struct extent_buffer *leaf;
@@ -3271,13 +3282,16 @@ int btrfs_truncate_item(struct btrfs_trans_handle *trans,
 	unsigned int old_size;
 	unsigned int size_diff;
 	int i;
+	struct btrfs_map_token token;
+
+	btrfs_init_map_token(&token);
 
 	leaf = path->nodes[0];
 	slot = path->slots[0];
 
 	old_size = btrfs_item_size_nr(leaf, slot);
 	if (old_size == new_size)
-		return 0;
+		return;
 
 	nritems = btrfs_header_nritems(leaf);
 	data_end = leaf_data_end(root, leaf);
@@ -3297,8 +3311,9 @@ int btrfs_truncate_item(struct btrfs_trans_handle *trans,
 		u32 ioff;
 		item = btrfs_item_nr(leaf, i);
 
-		ioff = btrfs_item_offset(leaf, item);
-		btrfs_set_item_offset(leaf, item, ioff + size_diff);
+		ioff = btrfs_token_item_offset(leaf, item, &token);
+		btrfs_set_token_item_offset(leaf, item,
+					    ioff + size_diff, &token);
 	}
 
 	/* shift the data */
@@ -3350,15 +3365,14 @@ int btrfs_truncate_item(struct btrfs_trans_handle *trans,
 		btrfs_print_leaf(root, leaf);
 		BUG();
 	}
-	return 0;
 }
 
 /*
  * make the item pointed to by the path bigger, data_size is the new size.
  */
-int btrfs_extend_item(struct btrfs_trans_handle *trans,
-		      struct btrfs_root *root, struct btrfs_path *path,
-		      u32 data_size)
+void btrfs_extend_item(struct btrfs_trans_handle *trans,
+		       struct btrfs_root *root, struct btrfs_path *path,
+		       u32 data_size)
 {
 	int slot;
 	struct extent_buffer *leaf;
@@ -3368,6 +3382,9 @@ int btrfs_extend_item(struct btrfs_trans_handle *trans,
 	unsigned int old_data;
 	unsigned int old_size;
 	int i;
+	struct btrfs_map_token token;
+
+	btrfs_init_map_token(&token);
 
 	leaf = path->nodes[0];
 
@@ -3397,8 +3414,9 @@ int btrfs_extend_item(struct btrfs_trans_handle *trans,
 		u32 ioff;
 		item = btrfs_item_nr(leaf, i);
 
-		ioff = btrfs_item_offset(leaf, item);
-		btrfs_set_item_offset(leaf, item, ioff - data_size);
+		ioff = btrfs_token_item_offset(leaf, item, &token);
+		btrfs_set_token_item_offset(leaf, item,
+					    ioff - data_size, &token);
 	}
 
 	/* shift the data */
@@ -3416,7 +3434,6 @@ int btrfs_extend_item(struct btrfs_trans_handle *trans,
 		btrfs_print_leaf(root, leaf);
 		BUG();
 	}
-	return 0;
 }
 
 /*
@@ -3441,6 +3458,9 @@ int btrfs_insert_some_items(struct btrfs_trans_handle *trans,
 	unsigned int data_end;
 	struct btrfs_disk_key disk_key;
 	struct btrfs_key found_key;
+	struct btrfs_map_token token;
+
+	btrfs_init_map_token(&token);
 
 	for (i = 0; i < nr; i++) {
 		if (total_size + data_size[i] + sizeof(struct btrfs_item) >
@@ -3506,8 +3526,9 @@ int btrfs_insert_some_items(struct btrfs_trans_handle *trans,
 			u32 ioff;
 
 			item = btrfs_item_nr(leaf, i);
-			ioff = btrfs_item_offset(leaf, item);
-			btrfs_set_item_offset(leaf, item, ioff - total_data);
+			ioff = btrfs_token_item_offset(leaf, item, &token);
+			btrfs_set_token_item_offset(leaf, item,
+						    ioff - total_data, &token);
 		}
 		/* shift the items */
 		memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr),
@@ -3534,9 +3555,10 @@ int btrfs_insert_some_items(struct btrfs_trans_handle *trans,
 		btrfs_cpu_key_to_disk(&disk_key, cpu_key + i);
 		btrfs_set_item_key(leaf, &disk_key, slot + i);
 		item = btrfs_item_nr(leaf, slot + i);
-		btrfs_set_item_offset(leaf, item, data_end - data_size[i]);
+		btrfs_set_token_item_offset(leaf, item,
+					    data_end - data_size[i], &token);
 		data_end -= data_size[i];
-		btrfs_set_item_size(leaf, item, data_size[i]);
+		btrfs_set_token_item_size(leaf, item, data_size[i], &token);
 	}
 	btrfs_set_header_nritems(leaf, nritems + nr);
 	btrfs_mark_buffer_dirty(leaf);
@@ -3544,7 +3566,7 @@ int btrfs_insert_some_items(struct btrfs_trans_handle *trans,
 	ret = 0;
 	if (slot == 0) {
 		btrfs_cpu_key_to_disk(&disk_key, cpu_key);
-		ret = fixup_low_keys(trans, root, path, &disk_key, 1);
+		fixup_low_keys(trans, root, path, &disk_key, 1);
 	}
 
 	if (btrfs_leaf_free_space(root, leaf) < 0) {
@@ -3562,19 +3584,21 @@ out:
  * to save stack depth by doing the bulk of the work in a function
  * that doesn't call btrfs_search_slot
  */
-int setup_items_for_insert(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root, struct btrfs_path *path,
-			   struct btrfs_key *cpu_key, u32 *data_size,
-			   u32 total_data, u32 total_size, int nr)
+void setup_items_for_insert(struct btrfs_trans_handle *trans,
+			    struct btrfs_root *root, struct btrfs_path *path,
+			    struct btrfs_key *cpu_key, u32 *data_size,
+			    u32 total_data, u32 total_size, int nr)
 {
 	struct btrfs_item *item;
 	int i;
 	u32 nritems;
 	unsigned int data_end;
 	struct btrfs_disk_key disk_key;
-	int ret;
 	struct extent_buffer *leaf;
 	int slot;
+	struct btrfs_map_token token;
+
+	btrfs_init_map_token(&token);
 
 	leaf = path->nodes[0];
 	slot = path->slots[0];
@@ -3606,8 +3630,9 @@ int setup_items_for_insert(struct btrfs_trans_handle *trans,
 			u32 ioff;
 
 			item = btrfs_item_nr(leaf, i);
-			ioff = btrfs_item_offset(leaf, item);
-			btrfs_set_item_offset(leaf, item, ioff - total_data);
+			ioff = btrfs_token_item_offset(leaf, item, &token);
+			btrfs_set_token_item_offset(leaf, item,
+						    ioff - total_data, &token);
 		}
 		/* shift the items */
 		memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot + nr),
@@ -3626,17 +3651,17 @@ int setup_items_for_insert(struct btrfs_trans_handle *trans,
 		btrfs_cpu_key_to_disk(&disk_key, cpu_key + i);
 		btrfs_set_item_key(leaf, &disk_key, slot + i);
 		item = btrfs_item_nr(leaf, slot + i);
-		btrfs_set_item_offset(leaf, item, data_end - data_size[i]);
+		btrfs_set_token_item_offset(leaf, item,
+					    data_end - data_size[i], &token);
 		data_end -= data_size[i];
-		btrfs_set_item_size(leaf, item, data_size[i]);
+		btrfs_set_token_item_size(leaf, item, data_size[i], &token);
 	}
 
 	btrfs_set_header_nritems(leaf, nritems + nr);
 
-	ret = 0;
 	if (slot == 0) {
 		btrfs_cpu_key_to_disk(&disk_key, cpu_key);
-		ret = fixup_low_keys(trans, root, path, &disk_key, 1);
+		fixup_low_keys(trans, root, path, &disk_key, 1);
 	}
 	btrfs_unlock_up_safe(path, 1);
 	btrfs_mark_buffer_dirty(leaf);
@@ -3645,7 +3670,6 @@ int setup_items_for_insert(struct btrfs_trans_handle *trans,
 		btrfs_print_leaf(root, leaf);
 		BUG();
 	}
-	return ret;
 }
 
 /*
@@ -3672,16 +3696,14 @@ int btrfs_insert_empty_items(struct btrfs_trans_handle *trans,
 	if (ret == 0)
 		return -EEXIST;
 	if (ret < 0)
-		goto out;
+		return ret;
 
 	slot = path->slots[0];
 	BUG_ON(slot < 0);
 
-	ret = setup_items_for_insert(trans, root, path, cpu_key, data_size,
+	setup_items_for_insert(trans, root, path, cpu_key, data_size,
 			       total_data, total_size, nr);
-
-out:
-	return ret;
+	return 0;
 }
 
 /*
@@ -3717,13 +3739,11 @@ int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root
  * the tree should have been previously balanced so the deletion does not
  * empty a node.
  */
-static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
-		   struct btrfs_path *path, int level, int slot)
+static void del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
+		    struct btrfs_path *path, int level, int slot)
 {
 	struct extent_buffer *parent = path->nodes[level];
 	u32 nritems;
-	int ret = 0;
-	int wret;
 
 	nritems = btrfs_header_nritems(parent);
 	if (slot != nritems - 1) {
@@ -3743,12 +3763,9 @@ static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		struct btrfs_disk_key disk_key;
 
 		btrfs_node_key(parent, &disk_key, 0);
-		wret = fixup_low_keys(trans, root, path, &disk_key, level + 1);
-		if (wret)
-			ret = wret;
+		fixup_low_keys(trans, root, path, &disk_key, level + 1);
 	}
 	btrfs_mark_buffer_dirty(parent);
-	return ret;
 }
 
 /*
@@ -3761,17 +3778,13 @@ static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
  * The path must have already been setup for deleting the leaf, including
  * all the proper balancing.  path->nodes[1] must be locked.
  */
-static noinline int btrfs_del_leaf(struct btrfs_trans_handle *trans,
-				   struct btrfs_root *root,
-				   struct btrfs_path *path,
-				   struct extent_buffer *leaf)
+static noinline void btrfs_del_leaf(struct btrfs_trans_handle *trans,
+				    struct btrfs_root *root,
+				    struct btrfs_path *path,
+				    struct extent_buffer *leaf)
 {
-	int ret;
-
 	WARN_ON(btrfs_header_generation(leaf) != trans->transid);
-	ret = del_ptr(trans, root, path, 1, path->slots[1]);
-	if (ret)
-		return ret;
+	del_ptr(trans, root, path, 1, path->slots[1]);
 
 	/*
 	 * btrfs_free_extent is expensive, we want to make sure we
@@ -3781,8 +3794,9 @@ static noinline int btrfs_del_leaf(struct btrfs_trans_handle *trans,
 
 	root_sub_used(root, leaf->len);
 
+	extent_buffer_get(leaf);
 	btrfs_free_tree_block(trans, root, leaf, 0, 1, 0);
-	return 0;
+	free_extent_buffer_stale(leaf);
 }
 /*
  * delete the item at the leaf level in path.  If that empties
@@ -3799,6 +3813,9 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 	int wret;
 	int i;
 	u32 nritems;
+	struct btrfs_map_token token;
+
+	btrfs_init_map_token(&token);
 
 	leaf = path->nodes[0];
 	last_off = btrfs_item_offset_nr(leaf, slot + nr - 1);
@@ -3820,8 +3837,9 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 			u32 ioff;
 
 			item = btrfs_item_nr(leaf, i);
-			ioff = btrfs_item_offset(leaf, item);
-			btrfs_set_item_offset(leaf, item, ioff + dsize);
+			ioff = btrfs_token_item_offset(leaf, item, &token);
+			btrfs_set_token_item_offset(leaf, item,
+						    ioff + dsize, &token);
 		}
 
 		memmove_extent_buffer(leaf, btrfs_item_nr_offset(slot),
@@ -3839,8 +3857,7 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		} else {
 			btrfs_set_path_blocking(path);
 			clean_tree_block(trans, root, leaf);
-			ret = btrfs_del_leaf(trans, root, path, leaf);
-			BUG_ON(ret);
+			btrfs_del_leaf(trans, root, path, leaf);
 		}
 	} else {
 		int used = leaf_space_used(leaf, 0, nritems);
@@ -3848,10 +3865,7 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 			struct btrfs_disk_key disk_key;
 
 			btrfs_item_key(leaf, &disk_key, 0);
-			wret = fixup_low_keys(trans, root, path,
-					      &disk_key, 1);
-			if (wret)
-				ret = wret;
+			fixup_low_keys(trans, root, path, &disk_key, 1);
 		}
 
 		/* delete the leaf if it is mostly empty */
@@ -3879,9 +3893,9 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 
 			if (btrfs_header_nritems(leaf) == 0) {
 				path->slots[1] = slot;
-				ret = btrfs_del_leaf(trans, root, path, leaf);
-				BUG_ON(ret);
+				btrfs_del_leaf(trans, root, path, leaf);
 				free_extent_buffer(leaf);
+				ret = 0;
 			} else {
 				/* if we're still in the path, make sure
 				 * we're dirty.  Otherwise, one of the
@@ -4059,18 +4073,18 @@ find_next_key:
 		path->slots[level] = slot;
 		if (level == path->lowest_level) {
 			ret = 0;
-			unlock_up(path, level, 1);
+			unlock_up(path, level, 1, 0, NULL);
 			goto out;
 		}
 		btrfs_set_path_blocking(path);
 		cur = read_node_slot(root, cur, slot);
-		BUG_ON(!cur);
+		BUG_ON(!cur); /* -ENOMEM */
 
 		btrfs_tree_read_lock(cur);
 
 		path->locks[level - 1] = BTRFS_READ_LOCK;
 		path->nodes[level - 1] = cur;
-		unlock_up(path, level, 1);
+		unlock_up(path, level, 1, 0, NULL);
 		btrfs_clear_path_blocking(path, NULL, 0);
 	}
 out:
@@ -4306,7 +4320,7 @@ again:
 	}
 	ret = 0;
 done:
-	unlock_up(path, 0, 1);
+	unlock_up(path, 0, 1, 0, NULL);
 	path->leave_spinning = old_spinning;
 	if (!old_spinning)
 		btrfs_set_path_blocking(path);
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 80b6486fd5e..5b8ef8eb352 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -48,6 +48,8 @@ struct btrfs_ordered_sum;
 
 #define BTRFS_MAGIC "_BHRfS_M"
 
+#define BTRFS_MAX_MIRRORS 2
+
 #define BTRFS_MAX_LEVEL 8
 
 #define BTRFS_COMPAT_EXTENT_TREE_V0
@@ -138,6 +140,12 @@ struct btrfs_ordered_sum;
 #define BTRFS_EMPTY_SUBVOL_DIR_OBJECTID 2
 
 /*
+ * the max metadata block size.  This limit is somewhat artificial,
+ * but the memmove costs go through the roof for larger blocks.
+ */
+#define BTRFS_MAX_METADATA_BLOCKSIZE 65536
+
+/*
  * we can actually store much bigger names, but lets not confuse the rest
  * of linux
  */
@@ -461,6 +469,19 @@ struct btrfs_super_block {
 #define BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL	(1ULL << 1)
 #define BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS	(1ULL << 2)
 #define BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO	(1ULL << 3)
+/*
+ * some patches floated around with a second compression method
+ * lets save that incompat here for when they do get in
+ * Note we don't actually support it, we're just reserving the
+ * number
+ */
+#define BTRFS_FEATURE_INCOMPAT_COMPRESS_LZOv2	(1ULL << 4)
+
+/*
+ * older kernels tried to do bigger metadata blocks, but the
+ * code was pretty buggy.  Lets not let them try anymore.
+ */
+#define BTRFS_FEATURE_INCOMPAT_BIG_METADATA	(1ULL << 5)
 
 #define BTRFS_FEATURE_COMPAT_SUPP		0ULL
 #define BTRFS_FEATURE_COMPAT_RO_SUPP		0ULL
@@ -468,6 +489,7 @@ struct btrfs_super_block {
 	(BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF |		\
 	 BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL |	\
 	 BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS |		\
+	 BTRFS_FEATURE_INCOMPAT_BIG_METADATA |		\
 	 BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO)
 
 /*
@@ -829,6 +851,21 @@ struct btrfs_csum_item {
  */
 #define BTRFS_AVAIL_ALLOC_BIT_SINGLE	(1ULL << 48)
 
+#define BTRFS_EXTENDED_PROFILE_MASK	(BTRFS_BLOCK_GROUP_PROFILE_MASK | \
+					 BTRFS_AVAIL_ALLOC_BIT_SINGLE)
+
+static inline u64 chunk_to_extended(u64 flags)
+{
+	if ((flags & BTRFS_BLOCK_GROUP_PROFILE_MASK) == 0)
+		flags |= BTRFS_AVAIL_ALLOC_BIT_SINGLE;
+
+	return flags;
+}
+static inline u64 extended_to_chunk(u64 flags)
+{
+	return flags & ~BTRFS_AVAIL_ALLOC_BIT_SINGLE;
+}
+
 struct btrfs_block_group_item {
 	__le64 used;
 	__le64 chunk_objectid;
@@ -1503,6 +1540,7 @@ struct btrfs_ioctl_defrag_range_args {
 #define BTRFS_MOUNT_SKIP_BALANCE	(1 << 19)
 #define BTRFS_MOUNT_CHECK_INTEGRITY	(1 << 20)
 #define BTRFS_MOUNT_CHECK_INTEGRITY_INCLUDING_EXTENT_DATA (1 << 21)
+#define BTRFS_MOUNT_PANIC_ON_FATAL_ERROR	(1 << 22)
 
 #define btrfs_clear_opt(o, opt)		((o) &= ~BTRFS_MOUNT_##opt)
 #define btrfs_set_opt(o, opt)		((o) |= BTRFS_MOUNT_##opt)
@@ -1526,6 +1564,17 @@ struct btrfs_ioctl_defrag_range_args {
 
 #define BTRFS_INODE_ROOT_ITEM_INIT	(1 << 31)
 
+struct btrfs_map_token {
+	struct extent_buffer *eb;
+	char *kaddr;
+	unsigned long offset;
+};
+
+static inline void btrfs_init_map_token (struct btrfs_map_token *token)
+{
+	memset(token, 0, sizeof(*token));
+}
+
 /* some macros to generate set/get funcs for the struct fields.  This
  * assumes there is a lefoo_to_cpu for every type, so lets make a simple
  * one for u8:
@@ -1549,20 +1598,22 @@ struct btrfs_ioctl_defrag_range_args {
 #ifndef BTRFS_SETGET_FUNCS
 #define BTRFS_SETGET_FUNCS(name, type, member, bits)			\
 u##bits btrfs_##name(struct extent_buffer *eb, type *s);		\
+u##bits btrfs_token_##name(struct extent_buffer *eb, type *s, struct btrfs_map_token *token);		\
+void btrfs_set_token_##name(struct extent_buffer *eb, type *s, u##bits val, struct btrfs_map_token *token);\
 void btrfs_set_##name(struct extent_buffer *eb, type *s, u##bits val);
 #endif
 
 #define BTRFS_SETGET_HEADER_FUNCS(name, type, member, bits)		\
 static inline u##bits btrfs_##name(struct extent_buffer *eb)		\
 {									\
-	type *p = page_address(eb->first_page);				\
+	type *p = page_address(eb->pages[0]);				\
 	u##bits res = le##bits##_to_cpu(p->member);			\
 	return res;							\
 }									\
 static inline void btrfs_set_##name(struct extent_buffer *eb,		\
 				    u##bits val)			\
 {									\
-	type *p = page_address(eb->first_page);				\
+	type *p = page_address(eb->pages[0]);				\
 	p->member = cpu_to_le##bits(val);				\
 }
 
@@ -2466,8 +2517,7 @@ int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
 				  struct btrfs_root *root,
 				  u64 num_bytes, u64 min_alloc_size,
 				  u64 empty_size, u64 hint_byte,
-				  u64 search_end, struct btrfs_key *ins,
-				  u64 data);
+				  struct btrfs_key *ins, u64 data);
 int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		  struct extent_buffer *buf, int full_backref, int for_cow);
 int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
@@ -2484,8 +2534,8 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans,
 int btrfs_free_reserved_extent(struct btrfs_root *root, u64 start, u64 len);
 int btrfs_free_and_pin_reserved_extent(struct btrfs_root *root,
 				       u64 start, u64 len);
-int btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root);
+void btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans,
+				 struct btrfs_root *root);
 int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root);
 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
@@ -2548,8 +2598,8 @@ void btrfs_block_rsv_release(struct btrfs_root *root,
 			     u64 num_bytes);
 int btrfs_set_block_group_ro(struct btrfs_root *root,
 			     struct btrfs_block_group_cache *cache);
-int btrfs_set_block_group_rw(struct btrfs_root *root,
-			     struct btrfs_block_group_cache *cache);
+void btrfs_set_block_group_rw(struct btrfs_root *root,
+			      struct btrfs_block_group_cache *cache);
 void btrfs_put_block_group_cache(struct btrfs_fs_info *info);
 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
 int btrfs_error_unpin_extent_range(struct btrfs_root *root,
@@ -2568,9 +2618,9 @@ int btrfs_comp_cpu_keys(struct btrfs_key *k1, struct btrfs_key *k2);
 int btrfs_previous_item(struct btrfs_root *root,
 			struct btrfs_path *path, u64 min_objectid,
 			int type);
-int btrfs_set_item_key_safe(struct btrfs_trans_handle *trans,
-			    struct btrfs_root *root, struct btrfs_path *path,
-			    struct btrfs_key *new_key);
+void btrfs_set_item_key_safe(struct btrfs_trans_handle *trans,
+			     struct btrfs_root *root, struct btrfs_path *path,
+			     struct btrfs_key *new_key);
 struct extent_buffer *btrfs_root_node(struct btrfs_root *root);
 struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root);
 int btrfs_find_next_key(struct btrfs_root *root, struct btrfs_path *path,
@@ -2590,12 +2640,13 @@ int btrfs_copy_root(struct btrfs_trans_handle *trans,
 		      struct extent_buffer **cow_ret, u64 new_root_objectid);
 int btrfs_block_can_be_shared(struct btrfs_root *root,
 			      struct extent_buffer *buf);
-int btrfs_extend_item(struct btrfs_trans_handle *trans, struct btrfs_root
-		      *root, struct btrfs_path *path, u32 data_size);
-int btrfs_truncate_item(struct btrfs_trans_handle *trans,
-			struct btrfs_root *root,
-			struct btrfs_path *path,
-			u32 new_size, int from_end);
+void btrfs_extend_item(struct btrfs_trans_handle *trans,
+		       struct btrfs_root *root, struct btrfs_path *path,
+		       u32 data_size);
+void btrfs_truncate_item(struct btrfs_trans_handle *trans,
+			 struct btrfs_root *root,
+			 struct btrfs_path *path,
+			 u32 new_size, int from_end);
 int btrfs_split_item(struct btrfs_trans_handle *trans,
 		     struct btrfs_root *root,
 		     struct btrfs_path *path,
@@ -2629,10 +2680,10 @@ static inline int btrfs_del_item(struct btrfs_trans_handle *trans,
 	return btrfs_del_items(trans, root, path, path->slots[0], 1);
 }
 
-int setup_items_for_insert(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root, struct btrfs_path *path,
-			   struct btrfs_key *cpu_key, u32 *data_size,
-			   u32 total_data, u32 total_size, int nr);
+void setup_items_for_insert(struct btrfs_trans_handle *trans,
+			    struct btrfs_root *root, struct btrfs_path *path,
+			    struct btrfs_key *cpu_key, u32 *data_size,
+			    u32 total_data, u32 total_size, int nr);
 int btrfs_insert_item(struct btrfs_trans_handle *trans, struct btrfs_root
 		      *root, struct btrfs_key *key, void *data, u32 data_size);
 int btrfs_insert_empty_items(struct btrfs_trans_handle *trans,
@@ -2659,9 +2710,9 @@ static inline int btrfs_next_item(struct btrfs_root *root, struct btrfs_path *p)
 }
 int btrfs_prev_leaf(struct btrfs_root *root, struct btrfs_path *path);
 int btrfs_leaf_free_space(struct btrfs_root *root, struct extent_buffer *leaf);
-void btrfs_drop_snapshot(struct btrfs_root *root,
-			 struct btrfs_block_rsv *block_rsv, int update_ref,
-			 int for_reloc);
+int __must_check btrfs_drop_snapshot(struct btrfs_root *root,
+				     struct btrfs_block_rsv *block_rsv,
+				     int update_ref, int for_reloc);
 int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
 			struct btrfs_root *root,
 			struct extent_buffer *node,
@@ -2687,24 +2738,6 @@ static inline void free_fs_info(struct btrfs_fs_info *fs_info)
 	kfree(fs_info->super_for_commit);
 	kfree(fs_info);
 }
-/**
- * profile_is_valid - tests whether a given profile is valid and reduced
- * @flags: profile to validate
- * @extended: if true @flags is treated as an extended profile
- */
-static inline int profile_is_valid(u64 flags, int extended)
-{
-	u64 mask = ~BTRFS_BLOCK_GROUP_PROFILE_MASK;
-
-	flags &= ~BTRFS_BLOCK_GROUP_TYPE_MASK;
-	if (extended)
-		mask &= ~BTRFS_AVAIL_ALLOC_BIT_SINGLE;
-
-	if (flags & mask)
-		return 0;
-	/* true if zero or exactly one bit set */
-	return (flags & (~flags + 1)) == flags;
-}
 
 /* root-item.c */
 int btrfs_find_root_ref(struct btrfs_root *tree_root,
@@ -2723,9 +2756,10 @@ int btrfs_del_root(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 int btrfs_insert_root(struct btrfs_trans_handle *trans, struct btrfs_root
 		      *root, struct btrfs_key *key, struct btrfs_root_item
 		      *item);
-int btrfs_update_root(struct btrfs_trans_handle *trans, struct btrfs_root
-		      *root, struct btrfs_key *key, struct btrfs_root_item
-		      *item);
+int __must_check btrfs_update_root(struct btrfs_trans_handle *trans,
+				   struct btrfs_root *root,
+				   struct btrfs_key *key,
+				   struct btrfs_root_item *item);
 int btrfs_find_last_root(struct btrfs_root *root, u64 objectid, struct
 			 btrfs_root_item *item, struct btrfs_key *key);
 int btrfs_find_dead_roots(struct btrfs_root *root, u64 objectid);
@@ -2909,7 +2943,7 @@ int btrfs_orphan_cleanup(struct btrfs_root *root);
 void btrfs_orphan_commit_root(struct btrfs_trans_handle *trans,
 			      struct btrfs_root *root);
 int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size);
-int btrfs_invalidate_inodes(struct btrfs_root *root);
+void btrfs_invalidate_inodes(struct btrfs_root *root);
 void btrfs_add_delayed_iput(struct inode *inode);
 void btrfs_run_delayed_iputs(struct btrfs_root *root);
 int btrfs_prealloc_file_range(struct inode *inode, int mode,
@@ -2961,13 +2995,41 @@ ssize_t btrfs_listxattr(struct dentry *dentry, char *buffer, size_t size);
 /* super.c */
 int btrfs_parse_options(struct btrfs_root *root, char *options);
 int btrfs_sync_fs(struct super_block *sb, int wait);
+void btrfs_printk(struct btrfs_fs_info *fs_info, const char *fmt, ...);
 void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
-		     unsigned int line, int errno);
+		     unsigned int line, int errno, const char *fmt, ...);
+
+void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
+			       struct btrfs_root *root, const char *function,
+			       unsigned int line, int errno);
+
+#define btrfs_abort_transaction(trans, root, errno)		\
+do {								\
+	__btrfs_abort_transaction(trans, root, __func__,	\
+				  __LINE__, errno);		\
+} while (0)
 
 #define btrfs_std_error(fs_info, errno)				\
 do {								\
 	if ((errno))						\
-		__btrfs_std_error((fs_info), __func__, __LINE__, (errno));\
+		__btrfs_std_error((fs_info), __func__,		\
+				   __LINE__, (errno), NULL);	\
+} while (0)
+
+#define btrfs_error(fs_info, errno, fmt, args...)		\
+do {								\
+	__btrfs_std_error((fs_info), __func__, __LINE__,	\
+			  (errno), fmt, ##args);		\
+} while (0)
+
+void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function,
+		   unsigned int line, int errno, const char *fmt, ...);
+
+#define btrfs_panic(fs_info, errno, fmt, args...)			\
+do {									\
+	struct btrfs_fs_info *_i = (fs_info);				\
+	__btrfs_panic(_i, __func__, __LINE__, errno, fmt, ##args);	\
+	BUG_ON(!(_i->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR));	\
 } while (0)
 
 /* acl.c */
@@ -3003,16 +3065,17 @@ void btrfs_reloc_cow_block(struct btrfs_trans_handle *trans,
 void btrfs_reloc_pre_snapshot(struct btrfs_trans_handle *trans,
 			      struct btrfs_pending_snapshot *pending,
 			      u64 *bytes_to_reserve);
-void btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
+int btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
 			      struct btrfs_pending_snapshot *pending);
 
 /* scrub.c */
 int btrfs_scrub_dev(struct btrfs_root *root, u64 devid, u64 start, u64 end,
 		    struct btrfs_scrub_progress *progress, int readonly);
-int btrfs_scrub_pause(struct btrfs_root *root);
-int btrfs_scrub_pause_super(struct btrfs_root *root);
-int btrfs_scrub_continue(struct btrfs_root *root);
-int btrfs_scrub_continue_super(struct btrfs_root *root);
+void btrfs_scrub_pause(struct btrfs_root *root);
+void btrfs_scrub_pause_super(struct btrfs_root *root);
+void btrfs_scrub_continue(struct btrfs_root *root);
+void btrfs_scrub_continue_super(struct btrfs_root *root);
+int __btrfs_scrub_cancel(struct btrfs_fs_info *info);
 int btrfs_scrub_cancel(struct btrfs_root *root);
 int btrfs_scrub_cancel_dev(struct btrfs_root *root, struct btrfs_device *dev);
 int btrfs_scrub_cancel_devid(struct btrfs_root *root, u64 devid);
diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c
index fe4cd0f1cef..03e3748d84d 100644
--- a/fs/btrfs/delayed-inode.c
+++ b/fs/btrfs/delayed-inode.c
@@ -115,6 +115,7 @@ static struct btrfs_delayed_node *btrfs_get_delayed_node(struct inode *inode)
 	return NULL;
 }
 
+/* Will return either the node or PTR_ERR(-ENOMEM) */
 static struct btrfs_delayed_node *btrfs_get_or_create_delayed_node(
 							struct inode *inode)
 {
@@ -836,10 +837,8 @@ static int btrfs_batch_insert_items(struct btrfs_trans_handle *trans,
 	btrfs_clear_path_blocking(path, NULL, 0);
 
 	/* insert the keys of the items */
-	ret = setup_items_for_insert(trans, root, path, keys, data_size,
-				     total_data_size, total_size, nitems);
-	if (ret)
-		goto error;
+	setup_items_for_insert(trans, root, path, keys, data_size,
+			       total_data_size, total_size, nitems);
 
 	/* insert the dir index items */
 	slot = path->slots[0];
@@ -1108,16 +1107,25 @@ static int btrfs_update_delayed_inode(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
-/* Called when committing the transaction. */
+/*
+ * Called when committing the transaction.
+ * Returns 0 on success.
+ * Returns < 0 on error and returns with an aborted transaction with any
+ * outstanding delayed items cleaned up.
+ */
 int btrfs_run_delayed_items(struct btrfs_trans_handle *trans,
 			    struct btrfs_root *root)
 {
+	struct btrfs_root *curr_root = root;
 	struct btrfs_delayed_root *delayed_root;
 	struct btrfs_delayed_node *curr_node, *prev_node;
 	struct btrfs_path *path;
 	struct btrfs_block_rsv *block_rsv;
 	int ret = 0;
 
+	if (trans->aborted)
+		return -EIO;
+
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
@@ -1130,17 +1138,18 @@ int btrfs_run_delayed_items(struct btrfs_trans_handle *trans,
 
 	curr_node = btrfs_first_delayed_node(delayed_root);
 	while (curr_node) {
-		root = curr_node->root;
-		ret = btrfs_insert_delayed_items(trans, path, root,
+		curr_root = curr_node->root;
+		ret = btrfs_insert_delayed_items(trans, path, curr_root,
 						 curr_node);
 		if (!ret)
-			ret = btrfs_delete_delayed_items(trans, path, root,
-							 curr_node);
+			ret = btrfs_delete_delayed_items(trans, path,
+						curr_root, curr_node);
 		if (!ret)
-			ret = btrfs_update_delayed_inode(trans, root, path,
-							 curr_node);
+			ret = btrfs_update_delayed_inode(trans, curr_root,
+						path, curr_node);
 		if (ret) {
 			btrfs_release_delayed_node(curr_node);
+			btrfs_abort_transaction(trans, root, ret);
 			break;
 		}
 
@@ -1151,6 +1160,7 @@ int btrfs_run_delayed_items(struct btrfs_trans_handle *trans,
 
 	btrfs_free_path(path);
 	trans->block_rsv = block_rsv;
+
 	return ret;
 }
 
@@ -1371,6 +1381,7 @@ void btrfs_balance_delayed_items(struct btrfs_root *root)
 	btrfs_wq_run_delayed_node(delayed_root, root, 0);
 }
 
+/* Will return 0 or -ENOMEM */
 int btrfs_insert_delayed_dir_index(struct btrfs_trans_handle *trans,
 				   struct btrfs_root *root, const char *name,
 				   int name_len, struct inode *dir,
diff --git a/fs/btrfs/delayed-ref.c b/fs/btrfs/delayed-ref.c
index 66e4f29505a..69f22e3ab3b 100644
--- a/fs/btrfs/delayed-ref.c
+++ b/fs/btrfs/delayed-ref.c
@@ -420,7 +420,7 @@ update_existing_head_ref(struct btrfs_delayed_ref_node *existing,
  * this does all the dirty work in terms of maintaining the correct
  * overall modification count.
  */
-static noinline int add_delayed_ref_head(struct btrfs_fs_info *fs_info,
+static noinline void add_delayed_ref_head(struct btrfs_fs_info *fs_info,
 					struct btrfs_trans_handle *trans,
 					struct btrfs_delayed_ref_node *ref,
 					u64 bytenr, u64 num_bytes,
@@ -487,20 +487,19 @@ static noinline int add_delayed_ref_head(struct btrfs_fs_info *fs_info,
 		 * we've updated the existing ref, free the newly
 		 * allocated ref
 		 */
-		kfree(ref);
+		kfree(head_ref);
 	} else {
 		delayed_refs->num_heads++;
 		delayed_refs->num_heads_ready++;
 		delayed_refs->num_entries++;
 		trans->delayed_ref_updates++;
 	}
-	return 0;
 }
 
 /*
  * helper to insert a delayed tree ref into the rbtree.
  */
-static noinline int add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
+static noinline void add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 					 struct btrfs_trans_handle *trans,
 					 struct btrfs_delayed_ref_node *ref,
 					 u64 bytenr, u64 num_bytes, u64 parent,
@@ -549,18 +548,17 @@ static noinline int add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 		 * we've updated the existing ref, free the newly
 		 * allocated ref
 		 */
-		kfree(ref);
+		kfree(full_ref);
 	} else {
 		delayed_refs->num_entries++;
 		trans->delayed_ref_updates++;
 	}
-	return 0;
 }
 
 /*
  * helper to insert a delayed data ref into the rbtree.
  */
-static noinline int add_delayed_data_ref(struct btrfs_fs_info *fs_info,
+static noinline void add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 					 struct btrfs_trans_handle *trans,
 					 struct btrfs_delayed_ref_node *ref,
 					 u64 bytenr, u64 num_bytes, u64 parent,
@@ -611,12 +609,11 @@ static noinline int add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 		 * we've updated the existing ref, free the newly
 		 * allocated ref
 		 */
-		kfree(ref);
+		kfree(full_ref);
 	} else {
 		delayed_refs->num_entries++;
 		trans->delayed_ref_updates++;
 	}
-	return 0;
 }
 
 /*
@@ -634,7 +631,6 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 	struct btrfs_delayed_tree_ref *ref;
 	struct btrfs_delayed_ref_head *head_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
-	int ret;
 
 	BUG_ON(extent_op && extent_op->is_data);
 	ref = kmalloc(sizeof(*ref), GFP_NOFS);
@@ -656,14 +652,12 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 	 * insert both the head node and the new ref without dropping
 	 * the spin lock
 	 */
-	ret = add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr,
+	add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr,
 				   num_bytes, action, 0);
-	BUG_ON(ret);
 
-	ret = add_delayed_tree_ref(fs_info, trans, &ref->node, bytenr,
+	add_delayed_tree_ref(fs_info, trans, &ref->node, bytenr,
 				   num_bytes, parent, ref_root, level, action,
 				   for_cow);
-	BUG_ON(ret);
 	if (!need_ref_seq(for_cow, ref_root) &&
 	    waitqueue_active(&delayed_refs->seq_wait))
 		wake_up(&delayed_refs->seq_wait);
@@ -685,7 +679,6 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 	struct btrfs_delayed_data_ref *ref;
 	struct btrfs_delayed_ref_head *head_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
-	int ret;
 
 	BUG_ON(extent_op && !extent_op->is_data);
 	ref = kmalloc(sizeof(*ref), GFP_NOFS);
@@ -707,14 +700,12 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 	 * insert both the head node and the new ref without dropping
 	 * the spin lock
 	 */
-	ret = add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr,
+	add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr,
 				   num_bytes, action, 1);
-	BUG_ON(ret);
 
-	ret = add_delayed_data_ref(fs_info, trans, &ref->node, bytenr,
+	add_delayed_data_ref(fs_info, trans, &ref->node, bytenr,
 				   num_bytes, parent, ref_root, owner, offset,
 				   action, for_cow);
-	BUG_ON(ret);
 	if (!need_ref_seq(for_cow, ref_root) &&
 	    waitqueue_active(&delayed_refs->seq_wait))
 		wake_up(&delayed_refs->seq_wait);
@@ -729,7 +720,6 @@ int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info,
 {
 	struct btrfs_delayed_ref_head *head_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
-	int ret;
 
 	head_ref = kmalloc(sizeof(*head_ref), GFP_NOFS);
 	if (!head_ref)
@@ -740,10 +730,9 @@ int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info,
 	delayed_refs = &trans->transaction->delayed_refs;
 	spin_lock(&delayed_refs->lock);
 
-	ret = add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr,
+	add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr,
 				   num_bytes, BTRFS_UPDATE_DELAYED_HEAD,
 				   extent_op->is_data);
-	BUG_ON(ret);
 
 	if (waitqueue_active(&delayed_refs->seq_wait))
 		wake_up(&delayed_refs->seq_wait);
diff --git a/fs/btrfs/dir-item.c b/fs/btrfs/dir-item.c
index 31d84e78129..c1a074d0696 100644
--- a/fs/btrfs/dir-item.c
+++ b/fs/btrfs/dir-item.c
@@ -49,9 +49,8 @@ static struct btrfs_dir_item *insert_with_overflow(struct btrfs_trans_handle
 		di = btrfs_match_dir_item_name(root, path, name, name_len);
 		if (di)
 			return ERR_PTR(-EEXIST);
-		ret = btrfs_extend_item(trans, root, path, data_size);
-	}
-	if (ret < 0)
+		btrfs_extend_item(trans, root, path, data_size);
+	} else if (ret < 0)
 		return ERR_PTR(ret);
 	WARN_ON(ret > 0);
 	leaf = path->nodes[0];
@@ -116,6 +115,7 @@ int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans,
  * 'location' is the key to stuff into the directory item, 'type' is the
  * type of the inode we're pointing to, and 'index' is the sequence number
  * to use for the second index (if one is created).
+ * Will return 0 or -ENOMEM
  */
 int btrfs_insert_dir_item(struct btrfs_trans_handle *trans, struct btrfs_root
 			  *root, const char *name, int name_len,
@@ -383,8 +383,8 @@ int btrfs_delete_one_dir_name(struct btrfs_trans_handle *trans,
 		start = btrfs_item_ptr_offset(leaf, path->slots[0]);
 		memmove_extent_buffer(leaf, ptr, ptr + sub_item_len,
 			item_len - (ptr + sub_item_len - start));
-		ret = btrfs_truncate_item(trans, root, path,
-					  item_len - sub_item_len, 1);
+		btrfs_truncate_item(trans, root, path,
+				    item_len - sub_item_len, 1);
 	}
 	return ret;
 }
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 534266fe505..20196f41120 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -48,20 +48,19 @@
 static struct extent_io_ops btree_extent_io_ops;
 static void end_workqueue_fn(struct btrfs_work *work);
 static void free_fs_root(struct btrfs_root *root);
-static void btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
+static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
 				    int read_only);
-static int btrfs_destroy_ordered_operations(struct btrfs_root *root);
-static int btrfs_destroy_ordered_extents(struct btrfs_root *root);
+static void btrfs_destroy_ordered_operations(struct btrfs_root *root);
+static void btrfs_destroy_ordered_extents(struct btrfs_root *root);
 static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 				      struct btrfs_root *root);
-static int btrfs_destroy_pending_snapshots(struct btrfs_transaction *t);
-static int btrfs_destroy_delalloc_inodes(struct btrfs_root *root);
+static void btrfs_destroy_pending_snapshots(struct btrfs_transaction *t);
+static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root);
 static int btrfs_destroy_marked_extents(struct btrfs_root *root,
 					struct extent_io_tree *dirty_pages,
 					int mark);
 static int btrfs_destroy_pinned_extent(struct btrfs_root *root,
 				       struct extent_io_tree *pinned_extents);
-static int btrfs_cleanup_transaction(struct btrfs_root *root);
 
 /*
  * end_io_wq structs are used to do processing in task context when an IO is
@@ -99,6 +98,7 @@ struct async_submit_bio {
 	 */
 	u64 bio_offset;
 	struct btrfs_work work;
+	int error;
 };
 
 /*
@@ -332,8 +332,8 @@ static int verify_parent_transid(struct extent_io_tree *io_tree,
 		return 0;
 
 	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
-			 0, &cached_state, GFP_NOFS);
-	if (extent_buffer_uptodate(io_tree, eb, cached_state) &&
+			 0, &cached_state);
+	if (extent_buffer_uptodate(eb) &&
 	    btrfs_header_generation(eb) == parent_transid) {
 		ret = 0;
 		goto out;
@@ -344,7 +344,7 @@ static int verify_parent_transid(struct extent_io_tree *io_tree,
 		       (unsigned long long)parent_transid,
 		       (unsigned long long)btrfs_header_generation(eb));
 	ret = 1;
-	clear_extent_buffer_uptodate(io_tree, eb, &cached_state);
+	clear_extent_buffer_uptodate(eb);
 out:
 	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
 			     &cached_state, GFP_NOFS);
@@ -360,9 +360,11 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root,
 					  u64 start, u64 parent_transid)
 {
 	struct extent_io_tree *io_tree;
+	int failed = 0;
 	int ret;
 	int num_copies = 0;
 	int mirror_num = 0;
+	int failed_mirror = 0;
 
 	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
 	io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
@@ -370,9 +372,8 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root,
 		ret = read_extent_buffer_pages(io_tree, eb, start,
 					       WAIT_COMPLETE,
 					       btree_get_extent, mirror_num);
-		if (!ret &&
-		    !verify_parent_transid(io_tree, eb, parent_transid))
-			return ret;
+		if (!ret && !verify_parent_transid(io_tree, eb, parent_transid))
+			break;
 
 		/*
 		 * This buffer's crc is fine, but its contents are corrupted, so
@@ -380,18 +381,31 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root,
 		 * any less wrong.
 		 */
 		if (test_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags))
-			return ret;
+			break;
+
+		if (!failed_mirror) {
+			failed = 1;
+			printk(KERN_ERR "failed mirror was %d\n", eb->failed_mirror);
+			failed_mirror = eb->failed_mirror;
+		}
 
 		num_copies = btrfs_num_copies(&root->fs_info->mapping_tree,
 					      eb->start, eb->len);
 		if (num_copies == 1)
-			return ret;
+			break;
 
 		mirror_num++;
+		if (mirror_num == failed_mirror)
+			mirror_num++;
+
 		if (mirror_num > num_copies)
-			return ret;
+			break;
 	}
-	return -EIO;
+
+	if (failed && !ret)
+		repair_eb_io_failure(root, eb, failed_mirror);
+
+	return ret;
 }
 
 /*
@@ -404,50 +418,27 @@ static int csum_dirty_buffer(struct btrfs_root *root, struct page *page)
 	struct extent_io_tree *tree;
 	u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
 	u64 found_start;
-	unsigned long len;
 	struct extent_buffer *eb;
-	int ret;
 
 	tree = &BTRFS_I(page->mapping->host)->io_tree;
 
-	if (page->private == EXTENT_PAGE_PRIVATE) {
-		WARN_ON(1);
-		goto out;
-	}
-	if (!page->private) {
-		WARN_ON(1);
-		goto out;
-	}
-	len = page->private >> 2;
-	WARN_ON(len == 0);
-
-	eb = alloc_extent_buffer(tree, start, len, page);
-	if (eb == NULL) {
-		WARN_ON(1);
-		goto out;
-	}
-	ret = btree_read_extent_buffer_pages(root, eb, start + PAGE_CACHE_SIZE,
-					     btrfs_header_generation(eb));
-	BUG_ON(ret);
-	WARN_ON(!btrfs_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN));
-
+	eb = (struct extent_buffer *)page->private;
+	if (page != eb->pages[0])
+		return 0;
 	found_start = btrfs_header_bytenr(eb);
 	if (found_start != start) {
 		WARN_ON(1);
-		goto err;
+		return 0;
 	}
-	if (eb->first_page != page) {
+	if (eb->pages[0] != page) {
 		WARN_ON(1);
-		goto err;
+		return 0;
 	}
 	if (!PageUptodate(page)) {
 		WARN_ON(1);
-		goto err;
+		return 0;
 	}
 	csum_tree_block(root, eb, 0);
-err:
-	free_extent_buffer(eb);
-out:
 	return 0;
 }
 
@@ -537,34 +528,74 @@ static noinline int check_leaf(struct btrfs_root *root,
 	return 0;
 }
 
+struct extent_buffer *find_eb_for_page(struct extent_io_tree *tree,
+				       struct page *page, int max_walk)
+{
+	struct extent_buffer *eb;
+	u64 start = page_offset(page);
+	u64 target = start;
+	u64 min_start;
+
+	if (start < max_walk)
+		min_start = 0;
+	else
+		min_start = start - max_walk;
+
+	while (start >= min_start) {
+		eb = find_extent_buffer(tree, start, 0);
+		if (eb) {
+			/*
+			 * we found an extent buffer and it contains our page
+			 * horray!
+			 */
+			if (eb->start <= target &&
+			    eb->start + eb->len > target)
+				return eb;
+
+			/* we found an extent buffer that wasn't for us */
+			free_extent_buffer(eb);
+			return NULL;
+		}
+		if (start == 0)
+			break;
+		start -= PAGE_CACHE_SIZE;
+	}
+	return NULL;
+}
+
 static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 			       struct extent_state *state)
 {
 	struct extent_io_tree *tree;
 	u64 found_start;
 	int found_level;
-	unsigned long len;
 	struct extent_buffer *eb;
 	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
 	int ret = 0;
+	int reads_done;
 
-	tree = &BTRFS_I(page->mapping->host)->io_tree;
-	if (page->private == EXTENT_PAGE_PRIVATE)
-		goto out;
 	if (!page->private)
 		goto out;
 
-	len = page->private >> 2;
-	WARN_ON(len == 0);
+	tree = &BTRFS_I(page->mapping->host)->io_tree;
+	eb = (struct extent_buffer *)page->private;
 
-	eb = alloc_extent_buffer(tree, start, len, page);
-	if (eb == NULL) {
+	/* the pending IO might have been the only thing that kept this buffer
+	 * in memory.  Make sure we have a ref for all this other checks
+	 */
+	extent_buffer_get(eb);
+
+	reads_done = atomic_dec_and_test(&eb->io_pages);
+	if (!reads_done)
+		goto err;
+
+	if (test_bit(EXTENT_BUFFER_IOERR, &eb->bflags)) {
 		ret = -EIO;
-		goto out;
+		goto err;
 	}
 
 	found_start = btrfs_header_bytenr(eb);
-	if (found_start != start) {
+	if (found_start != eb->start) {
 		printk_ratelimited(KERN_INFO "btrfs bad tree block start "
 			       "%llu %llu\n",
 			       (unsigned long long)found_start,
@@ -572,13 +603,6 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 		ret = -EIO;
 		goto err;
 	}
-	if (eb->first_page != page) {
-		printk(KERN_INFO "btrfs bad first page %lu %lu\n",
-		       eb->first_page->index, page->index);
-		WARN_ON(1);
-		ret = -EIO;
-		goto err;
-	}
 	if (check_tree_block_fsid(root, eb)) {
 		printk_ratelimited(KERN_INFO "btrfs bad fsid on block %llu\n",
 			       (unsigned long long)eb->start);
@@ -606,48 +630,31 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 		ret = -EIO;
 	}
 
-	end = min_t(u64, eb->len, PAGE_CACHE_SIZE);
-	end = eb->start + end - 1;
+	if (!ret)
+		set_extent_buffer_uptodate(eb);
 err:
 	if (test_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) {
 		clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags);
 		btree_readahead_hook(root, eb, eb->start, ret);
 	}
 
+	if (ret)
+		clear_extent_buffer_uptodate(eb);
 	free_extent_buffer(eb);
 out:
 	return ret;
 }
 
-static int btree_io_failed_hook(struct bio *failed_bio,
-			 struct page *page, u64 start, u64 end,
-			 int mirror_num, struct extent_state *state)
+static int btree_io_failed_hook(struct page *page, int failed_mirror)
 {
-	struct extent_io_tree *tree;
-	unsigned long len;
 	struct extent_buffer *eb;
 	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
 
-	tree = &BTRFS_I(page->mapping->host)->io_tree;
-	if (page->private == EXTENT_PAGE_PRIVATE)
-		goto out;
-	if (!page->private)
-		goto out;
-
-	len = page->private >> 2;
-	WARN_ON(len == 0);
-
-	eb = alloc_extent_buffer(tree, start, len, page);
-	if (eb == NULL)
-		goto out;
-
-	if (test_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) {
-		clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags);
+	eb = (struct extent_buffer *)page->private;
+	set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+	eb->failed_mirror = failed_mirror;
+	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
 		btree_readahead_hook(root, eb, eb->start, -EIO);
-	}
-	free_extent_buffer(eb);
-
-out:
 	return -EIO;	/* we fixed nothing */
 }
 
@@ -719,11 +726,14 @@ unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
 static void run_one_async_start(struct btrfs_work *work)
 {
 	struct async_submit_bio *async;
+	int ret;
 
 	async = container_of(work, struct  async_submit_bio, work);
-	async->submit_bio_start(async->inode, async->rw, async->bio,
-			       async->mirror_num, async->bio_flags,
-			       async->bio_offset);
+	ret = async->submit_bio_start(async->inode, async->rw, async->bio,
+				      async->mirror_num, async->bio_flags,
+				      async->bio_offset);
+	if (ret)
+		async->error = ret;
 }
 
 static void run_one_async_done(struct btrfs_work *work)
@@ -744,6 +754,12 @@ static void run_one_async_done(struct btrfs_work *work)
 	    waitqueue_active(&fs_info->async_submit_wait))
 		wake_up(&fs_info->async_submit_wait);
 
+	/* If an error occured we just want to clean up the bio and move on */
+	if (async->error) {
+		bio_endio(async->bio, async->error);
+		return;
+	}
+
 	async->submit_bio_done(async->inode, async->rw, async->bio,
 			       async->mirror_num, async->bio_flags,
 			       async->bio_offset);
@@ -785,6 +801,8 @@ int btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct inode *inode,
 	async->bio_flags = bio_flags;
 	async->bio_offset = bio_offset;
 
+	async->error = 0;
+
 	atomic_inc(&fs_info->nr_async_submits);
 
 	if (rw & REQ_SYNC)
@@ -806,15 +824,18 @@ static int btree_csum_one_bio(struct bio *bio)
 	struct bio_vec *bvec = bio->bi_io_vec;
 	int bio_index = 0;
 	struct btrfs_root *root;
+	int ret = 0;
 
 	WARN_ON(bio->bi_vcnt <= 0);
 	while (bio_index < bio->bi_vcnt) {
 		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
-		csum_dirty_buffer(root, bvec->bv_page);
+		ret = csum_dirty_buffer(root, bvec->bv_page);
+		if (ret)
+			break;
 		bio_index++;
 		bvec++;
 	}
-	return 0;
+	return ret;
 }
 
 static int __btree_submit_bio_start(struct inode *inode, int rw,
@@ -826,8 +847,7 @@ static int __btree_submit_bio_start(struct inode *inode, int rw,
 	 * when we're called for a write, we're already in the async
 	 * submission context.  Just jump into btrfs_map_bio
 	 */
-	btree_csum_one_bio(bio);
-	return 0;
+	return btree_csum_one_bio(bio);
 }
 
 static int __btree_submit_bio_done(struct inode *inode, int rw, struct bio *bio,
@@ -847,15 +867,16 @@ static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
 {
 	int ret;
 
-	ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
-					  bio, 1);
-	BUG_ON(ret);
-
 	if (!(rw & REQ_WRITE)) {
+
 		/*
 		 * called for a read, do the setup so that checksum validation
 		 * can happen in the async kernel threads
 		 */
+		ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
+					  bio, 1);
+		if (ret)
+			return ret;
 		return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
 				     mirror_num, 0);
 	}
@@ -893,34 +914,6 @@ static int btree_migratepage(struct address_space *mapping,
 }
 #endif
 
-static int btree_writepage(struct page *page, struct writeback_control *wbc)
-{
-	struct extent_io_tree *tree;
-	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
-	struct extent_buffer *eb;
-	int was_dirty;
-
-	tree = &BTRFS_I(page->mapping->host)->io_tree;
-	if (!(current->flags & PF_MEMALLOC)) {
-		return extent_write_full_page(tree, page,
-					      btree_get_extent, wbc);
-	}
-
-	redirty_page_for_writepage(wbc, page);
-	eb = btrfs_find_tree_block(root, page_offset(page), PAGE_CACHE_SIZE);
-	WARN_ON(!eb);
-
-	was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags);
-	if (!was_dirty) {
-		spin_lock(&root->fs_info->delalloc_lock);
-		root->fs_info->dirty_metadata_bytes += PAGE_CACHE_SIZE;
-		spin_unlock(&root->fs_info->delalloc_lock);
-	}
-	free_extent_buffer(eb);
-
-	unlock_page(page);
-	return 0;
-}
 
 static int btree_writepages(struct address_space *mapping,
 			    struct writeback_control *wbc)
@@ -940,7 +933,7 @@ static int btree_writepages(struct address_space *mapping,
 		if (num_dirty < thresh)
 			return 0;
 	}
-	return extent_writepages(tree, mapping, btree_get_extent, wbc);
+	return btree_write_cache_pages(mapping, wbc);
 }
 
 static int btree_readpage(struct file *file, struct page *page)
@@ -952,16 +945,8 @@ static int btree_readpage(struct file *file, struct page *page)
 
 static int btree_releasepage(struct page *page, gfp_t gfp_flags)
 {
-	struct extent_io_tree *tree;
-	struct extent_map_tree *map;
-	int ret;
-
 	if (PageWriteback(page) || PageDirty(page))
 		return 0;
-
-	tree = &BTRFS_I(page->mapping->host)->io_tree;
-	map = &BTRFS_I(page->mapping->host)->extent_tree;
-
 	/*
 	 * We need to mask out eg. __GFP_HIGHMEM and __GFP_DMA32 as we're doing
 	 * slab allocation from alloc_extent_state down the callchain where
@@ -969,18 +954,7 @@ static int btree_releasepage(struct page *page, gfp_t gfp_flags)
 	 */
 	gfp_flags &= ~GFP_SLAB_BUG_MASK;
 
-	ret = try_release_extent_state(map, tree, page, gfp_flags);
-	if (!ret)
-		return 0;
-
-	ret = try_release_extent_buffer(tree, page);
-	if (ret == 1) {
-		ClearPagePrivate(page);
-		set_page_private(page, 0);
-		page_cache_release(page);
-	}
-
-	return ret;
+	return try_release_extent_buffer(page, gfp_flags);
 }
 
 static void btree_invalidatepage(struct page *page, unsigned long offset)
@@ -998,15 +972,28 @@ static void btree_invalidatepage(struct page *page, unsigned long offset)
 	}
 }
 
+static int btree_set_page_dirty(struct page *page)
+{
+	struct extent_buffer *eb;
+
+	BUG_ON(!PagePrivate(page));
+	eb = (struct extent_buffer *)page->private;
+	BUG_ON(!eb);
+	BUG_ON(!test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
+	BUG_ON(!atomic_read(&eb->refs));
+	btrfs_assert_tree_locked(eb);
+	return __set_page_dirty_nobuffers(page);
+}
+
 static const struct address_space_operations btree_aops = {
 	.readpage	= btree_readpage,
-	.writepage	= btree_writepage,
 	.writepages	= btree_writepages,
 	.releasepage	= btree_releasepage,
 	.invalidatepage = btree_invalidatepage,
 #ifdef CONFIG_MIGRATION
 	.migratepage	= btree_migratepage,
 #endif
+	.set_page_dirty = btree_set_page_dirty,
 };
 
 int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize,
@@ -1049,7 +1036,7 @@ int reada_tree_block_flagged(struct btrfs_root *root, u64 bytenr, u32 blocksize,
 	if (test_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags)) {
 		free_extent_buffer(buf);
 		return -EIO;
-	} else if (extent_buffer_uptodate(io_tree, buf, NULL)) {
+	} else if (extent_buffer_uptodate(buf)) {
 		*eb = buf;
 	} else {
 		free_extent_buffer(buf);
@@ -1074,20 +1061,20 @@ struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
 	struct extent_buffer *eb;
 
 	eb = alloc_extent_buffer(&BTRFS_I(btree_inode)->io_tree,
-				 bytenr, blocksize, NULL);
+				 bytenr, blocksize);
 	return eb;
 }
 
 
 int btrfs_write_tree_block(struct extent_buffer *buf)
 {
-	return filemap_fdatawrite_range(buf->first_page->mapping, buf->start,
+	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
 					buf->start + buf->len - 1);
 }
 
 int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
 {
-	return filemap_fdatawait_range(buf->first_page->mapping,
+	return filemap_fdatawait_range(buf->pages[0]->mapping,
 				       buf->start, buf->start + buf->len - 1);
 }
 
@@ -1102,17 +1089,13 @@ struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
 		return NULL;
 
 	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
-
-	if (ret == 0)
-		set_bit(EXTENT_BUFFER_UPTODATE, &buf->bflags);
 	return buf;
 
 }
 
-int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
-		     struct extent_buffer *buf)
+void clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
+		      struct extent_buffer *buf)
 {
-	struct inode *btree_inode = root->fs_info->btree_inode;
 	if (btrfs_header_generation(buf) ==
 	    root->fs_info->running_transaction->transid) {
 		btrfs_assert_tree_locked(buf);
@@ -1121,23 +1104,27 @@ int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 			spin_lock(&root->fs_info->delalloc_lock);
 			if (root->fs_info->dirty_metadata_bytes >= buf->len)
 				root->fs_info->dirty_metadata_bytes -= buf->len;
-			else
-				WARN_ON(1);
+			else {
+				spin_unlock(&root->fs_info->delalloc_lock);
+				btrfs_panic(root->fs_info, -EOVERFLOW,
+					  "Can't clear %lu bytes from "
+					  " dirty_mdatadata_bytes (%lu)",
+					  buf->len,
+					  root->fs_info->dirty_metadata_bytes);
+			}
 			spin_unlock(&root->fs_info->delalloc_lock);
 		}
 
 		/* ugh, clear_extent_buffer_dirty needs to lock the page */
 		btrfs_set_lock_blocking(buf);
-		clear_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree,
-					  buf);
+		clear_extent_buffer_dirty(buf);
 	}
-	return 0;
 }
 
-static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
-			u32 stripesize, struct btrfs_root *root,
-			struct btrfs_fs_info *fs_info,
-			u64 objectid)
+static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
+			 u32 stripesize, struct btrfs_root *root,
+			 struct btrfs_fs_info *fs_info,
+			 u64 objectid)
 {
 	root->node = NULL;
 	root->commit_root = NULL;
@@ -1189,13 +1176,12 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
 	root->defrag_running = 0;
 	root->root_key.objectid = objectid;
 	root->anon_dev = 0;
-	return 0;
 }
 
-static int find_and_setup_root(struct btrfs_root *tree_root,
-			       struct btrfs_fs_info *fs_info,
-			       u64 objectid,
-			       struct btrfs_root *root)
+static int __must_check find_and_setup_root(struct btrfs_root *tree_root,
+					    struct btrfs_fs_info *fs_info,
+					    u64 objectid,
+					    struct btrfs_root *root)
 {
 	int ret;
 	u32 blocksize;
@@ -1208,7 +1194,8 @@ static int find_and_setup_root(struct btrfs_root *tree_root,
 				   &root->root_item, &root->root_key);
 	if (ret > 0)
 		return -ENOENT;
-	BUG_ON(ret);
+	else if (ret < 0)
+		return ret;
 
 	generation = btrfs_root_generation(&root->root_item);
 	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
@@ -1377,7 +1364,7 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root,
 	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
 				     blocksize, generation);
 	root->commit_root = btrfs_root_node(root);
-	BUG_ON(!root->node);
+	BUG_ON(!root->node); /* -ENOMEM */
 out:
 	if (location->objectid != BTRFS_TREE_LOG_OBJECTID) {
 		root->ref_cows = 1;
@@ -1513,41 +1500,6 @@ static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
 	return 0;
 }
 
-static int bio_ready_for_csum(struct bio *bio)
-{
-	u64 length = 0;
-	u64 buf_len = 0;
-	u64 start = 0;
-	struct page *page;
-	struct extent_io_tree *io_tree = NULL;
-	struct bio_vec *bvec;
-	int i;
-	int ret;
-
-	bio_for_each_segment(bvec, bio, i) {
-		page = bvec->bv_page;
-		if (page->private == EXTENT_PAGE_PRIVATE) {
-			length += bvec->bv_len;
-			continue;
-		}
-		if (!page->private) {
-			length += bvec->bv_len;
-			continue;
-		}
-		length = bvec->bv_len;
-		buf_len = page->private >> 2;
-		start = page_offset(page) + bvec->bv_offset;
-		io_tree = &BTRFS_I(page->mapping->host)->io_tree;
-	}
-	/* are we fully contained in this bio? */
-	if (buf_len <= length)
-		return 1;
-
-	ret = extent_range_uptodate(io_tree, start + length,
-				    start + buf_len - 1);
-	return ret;
-}
-
 /*
  * called by the kthread helper functions to finally call the bio end_io
  * functions.  This is where read checksum verification actually happens
@@ -1563,17 +1515,6 @@ static void end_workqueue_fn(struct btrfs_work *work)
 	bio = end_io_wq->bio;
 	fs_info = end_io_wq->info;
 
-	/* metadata bio reads are special because the whole tree block must
-	 * be checksummed at once.  This makes sure the entire block is in
-	 * ram and up to date before trying to verify things.  For
-	 * blocksize <= pagesize, it is basically a noop
-	 */
-	if (!(bio->bi_rw & REQ_WRITE) && end_io_wq->metadata &&
-	    !bio_ready_for_csum(bio)) {
-		btrfs_queue_worker(&fs_info->endio_meta_workers,
-				   &end_io_wq->work);
-		return;
-	}
 	error = end_io_wq->error;
 	bio->bi_private = end_io_wq->private;
 	bio->bi_end_io = end_io_wq->end_io;
@@ -1614,9 +1555,10 @@ static int transaction_kthread(void *arg)
 	u64 transid;
 	unsigned long now;
 	unsigned long delay;
-	int ret;
+	bool cannot_commit;
 
 	do {
+		cannot_commit = false;
 		delay = HZ * 30;
 		vfs_check_frozen(root->fs_info->sb, SB_FREEZE_WRITE);
 		mutex_lock(&root->fs_info->transaction_kthread_mutex);
@@ -1638,11 +1580,14 @@ static int transaction_kthread(void *arg)
 		transid = cur->transid;
 		spin_unlock(&root->fs_info->trans_lock);
 
+		/* If the file system is aborted, this will always fail. */
 		trans = btrfs_join_transaction(root);
-		BUG_ON(IS_ERR(trans));
+		if (IS_ERR(trans)) {
+			cannot_commit = true;
+			goto sleep;
+		}
 		if (transid == trans->transid) {
-			ret = btrfs_commit_transaction(trans, root);
-			BUG_ON(ret);
+			btrfs_commit_transaction(trans, root);
 		} else {
 			btrfs_end_transaction(trans, root);
 		}
@@ -1653,7 +1598,8 @@ sleep:
 		if (!try_to_freeze()) {
 			set_current_state(TASK_INTERRUPTIBLE);
 			if (!kthread_should_stop() &&
-			    !btrfs_transaction_blocked(root->fs_info))
+			    (!btrfs_transaction_blocked(root->fs_info) ||
+			     cannot_commit))
 				schedule_timeout(delay);
 			__set_current_state(TASK_RUNNING);
 		}
@@ -2042,6 +1988,7 @@ int open_ctree(struct super_block *sb,
 	RB_CLEAR_NODE(&BTRFS_I(fs_info->btree_inode)->rb_node);
 	extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree,
 			     fs_info->btree_inode->i_mapping);
+	BTRFS_I(fs_info->btree_inode)->io_tree.track_uptodate = 0;
 	extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree);
 
 	BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;
@@ -2084,6 +2031,7 @@ int open_ctree(struct super_block *sb,
 	__setup_root(4096, 4096, 4096, 4096, tree_root,
 		     fs_info, BTRFS_ROOT_TREE_OBJECTID);
 
+	invalidate_bdev(fs_devices->latest_bdev);
 	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
 	if (!bh) {
 		err = -EINVAL;
@@ -2104,7 +2052,12 @@ int open_ctree(struct super_block *sb,
 	/* check FS state, whether FS is broken. */
 	fs_info->fs_state |= btrfs_super_flags(disk_super);
 
-	btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);
+	ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);
+	if (ret) {
+		printk(KERN_ERR "btrfs: superblock contains fatal errors\n");
+		err = ret;
+		goto fail_alloc;
+	}
 
 	/*
 	 * run through our array of backup supers and setup
@@ -2135,10 +2088,55 @@ int open_ctree(struct super_block *sb,
 		goto fail_alloc;
 	}
 
+	if (btrfs_super_leafsize(disk_super) !=
+	    btrfs_super_nodesize(disk_super)) {
+		printk(KERN_ERR "BTRFS: couldn't mount because metadata "
+		       "blocksizes don't match.  node %d leaf %d\n",
+		       btrfs_super_nodesize(disk_super),
+		       btrfs_super_leafsize(disk_super));
+		err = -EINVAL;
+		goto fail_alloc;
+	}
+	if (btrfs_super_leafsize(disk_super) > BTRFS_MAX_METADATA_BLOCKSIZE) {
+		printk(KERN_ERR "BTRFS: couldn't mount because metadata "
+		       "blocksize (%d) was too large\n",
+		       btrfs_super_leafsize(disk_super));
+		err = -EINVAL;
+		goto fail_alloc;
+	}
+
 	features = btrfs_super_incompat_flags(disk_super);
 	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
 	if (tree_root->fs_info->compress_type & BTRFS_COMPRESS_LZO)
 		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
+
+	/*
+	 * flag our filesystem as having big metadata blocks if
+	 * they are bigger than the page size
+	 */
+	if (btrfs_super_leafsize(disk_super) > PAGE_CACHE_SIZE) {
+		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
+			printk(KERN_INFO "btrfs flagging fs with big metadata feature\n");
+		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
+	}
+
+	nodesize = btrfs_super_nodesize(disk_super);
+	leafsize = btrfs_super_leafsize(disk_super);
+	sectorsize = btrfs_super_sectorsize(disk_super);
+	stripesize = btrfs_super_stripesize(disk_super);
+
+	/*
+	 * mixed block groups end up with duplicate but slightly offset
+	 * extent buffers for the same range.  It leads to corruptions
+	 */
+	if ((features & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) &&
+	    (sectorsize != leafsize)) {
+		printk(KERN_WARNING "btrfs: unequal leaf/node/sector sizes "
+				"are not allowed for mixed block groups on %s\n",
+				sb->s_id);
+		goto fail_alloc;
+	}
+
 	btrfs_set_super_incompat_flags(disk_super, features);
 
 	features = btrfs_super_compat_ro_flags(disk_super) &
@@ -2242,10 +2240,6 @@ int open_ctree(struct super_block *sb,
 	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
 				    4 * 1024 * 1024 / PAGE_CACHE_SIZE);
 
-	nodesize = btrfs_super_nodesize(disk_super);
-	leafsize = btrfs_super_leafsize(disk_super);
-	sectorsize = btrfs_super_sectorsize(disk_super);
-	stripesize = btrfs_super_stripesize(disk_super);
 	tree_root->nodesize = nodesize;
 	tree_root->leafsize = leafsize;
 	tree_root->sectorsize = sectorsize;
@@ -2285,7 +2279,7 @@ int open_ctree(struct super_block *sb,
 	chunk_root->node = read_tree_block(chunk_root,
 					   btrfs_super_chunk_root(disk_super),
 					   blocksize, generation);
-	BUG_ON(!chunk_root->node);
+	BUG_ON(!chunk_root->node); /* -ENOMEM */
 	if (!test_bit(EXTENT_BUFFER_UPTODATE, &chunk_root->node->bflags)) {
 		printk(KERN_WARNING "btrfs: failed to read chunk root on %s\n",
 		       sb->s_id);
@@ -2425,21 +2419,31 @@ retry_root_backup:
 		log_tree_root->node = read_tree_block(tree_root, bytenr,
 						      blocksize,
 						      generation + 1);
+		/* returns with log_tree_root freed on success */
 		ret = btrfs_recover_log_trees(log_tree_root);
-		BUG_ON(ret);
+		if (ret) {
+			btrfs_error(tree_root->fs_info, ret,
+				    "Failed to recover log tree");
+			free_extent_buffer(log_tree_root->node);
+			kfree(log_tree_root);
+			goto fail_trans_kthread;
+		}
 
 		if (sb->s_flags & MS_RDONLY) {
-			ret =  btrfs_commit_super(tree_root);
-			BUG_ON(ret);
+			ret = btrfs_commit_super(tree_root);
+			if (ret)
+				goto fail_trans_kthread;
 		}
 	}
 
 	ret = btrfs_find_orphan_roots(tree_root);
-	BUG_ON(ret);
+	if (ret)
+		goto fail_trans_kthread;
 
 	if (!(sb->s_flags & MS_RDONLY)) {
 		ret = btrfs_cleanup_fs_roots(fs_info);
-		BUG_ON(ret);
+		if (ret) {
+			}
 
 		ret = btrfs_recover_relocation(tree_root);
 		if (ret < 0) {
@@ -2859,6 +2863,8 @@ int write_all_supers(struct btrfs_root *root, int max_mirrors)
 	if (total_errors > max_errors) {
 		printk(KERN_ERR "btrfs: %d errors while writing supers\n",
 		       total_errors);
+
+		/* This shouldn't happen. FUA is masked off if unsupported */
 		BUG();
 	}
 
@@ -2875,9 +2881,9 @@ int write_all_supers(struct btrfs_root *root, int max_mirrors)
 	}
 	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
 	if (total_errors > max_errors) {
-		printk(KERN_ERR "btrfs: %d errors while writing supers\n",
-		       total_errors);
-		BUG();
+		btrfs_error(root->fs_info, -EIO,
+			    "%d errors while writing supers", total_errors);
+		return -EIO;
 	}
 	return 0;
 }
@@ -2891,7 +2897,20 @@ int write_ctree_super(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
-int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
+/* Kill all outstanding I/O */
+void btrfs_abort_devices(struct btrfs_root *root)
+{
+	struct list_head *head;
+	struct btrfs_device *dev;
+	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
+	head = &root->fs_info->fs_devices->devices;
+	list_for_each_entry_rcu(dev, head, dev_list) {
+		blk_abort_queue(dev->bdev->bd_disk->queue);
+	}
+	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
+}
+
+void btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
 {
 	spin_lock(&fs_info->fs_roots_radix_lock);
 	radix_tree_delete(&fs_info->fs_roots_radix,
@@ -2904,7 +2923,6 @@ int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
 	__btrfs_remove_free_space_cache(root->free_ino_pinned);
 	__btrfs_remove_free_space_cache(root->free_ino_ctl);
 	free_fs_root(root);
-	return 0;
 }
 
 static void free_fs_root(struct btrfs_root *root)
@@ -2921,7 +2939,7 @@ static void free_fs_root(struct btrfs_root *root)
 	kfree(root);
 }
 
-static int del_fs_roots(struct btrfs_fs_info *fs_info)
+static void del_fs_roots(struct btrfs_fs_info *fs_info)
 {
 	int ret;
 	struct btrfs_root *gang[8];
@@ -2950,7 +2968,6 @@ static int del_fs_roots(struct btrfs_fs_info *fs_info)
 		for (i = 0; i < ret; i++)
 			btrfs_free_fs_root(fs_info, gang[i]);
 	}
-	return 0;
 }
 
 int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
@@ -2999,14 +3016,21 @@ int btrfs_commit_super(struct btrfs_root *root)
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
 	ret = btrfs_commit_transaction(trans, root);
-	BUG_ON(ret);
+	if (ret)
+		return ret;
 	/* run commit again to drop the original snapshot */
 	trans = btrfs_join_transaction(root);
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
-	btrfs_commit_transaction(trans, root);
+	ret = btrfs_commit_transaction(trans, root);
+	if (ret)
+		return ret;
 	ret = btrfs_write_and_wait_transaction(NULL, root);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_error(root->fs_info, ret,
+			    "Failed to sync btree inode to disk.");
+		return ret;
+	}
 
 	ret = write_ctree_super(NULL, root, 0);
 	return ret;
@@ -3122,10 +3146,9 @@ int close_ctree(struct btrfs_root *root)
 int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid)
 {
 	int ret;
-	struct inode *btree_inode = buf->first_page->mapping->host;
+	struct inode *btree_inode = buf->pages[0]->mapping->host;
 
-	ret = extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf,
-				     NULL);
+	ret = extent_buffer_uptodate(buf);
 	if (!ret)
 		return ret;
 
@@ -3136,16 +3159,13 @@ int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid)
 
 int btrfs_set_buffer_uptodate(struct extent_buffer *buf)
 {
-	struct inode *btree_inode = buf->first_page->mapping->host;
-	return set_extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree,
-					  buf);
+	return set_extent_buffer_uptodate(buf);
 }
 
 void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
 {
-	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
+	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
 	u64 transid = btrfs_header_generation(buf);
-	struct inode *btree_inode = root->fs_info->btree_inode;
 	int was_dirty;
 
 	btrfs_assert_tree_locked(buf);
@@ -3157,8 +3177,7 @@ void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
 			(unsigned long long)root->fs_info->generation);
 		WARN_ON(1);
 	}
-	was_dirty = set_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree,
-					    buf);
+	was_dirty = set_extent_buffer_dirty(buf);
 	if (!was_dirty) {
 		spin_lock(&root->fs_info->delalloc_lock);
 		root->fs_info->dirty_metadata_bytes += buf->len;
@@ -3212,12 +3231,8 @@ void __btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr)
 
 int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
 {
-	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
-	int ret;
-	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
-	if (ret == 0)
-		set_bit(EXTENT_BUFFER_UPTODATE, &buf->bflags);
-	return ret;
+	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
+	return btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
 }
 
 static int btree_lock_page_hook(struct page *page, void *data,
@@ -3225,17 +3240,21 @@ static int btree_lock_page_hook(struct page *page, void *data,
 {
 	struct inode *inode = page->mapping->host;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	struct extent_buffer *eb;
-	unsigned long len;
-	u64 bytenr = page_offset(page);
 
-	if (page->private == EXTENT_PAGE_PRIVATE)
+	/*
+	 * We culled this eb but the page is still hanging out on the mapping,
+	 * carry on.
+	 */
+	if (!PagePrivate(page))
 		goto out;
 
-	len = page->private >> 2;
-	eb = find_extent_buffer(io_tree, bytenr, len);
-	if (!eb)
+	eb = (struct extent_buffer *)page->private;
+	if (!eb) {
+		WARN_ON(1);
+		goto out;
+	}
+	if (page != eb->pages[0])
 		goto out;
 
 	if (!btrfs_try_tree_write_lock(eb)) {
@@ -3254,7 +3273,6 @@ static int btree_lock_page_hook(struct page *page, void *data,
 	}
 
 	btrfs_tree_unlock(eb);
-	free_extent_buffer(eb);
 out:
 	if (!trylock_page(page)) {
 		flush_fn(data);
@@ -3263,15 +3281,23 @@ out:
 	return 0;
 }
 
-static void btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
+static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
 			      int read_only)
 {
+	if (btrfs_super_csum_type(fs_info->super_copy) >= ARRAY_SIZE(btrfs_csum_sizes)) {
+		printk(KERN_ERR "btrfs: unsupported checksum algorithm\n");
+		return -EINVAL;
+	}
+
 	if (read_only)
-		return;
+		return 0;
 
-	if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR)
+	if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) {
 		printk(KERN_WARNING "warning: mount fs with errors, "
 		       "running btrfsck is recommended\n");
+	}
+
+	return 0;
 }
 
 int btrfs_error_commit_super(struct btrfs_root *root)
@@ -3293,7 +3319,7 @@ int btrfs_error_commit_super(struct btrfs_root *root)
 	return ret;
 }
 
-static int btrfs_destroy_ordered_operations(struct btrfs_root *root)
+static void btrfs_destroy_ordered_operations(struct btrfs_root *root)
 {
 	struct btrfs_inode *btrfs_inode;
 	struct list_head splice;
@@ -3315,11 +3341,9 @@ static int btrfs_destroy_ordered_operations(struct btrfs_root *root)
 
 	spin_unlock(&root->fs_info->ordered_extent_lock);
 	mutex_unlock(&root->fs_info->ordered_operations_mutex);
-
-	return 0;
 }
 
-static int btrfs_destroy_ordered_extents(struct btrfs_root *root)
+static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
 {
 	struct list_head splice;
 	struct btrfs_ordered_extent *ordered;
@@ -3351,12 +3375,10 @@ static int btrfs_destroy_ordered_extents(struct btrfs_root *root)
 	}
 
 	spin_unlock(&root->fs_info->ordered_extent_lock);
-
-	return 0;
 }
 
-static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
-				      struct btrfs_root *root)
+int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
+			       struct btrfs_root *root)
 {
 	struct rb_node *node;
 	struct btrfs_delayed_ref_root *delayed_refs;
@@ -3365,6 +3387,7 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 
 	delayed_refs = &trans->delayed_refs;
 
+again:
 	spin_lock(&delayed_refs->lock);
 	if (delayed_refs->num_entries == 0) {
 		spin_unlock(&delayed_refs->lock);
@@ -3386,6 +3409,7 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 			struct btrfs_delayed_ref_head *head;
 
 			head = btrfs_delayed_node_to_head(ref);
+			spin_unlock(&delayed_refs->lock);
 			mutex_lock(&head->mutex);
 			kfree(head->extent_op);
 			delayed_refs->num_heads--;
@@ -3393,8 +3417,9 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 				delayed_refs->num_heads_ready--;
 			list_del_init(&head->cluster);
 			mutex_unlock(&head->mutex);
+			btrfs_put_delayed_ref(ref);
+			goto again;
 		}
-
 		spin_unlock(&delayed_refs->lock);
 		btrfs_put_delayed_ref(ref);
 
@@ -3407,7 +3432,7 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 	return ret;
 }
 
-static int btrfs_destroy_pending_snapshots(struct btrfs_transaction *t)
+static void btrfs_destroy_pending_snapshots(struct btrfs_transaction *t)
 {
 	struct btrfs_pending_snapshot *snapshot;
 	struct list_head splice;
@@ -3425,11 +3450,9 @@ static int btrfs_destroy_pending_snapshots(struct btrfs_transaction *t)
 
 		kfree(snapshot);
 	}
-
-	return 0;
 }
 
-static int btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
+static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
 {
 	struct btrfs_inode *btrfs_inode;
 	struct list_head splice;
@@ -3449,8 +3472,6 @@ static int btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
 	}
 
 	spin_unlock(&root->fs_info->delalloc_lock);
-
-	return 0;
 }
 
 static int btrfs_destroy_marked_extents(struct btrfs_root *root,
@@ -3541,13 +3562,43 @@ static int btrfs_destroy_pinned_extent(struct btrfs_root *root,
 	return 0;
 }
 
-static int btrfs_cleanup_transaction(struct btrfs_root *root)
+void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
+				   struct btrfs_root *root)
+{
+	btrfs_destroy_delayed_refs(cur_trans, root);
+	btrfs_block_rsv_release(root, &root->fs_info->trans_block_rsv,
+				cur_trans->dirty_pages.dirty_bytes);
+
+	/* FIXME: cleanup wait for commit */
+	cur_trans->in_commit = 1;
+	cur_trans->blocked = 1;
+	if (waitqueue_active(&root->fs_info->transaction_blocked_wait))
+		wake_up(&root->fs_info->transaction_blocked_wait);
+
+	cur_trans->blocked = 0;
+	if (waitqueue_active(&root->fs_info->transaction_wait))
+		wake_up(&root->fs_info->transaction_wait);
+
+	cur_trans->commit_done = 1;
+	if (waitqueue_active(&cur_trans->commit_wait))
+		wake_up(&cur_trans->commit_wait);
+
+	btrfs_destroy_pending_snapshots(cur_trans);
+
+	btrfs_destroy_marked_extents(root, &cur_trans->dirty_pages,
+				     EXTENT_DIRTY);
+
+	/*
+	memset(cur_trans, 0, sizeof(*cur_trans));
+	kmem_cache_free(btrfs_transaction_cachep, cur_trans);
+	*/
+}
+
+int btrfs_cleanup_transaction(struct btrfs_root *root)
 {
 	struct btrfs_transaction *t;
 	LIST_HEAD(list);
 
-	WARN_ON(1);
-
 	mutex_lock(&root->fs_info->transaction_kthread_mutex);
 
 	spin_lock(&root->fs_info->trans_lock);
@@ -3612,6 +3663,17 @@ static int btrfs_cleanup_transaction(struct btrfs_root *root)
 	return 0;
 }
 
+static int btree_writepage_io_failed_hook(struct bio *bio, struct page *page,
+					  u64 start, u64 end,
+					  struct extent_state *state)
+{
+	struct super_block *sb = page->mapping->host->i_sb;
+	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
+	btrfs_error(fs_info, -EIO,
+		    "Error occured while writing out btree at %llu", start);
+	return -EIO;
+}
+
 static struct extent_io_ops btree_extent_io_ops = {
 	.write_cache_pages_lock_hook = btree_lock_page_hook,
 	.readpage_end_io_hook = btree_readpage_end_io_hook,
@@ -3619,4 +3681,5 @@ static struct extent_io_ops btree_extent_io_ops = {
 	.submit_bio_hook = btree_submit_bio_hook,
 	/* note we're sharing with inode.c for the merge bio hook */
 	.merge_bio_hook = btrfs_merge_bio_hook,
+	.writepage_io_failed_hook = btree_writepage_io_failed_hook,
 };
diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h
index e4bc4741319..a7ace1a2dd1 100644
--- a/fs/btrfs/disk-io.h
+++ b/fs/btrfs/disk-io.h
@@ -44,8 +44,8 @@ int reada_tree_block_flagged(struct btrfs_root *root, u64 bytenr, u32 blocksize,
 			 int mirror_num, struct extent_buffer **eb);
 struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
 						   u64 bytenr, u32 blocksize);
-int clean_tree_block(struct btrfs_trans_handle *trans,
-		     struct btrfs_root *root, struct extent_buffer *buf);
+void clean_tree_block(struct btrfs_trans_handle *trans,
+		      struct btrfs_root *root, struct extent_buffer *buf);
 int open_ctree(struct super_block *sb,
 	       struct btrfs_fs_devices *fs_devices,
 	       char *options);
@@ -64,7 +64,7 @@ struct btrfs_root *btrfs_read_fs_root_no_name(struct btrfs_fs_info *fs_info,
 int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info);
 void btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr);
 void __btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr);
-int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root);
+void btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root);
 void btrfs_mark_buffer_dirty(struct extent_buffer *buf);
 int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid);
 int btrfs_set_buffer_uptodate(struct extent_buffer *buf);
@@ -85,6 +85,10 @@ int btrfs_init_log_root_tree(struct btrfs_trans_handle *trans,
 			     struct btrfs_fs_info *fs_info);
 int btrfs_add_log_tree(struct btrfs_trans_handle *trans,
 		       struct btrfs_root *root);
+int btrfs_cleanup_transaction(struct btrfs_root *root);
+void btrfs_cleanup_one_transaction(struct btrfs_transaction *trans,
+				  struct btrfs_root *root);
+void btrfs_abort_devices(struct btrfs_root *root);
 
 #ifdef CONFIG_DEBUG_LOCK_ALLOC
 void btrfs_init_lockdep(void);
diff --git a/fs/btrfs/export.c b/fs/btrfs/export.c
index 5f77166fd01..e887ee62b6d 100644
--- a/fs/btrfs/export.c
+++ b/fs/btrfs/export.c
@@ -193,7 +193,7 @@ static struct dentry *btrfs_get_parent(struct dentry *child)
 	if (ret < 0)
 		goto fail;
 
-	BUG_ON(ret == 0);
+	BUG_ON(ret == 0); /* Key with offset of -1 found */
 	if (path->slots[0] == 0) {
 		ret = -ENOENT;
 		goto fail;
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 37e0a800d34..a84420491c1 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -245,7 +245,7 @@ static int exclude_super_stripes(struct btrfs_root *root,
 		cache->bytes_super += stripe_len;
 		ret = add_excluded_extent(root, cache->key.objectid,
 					  stripe_len);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 	}
 
 	for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
@@ -253,13 +253,13 @@ static int exclude_super_stripes(struct btrfs_root *root,
 		ret = btrfs_rmap_block(&root->fs_info->mapping_tree,
 				       cache->key.objectid, bytenr,
 				       0, &logical, &nr, &stripe_len);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 
 		while (nr--) {
 			cache->bytes_super += stripe_len;
 			ret = add_excluded_extent(root, logical[nr],
 						  stripe_len);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 		}
 
 		kfree(logical);
@@ -321,7 +321,7 @@ static u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
 			total_added += size;
 			ret = btrfs_add_free_space(block_group, start,
 						   size);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM or logic error */
 			start = extent_end + 1;
 		} else {
 			break;
@@ -332,7 +332,7 @@ static u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
 		size = end - start;
 		total_added += size;
 		ret = btrfs_add_free_space(block_group, start, size);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM or logic error */
 	}
 
 	return total_added;
@@ -474,7 +474,8 @@ static int cache_block_group(struct btrfs_block_group_cache *cache,
 	int ret = 0;
 
 	caching_ctl = kzalloc(sizeof(*caching_ctl), GFP_NOFS);
-	BUG_ON(!caching_ctl);
+	if (!caching_ctl)
+		return -ENOMEM;
 
 	INIT_LIST_HEAD(&caching_ctl->list);
 	mutex_init(&caching_ctl->mutex);
@@ -982,7 +983,7 @@ static int convert_extent_item_v0(struct btrfs_trans_handle *trans,
 				ret = btrfs_next_leaf(root, path);
 				if (ret < 0)
 					return ret;
-				BUG_ON(ret > 0);
+				BUG_ON(ret > 0); /* Corruption */
 				leaf = path->nodes[0];
 			}
 			btrfs_item_key_to_cpu(leaf, &found_key,
@@ -1008,9 +1009,9 @@ static int convert_extent_item_v0(struct btrfs_trans_handle *trans,
 				new_size + extra_size, 1);
 	if (ret < 0)
 		return ret;
-	BUG_ON(ret);
+	BUG_ON(ret); /* Corruption */
 
-	ret = btrfs_extend_item(trans, root, path, new_size);
+	btrfs_extend_item(trans, root, path, new_size);
 
 	leaf = path->nodes[0];
 	item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
@@ -1478,7 +1479,11 @@ int lookup_inline_extent_backref(struct btrfs_trans_handle *trans,
 		err = ret;
 		goto out;
 	}
-	BUG_ON(ret);
+	if (ret && !insert) {
+		err = -ENOENT;
+		goto out;
+	}
+	BUG_ON(ret); /* Corruption */
 
 	leaf = path->nodes[0];
 	item_size = btrfs_item_size_nr(leaf, path->slots[0]);
@@ -1592,13 +1597,13 @@ out:
  * helper to add new inline back ref
  */
 static noinline_for_stack
-int setup_inline_extent_backref(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root,
-				struct btrfs_path *path,
-				struct btrfs_extent_inline_ref *iref,
-				u64 parent, u64 root_objectid,
-				u64 owner, u64 offset, int refs_to_add,
-				struct btrfs_delayed_extent_op *extent_op)
+void setup_inline_extent_backref(struct btrfs_trans_handle *trans,
+				 struct btrfs_root *root,
+				 struct btrfs_path *path,
+				 struct btrfs_extent_inline_ref *iref,
+				 u64 parent, u64 root_objectid,
+				 u64 owner, u64 offset, int refs_to_add,
+				 struct btrfs_delayed_extent_op *extent_op)
 {
 	struct extent_buffer *leaf;
 	struct btrfs_extent_item *ei;
@@ -1608,7 +1613,6 @@ int setup_inline_extent_backref(struct btrfs_trans_handle *trans,
 	u64 refs;
 	int size;
 	int type;
-	int ret;
 
 	leaf = path->nodes[0];
 	ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
@@ -1617,7 +1621,7 @@ int setup_inline_extent_backref(struct btrfs_trans_handle *trans,
 	type = extent_ref_type(parent, owner);
 	size = btrfs_extent_inline_ref_size(type);
 
-	ret = btrfs_extend_item(trans, root, path, size);
+	btrfs_extend_item(trans, root, path, size);
 
 	ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
 	refs = btrfs_extent_refs(leaf, ei);
@@ -1652,7 +1656,6 @@ int setup_inline_extent_backref(struct btrfs_trans_handle *trans,
 		btrfs_set_extent_inline_ref_offset(leaf, iref, root_objectid);
 	}
 	btrfs_mark_buffer_dirty(leaf);
-	return 0;
 }
 
 static int lookup_extent_backref(struct btrfs_trans_handle *trans,
@@ -1687,12 +1690,12 @@ static int lookup_extent_backref(struct btrfs_trans_handle *trans,
  * helper to update/remove inline back ref
  */
 static noinline_for_stack
-int update_inline_extent_backref(struct btrfs_trans_handle *trans,
-				 struct btrfs_root *root,
-				 struct btrfs_path *path,
-				 struct btrfs_extent_inline_ref *iref,
-				 int refs_to_mod,
-				 struct btrfs_delayed_extent_op *extent_op)
+void update_inline_extent_backref(struct btrfs_trans_handle *trans,
+				  struct btrfs_root *root,
+				  struct btrfs_path *path,
+				  struct btrfs_extent_inline_ref *iref,
+				  int refs_to_mod,
+				  struct btrfs_delayed_extent_op *extent_op)
 {
 	struct extent_buffer *leaf;
 	struct btrfs_extent_item *ei;
@@ -1703,7 +1706,6 @@ int update_inline_extent_backref(struct btrfs_trans_handle *trans,
 	u32 item_size;
 	int size;
 	int type;
-	int ret;
 	u64 refs;
 
 	leaf = path->nodes[0];
@@ -1745,10 +1747,9 @@ int update_inline_extent_backref(struct btrfs_trans_handle *trans,
 			memmove_extent_buffer(leaf, ptr, ptr + size,
 					      end - ptr - size);
 		item_size -= size;
-		ret = btrfs_truncate_item(trans, root, path, item_size, 1);
+		btrfs_truncate_item(trans, root, path, item_size, 1);
 	}
 	btrfs_mark_buffer_dirty(leaf);
-	return 0;
 }
 
 static noinline_for_stack
@@ -1768,13 +1769,13 @@ int insert_inline_extent_backref(struct btrfs_trans_handle *trans,
 					   root_objectid, owner, offset, 1);
 	if (ret == 0) {
 		BUG_ON(owner < BTRFS_FIRST_FREE_OBJECTID);
-		ret = update_inline_extent_backref(trans, root, path, iref,
-						   refs_to_add, extent_op);
+		update_inline_extent_backref(trans, root, path, iref,
+					     refs_to_add, extent_op);
 	} else if (ret == -ENOENT) {
-		ret = setup_inline_extent_backref(trans, root, path, iref,
-						  parent, root_objectid,
-						  owner, offset, refs_to_add,
-						  extent_op);
+		setup_inline_extent_backref(trans, root, path, iref, parent,
+					    root_objectid, owner, offset,
+					    refs_to_add, extent_op);
+		ret = 0;
 	}
 	return ret;
 }
@@ -1804,12 +1805,12 @@ static int remove_extent_backref(struct btrfs_trans_handle *trans,
 				 struct btrfs_extent_inline_ref *iref,
 				 int refs_to_drop, int is_data)
 {
-	int ret;
+	int ret = 0;
 
 	BUG_ON(!is_data && refs_to_drop != 1);
 	if (iref) {
-		ret = update_inline_extent_backref(trans, root, path, iref,
-						   -refs_to_drop, NULL);
+		update_inline_extent_backref(trans, root, path, iref,
+					     -refs_to_drop, NULL);
 	} else if (is_data) {
 		ret = remove_extent_data_ref(trans, root, path, refs_to_drop);
 	} else {
@@ -1835,6 +1836,7 @@ static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
 	/* Tell the block device(s) that the sectors can be discarded */
 	ret = btrfs_map_block(&root->fs_info->mapping_tree, REQ_DISCARD,
 			      bytenr, &num_bytes, &bbio, 0);
+	/* Error condition is -ENOMEM */
 	if (!ret) {
 		struct btrfs_bio_stripe *stripe = bbio->stripes;
 		int i;
@@ -1850,7 +1852,7 @@ static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
 			if (!ret)
 				discarded_bytes += stripe->length;
 			else if (ret != -EOPNOTSUPP)
-				break;
+				break; /* Logic errors or -ENOMEM, or -EIO but I don't know how that could happen JDM */
 
 			/*
 			 * Just in case we get back EOPNOTSUPP for some reason,
@@ -1869,6 +1871,7 @@ static int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
 	return ret;
 }
 
+/* Can return -ENOMEM */
 int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 			 struct btrfs_root *root,
 			 u64 bytenr, u64 num_bytes, u64 parent,
@@ -1944,7 +1947,8 @@ static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 	ret = insert_extent_backref(trans, root->fs_info->extent_root,
 				    path, bytenr, parent, root_objectid,
 				    owner, offset, refs_to_add);
-	BUG_ON(ret);
+	if (ret)
+		btrfs_abort_transaction(trans, root, ret);
 out:
 	btrfs_free_path(path);
 	return err;
@@ -2031,6 +2035,9 @@ static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
 	int ret;
 	int err = 0;
 
+	if (trans->aborted)
+		return 0;
+
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
@@ -2128,7 +2135,11 @@ static int run_one_delayed_ref(struct btrfs_trans_handle *trans,
 			       struct btrfs_delayed_extent_op *extent_op,
 			       int insert_reserved)
 {
-	int ret;
+	int ret = 0;
+
+	if (trans->aborted)
+		return 0;
+
 	if (btrfs_delayed_ref_is_head(node)) {
 		struct btrfs_delayed_ref_head *head;
 		/*
@@ -2146,11 +2157,10 @@ static int run_one_delayed_ref(struct btrfs_trans_handle *trans,
 				ret = btrfs_del_csums(trans, root,
 						      node->bytenr,
 						      node->num_bytes);
-				BUG_ON(ret);
 			}
 		}
 		mutex_unlock(&head->mutex);
-		return 0;
+		return ret;
 	}
 
 	if (node->type == BTRFS_TREE_BLOCK_REF_KEY ||
@@ -2197,6 +2207,10 @@ again:
 	return NULL;
 }
 
+/*
+ * Returns 0 on success or if called with an already aborted transaction.
+ * Returns -ENOMEM or -EIO on failure and will abort the transaction.
+ */
 static noinline int run_clustered_refs(struct btrfs_trans_handle *trans,
 				       struct btrfs_root *root,
 				       struct list_head *cluster)
@@ -2285,9 +2299,13 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans,
 
 				ret = run_delayed_extent_op(trans, root,
 							    ref, extent_op);
-				BUG_ON(ret);
 				kfree(extent_op);
 
+				if (ret) {
+					printk(KERN_DEBUG "btrfs: run_delayed_extent_op returned %d\n", ret);
+					return ret;
+				}
+
 				goto next;
 			}
 
@@ -2308,11 +2326,16 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans,
 
 		ret = run_one_delayed_ref(trans, root, ref, extent_op,
 					  must_insert_reserved);
-		BUG_ON(ret);
 
 		btrfs_put_delayed_ref(ref);
 		kfree(extent_op);
 		count++;
+
+		if (ret) {
+			printk(KERN_DEBUG "btrfs: run_one_delayed_ref returned %d\n", ret);
+			return ret;
+		}
+
 next:
 		do_chunk_alloc(trans, root->fs_info->extent_root,
 			       2 * 1024 * 1024,
@@ -2347,6 +2370,9 @@ static void wait_for_more_refs(struct btrfs_delayed_ref_root *delayed_refs,
  * 0, which means to process everything in the tree at the start
  * of the run (but not newly added entries), or it can be some target
  * number you'd like to process.
+ *
+ * Returns 0 on success or if called with an aborted transaction
+ * Returns <0 on error and aborts the transaction
  */
 int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 			   struct btrfs_root *root, unsigned long count)
@@ -2362,6 +2388,10 @@ int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 	unsigned long num_refs = 0;
 	int consider_waiting;
 
+	/* We'll clean this up in btrfs_cleanup_transaction */
+	if (trans->aborted)
+		return 0;
+
 	if (root == root->fs_info->extent_root)
 		root = root->fs_info->tree_root;
 
@@ -2419,7 +2449,11 @@ again:
 		}
 
 		ret = run_clustered_refs(trans, root, &cluster);
-		BUG_ON(ret < 0);
+		if (ret < 0) {
+			spin_unlock(&delayed_refs->lock);
+			btrfs_abort_transaction(trans, root, ret);
+			return ret;
+		}
 
 		count -= min_t(unsigned long, ret, count);
 
@@ -2584,7 +2618,7 @@ static noinline int check_committed_ref(struct btrfs_trans_handle *trans,
 	ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
 	if (ret < 0)
 		goto out;
-	BUG_ON(ret == 0);
+	BUG_ON(ret == 0); /* Corruption */
 
 	ret = -ENOENT;
 	if (path->slots[0] == 0)
@@ -2738,7 +2772,6 @@ static int __btrfs_mod_ref(struct btrfs_trans_handle *trans,
 	}
 	return 0;
 fail:
-	BUG();
 	return ret;
 }
 
@@ -2767,7 +2800,7 @@ static int write_one_cache_group(struct btrfs_trans_handle *trans,
 	ret = btrfs_search_slot(trans, extent_root, &cache->key, path, 0, 1);
 	if (ret < 0)
 		goto fail;
-	BUG_ON(ret);
+	BUG_ON(ret); /* Corruption */
 
 	leaf = path->nodes[0];
 	bi = btrfs_item_ptr_offset(leaf, path->slots[0]);
@@ -2775,8 +2808,10 @@ static int write_one_cache_group(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_release_path(path);
 fail:
-	if (ret)
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
 		return ret;
+	}
 	return 0;
 
 }
@@ -2949,7 +2984,8 @@ again:
 		if (last == 0) {
 			err = btrfs_run_delayed_refs(trans, root,
 						     (unsigned long)-1);
-			BUG_ON(err);
+			if (err) /* File system offline */
+				goto out;
 		}
 
 		cache = btrfs_lookup_first_block_group(root->fs_info, last);
@@ -2976,7 +3012,9 @@ again:
 		last = cache->key.objectid + cache->key.offset;
 
 		err = write_one_cache_group(trans, root, path, cache);
-		BUG_ON(err);
+		if (err) /* File system offline */
+			goto out;
+
 		btrfs_put_block_group(cache);
 	}
 
@@ -2989,7 +3027,8 @@ again:
 		if (last == 0) {
 			err = btrfs_run_delayed_refs(trans, root,
 						     (unsigned long)-1);
-			BUG_ON(err);
+			if (err) /* File system offline */
+				goto out;
 		}
 
 		cache = btrfs_lookup_first_block_group(root->fs_info, last);
@@ -3014,20 +3053,21 @@ again:
 			continue;
 		}
 
-		btrfs_write_out_cache(root, trans, cache, path);
+		err = btrfs_write_out_cache(root, trans, cache, path);
 
 		/*
 		 * If we didn't have an error then the cache state is still
 		 * NEED_WRITE, so we can set it to WRITTEN.
 		 */
-		if (cache->disk_cache_state == BTRFS_DC_NEED_WRITE)
+		if (!err && cache->disk_cache_state == BTRFS_DC_NEED_WRITE)
 			cache->disk_cache_state = BTRFS_DC_WRITTEN;
 		last = cache->key.objectid + cache->key.offset;
 		btrfs_put_block_group(cache);
 	}
+out:
 
 	btrfs_free_path(path);
-	return 0;
+	return err;
 }
 
 int btrfs_extent_readonly(struct btrfs_root *root, u64 bytenr)
@@ -3098,11 +3138,8 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags,
 
 static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
 {
-	u64 extra_flags = flags & BTRFS_BLOCK_GROUP_PROFILE_MASK;
-
-	/* chunk -> extended profile */
-	if (extra_flags == 0)
-		extra_flags = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
+	u64 extra_flags = chunk_to_extended(flags) &
+				BTRFS_EXTENDED_PROFILE_MASK;
 
 	if (flags & BTRFS_BLOCK_GROUP_DATA)
 		fs_info->avail_data_alloc_bits |= extra_flags;
@@ -3113,6 +3150,35 @@ static void set_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
 }
 
 /*
+ * returns target flags in extended format or 0 if restripe for this
+ * chunk_type is not in progress
+ */
+static u64 get_restripe_target(struct btrfs_fs_info *fs_info, u64 flags)
+{
+	struct btrfs_balance_control *bctl = fs_info->balance_ctl;
+	u64 target = 0;
+
+	BUG_ON(!mutex_is_locked(&fs_info->volume_mutex) &&
+	       !spin_is_locked(&fs_info->balance_lock));
+
+	if (!bctl)
+		return 0;
+
+	if (flags & BTRFS_BLOCK_GROUP_DATA &&
+	    bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) {
+		target = BTRFS_BLOCK_GROUP_DATA | bctl->data.target;
+	} else if (flags & BTRFS_BLOCK_GROUP_SYSTEM &&
+		   bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
+		target = BTRFS_BLOCK_GROUP_SYSTEM | bctl->sys.target;
+	} else if (flags & BTRFS_BLOCK_GROUP_METADATA &&
+		   bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) {
+		target = BTRFS_BLOCK_GROUP_METADATA | bctl->meta.target;
+	}
+
+	return target;
+}
+
+/*
  * @flags: available profiles in extended format (see ctree.h)
  *
  * Returns reduced profile in chunk format.  If profile changing is in
@@ -3128,31 +3194,19 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags)
 	 */
 	u64 num_devices = root->fs_info->fs_devices->rw_devices +
 		root->fs_info->fs_devices->missing_devices;
+	u64 target;
 
-	/* pick restriper's target profile if it's available */
+	/*
+	 * see if restripe for this chunk_type is in progress, if so
+	 * try to reduce to the target profile
+	 */
 	spin_lock(&root->fs_info->balance_lock);
-	if (root->fs_info->balance_ctl) {
-		struct btrfs_balance_control *bctl = root->fs_info->balance_ctl;
-		u64 tgt = 0;
-
-		if ((flags & BTRFS_BLOCK_GROUP_DATA) &&
-		    (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
-		    (flags & bctl->data.target)) {
-			tgt = BTRFS_BLOCK_GROUP_DATA | bctl->data.target;
-		} else if ((flags & BTRFS_BLOCK_GROUP_SYSTEM) &&
-			   (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
-			   (flags & bctl->sys.target)) {
-			tgt = BTRFS_BLOCK_GROUP_SYSTEM | bctl->sys.target;
-		} else if ((flags & BTRFS_BLOCK_GROUP_METADATA) &&
-			   (bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
-			   (flags & bctl->meta.target)) {
-			tgt = BTRFS_BLOCK_GROUP_METADATA | bctl->meta.target;
-		}
-
-		if (tgt) {
+	target = get_restripe_target(root->fs_info, flags);
+	if (target) {
+		/* pick target profile only if it's already available */
+		if ((flags & target) & BTRFS_EXTENDED_PROFILE_MASK) {
 			spin_unlock(&root->fs_info->balance_lock);
-			flags = tgt;
-			goto out;
+			return extended_to_chunk(target);
 		}
 	}
 	spin_unlock(&root->fs_info->balance_lock);
@@ -3180,10 +3234,7 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags)
 		flags &= ~BTRFS_BLOCK_GROUP_RAID0;
 	}
 
-out:
-	/* extended -> chunk profile */
-	flags &= ~BTRFS_AVAIL_ALLOC_BIT_SINGLE;
-	return flags;
+	return extended_to_chunk(flags);
 }
 
 static u64 get_alloc_profile(struct btrfs_root *root, u64 flags)
@@ -3312,8 +3363,7 @@ commit_trans:
 	}
 	data_sinfo->bytes_may_use += bytes;
 	trace_btrfs_space_reservation(root->fs_info, "space_info",
-				      (u64)(unsigned long)data_sinfo,
-				      bytes, 1);
+				      data_sinfo->flags, bytes, 1);
 	spin_unlock(&data_sinfo->lock);
 
 	return 0;
@@ -3334,8 +3384,7 @@ void btrfs_free_reserved_data_space(struct inode *inode, u64 bytes)
 	spin_lock(&data_sinfo->lock);
 	data_sinfo->bytes_may_use -= bytes;
 	trace_btrfs_space_reservation(root->fs_info, "space_info",
-				      (u64)(unsigned long)data_sinfo,
-				      bytes, 0);
+				      data_sinfo->flags, bytes, 0);
 	spin_unlock(&data_sinfo->lock);
 }
 
@@ -3396,6 +3445,50 @@ static int should_alloc_chunk(struct btrfs_root *root,
 	return 1;
 }
 
+static u64 get_system_chunk_thresh(struct btrfs_root *root, u64 type)
+{
+	u64 num_dev;
+
+	if (type & BTRFS_BLOCK_GROUP_RAID10 ||
+	    type & BTRFS_BLOCK_GROUP_RAID0)
+		num_dev = root->fs_info->fs_devices->rw_devices;
+	else if (type & BTRFS_BLOCK_GROUP_RAID1)
+		num_dev = 2;
+	else
+		num_dev = 1;	/* DUP or single */
+
+	/* metadata for updaing devices and chunk tree */
+	return btrfs_calc_trans_metadata_size(root, num_dev + 1);
+}
+
+static void check_system_chunk(struct btrfs_trans_handle *trans,
+			       struct btrfs_root *root, u64 type)
+{
+	struct btrfs_space_info *info;
+	u64 left;
+	u64 thresh;
+
+	info = __find_space_info(root->fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
+	spin_lock(&info->lock);
+	left = info->total_bytes - info->bytes_used - info->bytes_pinned -
+		info->bytes_reserved - info->bytes_readonly;
+	spin_unlock(&info->lock);
+
+	thresh = get_system_chunk_thresh(root, type);
+	if (left < thresh && btrfs_test_opt(root, ENOSPC_DEBUG)) {
+		printk(KERN_INFO "left=%llu, need=%llu, flags=%llu\n",
+		       left, thresh, type);
+		dump_space_info(info, 0, 0);
+	}
+
+	if (left < thresh) {
+		u64 flags;
+
+		flags = btrfs_get_alloc_profile(root->fs_info->chunk_root, 0);
+		btrfs_alloc_chunk(trans, root, flags);
+	}
+}
+
 static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *extent_root, u64 alloc_bytes,
 			  u64 flags, int force)
@@ -3405,15 +3498,13 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 	int wait_for_alloc = 0;
 	int ret = 0;
 
-	BUG_ON(!profile_is_valid(flags, 0));
-
 	space_info = __find_space_info(extent_root->fs_info, flags);
 	if (!space_info) {
 		ret = update_space_info(extent_root->fs_info, flags,
 					0, 0, &space_info);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 	}
-	BUG_ON(!space_info);
+	BUG_ON(!space_info); /* Logic error */
 
 again:
 	spin_lock(&space_info->lock);
@@ -3468,6 +3559,12 @@ again:
 			force_metadata_allocation(fs_info);
 	}
 
+	/*
+	 * Check if we have enough space in SYSTEM chunk because we may need
+	 * to update devices.
+	 */
+	check_system_chunk(trans, extent_root, flags);
+
 	ret = btrfs_alloc_chunk(trans, extent_root, flags);
 	if (ret < 0 && ret != -ENOSPC)
 		goto out;
@@ -3678,8 +3775,10 @@ again:
 		ret = wait_event_interruptible(space_info->wait,
 					       !space_info->flush);
 		/* Must have been interrupted, return */
-		if (ret)
+		if (ret) {
+			printk(KERN_DEBUG "btrfs: %s returning -EINTR\n", __func__);
 			return -EINTR;
+		}
 
 		spin_lock(&space_info->lock);
 	}
@@ -3700,9 +3799,7 @@ again:
 		if (used + orig_bytes <= space_info->total_bytes) {
 			space_info->bytes_may_use += orig_bytes;
 			trace_btrfs_space_reservation(root->fs_info,
-					      "space_info",
-					      (u64)(unsigned long)space_info,
-					      orig_bytes, 1);
+				"space_info", space_info->flags, orig_bytes, 1);
 			ret = 0;
 		} else {
 			/*
@@ -3771,9 +3868,7 @@ again:
 		if (used + num_bytes < space_info->total_bytes + avail) {
 			space_info->bytes_may_use += orig_bytes;
 			trace_btrfs_space_reservation(root->fs_info,
-					      "space_info",
-					      (u64)(unsigned long)space_info,
-					      orig_bytes, 1);
+				"space_info", space_info->flags, orig_bytes, 1);
 			ret = 0;
 		} else {
 			wait_ordered = true;
@@ -3836,8 +3931,9 @@ out:
 	return ret;
 }
 
-static struct btrfs_block_rsv *get_block_rsv(struct btrfs_trans_handle *trans,
-					     struct btrfs_root *root)
+static struct btrfs_block_rsv *get_block_rsv(
+					const struct btrfs_trans_handle *trans,
+					const struct btrfs_root *root)
 {
 	struct btrfs_block_rsv *block_rsv = NULL;
 
@@ -3918,8 +4014,7 @@ static void block_rsv_release_bytes(struct btrfs_fs_info *fs_info,
 			spin_lock(&space_info->lock);
 			space_info->bytes_may_use -= num_bytes;
 			trace_btrfs_space_reservation(fs_info, "space_info",
-					      (u64)(unsigned long)space_info,
-					      num_bytes, 0);
+					space_info->flags, num_bytes, 0);
 			space_info->reservation_progress++;
 			spin_unlock(&space_info->lock);
 		}
@@ -4137,14 +4232,14 @@ static void update_global_block_rsv(struct btrfs_fs_info *fs_info)
 		block_rsv->reserved += num_bytes;
 		sinfo->bytes_may_use += num_bytes;
 		trace_btrfs_space_reservation(fs_info, "space_info",
-				      (u64)(unsigned long)sinfo, num_bytes, 1);
+				      sinfo->flags, num_bytes, 1);
 	}
 
 	if (block_rsv->reserved >= block_rsv->size) {
 		num_bytes = block_rsv->reserved - block_rsv->size;
 		sinfo->bytes_may_use -= num_bytes;
 		trace_btrfs_space_reservation(fs_info, "space_info",
-				      (u64)(unsigned long)sinfo, num_bytes, 0);
+				      sinfo->flags, num_bytes, 0);
 		sinfo->reservation_progress++;
 		block_rsv->reserved = block_rsv->size;
 		block_rsv->full = 1;
@@ -4198,12 +4293,12 @@ void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
 		return;
 
 	trace_btrfs_space_reservation(root->fs_info, "transaction",
-				      (u64)(unsigned long)trans,
-				      trans->bytes_reserved, 0);
+				      trans->transid, trans->bytes_reserved, 0);
 	btrfs_block_rsv_release(root, trans->block_rsv, trans->bytes_reserved);
 	trans->bytes_reserved = 0;
 }
 
+/* Can only return 0 or -ENOSPC */
 int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans,
 				  struct inode *inode)
 {
@@ -4540,7 +4635,7 @@ static int update_block_group(struct btrfs_trans_handle *trans,
 	while (total) {
 		cache = btrfs_lookup_block_group(info, bytenr);
 		if (!cache)
-			return -1;
+			return -ENOENT;
 		if (cache->flags & (BTRFS_BLOCK_GROUP_DUP |
 				    BTRFS_BLOCK_GROUP_RAID1 |
 				    BTRFS_BLOCK_GROUP_RAID10))
@@ -4643,7 +4738,7 @@ int btrfs_pin_extent(struct btrfs_root *root,
 	struct btrfs_block_group_cache *cache;
 
 	cache = btrfs_lookup_block_group(root->fs_info, bytenr);
-	BUG_ON(!cache);
+	BUG_ON(!cache); /* Logic error */
 
 	pin_down_extent(root, cache, bytenr, num_bytes, reserved);
 
@@ -4661,7 +4756,7 @@ int btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle *trans,
 	struct btrfs_block_group_cache *cache;
 
 	cache = btrfs_lookup_block_group(root->fs_info, bytenr);
-	BUG_ON(!cache);
+	BUG_ON(!cache); /* Logic error */
 
 	/*
 	 * pull in the free space cache (if any) so that our pin
@@ -4706,6 +4801,7 @@ static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache,
 {
 	struct btrfs_space_info *space_info = cache->space_info;
 	int ret = 0;
+
 	spin_lock(&space_info->lock);
 	spin_lock(&cache->lock);
 	if (reserve != RESERVE_FREE) {
@@ -4716,9 +4812,8 @@ static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache,
 			space_info->bytes_reserved += num_bytes;
 			if (reserve == RESERVE_ALLOC) {
 				trace_btrfs_space_reservation(cache->fs_info,
-					      "space_info",
-					      (u64)(unsigned long)space_info,
-					      num_bytes, 0);
+						"space_info", space_info->flags,
+						num_bytes, 0);
 				space_info->bytes_may_use -= num_bytes;
 			}
 		}
@@ -4734,7 +4829,7 @@ static int btrfs_update_reserved_bytes(struct btrfs_block_group_cache *cache,
 	return ret;
 }
 
-int btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans,
+void btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -4764,7 +4859,6 @@ int btrfs_prepare_extent_commit(struct btrfs_trans_handle *trans,
 	up_write(&fs_info->extent_commit_sem);
 
 	update_global_block_rsv(fs_info);
-	return 0;
 }
 
 static int unpin_extent_range(struct btrfs_root *root, u64 start, u64 end)
@@ -4779,7 +4873,7 @@ static int unpin_extent_range(struct btrfs_root *root, u64 start, u64 end)
 			if (cache)
 				btrfs_put_block_group(cache);
 			cache = btrfs_lookup_block_group(fs_info, start);
-			BUG_ON(!cache);
+			BUG_ON(!cache); /* Logic error */
 		}
 
 		len = cache->key.objectid + cache->key.offset - start;
@@ -4816,6 +4910,9 @@ int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans,
 	u64 end;
 	int ret;
 
+	if (trans->aborted)
+		return 0;
+
 	if (fs_info->pinned_extents == &fs_info->freed_extents[0])
 		unpin = &fs_info->freed_extents[1];
 	else
@@ -4901,7 +4998,8 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 			ret = remove_extent_backref(trans, extent_root, path,
 						    NULL, refs_to_drop,
 						    is_data);
-			BUG_ON(ret);
+			if (ret)
+				goto abort;
 			btrfs_release_path(path);
 			path->leave_spinning = 1;
 
@@ -4919,10 +5017,11 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 					btrfs_print_leaf(extent_root,
 							 path->nodes[0]);
 			}
-			BUG_ON(ret);
+			if (ret < 0)
+				goto abort;
 			extent_slot = path->slots[0];
 		}
-	} else {
+	} else if (ret == -ENOENT) {
 		btrfs_print_leaf(extent_root, path->nodes[0]);
 		WARN_ON(1);
 		printk(KERN_ERR "btrfs unable to find ref byte nr %llu "
@@ -4932,6 +5031,8 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 		       (unsigned long long)root_objectid,
 		       (unsigned long long)owner_objectid,
 		       (unsigned long long)owner_offset);
+	} else {
+		goto abort;
 	}
 
 	leaf = path->nodes[0];
@@ -4941,7 +5042,8 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 		BUG_ON(found_extent || extent_slot != path->slots[0]);
 		ret = convert_extent_item_v0(trans, extent_root, path,
 					     owner_objectid, 0);
-		BUG_ON(ret < 0);
+		if (ret < 0)
+			goto abort;
 
 		btrfs_release_path(path);
 		path->leave_spinning = 1;
@@ -4958,7 +5060,8 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 			       (unsigned long long)bytenr);
 			btrfs_print_leaf(extent_root, path->nodes[0]);
 		}
-		BUG_ON(ret);
+		if (ret < 0)
+			goto abort;
 		extent_slot = path->slots[0];
 		leaf = path->nodes[0];
 		item_size = btrfs_item_size_nr(leaf, extent_slot);
@@ -4995,7 +5098,8 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 			ret = remove_extent_backref(trans, extent_root, path,
 						    iref, refs_to_drop,
 						    is_data);
-			BUG_ON(ret);
+			if (ret)
+				goto abort;
 		}
 	} else {
 		if (found_extent) {
@@ -5012,23 +5116,27 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 
 		ret = btrfs_del_items(trans, extent_root, path, path->slots[0],
 				      num_to_del);
-		BUG_ON(ret);
+		if (ret)
+			goto abort;
 		btrfs_release_path(path);
 
 		if (is_data) {
 			ret = btrfs_del_csums(trans, root, bytenr, num_bytes);
-			BUG_ON(ret);
-		} else {
-			invalidate_mapping_pages(info->btree_inode->i_mapping,
-			     bytenr >> PAGE_CACHE_SHIFT,
-			     (bytenr + num_bytes - 1) >> PAGE_CACHE_SHIFT);
+			if (ret)
+				goto abort;
 		}
 
 		ret = update_block_group(trans, root, bytenr, num_bytes, 0);
-		BUG_ON(ret);
+		if (ret)
+			goto abort;
 	}
+out:
 	btrfs_free_path(path);
 	return ret;
+
+abort:
+	btrfs_abort_transaction(trans, extent_root, ret);
+	goto out;
 }
 
 /*
@@ -5124,7 +5232,7 @@ void btrfs_free_tree_block(struct btrfs_trans_handle *trans,
 					parent, root->root_key.objectid,
 					btrfs_header_level(buf),
 					BTRFS_DROP_DELAYED_REF, NULL, for_cow);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 	}
 
 	if (!last_ref)
@@ -5158,6 +5266,7 @@ out:
 	btrfs_put_block_group(cache);
 }
 
+/* Can return -ENOMEM */
 int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		      u64 bytenr, u64 num_bytes, u64 parent, u64 root_objectid,
 		      u64 owner, u64 offset, int for_cow)
@@ -5179,14 +5288,12 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 					num_bytes,
 					parent, root_objectid, (int)owner,
 					BTRFS_DROP_DELAYED_REF, NULL, for_cow);
-		BUG_ON(ret);
 	} else {
 		ret = btrfs_add_delayed_data_ref(fs_info, trans, bytenr,
 						num_bytes,
 						parent, root_objectid, owner,
 						offset, BTRFS_DROP_DELAYED_REF,
 						NULL, for_cow);
-		BUG_ON(ret);
 	}
 	return ret;
 }
@@ -5243,28 +5350,34 @@ wait_block_group_cache_done(struct btrfs_block_group_cache *cache)
 	return 0;
 }
 
-static int get_block_group_index(struct btrfs_block_group_cache *cache)
+static int __get_block_group_index(u64 flags)
 {
 	int index;
-	if (cache->flags & BTRFS_BLOCK_GROUP_RAID10)
+
+	if (flags & BTRFS_BLOCK_GROUP_RAID10)
 		index = 0;
-	else if (cache->flags & BTRFS_BLOCK_GROUP_RAID1)
+	else if (flags & BTRFS_BLOCK_GROUP_RAID1)
 		index = 1;
-	else if (cache->flags & BTRFS_BLOCK_GROUP_DUP)
+	else if (flags & BTRFS_BLOCK_GROUP_DUP)
 		index = 2;
-	else if (cache->flags & BTRFS_BLOCK_GROUP_RAID0)
+	else if (flags & BTRFS_BLOCK_GROUP_RAID0)
 		index = 3;
 	else
 		index = 4;
+
 	return index;
 }
 
+static int get_block_group_index(struct btrfs_block_group_cache *cache)
+{
+	return __get_block_group_index(cache->flags);
+}
+
 enum btrfs_loop_type {
-	LOOP_FIND_IDEAL = 0,
-	LOOP_CACHING_NOWAIT = 1,
-	LOOP_CACHING_WAIT = 2,
-	LOOP_ALLOC_CHUNK = 3,
-	LOOP_NO_EMPTY_SIZE = 4,
+	LOOP_CACHING_NOWAIT = 0,
+	LOOP_CACHING_WAIT = 1,
+	LOOP_ALLOC_CHUNK = 2,
+	LOOP_NO_EMPTY_SIZE = 3,
 };
 
 /*
@@ -5278,7 +5391,6 @@ enum btrfs_loop_type {
 static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 				     struct btrfs_root *orig_root,
 				     u64 num_bytes, u64 empty_size,
-				     u64 search_start, u64 search_end,
 				     u64 hint_byte, struct btrfs_key *ins,
 				     u64 data)
 {
@@ -5287,6 +5399,7 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 	struct btrfs_free_cluster *last_ptr = NULL;
 	struct btrfs_block_group_cache *block_group = NULL;
 	struct btrfs_block_group_cache *used_block_group;
+	u64 search_start = 0;
 	int empty_cluster = 2 * 1024 * 1024;
 	int allowed_chunk_alloc = 0;
 	int done_chunk_alloc = 0;
@@ -5300,8 +5413,6 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 	bool failed_alloc = false;
 	bool use_cluster = true;
 	bool have_caching_bg = false;
-	u64 ideal_cache_percent = 0;
-	u64 ideal_cache_offset = 0;
 
 	WARN_ON(num_bytes < root->sectorsize);
 	btrfs_set_key_type(ins, BTRFS_EXTENT_ITEM_KEY);
@@ -5351,7 +5462,6 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 		empty_cluster = 0;
 
 	if (search_start == hint_byte) {
-ideal_cache:
 		block_group = btrfs_lookup_block_group(root->fs_info,
 						       search_start);
 		used_block_group = block_group;
@@ -5363,8 +5473,7 @@ ideal_cache:
 		 * picked out then we don't care that the block group is cached.
 		 */
 		if (block_group && block_group_bits(block_group, data) &&
-		    (block_group->cached != BTRFS_CACHE_NO ||
-		     search_start == ideal_cache_offset)) {
+		    block_group->cached != BTRFS_CACHE_NO) {
 			down_read(&space_info->groups_sem);
 			if (list_empty(&block_group->list) ||
 			    block_group->ro) {
@@ -5418,44 +5527,13 @@ search:
 have_block_group:
 		cached = block_group_cache_done(block_group);
 		if (unlikely(!cached)) {
-			u64 free_percent;
-
 			found_uncached_bg = true;
 			ret = cache_block_group(block_group, trans,
-						orig_root, 1);
-			if (block_group->cached == BTRFS_CACHE_FINISHED)
-				goto alloc;
-
-			free_percent = btrfs_block_group_used(&block_group->item);
-			free_percent *= 100;
-			free_percent = div64_u64(free_percent,
-						 block_group->key.offset);
-			free_percent = 100 - free_percent;
-			if (free_percent > ideal_cache_percent &&
-			    likely(!block_group->ro)) {
-				ideal_cache_offset = block_group->key.objectid;
-				ideal_cache_percent = free_percent;
-			}
-
-			/*
-			 * The caching workers are limited to 2 threads, so we
-			 * can queue as much work as we care to.
-			 */
-			if (loop > LOOP_FIND_IDEAL) {
-				ret = cache_block_group(block_group, trans,
-							orig_root, 0);
-				BUG_ON(ret);
-			}
-
-			/*
-			 * If loop is set for cached only, try the next block
-			 * group.
-			 */
-			if (loop == LOOP_FIND_IDEAL)
-				goto loop;
+						orig_root, 0);
+			BUG_ON(ret < 0);
+			ret = 0;
 		}
 
-alloc:
 		if (unlikely(block_group->ro))
 			goto loop;
 
@@ -5606,11 +5684,6 @@ unclustered_alloc:
 		}
 checks:
 		search_start = stripe_align(root, offset);
-		/* move on to the next group */
-		if (search_start + num_bytes >= search_end) {
-			btrfs_add_free_space(used_block_group, offset, num_bytes);
-			goto loop;
-		}
 
 		/* move on to the next group */
 		if (search_start + num_bytes >
@@ -5661,9 +5734,7 @@ loop:
 	if (!ins->objectid && ++index < BTRFS_NR_RAID_TYPES)
 		goto search;
 
-	/* LOOP_FIND_IDEAL, only search caching/cached bg's, and don't wait for
-	 *			for them to make caching progress.  Also
-	 *			determine the best possible bg to cache
+	/*
 	 * LOOP_CACHING_NOWAIT, search partially cached block groups, kicking
 	 *			caching kthreads as we move along
 	 * LOOP_CACHING_WAIT, search everything, and wait if our bg is caching
@@ -5673,50 +5744,17 @@ loop:
 	 */
 	if (!ins->objectid && loop < LOOP_NO_EMPTY_SIZE) {
 		index = 0;
-		if (loop == LOOP_FIND_IDEAL && found_uncached_bg) {
-			found_uncached_bg = false;
-			loop++;
-			if (!ideal_cache_percent)
-				goto search;
-
-			/*
-			 * 1 of the following 2 things have happened so far
-			 *
-			 * 1) We found an ideal block group for caching that
-			 * is mostly full and will cache quickly, so we might
-			 * as well wait for it.
-			 *
-			 * 2) We searched for cached only and we didn't find
-			 * anything, and we didn't start any caching kthreads
-			 * either, so chances are we will loop through and
-			 * start a couple caching kthreads, and then come back
-			 * around and just wait for them.  This will be slower
-			 * because we will have 2 caching kthreads reading at
-			 * the same time when we could have just started one
-			 * and waited for it to get far enough to give us an
-			 * allocation, so go ahead and go to the wait caching
-			 * loop.
-			 */
-			loop = LOOP_CACHING_WAIT;
-			search_start = ideal_cache_offset;
-			ideal_cache_percent = 0;
-			goto ideal_cache;
-		} else if (loop == LOOP_FIND_IDEAL) {
-			/*
-			 * Didn't find a uncached bg, wait on anything we find
-			 * next.
-			 */
-			loop = LOOP_CACHING_WAIT;
-			goto search;
-		}
-
 		loop++;
-
 		if (loop == LOOP_ALLOC_CHUNK) {
 		       if (allowed_chunk_alloc) {
 				ret = do_chunk_alloc(trans, root, num_bytes +
 						     2 * 1024 * 1024, data,
 						     CHUNK_ALLOC_LIMITED);
+				if (ret < 0) {
+					btrfs_abort_transaction(trans,
+								root, ret);
+					goto out;
+				}
 				allowed_chunk_alloc = 0;
 				if (ret == 1)
 					done_chunk_alloc = 1;
@@ -5745,6 +5783,7 @@ loop:
 	} else if (ins->objectid) {
 		ret = 0;
 	}
+out:
 
 	return ret;
 }
@@ -5798,12 +5837,10 @@ int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
 			 struct btrfs_root *root,
 			 u64 num_bytes, u64 min_alloc_size,
 			 u64 empty_size, u64 hint_byte,
-			 u64 search_end, struct btrfs_key *ins,
-			 u64 data)
+			 struct btrfs_key *ins, u64 data)
 {
 	bool final_tried = false;
 	int ret;
-	u64 search_start = 0;
 
 	data = btrfs_get_alloc_profile(root, data);
 again:
@@ -5811,23 +5848,31 @@ again:
 	 * the only place that sets empty_size is btrfs_realloc_node, which
 	 * is not called recursively on allocations
 	 */
-	if (empty_size || root->ref_cows)
+	if (empty_size || root->ref_cows) {
 		ret = do_chunk_alloc(trans, root->fs_info->extent_root,
 				     num_bytes + 2 * 1024 * 1024, data,
 				     CHUNK_ALLOC_NO_FORCE);
+		if (ret < 0 && ret != -ENOSPC) {
+			btrfs_abort_transaction(trans, root, ret);
+			return ret;
+		}
+	}
 
 	WARN_ON(num_bytes < root->sectorsize);
 	ret = find_free_extent(trans, root, num_bytes, empty_size,
-			       search_start, search_end, hint_byte,
-			       ins, data);
+			       hint_byte, ins, data);
 
 	if (ret == -ENOSPC) {
 		if (!final_tried) {
 			num_bytes = num_bytes >> 1;
 			num_bytes = num_bytes & ~(root->sectorsize - 1);
 			num_bytes = max(num_bytes, min_alloc_size);
-			do_chunk_alloc(trans, root->fs_info->extent_root,
+			ret = do_chunk_alloc(trans, root->fs_info->extent_root,
 				       num_bytes, data, CHUNK_ALLOC_FORCE);
+			if (ret < 0 && ret != -ENOSPC) {
+				btrfs_abort_transaction(trans, root, ret);
+				return ret;
+			}
 			if (num_bytes == min_alloc_size)
 				final_tried = true;
 			goto again;
@@ -5838,7 +5883,8 @@ again:
 			printk(KERN_ERR "btrfs allocation failed flags %llu, "
 			       "wanted %llu\n", (unsigned long long)data,
 			       (unsigned long long)num_bytes);
-			dump_space_info(sinfo, num_bytes, 1);
+			if (sinfo)
+				dump_space_info(sinfo, num_bytes, 1);
 		}
 	}
 
@@ -5917,7 +5963,10 @@ static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
 	path->leave_spinning = 1;
 	ret = btrfs_insert_empty_item(trans, fs_info->extent_root, path,
 				      ins, size);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_free_path(path);
+		return ret;
+	}
 
 	leaf = path->nodes[0];
 	extent_item = btrfs_item_ptr(leaf, path->slots[0],
@@ -5947,7 +5996,7 @@ static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
 	btrfs_free_path(path);
 
 	ret = update_block_group(trans, root, ins->objectid, ins->offset, 1);
-	if (ret) {
+	if (ret) { /* -ENOENT, logic error */
 		printk(KERN_ERR "btrfs update block group failed for %llu "
 		       "%llu\n", (unsigned long long)ins->objectid,
 		       (unsigned long long)ins->offset);
@@ -5978,7 +6027,10 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 	path->leave_spinning = 1;
 	ret = btrfs_insert_empty_item(trans, fs_info->extent_root, path,
 				      ins, size);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_free_path(path);
+		return ret;
+	}
 
 	leaf = path->nodes[0];
 	extent_item = btrfs_item_ptr(leaf, path->slots[0],
@@ -6008,7 +6060,7 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 	btrfs_free_path(path);
 
 	ret = update_block_group(trans, root, ins->objectid, ins->offset, 1);
-	if (ret) {
+	if (ret) { /* -ENOENT, logic error */
 		printk(KERN_ERR "btrfs update block group failed for %llu "
 		       "%llu\n", (unsigned long long)ins->objectid,
 		       (unsigned long long)ins->offset);
@@ -6056,28 +6108,28 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans,
 	if (!caching_ctl) {
 		BUG_ON(!block_group_cache_done(block_group));
 		ret = btrfs_remove_free_space(block_group, start, num_bytes);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 	} else {
 		mutex_lock(&caching_ctl->mutex);
 
 		if (start >= caching_ctl->progress) {
 			ret = add_excluded_extent(root, start, num_bytes);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 		} else if (start + num_bytes <= caching_ctl->progress) {
 			ret = btrfs_remove_free_space(block_group,
 						      start, num_bytes);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 		} else {
 			num_bytes = caching_ctl->progress - start;
 			ret = btrfs_remove_free_space(block_group,
 						      start, num_bytes);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 
 			start = caching_ctl->progress;
 			num_bytes = ins->objectid + ins->offset -
 				    caching_ctl->progress;
 			ret = add_excluded_extent(root, start, num_bytes);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 		}
 
 		mutex_unlock(&caching_ctl->mutex);
@@ -6086,7 +6138,7 @@ int btrfs_alloc_logged_file_extent(struct btrfs_trans_handle *trans,
 
 	ret = btrfs_update_reserved_bytes(block_group, ins->offset,
 					  RESERVE_ALLOC_NO_ACCOUNT);
-	BUG_ON(ret);
+	BUG_ON(ret); /* logic error */
 	btrfs_put_block_group(block_group);
 	ret = alloc_reserved_file_extent(trans, root, 0, root_objectid,
 					 0, owner, offset, ins, 1);
@@ -6107,6 +6159,7 @@ struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans,
 	btrfs_set_buffer_lockdep_class(root->root_key.objectid, buf, level);
 	btrfs_tree_lock(buf);
 	clean_tree_block(trans, root, buf);
+	clear_bit(EXTENT_BUFFER_STALE, &buf->bflags);
 
 	btrfs_set_lock_blocking(buf);
 	btrfs_set_buffer_uptodate(buf);
@@ -6214,7 +6267,7 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
 		return ERR_CAST(block_rsv);
 
 	ret = btrfs_reserve_extent(trans, root, blocksize, blocksize,
-				   empty_size, hint, (u64)-1, &ins, 0);
+				   empty_size, hint, &ins, 0);
 	if (ret) {
 		unuse_block_rsv(root->fs_info, block_rsv, blocksize);
 		return ERR_PTR(ret);
@@ -6222,7 +6275,7 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
 
 	buf = btrfs_init_new_buffer(trans, root, ins.objectid,
 				    blocksize, level);
-	BUG_ON(IS_ERR(buf));
+	BUG_ON(IS_ERR(buf)); /* -ENOMEM */
 
 	if (root_objectid == BTRFS_TREE_RELOC_OBJECTID) {
 		if (parent == 0)
@@ -6234,7 +6287,7 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
 	if (root_objectid != BTRFS_TREE_LOG_OBJECTID) {
 		struct btrfs_delayed_extent_op *extent_op;
 		extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
-		BUG_ON(!extent_op);
+		BUG_ON(!extent_op); /* -ENOMEM */
 		if (key)
 			memcpy(&extent_op->key, key, sizeof(extent_op->key));
 		else
@@ -6249,7 +6302,7 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
 					ins.offset, parent, root_objectid,
 					level, BTRFS_ADD_DELAYED_EXTENT,
 					extent_op, for_cow);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 	}
 	return buf;
 }
@@ -6319,7 +6372,9 @@ static noinline void reada_walk_down(struct btrfs_trans_handle *trans,
 		/* We don't lock the tree block, it's OK to be racy here */
 		ret = btrfs_lookup_extent_info(trans, root, bytenr, blocksize,
 					       &refs, &flags);
-		BUG_ON(ret);
+		/* We don't care about errors in readahead. */
+		if (ret < 0)
+			continue;
 		BUG_ON(refs == 0);
 
 		if (wc->stage == DROP_REFERENCE) {
@@ -6386,7 +6441,9 @@ static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
 					       eb->start, eb->len,
 					       &wc->refs[level],
 					       &wc->flags[level]);
-		BUG_ON(ret);
+		BUG_ON(ret == -ENOMEM);
+		if (ret)
+			return ret;
 		BUG_ON(wc->refs[level] == 0);
 	}
 
@@ -6405,12 +6462,12 @@ static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
 	if (!(wc->flags[level] & flag)) {
 		BUG_ON(!path->locks[level]);
 		ret = btrfs_inc_ref(trans, root, eb, 1, wc->for_reloc);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 		ret = btrfs_dec_ref(trans, root, eb, 0, wc->for_reloc);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 		ret = btrfs_set_disk_extent_flags(trans, root, eb->start,
 						  eb->len, flag, 0);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 		wc->flags[level] |= flag;
 	}
 
@@ -6482,7 +6539,11 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans,
 	ret = btrfs_lookup_extent_info(trans, root, bytenr, blocksize,
 				       &wc->refs[level - 1],
 				       &wc->flags[level - 1]);
-	BUG_ON(ret);
+	if (ret < 0) {
+		btrfs_tree_unlock(next);
+		return ret;
+	}
+
 	BUG_ON(wc->refs[level - 1] == 0);
 	*lookup_info = 0;
 
@@ -6551,7 +6612,7 @@ skip:
 
 		ret = btrfs_free_extent(trans, root, bytenr, blocksize, parent,
 				root->root_key.objectid, level - 1, 0, 0);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 	}
 	btrfs_tree_unlock(next);
 	free_extent_buffer(next);
@@ -6609,7 +6670,10 @@ static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
 						       eb->start, eb->len,
 						       &wc->refs[level],
 						       &wc->flags[level]);
-			BUG_ON(ret);
+			if (ret < 0) {
+				btrfs_tree_unlock_rw(eb, path->locks[level]);
+				return ret;
+			}
 			BUG_ON(wc->refs[level] == 0);
 			if (wc->refs[level] == 1) {
 				btrfs_tree_unlock_rw(eb, path->locks[level]);
@@ -6629,7 +6693,7 @@ static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
 			else
 				ret = btrfs_dec_ref(trans, root, eb, 0,
 						    wc->for_reloc);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM */
 		}
 		/* make block locked assertion in clean_tree_block happy */
 		if (!path->locks[level] &&
@@ -6738,7 +6802,7 @@ static noinline int walk_up_tree(struct btrfs_trans_handle *trans,
  * also make sure backrefs for the shared block and all lower level
  * blocks are properly updated.
  */
-void btrfs_drop_snapshot(struct btrfs_root *root,
+int btrfs_drop_snapshot(struct btrfs_root *root,
 			 struct btrfs_block_rsv *block_rsv, int update_ref,
 			 int for_reloc)
 {
@@ -6766,7 +6830,10 @@ void btrfs_drop_snapshot(struct btrfs_root *root,
 	}
 
 	trans = btrfs_start_transaction(tree_root, 0);
-	BUG_ON(IS_ERR(trans));
+	if (IS_ERR(trans)) {
+		err = PTR_ERR(trans);
+		goto out_free;
+	}
 
 	if (block_rsv)
 		trans->block_rsv = block_rsv;
@@ -6791,7 +6858,7 @@ void btrfs_drop_snapshot(struct btrfs_root *root,
 		path->lowest_level = 0;
 		if (ret < 0) {
 			err = ret;
-			goto out_free;
+			goto out_end_trans;
 		}
 		WARN_ON(ret > 0);
 
@@ -6811,7 +6878,10 @@ void btrfs_drop_snapshot(struct btrfs_root *root,
 						path->nodes[level]->len,
 						&wc->refs[level],
 						&wc->flags[level]);
-			BUG_ON(ret);
+			if (ret < 0) {
+				err = ret;
+				goto out_end_trans;
+			}
 			BUG_ON(wc->refs[level] == 0);
 
 			if (level == root_item->drop_level)
@@ -6862,26 +6932,40 @@ void btrfs_drop_snapshot(struct btrfs_root *root,
 			ret = btrfs_update_root(trans, tree_root,
 						&root->root_key,
 						root_item);
-			BUG_ON(ret);
+			if (ret) {
+				btrfs_abort_transaction(trans, tree_root, ret);
+				err = ret;
+				goto out_end_trans;
+			}
 
 			btrfs_end_transaction_throttle(trans, tree_root);
 			trans = btrfs_start_transaction(tree_root, 0);
-			BUG_ON(IS_ERR(trans));
+			if (IS_ERR(trans)) {
+				err = PTR_ERR(trans);
+				goto out_free;
+			}
 			if (block_rsv)
 				trans->block_rsv = block_rsv;
 		}
 	}
 	btrfs_release_path(path);
-	BUG_ON(err);
+	if (err)
+		goto out_end_trans;
 
 	ret = btrfs_del_root(trans, tree_root, &root->root_key);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, tree_root, ret);
+		goto out_end_trans;
+	}
 
 	if (root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID) {
 		ret = btrfs_find_last_root(tree_root, root->root_key.objectid,
 					   NULL, NULL);
-		BUG_ON(ret < 0);
-		if (ret > 0) {
+		if (ret < 0) {
+			btrfs_abort_transaction(trans, tree_root, ret);
+			err = ret;
+			goto out_end_trans;
+		} else if (ret > 0) {
 			/* if we fail to delete the orphan item this time
 			 * around, it'll get picked up the next time.
 			 *
@@ -6899,14 +6983,15 @@ void btrfs_drop_snapshot(struct btrfs_root *root,
 		free_extent_buffer(root->commit_root);
 		kfree(root);
 	}
-out_free:
+out_end_trans:
 	btrfs_end_transaction_throttle(trans, tree_root);
+out_free:
 	kfree(wc);
 	btrfs_free_path(path);
 out:
 	if (err)
 		btrfs_std_error(root->fs_info, err);
-	return;
+	return err;
 }
 
 /*
@@ -6983,31 +7068,15 @@ int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
 static u64 update_block_group_flags(struct btrfs_root *root, u64 flags)
 {
 	u64 num_devices;
-	u64 stripped = BTRFS_BLOCK_GROUP_RAID0 |
-		BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10;
+	u64 stripped;
 
-	if (root->fs_info->balance_ctl) {
-		struct btrfs_balance_control *bctl = root->fs_info->balance_ctl;
-		u64 tgt = 0;
-
-		/* pick restriper's target profile and return */
-		if (flags & BTRFS_BLOCK_GROUP_DATA &&
-		    bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) {
-			tgt = BTRFS_BLOCK_GROUP_DATA | bctl->data.target;
-		} else if (flags & BTRFS_BLOCK_GROUP_SYSTEM &&
-			   bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
-			tgt = BTRFS_BLOCK_GROUP_SYSTEM | bctl->sys.target;
-		} else if (flags & BTRFS_BLOCK_GROUP_METADATA &&
-			   bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) {
-			tgt = BTRFS_BLOCK_GROUP_METADATA | bctl->meta.target;
-		}
-
-		if (tgt) {
-			/* extended -> chunk profile */
-			tgt &= ~BTRFS_AVAIL_ALLOC_BIT_SINGLE;
-			return tgt;
-		}
-	}
+	/*
+	 * if restripe for this chunk_type is on pick target profile and
+	 * return, otherwise do the usual balance
+	 */
+	stripped = get_restripe_target(root->fs_info, flags);
+	if (stripped)
+		return extended_to_chunk(stripped);
 
 	/*
 	 * we add in the count of missing devices because we want
@@ -7017,6 +7086,9 @@ static u64 update_block_group_flags(struct btrfs_root *root, u64 flags)
 	num_devices = root->fs_info->fs_devices->rw_devices +
 		root->fs_info->fs_devices->missing_devices;
 
+	stripped = BTRFS_BLOCK_GROUP_RAID0 |
+		BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10;
+
 	if (num_devices == 1) {
 		stripped |= BTRFS_BLOCK_GROUP_DUP;
 		stripped = flags & ~stripped;
@@ -7029,7 +7101,6 @@ static u64 update_block_group_flags(struct btrfs_root *root, u64 flags)
 		if (flags & (BTRFS_BLOCK_GROUP_RAID1 |
 			     BTRFS_BLOCK_GROUP_RAID10))
 			return stripped | BTRFS_BLOCK_GROUP_DUP;
-		return flags;
 	} else {
 		/* they already had raid on here, just return */
 		if (flags & stripped)
@@ -7042,9 +7113,9 @@ static u64 update_block_group_flags(struct btrfs_root *root, u64 flags)
 		if (flags & BTRFS_BLOCK_GROUP_DUP)
 			return stripped | BTRFS_BLOCK_GROUP_RAID1;
 
-		/* turn single device chunks into raid0 */
-		return stripped | BTRFS_BLOCK_GROUP_RAID0;
+		/* this is drive concat, leave it alone */
 	}
+
 	return flags;
 }
 
@@ -7103,12 +7174,16 @@ int btrfs_set_block_group_ro(struct btrfs_root *root,
 	BUG_ON(cache->ro);
 
 	trans = btrfs_join_transaction(root);
-	BUG_ON(IS_ERR(trans));
+	if (IS_ERR(trans))
+		return PTR_ERR(trans);
 
 	alloc_flags = update_block_group_flags(root, cache->flags);
-	if (alloc_flags != cache->flags)
-		do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags,
-			       CHUNK_ALLOC_FORCE);
+	if (alloc_flags != cache->flags) {
+		ret = do_chunk_alloc(trans, root, 2 * 1024 * 1024, alloc_flags,
+				     CHUNK_ALLOC_FORCE);
+		if (ret < 0)
+			goto out;
+	}
 
 	ret = set_block_group_ro(cache, 0);
 	if (!ret)
@@ -7188,7 +7263,7 @@ u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo)
 	return free_bytes;
 }
 
-int btrfs_set_block_group_rw(struct btrfs_root *root,
+void btrfs_set_block_group_rw(struct btrfs_root *root,
 			      struct btrfs_block_group_cache *cache)
 {
 	struct btrfs_space_info *sinfo = cache->space_info;
@@ -7204,7 +7279,6 @@ int btrfs_set_block_group_rw(struct btrfs_root *root,
 	cache->ro = 0;
 	spin_unlock(&cache->lock);
 	spin_unlock(&sinfo->lock);
-	return 0;
 }
 
 /*
@@ -7222,6 +7296,7 @@ int btrfs_can_relocate(struct btrfs_root *root, u64 bytenr)
 	u64 min_free;
 	u64 dev_min = 1;
 	u64 dev_nr = 0;
+	u64 target;
 	int index;
 	int full = 0;
 	int ret = 0;
@@ -7262,13 +7337,11 @@ int btrfs_can_relocate(struct btrfs_root *root, u64 bytenr)
 	/*
 	 * ok we don't have enough space, but maybe we have free space on our
 	 * devices to allocate new chunks for relocation, so loop through our
-	 * alloc devices and guess if we have enough space.  However, if we
-	 * were marked as full, then we know there aren't enough chunks, and we
-	 * can just return.
+	 * alloc devices and guess if we have enough space.  if this block
+	 * group is going to be restriped, run checks against the target
+	 * profile instead of the current one.
 	 */
 	ret = -1;
-	if (full)
-		goto out;
 
 	/*
 	 * index:
@@ -7278,7 +7351,20 @@ int btrfs_can_relocate(struct btrfs_root *root, u64 bytenr)
 	 *      3: raid0
 	 *      4: single
 	 */
-	index = get_block_group_index(block_group);
+	target = get_restripe_target(root->fs_info, block_group->flags);
+	if (target) {
+		index = __get_block_group_index(extended_to_chunk(target));
+	} else {
+		/*
+		 * this is just a balance, so if we were marked as full
+		 * we know there is no space for a new chunk
+		 */
+		if (full)
+			goto out;
+
+		index = get_block_group_index(block_group);
+	}
+
 	if (index == 0) {
 		dev_min = 4;
 		/* Divide by 2 */
@@ -7572,7 +7658,7 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 		ret = update_space_info(info, cache->flags, found_key.offset,
 					btrfs_block_group_used(&cache->item),
 					&space_info);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 		cache->space_info = space_info;
 		spin_lock(&cache->space_info->lock);
 		cache->space_info->bytes_readonly += cache->bytes_super;
@@ -7581,7 +7667,7 @@ int btrfs_read_block_groups(struct btrfs_root *root)
 		__link_block_group(space_info, cache);
 
 		ret = btrfs_add_block_group_cache(root->fs_info, cache);
-		BUG_ON(ret);
+		BUG_ON(ret); /* Logic error */
 
 		set_avail_alloc_bits(root->fs_info, cache->flags);
 		if (btrfs_chunk_readonly(root, cache->key.objectid))
@@ -7663,7 +7749,7 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 
 	ret = update_space_info(root->fs_info, cache->flags, size, bytes_used,
 				&cache->space_info);
-	BUG_ON(ret);
+	BUG_ON(ret); /* -ENOMEM */
 	update_global_block_rsv(root->fs_info);
 
 	spin_lock(&cache->space_info->lock);
@@ -7673,11 +7759,14 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 	__link_block_group(cache->space_info, cache);
 
 	ret = btrfs_add_block_group_cache(root->fs_info, cache);
-	BUG_ON(ret);
+	BUG_ON(ret); /* Logic error */
 
 	ret = btrfs_insert_item(trans, extent_root, &cache->key, &cache->item,
 				sizeof(cache->item));
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, extent_root, ret);
+		return ret;
+	}
 
 	set_avail_alloc_bits(extent_root->fs_info, type);
 
@@ -7686,11 +7775,8 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 
 static void clear_avail_alloc_bits(struct btrfs_fs_info *fs_info, u64 flags)
 {
-	u64 extra_flags = flags & BTRFS_BLOCK_GROUP_PROFILE_MASK;
-
-	/* chunk -> extended profile */
-	if (extra_flags == 0)
-		extra_flags = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
+	u64 extra_flags = chunk_to_extended(flags) &
+				BTRFS_EXTENDED_PROFILE_MASK;
 
 	if (flags & BTRFS_BLOCK_GROUP_DATA)
 		fs_info->avail_data_alloc_bits &= ~extra_flags;
@@ -7758,7 +7844,10 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 	inode = lookup_free_space_inode(tree_root, block_group, path);
 	if (!IS_ERR(inode)) {
 		ret = btrfs_orphan_add(trans, inode);
-		BUG_ON(ret);
+		if (ret) {
+			btrfs_add_delayed_iput(inode);
+			goto out;
+		}
 		clear_nlink(inode);
 		/* One for the block groups ref */
 		spin_lock(&block_group->lock);
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 2862454bcdb..8d904dd7ea9 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -19,6 +19,7 @@
 #include "btrfs_inode.h"
 #include "volumes.h"
 #include "check-integrity.h"
+#include "locking.h"
 
 static struct kmem_cache *extent_state_cache;
 static struct kmem_cache *extent_buffer_cache;
@@ -53,6 +54,13 @@ struct extent_page_data {
 	unsigned int sync_io:1;
 };
 
+static noinline void flush_write_bio(void *data);
+static inline struct btrfs_fs_info *
+tree_fs_info(struct extent_io_tree *tree)
+{
+	return btrfs_sb(tree->mapping->host->i_sb);
+}
+
 int __init extent_io_init(void)
 {
 	extent_state_cache = kmem_cache_create("extent_state",
@@ -136,6 +144,7 @@ static struct extent_state *alloc_extent_state(gfp_t mask)
 #endif
 	atomic_set(&state->refs, 1);
 	init_waitqueue_head(&state->wq);
+	trace_alloc_extent_state(state, mask, _RET_IP_);
 	return state;
 }
 
@@ -153,6 +162,7 @@ void free_extent_state(struct extent_state *state)
 		list_del(&state->leak_list);
 		spin_unlock_irqrestore(&leak_lock, flags);
 #endif
+		trace_free_extent_state(state, _RET_IP_);
 		kmem_cache_free(extent_state_cache, state);
 	}
 }
@@ -439,6 +449,13 @@ alloc_extent_state_atomic(struct extent_state *prealloc)
 	return prealloc;
 }
 
+void extent_io_tree_panic(struct extent_io_tree *tree, int err)
+{
+	btrfs_panic(tree_fs_info(tree), err, "Locking error: "
+		    "Extent tree was modified by another "
+		    "thread while locked.");
+}
+
 /*
  * clear some bits on a range in the tree.  This may require splitting
  * or inserting elements in the tree, so the gfp mask is used to
@@ -449,8 +466,7 @@ alloc_extent_state_atomic(struct extent_state *prealloc)
  *
  * the range [start, end] is inclusive.
  *
- * This takes the tree lock, and returns < 0 on error, > 0 if any of the
- * bits were already set, or zero if none of the bits were already set.
+ * This takes the tree lock, and returns 0 on success and < 0 on error.
  */
 int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 		     int bits, int wake, int delete,
@@ -464,7 +480,6 @@ int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 	struct rb_node *node;
 	u64 last_end;
 	int err;
-	int set = 0;
 	int clear = 0;
 
 	if (delete)
@@ -542,12 +557,14 @@ hit_next:
 		prealloc = alloc_extent_state_atomic(prealloc);
 		BUG_ON(!prealloc);
 		err = split_state(tree, state, prealloc, start);
-		BUG_ON(err == -EEXIST);
+		if (err)
+			extent_io_tree_panic(tree, err);
+
 		prealloc = NULL;
 		if (err)
 			goto out;
 		if (state->end <= end) {
-			set |= clear_state_bit(tree, state, &bits, wake);
+			clear_state_bit(tree, state, &bits, wake);
 			if (last_end == (u64)-1)
 				goto out;
 			start = last_end + 1;
@@ -564,17 +581,19 @@ hit_next:
 		prealloc = alloc_extent_state_atomic(prealloc);
 		BUG_ON(!prealloc);
 		err = split_state(tree, state, prealloc, end + 1);
-		BUG_ON(err == -EEXIST);
+		if (err)
+			extent_io_tree_panic(tree, err);
+
 		if (wake)
 			wake_up(&state->wq);
 
-		set |= clear_state_bit(tree, prealloc, &bits, wake);
+		clear_state_bit(tree, prealloc, &bits, wake);
 
 		prealloc = NULL;
 		goto out;
 	}
 
-	set |= clear_state_bit(tree, state, &bits, wake);
+	clear_state_bit(tree, state, &bits, wake);
 next:
 	if (last_end == (u64)-1)
 		goto out;
@@ -591,7 +610,7 @@ out:
 	if (prealloc)
 		free_extent_state(prealloc);
 
-	return set;
+	return 0;
 
 search_again:
 	if (start > end)
@@ -602,8 +621,8 @@ search_again:
 	goto again;
 }
 
-static int wait_on_state(struct extent_io_tree *tree,
-			 struct extent_state *state)
+static void wait_on_state(struct extent_io_tree *tree,
+			  struct extent_state *state)
 		__releases(tree->lock)
 		__acquires(tree->lock)
 {
@@ -613,7 +632,6 @@ static int wait_on_state(struct extent_io_tree *tree,
 	schedule();
 	spin_lock(&tree->lock);
 	finish_wait(&state->wq, &wait);
-	return 0;
 }
 
 /*
@@ -621,7 +639,7 @@ static int wait_on_state(struct extent_io_tree *tree,
  * The range [start, end] is inclusive.
  * The tree lock is taken by this function
  */
-int wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits)
+void wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits)
 {
 	struct extent_state *state;
 	struct rb_node *node;
@@ -658,7 +676,6 @@ again:
 	}
 out:
 	spin_unlock(&tree->lock);
-	return 0;
 }
 
 static void set_state_bits(struct extent_io_tree *tree,
@@ -706,9 +723,10 @@ static void uncache_state(struct extent_state **cached_ptr)
  * [start, end] is inclusive This takes the tree lock.
  */
 
-int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		   int bits, int exclusive_bits, u64 *failed_start,
-		   struct extent_state **cached_state, gfp_t mask)
+static int __must_check
+__set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
+		 int bits, int exclusive_bits, u64 *failed_start,
+		 struct extent_state **cached_state, gfp_t mask)
 {
 	struct extent_state *state;
 	struct extent_state *prealloc = NULL;
@@ -742,8 +760,10 @@ again:
 		prealloc = alloc_extent_state_atomic(prealloc);
 		BUG_ON(!prealloc);
 		err = insert_state(tree, prealloc, start, end, &bits);
+		if (err)
+			extent_io_tree_panic(tree, err);
+
 		prealloc = NULL;
-		BUG_ON(err == -EEXIST);
 		goto out;
 	}
 	state = rb_entry(node, struct extent_state, rb_node);
@@ -809,7 +829,9 @@ hit_next:
 		prealloc = alloc_extent_state_atomic(prealloc);
 		BUG_ON(!prealloc);
 		err = split_state(tree, state, prealloc, start);
-		BUG_ON(err == -EEXIST);
+		if (err)
+			extent_io_tree_panic(tree, err);
+
 		prealloc = NULL;
 		if (err)
 			goto out;
@@ -846,12 +868,9 @@ hit_next:
 		 */
 		err = insert_state(tree, prealloc, start, this_end,
 				   &bits);
-		BUG_ON(err == -EEXIST);
-		if (err) {
-			free_extent_state(prealloc);
-			prealloc = NULL;
-			goto out;
-		}
+		if (err)
+			extent_io_tree_panic(tree, err);
+
 		cache_state(prealloc, cached_state);
 		prealloc = NULL;
 		start = this_end + 1;
@@ -873,7 +892,8 @@ hit_next:
 		prealloc = alloc_extent_state_atomic(prealloc);
 		BUG_ON(!prealloc);
 		err = split_state(tree, state, prealloc, end + 1);
-		BUG_ON(err == -EEXIST);
+		if (err)
+			extent_io_tree_panic(tree, err);
 
 		set_state_bits(tree, prealloc, &bits);
 		cache_state(prealloc, cached_state);
@@ -900,6 +920,15 @@ search_again:
 	goto again;
 }
 
+int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits,
+		   u64 *failed_start, struct extent_state **cached_state,
+		   gfp_t mask)
+{
+	return __set_extent_bit(tree, start, end, bits, 0, failed_start,
+				cached_state, mask);
+}
+
+
 /**
  * convert_extent - convert all bits in a given range from one bit to another
  * @tree:	the io tree to search
@@ -946,7 +975,8 @@ again:
 		}
 		err = insert_state(tree, prealloc, start, end, &bits);
 		prealloc = NULL;
-		BUG_ON(err == -EEXIST);
+		if (err)
+			extent_io_tree_panic(tree, err);
 		goto out;
 	}
 	state = rb_entry(node, struct extent_state, rb_node);
@@ -1002,7 +1032,8 @@ hit_next:
 			goto out;
 		}
 		err = split_state(tree, state, prealloc, start);
-		BUG_ON(err == -EEXIST);
+		if (err)
+			extent_io_tree_panic(tree, err);
 		prealloc = NULL;
 		if (err)
 			goto out;
@@ -1041,12 +1072,8 @@ hit_next:
 		 */
 		err = insert_state(tree, prealloc, start, this_end,
 				   &bits);
-		BUG_ON(err == -EEXIST);
-		if (err) {
-			free_extent_state(prealloc);
-			prealloc = NULL;
-			goto out;
-		}
+		if (err)
+			extent_io_tree_panic(tree, err);
 		prealloc = NULL;
 		start = this_end + 1;
 		goto search_again;
@@ -1065,7 +1092,8 @@ hit_next:
 		}
 
 		err = split_state(tree, state, prealloc, end + 1);
-		BUG_ON(err == -EEXIST);
+		if (err)
+			extent_io_tree_panic(tree, err);
 
 		set_state_bits(tree, prealloc, &bits);
 		clear_state_bit(tree, prealloc, &clear_bits, 0);
@@ -1095,14 +1123,14 @@ search_again:
 int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 		     gfp_t mask)
 {
-	return set_extent_bit(tree, start, end, EXTENT_DIRTY, 0, NULL,
+	return set_extent_bit(tree, start, end, EXTENT_DIRTY, NULL,
 			      NULL, mask);
 }
 
 int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 		    int bits, gfp_t mask)
 {
-	return set_extent_bit(tree, start, end, bits, 0, NULL,
+	return set_extent_bit(tree, start, end, bits, NULL,
 			      NULL, mask);
 }
 
@@ -1117,7 +1145,7 @@ int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
 {
 	return set_extent_bit(tree, start, end,
 			      EXTENT_DELALLOC | EXTENT_UPTODATE,
-			      0, NULL, cached_state, mask);
+			      NULL, cached_state, mask);
 }
 
 int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
@@ -1131,7 +1159,7 @@ int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
 		     gfp_t mask)
 {
-	return set_extent_bit(tree, start, end, EXTENT_NEW, 0, NULL,
+	return set_extent_bit(tree, start, end, EXTENT_NEW, NULL,
 			      NULL, mask);
 }
 
@@ -1139,7 +1167,7 @@ int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end,
 			struct extent_state **cached_state, gfp_t mask)
 {
 	return set_extent_bit(tree, start, end, EXTENT_UPTODATE, 0,
-			      NULL, cached_state, mask);
+			      cached_state, mask);
 }
 
 static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start,
@@ -1155,42 +1183,40 @@ static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start,
  * us if waiting is desired.
  */
 int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
-		     int bits, struct extent_state **cached_state, gfp_t mask)
+		     int bits, struct extent_state **cached_state)
 {
 	int err;
 	u64 failed_start;
 	while (1) {
-		err = set_extent_bit(tree, start, end, EXTENT_LOCKED | bits,
-				     EXTENT_LOCKED, &failed_start,
-				     cached_state, mask);
-		if (err == -EEXIST && (mask & __GFP_WAIT)) {
+		err = __set_extent_bit(tree, start, end, EXTENT_LOCKED | bits,
+				       EXTENT_LOCKED, &failed_start,
+				       cached_state, GFP_NOFS);
+		if (err == -EEXIST) {
 			wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED);
 			start = failed_start;
-		} else {
+		} else
 			break;
-		}
 		WARN_ON(start > end);
 	}
 	return err;
 }
 
-int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
+int lock_extent(struct extent_io_tree *tree, u64 start, u64 end)
 {
-	return lock_extent_bits(tree, start, end, 0, NULL, mask);
+	return lock_extent_bits(tree, start, end, 0, NULL);
 }
 
-int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
-		    gfp_t mask)
+int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end)
 {
 	int err;
 	u64 failed_start;
 
-	err = set_extent_bit(tree, start, end, EXTENT_LOCKED, EXTENT_LOCKED,
-			     &failed_start, NULL, mask);
+	err = __set_extent_bit(tree, start, end, EXTENT_LOCKED, EXTENT_LOCKED,
+			       &failed_start, NULL, GFP_NOFS);
 	if (err == -EEXIST) {
 		if (failed_start > start)
 			clear_extent_bit(tree, start, failed_start - 1,
-					 EXTENT_LOCKED, 1, 0, NULL, mask);
+					 EXTENT_LOCKED, 1, 0, NULL, GFP_NOFS);
 		return 0;
 	}
 	return 1;
@@ -1203,10 +1229,10 @@ int unlock_extent_cached(struct extent_io_tree *tree, u64 start, u64 end,
 				mask);
 }
 
-int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
+int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end)
 {
 	return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, NULL,
-				mask);
+				GFP_NOFS);
 }
 
 /*
@@ -1220,7 +1246,7 @@ static int set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end)
 
 	while (index <= end_index) {
 		page = find_get_page(tree->mapping, index);
-		BUG_ON(!page);
+		BUG_ON(!page); /* Pages should be in the extent_io_tree */
 		set_page_writeback(page);
 		page_cache_release(page);
 		index++;
@@ -1343,9 +1369,9 @@ out:
 	return found;
 }
 
-static noinline int __unlock_for_delalloc(struct inode *inode,
-					  struct page *locked_page,
-					  u64 start, u64 end)
+static noinline void __unlock_for_delalloc(struct inode *inode,
+					   struct page *locked_page,
+					   u64 start, u64 end)
 {
 	int ret;
 	struct page *pages[16];
@@ -1355,7 +1381,7 @@ static noinline int __unlock_for_delalloc(struct inode *inode,
 	int i;
 
 	if (index == locked_page->index && end_index == index)
-		return 0;
+		return;
 
 	while (nr_pages > 0) {
 		ret = find_get_pages_contig(inode->i_mapping, index,
@@ -1370,7 +1396,6 @@ static noinline int __unlock_for_delalloc(struct inode *inode,
 		index += ret;
 		cond_resched();
 	}
-	return 0;
 }
 
 static noinline int lock_delalloc_pages(struct inode *inode,
@@ -1500,11 +1525,10 @@ again:
 			goto out_failed;
 		}
 	}
-	BUG_ON(ret);
+	BUG_ON(ret); /* Only valid values are 0 and -EAGAIN */
 
 	/* step three, lock the state bits for the whole range */
-	lock_extent_bits(tree, delalloc_start, delalloc_end,
-			 0, &cached_state, GFP_NOFS);
+	lock_extent_bits(tree, delalloc_start, delalloc_end, 0, &cached_state);
 
 	/* then test to make sure it is all still delalloc */
 	ret = test_range_bit(tree, delalloc_start, delalloc_end,
@@ -1761,39 +1785,34 @@ int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
  * helper function to set a given page up to date if all the
  * extents in the tree for that page are up to date
  */
-static int check_page_uptodate(struct extent_io_tree *tree,
-			       struct page *page)
+static void check_page_uptodate(struct extent_io_tree *tree, struct page *page)
 {
 	u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
 	u64 end = start + PAGE_CACHE_SIZE - 1;
 	if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL))
 		SetPageUptodate(page);
-	return 0;
 }
 
 /*
  * helper function to unlock a page if all the extents in the tree
  * for that page are unlocked
  */
-static int check_page_locked(struct extent_io_tree *tree,
-			     struct page *page)
+static void check_page_locked(struct extent_io_tree *tree, struct page *page)
 {
 	u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
 	u64 end = start + PAGE_CACHE_SIZE - 1;
 	if (!test_range_bit(tree, start, end, EXTENT_LOCKED, 0, NULL))
 		unlock_page(page);
-	return 0;
 }
 
 /*
  * helper function to end page writeback if all the extents
  * in the tree for that page are done with writeback
  */
-static int check_page_writeback(struct extent_io_tree *tree,
-			     struct page *page)
+static void check_page_writeback(struct extent_io_tree *tree,
+				 struct page *page)
 {
 	end_page_writeback(page);
-	return 0;
 }
 
 /*
@@ -1912,6 +1931,26 @@ int repair_io_failure(struct btrfs_mapping_tree *map_tree, u64 start,
 	return 0;
 }
 
+int repair_eb_io_failure(struct btrfs_root *root, struct extent_buffer *eb,
+			 int mirror_num)
+{
+	struct btrfs_mapping_tree *map_tree = &root->fs_info->mapping_tree;
+	u64 start = eb->start;
+	unsigned long i, num_pages = num_extent_pages(eb->start, eb->len);
+	int ret;
+
+	for (i = 0; i < num_pages; i++) {
+		struct page *p = extent_buffer_page(eb, i);
+		ret = repair_io_failure(map_tree, start, PAGE_CACHE_SIZE,
+					start, p, mirror_num);
+		if (ret)
+			break;
+		start += PAGE_CACHE_SIZE;
+	}
+
+	return ret;
+}
+
 /*
  * each time an IO finishes, we do a fast check in the IO failure tree
  * to see if we need to process or clean up an io_failure_record
@@ -2258,6 +2297,7 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 	u64 start;
 	u64 end;
 	int whole_page;
+	int failed_mirror;
 	int ret;
 
 	if (err)
@@ -2304,9 +2344,16 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 			else
 				clean_io_failure(start, page);
 		}
-		if (!uptodate) {
-			int failed_mirror;
+
+		if (!uptodate)
 			failed_mirror = (int)(unsigned long)bio->bi_bdev;
+
+		if (!uptodate && tree->ops && tree->ops->readpage_io_failed_hook) {
+			ret = tree->ops->readpage_io_failed_hook(page, failed_mirror);
+			if (!ret && !err &&
+			    test_bit(BIO_UPTODATE, &bio->bi_flags))
+				uptodate = 1;
+		} else if (!uptodate) {
 			/*
 			 * The generic bio_readpage_error handles errors the
 			 * following way: If possible, new read requests are
@@ -2320,7 +2367,6 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 			ret = bio_readpage_error(bio, page, start, end,
 							failed_mirror, NULL);
 			if (ret == 0) {
-error_handled:
 				uptodate =
 					test_bit(BIO_UPTODATE, &bio->bi_flags);
 				if (err)
@@ -2328,16 +2374,9 @@ error_handled:
 				uncache_state(&cached);
 				continue;
 			}
-			if (tree->ops && tree->ops->readpage_io_failed_hook) {
-				ret = tree->ops->readpage_io_failed_hook(
-							bio, page, start, end,
-							failed_mirror, state);
-				if (ret == 0)
-					goto error_handled;
-			}
 		}
 
-		if (uptodate) {
+		if (uptodate && tree->track_uptodate) {
 			set_extent_uptodate(tree, start, end, &cached,
 					    GFP_ATOMIC);
 		}
@@ -2386,8 +2425,12 @@ btrfs_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs,
 	return bio;
 }
 
-static int submit_one_bio(int rw, struct bio *bio, int mirror_num,
-			  unsigned long bio_flags)
+/*
+ * Since writes are async, they will only return -ENOMEM.
+ * Reads can return the full range of I/O error conditions.
+ */
+static int __must_check submit_one_bio(int rw, struct bio *bio,
+				       int mirror_num, unsigned long bio_flags)
 {
 	int ret = 0;
 	struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
@@ -2413,6 +2456,19 @@ static int submit_one_bio(int rw, struct bio *bio, int mirror_num,
 	return ret;
 }
 
+static int merge_bio(struct extent_io_tree *tree, struct page *page,
+		     unsigned long offset, size_t size, struct bio *bio,
+		     unsigned long bio_flags)
+{
+	int ret = 0;
+	if (tree->ops && tree->ops->merge_bio_hook)
+		ret = tree->ops->merge_bio_hook(page, offset, size, bio,
+						bio_flags);
+	BUG_ON(ret < 0);
+	return ret;
+
+}
+
 static int submit_extent_page(int rw, struct extent_io_tree *tree,
 			      struct page *page, sector_t sector,
 			      size_t size, unsigned long offset,
@@ -2441,12 +2497,12 @@ static int submit_extent_page(int rw, struct extent_io_tree *tree,
 				sector;
 
 		if (prev_bio_flags != bio_flags || !contig ||
-		    (tree->ops && tree->ops->merge_bio_hook &&
-		     tree->ops->merge_bio_hook(page, offset, page_size, bio,
-					       bio_flags)) ||
+		    merge_bio(tree, page, offset, page_size, bio, bio_flags) ||
 		    bio_add_page(bio, page, page_size, offset) < page_size) {
 			ret = submit_one_bio(rw, bio, mirror_num,
 					     prev_bio_flags);
+			if (ret < 0)
+				return ret;
 			bio = NULL;
 		} else {
 			return 0;
@@ -2473,25 +2529,31 @@ static int submit_extent_page(int rw, struct extent_io_tree *tree,
 	return ret;
 }
 
-void set_page_extent_mapped(struct page *page)
+void attach_extent_buffer_page(struct extent_buffer *eb, struct page *page)
 {
 	if (!PagePrivate(page)) {
 		SetPagePrivate(page);
 		page_cache_get(page);
-		set_page_private(page, EXTENT_PAGE_PRIVATE);
+		set_page_private(page, (unsigned long)eb);
+	} else {
+		WARN_ON(page->private != (unsigned long)eb);
 	}
 }
 
-static void set_page_extent_head(struct page *page, unsigned long len)
+void set_page_extent_mapped(struct page *page)
 {
-	WARN_ON(!PagePrivate(page));
-	set_page_private(page, EXTENT_PAGE_PRIVATE_FIRST_PAGE | len << 2);
+	if (!PagePrivate(page)) {
+		SetPagePrivate(page);
+		page_cache_get(page);
+		set_page_private(page, EXTENT_PAGE_PRIVATE);
+	}
 }
 
 /*
  * basic readpage implementation.  Locked extent state structs are inserted
  * into the tree that are removed when the IO is done (by the end_io
  * handlers)
+ * XXX JDM: This needs looking at to ensure proper page locking
  */
 static int __extent_read_full_page(struct extent_io_tree *tree,
 				   struct page *page,
@@ -2531,11 +2593,11 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 
 	end = page_end;
 	while (1) {
-		lock_extent(tree, start, end, GFP_NOFS);
+		lock_extent(tree, start, end);
 		ordered = btrfs_lookup_ordered_extent(inode, start);
 		if (!ordered)
 			break;
-		unlock_extent(tree, start, end, GFP_NOFS);
+		unlock_extent(tree, start, end);
 		btrfs_start_ordered_extent(inode, ordered, 1);
 		btrfs_put_ordered_extent(ordered);
 	}
@@ -2572,7 +2634,7 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 				end - cur + 1, 0);
 		if (IS_ERR_OR_NULL(em)) {
 			SetPageError(page);
-			unlock_extent(tree, cur, end, GFP_NOFS);
+			unlock_extent(tree, cur, end);
 			break;
 		}
 		extent_offset = cur - em->start;
@@ -2624,7 +2686,7 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 		if (test_range_bit(tree, cur, cur_end,
 				   EXTENT_UPTODATE, 1, NULL)) {
 			check_page_uptodate(tree, page);
-			unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
+			unlock_extent(tree, cur, cur + iosize - 1);
 			cur = cur + iosize;
 			pg_offset += iosize;
 			continue;
@@ -2634,7 +2696,7 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 		 */
 		if (block_start == EXTENT_MAP_INLINE) {
 			SetPageError(page);
-			unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
+			unlock_extent(tree, cur, cur + iosize - 1);
 			cur = cur + iosize;
 			pg_offset += iosize;
 			continue;
@@ -2654,6 +2716,7 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 					 end_bio_extent_readpage, mirror_num,
 					 *bio_flags,
 					 this_bio_flag);
+			BUG_ON(ret == -ENOMEM);
 			nr++;
 			*bio_flags = this_bio_flag;
 		}
@@ -2795,7 +2858,11 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 						       delalloc_end,
 						       &page_started,
 						       &nr_written);
-			BUG_ON(ret);
+			/* File system has been set read-only */
+			if (ret) {
+				SetPageError(page);
+				goto done;
+			}
 			/*
 			 * delalloc_end is already one less than the total
 			 * length, so we don't subtract one from
@@ -2968,6 +3035,275 @@ done_unlocked:
 	return 0;
 }
 
+static int eb_wait(void *word)
+{
+	io_schedule();
+	return 0;
+}
+
+static void wait_on_extent_buffer_writeback(struct extent_buffer *eb)
+{
+	wait_on_bit(&eb->bflags, EXTENT_BUFFER_WRITEBACK, eb_wait,
+		    TASK_UNINTERRUPTIBLE);
+}
+
+static int lock_extent_buffer_for_io(struct extent_buffer *eb,
+				     struct btrfs_fs_info *fs_info,
+				     struct extent_page_data *epd)
+{
+	unsigned long i, num_pages;
+	int flush = 0;
+	int ret = 0;
+
+	if (!btrfs_try_tree_write_lock(eb)) {
+		flush = 1;
+		flush_write_bio(epd);
+		btrfs_tree_lock(eb);
+	}
+
+	if (test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags)) {
+		btrfs_tree_unlock(eb);
+		if (!epd->sync_io)
+			return 0;
+		if (!flush) {
+			flush_write_bio(epd);
+			flush = 1;
+		}
+		while (1) {
+			wait_on_extent_buffer_writeback(eb);
+			btrfs_tree_lock(eb);
+			if (!test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags))
+				break;
+			btrfs_tree_unlock(eb);
+		}
+	}
+
+	if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
+		set_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags);
+		btrfs_set_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN);
+		spin_lock(&fs_info->delalloc_lock);
+		if (fs_info->dirty_metadata_bytes >= eb->len)
+			fs_info->dirty_metadata_bytes -= eb->len;
+		else
+			WARN_ON(1);
+		spin_unlock(&fs_info->delalloc_lock);
+		ret = 1;
+	}
+
+	btrfs_tree_unlock(eb);
+
+	if (!ret)
+		return ret;
+
+	num_pages = num_extent_pages(eb->start, eb->len);
+	for (i = 0; i < num_pages; i++) {
+		struct page *p = extent_buffer_page(eb, i);
+
+		if (!trylock_page(p)) {
+			if (!flush) {
+				flush_write_bio(epd);
+				flush = 1;
+			}
+			lock_page(p);
+		}
+	}
+
+	return ret;
+}
+
+static void end_extent_buffer_writeback(struct extent_buffer *eb)
+{
+	clear_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags);
+	smp_mb__after_clear_bit();
+	wake_up_bit(&eb->bflags, EXTENT_BUFFER_WRITEBACK);
+}
+
+static void end_bio_extent_buffer_writepage(struct bio *bio, int err)
+{
+	int uptodate = err == 0;
+	struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
+	struct extent_buffer *eb;
+	int done;
+
+	do {
+		struct page *page = bvec->bv_page;
+
+		bvec--;
+		eb = (struct extent_buffer *)page->private;
+		BUG_ON(!eb);
+		done = atomic_dec_and_test(&eb->io_pages);
+
+		if (!uptodate || test_bit(EXTENT_BUFFER_IOERR, &eb->bflags)) {
+			set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+			ClearPageUptodate(page);
+			SetPageError(page);
+		}
+
+		end_page_writeback(page);
+
+		if (!done)
+			continue;
+
+		end_extent_buffer_writeback(eb);
+	} while (bvec >= bio->bi_io_vec);
+
+	bio_put(bio);
+
+}
+
+static int write_one_eb(struct extent_buffer *eb,
+			struct btrfs_fs_info *fs_info,
+			struct writeback_control *wbc,
+			struct extent_page_data *epd)
+{
+	struct block_device *bdev = fs_info->fs_devices->latest_bdev;
+	u64 offset = eb->start;
+	unsigned long i, num_pages;
+	int rw = (epd->sync_io ? WRITE_SYNC : WRITE);
+	int ret;
+
+	clear_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+	num_pages = num_extent_pages(eb->start, eb->len);
+	atomic_set(&eb->io_pages, num_pages);
+	for (i = 0; i < num_pages; i++) {
+		struct page *p = extent_buffer_page(eb, i);
+
+		clear_page_dirty_for_io(p);
+		set_page_writeback(p);
+		ret = submit_extent_page(rw, eb->tree, p, offset >> 9,
+					 PAGE_CACHE_SIZE, 0, bdev, &epd->bio,
+					 -1, end_bio_extent_buffer_writepage,
+					 0, 0, 0);
+		if (ret) {
+			set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+			SetPageError(p);
+			if (atomic_sub_and_test(num_pages - i, &eb->io_pages))
+				end_extent_buffer_writeback(eb);
+			ret = -EIO;
+			break;
+		}
+		offset += PAGE_CACHE_SIZE;
+		update_nr_written(p, wbc, 1);
+		unlock_page(p);
+	}
+
+	if (unlikely(ret)) {
+		for (; i < num_pages; i++) {
+			struct page *p = extent_buffer_page(eb, i);
+			unlock_page(p);
+		}
+	}
+
+	return ret;
+}
+
+int btree_write_cache_pages(struct address_space *mapping,
+				   struct writeback_control *wbc)
+{
+	struct extent_io_tree *tree = &BTRFS_I(mapping->host)->io_tree;
+	struct btrfs_fs_info *fs_info = BTRFS_I(mapping->host)->root->fs_info;
+	struct extent_buffer *eb, *prev_eb = NULL;
+	struct extent_page_data epd = {
+		.bio = NULL,
+		.tree = tree,
+		.extent_locked = 0,
+		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
+	};
+	int ret = 0;
+	int done = 0;
+	int nr_to_write_done = 0;
+	struct pagevec pvec;
+	int nr_pages;
+	pgoff_t index;
+	pgoff_t end;		/* Inclusive */
+	int scanned = 0;
+	int tag;
+
+	pagevec_init(&pvec, 0);
+	if (wbc->range_cyclic) {
+		index = mapping->writeback_index; /* Start from prev offset */
+		end = -1;
+	} else {
+		index = wbc->range_start >> PAGE_CACHE_SHIFT;
+		end = wbc->range_end >> PAGE_CACHE_SHIFT;
+		scanned = 1;
+	}
+	if (wbc->sync_mode == WB_SYNC_ALL)
+		tag = PAGECACHE_TAG_TOWRITE;
+	else
+		tag = PAGECACHE_TAG_DIRTY;
+retry:
+	if (wbc->sync_mode == WB_SYNC_ALL)
+		tag_pages_for_writeback(mapping, index, end);
+	while (!done && !nr_to_write_done && (index <= end) &&
+	       (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, tag,
+			min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
+		unsigned i;
+
+		scanned = 1;
+		for (i = 0; i < nr_pages; i++) {
+			struct page *page = pvec.pages[i];
+
+			if (!PagePrivate(page))
+				continue;
+
+			if (!wbc->range_cyclic && page->index > end) {
+				done = 1;
+				break;
+			}
+
+			eb = (struct extent_buffer *)page->private;
+			if (!eb) {
+				WARN_ON(1);
+				continue;
+			}
+
+			if (eb == prev_eb)
+				continue;
+
+			if (!atomic_inc_not_zero(&eb->refs)) {
+				WARN_ON(1);
+				continue;
+			}
+
+			prev_eb = eb;
+			ret = lock_extent_buffer_for_io(eb, fs_info, &epd);
+			if (!ret) {
+				free_extent_buffer(eb);
+				continue;
+			}
+
+			ret = write_one_eb(eb, fs_info, wbc, &epd);
+			if (ret) {
+				done = 1;
+				free_extent_buffer(eb);
+				break;
+			}
+			free_extent_buffer(eb);
+
+			/*
+			 * the filesystem may choose to bump up nr_to_write.
+			 * We have to make sure to honor the new nr_to_write
+			 * at any time
+			 */
+			nr_to_write_done = wbc->nr_to_write <= 0;
+		}
+		pagevec_release(&pvec);
+		cond_resched();
+	}
+	if (!scanned && !done) {
+		/*
+		 * We hit the last page and there is more work to be done: wrap
+		 * back to the start of the file
+		 */
+		scanned = 1;
+		index = 0;
+		goto retry;
+	}
+	flush_write_bio(&epd);
+	return ret;
+}
+
 /**
  * write_cache_pages - walk the list of dirty pages of the given address space and write all of them.
  * @mapping: address space structure to write
@@ -3099,10 +3435,14 @@ retry:
 static void flush_epd_write_bio(struct extent_page_data *epd)
 {
 	if (epd->bio) {
+		int rw = WRITE;
+		int ret;
+
 		if (epd->sync_io)
-			submit_one_bio(WRITE_SYNC, epd->bio, 0, 0);
-		else
-			submit_one_bio(WRITE, epd->bio, 0, 0);
+			rw = WRITE_SYNC;
+
+		ret = submit_one_bio(rw, epd->bio, 0, 0);
+		BUG_ON(ret < 0); /* -ENOMEM */
 		epd->bio = NULL;
 	}
 }
@@ -3219,7 +3559,7 @@ int extent_readpages(struct extent_io_tree *tree,
 	}
 	BUG_ON(!list_empty(pages));
 	if (bio)
-		submit_one_bio(READ, bio, 0, bio_flags);
+		return submit_one_bio(READ, bio, 0, bio_flags);
 	return 0;
 }
 
@@ -3240,7 +3580,7 @@ int extent_invalidatepage(struct extent_io_tree *tree,
 	if (start > end)
 		return 0;
 
-	lock_extent_bits(tree, start, end, 0, &cached_state, GFP_NOFS);
+	lock_extent_bits(tree, start, end, 0, &cached_state);
 	wait_on_page_writeback(page);
 	clear_extent_bit(tree, start, end,
 			 EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
@@ -3454,7 +3794,7 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 	}
 
 	lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len, 0,
-			 &cached_state, GFP_NOFS);
+			 &cached_state);
 
 	em = get_extent_skip_holes(inode, start, last_for_get_extent,
 				   get_extent);
@@ -3548,26 +3888,7 @@ out:
 inline struct page *extent_buffer_page(struct extent_buffer *eb,
 					      unsigned long i)
 {
-	struct page *p;
-	struct address_space *mapping;
-
-	if (i == 0)
-		return eb->first_page;
-	i += eb->start >> PAGE_CACHE_SHIFT;
-	mapping = eb->first_page->mapping;
-	if (!mapping)
-		return NULL;
-
-	/*
-	 * extent_buffer_page is only called after pinning the page
-	 * by increasing the reference count.  So we know the page must
-	 * be in the radix tree.
-	 */
-	rcu_read_lock();
-	p = radix_tree_lookup(&mapping->page_tree, i);
-	rcu_read_unlock();
-
-	return p;
+	return eb->pages[i];
 }
 
 inline unsigned long num_extent_pages(u64 start, u64 len)
@@ -3576,6 +3897,19 @@ inline unsigned long num_extent_pages(u64 start, u64 len)
 		(start >> PAGE_CACHE_SHIFT);
 }
 
+static void __free_extent_buffer(struct extent_buffer *eb)
+{
+#if LEAK_DEBUG
+	unsigned long flags;
+	spin_lock_irqsave(&leak_lock, flags);
+	list_del(&eb->leak_list);
+	spin_unlock_irqrestore(&leak_lock, flags);
+#endif
+	if (eb->pages && eb->pages != eb->inline_pages)
+		kfree(eb->pages);
+	kmem_cache_free(extent_buffer_cache, eb);
+}
+
 static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
 						   u64 start,
 						   unsigned long len,
@@ -3591,6 +3925,7 @@ static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
 		return NULL;
 	eb->start = start;
 	eb->len = len;
+	eb->tree = tree;
 	rwlock_init(&eb->lock);
 	atomic_set(&eb->write_locks, 0);
 	atomic_set(&eb->read_locks, 0);
@@ -3607,20 +3942,32 @@ static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
 	list_add(&eb->leak_list, &buffers);
 	spin_unlock_irqrestore(&leak_lock, flags);
 #endif
+	spin_lock_init(&eb->refs_lock);
 	atomic_set(&eb->refs, 1);
+	atomic_set(&eb->io_pages, 0);
+
+	if (len > MAX_INLINE_EXTENT_BUFFER_SIZE) {
+		struct page **pages;
+		int num_pages = (len + PAGE_CACHE_SIZE - 1) >>
+			PAGE_CACHE_SHIFT;
+		pages = kzalloc(num_pages, mask);
+		if (!pages) {
+			__free_extent_buffer(eb);
+			return NULL;
+		}
+		eb->pages = pages;
+	} else {
+		eb->pages = eb->inline_pages;
+	}
 
 	return eb;
 }
 
-static void __free_extent_buffer(struct extent_buffer *eb)
+static int extent_buffer_under_io(struct extent_buffer *eb)
 {
-#if LEAK_DEBUG
-	unsigned long flags;
-	spin_lock_irqsave(&leak_lock, flags);
-	list_del(&eb->leak_list);
-	spin_unlock_irqrestore(&leak_lock, flags);
-#endif
-	kmem_cache_free(extent_buffer_cache, eb);
+	return (atomic_read(&eb->io_pages) ||
+		test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags) ||
+		test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
 }
 
 /*
@@ -3632,8 +3979,7 @@ static void btrfs_release_extent_buffer_page(struct extent_buffer *eb,
 	unsigned long index;
 	struct page *page;
 
-	if (!eb->first_page)
-		return;
+	BUG_ON(extent_buffer_under_io(eb));
 
 	index = num_extent_pages(eb->start, eb->len);
 	if (start_idx >= index)
@@ -3642,8 +3988,34 @@ static void btrfs_release_extent_buffer_page(struct extent_buffer *eb,
 	do {
 		index--;
 		page = extent_buffer_page(eb, index);
-		if (page)
+		if (page) {
+			spin_lock(&page->mapping->private_lock);
+			/*
+			 * We do this since we'll remove the pages after we've
+			 * removed the eb from the radix tree, so we could race
+			 * and have this page now attached to the new eb.  So
+			 * only clear page_private if it's still connected to
+			 * this eb.
+			 */
+			if (PagePrivate(page) &&
+			    page->private == (unsigned long)eb) {
+				BUG_ON(test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
+				BUG_ON(PageDirty(page));
+				BUG_ON(PageWriteback(page));
+				/*
+				 * We need to make sure we haven't be attached
+				 * to a new eb.
+				 */
+				ClearPagePrivate(page);
+				set_page_private(page, 0);
+				/* One for the page private */
+				page_cache_release(page);
+			}
+			spin_unlock(&page->mapping->private_lock);
+
+			/* One for when we alloced the page */
 			page_cache_release(page);
+		}
 	} while (index != start_idx);
 }
 
@@ -3656,9 +4028,50 @@ static inline void btrfs_release_extent_buffer(struct extent_buffer *eb)
 	__free_extent_buffer(eb);
 }
 
+static void check_buffer_tree_ref(struct extent_buffer *eb)
+{
+	/* the ref bit is tricky.  We have to make sure it is set
+	 * if we have the buffer dirty.   Otherwise the
+	 * code to free a buffer can end up dropping a dirty
+	 * page
+	 *
+	 * Once the ref bit is set, it won't go away while the
+	 * buffer is dirty or in writeback, and it also won't
+	 * go away while we have the reference count on the
+	 * eb bumped.
+	 *
+	 * We can't just set the ref bit without bumping the
+	 * ref on the eb because free_extent_buffer might
+	 * see the ref bit and try to clear it.  If this happens
+	 * free_extent_buffer might end up dropping our original
+	 * ref by mistake and freeing the page before we are able
+	 * to add one more ref.
+	 *
+	 * So bump the ref count first, then set the bit.  If someone
+	 * beat us to it, drop the ref we added.
+	 */
+	if (!test_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags)) {
+		atomic_inc(&eb->refs);
+		if (test_and_set_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
+			atomic_dec(&eb->refs);
+	}
+}
+
+static void mark_extent_buffer_accessed(struct extent_buffer *eb)
+{
+	unsigned long num_pages, i;
+
+	check_buffer_tree_ref(eb);
+
+	num_pages = num_extent_pages(eb->start, eb->len);
+	for (i = 0; i < num_pages; i++) {
+		struct page *p = extent_buffer_page(eb, i);
+		mark_page_accessed(p);
+	}
+}
+
 struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
-					  u64 start, unsigned long len,
-					  struct page *page0)
+					  u64 start, unsigned long len)
 {
 	unsigned long num_pages = num_extent_pages(start, len);
 	unsigned long i;
@@ -3674,7 +4087,7 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
 	eb = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT);
 	if (eb && atomic_inc_not_zero(&eb->refs)) {
 		rcu_read_unlock();
-		mark_page_accessed(eb->first_page);
+		mark_extent_buffer_accessed(eb);
 		return eb;
 	}
 	rcu_read_unlock();
@@ -3683,32 +4096,43 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
 	if (!eb)
 		return NULL;
 
-	if (page0) {
-		eb->first_page = page0;
-		i = 1;
-		index++;
-		page_cache_get(page0);
-		mark_page_accessed(page0);
-		set_page_extent_mapped(page0);
-		set_page_extent_head(page0, len);
-		uptodate = PageUptodate(page0);
-	} else {
-		i = 0;
-	}
-	for (; i < num_pages; i++, index++) {
+	for (i = 0; i < num_pages; i++, index++) {
 		p = find_or_create_page(mapping, index, GFP_NOFS);
 		if (!p) {
 			WARN_ON(1);
 			goto free_eb;
 		}
-		set_page_extent_mapped(p);
-		mark_page_accessed(p);
-		if (i == 0) {
-			eb->first_page = p;
-			set_page_extent_head(p, len);
-		} else {
-			set_page_private(p, EXTENT_PAGE_PRIVATE);
+
+		spin_lock(&mapping->private_lock);
+		if (PagePrivate(p)) {
+			/*
+			 * We could have already allocated an eb for this page
+			 * and attached one so lets see if we can get a ref on
+			 * the existing eb, and if we can we know it's good and
+			 * we can just return that one, else we know we can just
+			 * overwrite page->private.
+			 */
+			exists = (struct extent_buffer *)p->private;
+			if (atomic_inc_not_zero(&exists->refs)) {
+				spin_unlock(&mapping->private_lock);
+				unlock_page(p);
+				mark_extent_buffer_accessed(exists);
+				goto free_eb;
+			}
+
+			/*
+			 * Do this so attach doesn't complain and we need to
+			 * drop the ref the old guy had.
+			 */
+			ClearPagePrivate(p);
+			WARN_ON(PageDirty(p));
+			page_cache_release(p);
 		}
+		attach_extent_buffer_page(eb, p);
+		spin_unlock(&mapping->private_lock);
+		WARN_ON(PageDirty(p));
+		mark_page_accessed(p);
+		eb->pages[i] = p;
 		if (!PageUptodate(p))
 			uptodate = 0;
 
@@ -3716,12 +4140,10 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
 		 * see below about how we avoid a nasty race with release page
 		 * and why we unlock later
 		 */
-		if (i != 0)
-			unlock_page(p);
 	}
 	if (uptodate)
 		set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
-
+again:
 	ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
 	if (ret)
 		goto free_eb;
@@ -3731,14 +4153,21 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
 	if (ret == -EEXIST) {
 		exists = radix_tree_lookup(&tree->buffer,
 						start >> PAGE_CACHE_SHIFT);
-		/* add one reference for the caller */
-		atomic_inc(&exists->refs);
+		if (!atomic_inc_not_zero(&exists->refs)) {
+			spin_unlock(&tree->buffer_lock);
+			radix_tree_preload_end();
+			exists = NULL;
+			goto again;
+		}
 		spin_unlock(&tree->buffer_lock);
 		radix_tree_preload_end();
+		mark_extent_buffer_accessed(exists);
 		goto free_eb;
 	}
 	/* add one reference for the tree */
-	atomic_inc(&eb->refs);
+	spin_lock(&eb->refs_lock);
+	check_buffer_tree_ref(eb);
+	spin_unlock(&eb->refs_lock);
 	spin_unlock(&tree->buffer_lock);
 	radix_tree_preload_end();
 
@@ -3751,15 +4180,20 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
 	 * after the extent buffer is in the radix tree so
 	 * it doesn't get lost
 	 */
-	set_page_extent_mapped(eb->first_page);
-	set_page_extent_head(eb->first_page, eb->len);
-	if (!page0)
-		unlock_page(eb->first_page);
+	SetPageChecked(eb->pages[0]);
+	for (i = 1; i < num_pages; i++) {
+		p = extent_buffer_page(eb, i);
+		ClearPageChecked(p);
+		unlock_page(p);
+	}
+	unlock_page(eb->pages[0]);
 	return eb;
 
 free_eb:
-	if (eb->first_page && !page0)
-		unlock_page(eb->first_page);
+	for (i = 0; i < num_pages; i++) {
+		if (eb->pages[i])
+			unlock_page(eb->pages[i]);
+	}
 
 	if (!atomic_dec_and_test(&eb->refs))
 		return exists;
@@ -3776,7 +4210,7 @@ struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree,
 	eb = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT);
 	if (eb && atomic_inc_not_zero(&eb->refs)) {
 		rcu_read_unlock();
-		mark_page_accessed(eb->first_page);
+		mark_extent_buffer_accessed(eb);
 		return eb;
 	}
 	rcu_read_unlock();
@@ -3784,19 +4218,71 @@ struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree,
 	return NULL;
 }
 
+static inline void btrfs_release_extent_buffer_rcu(struct rcu_head *head)
+{
+	struct extent_buffer *eb =
+			container_of(head, struct extent_buffer, rcu_head);
+
+	__free_extent_buffer(eb);
+}
+
+/* Expects to have eb->eb_lock already held */
+static void release_extent_buffer(struct extent_buffer *eb, gfp_t mask)
+{
+	WARN_ON(atomic_read(&eb->refs) == 0);
+	if (atomic_dec_and_test(&eb->refs)) {
+		struct extent_io_tree *tree = eb->tree;
+
+		spin_unlock(&eb->refs_lock);
+
+		spin_lock(&tree->buffer_lock);
+		radix_tree_delete(&tree->buffer,
+				  eb->start >> PAGE_CACHE_SHIFT);
+		spin_unlock(&tree->buffer_lock);
+
+		/* Should be safe to release our pages at this point */
+		btrfs_release_extent_buffer_page(eb, 0);
+
+		call_rcu(&eb->rcu_head, btrfs_release_extent_buffer_rcu);
+		return;
+	}
+	spin_unlock(&eb->refs_lock);
+}
+
 void free_extent_buffer(struct extent_buffer *eb)
 {
 	if (!eb)
 		return;
 
-	if (!atomic_dec_and_test(&eb->refs))
+	spin_lock(&eb->refs_lock);
+	if (atomic_read(&eb->refs) == 2 &&
+	    test_bit(EXTENT_BUFFER_STALE, &eb->bflags) &&
+	    !extent_buffer_under_io(eb) &&
+	    test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
+		atomic_dec(&eb->refs);
+
+	/*
+	 * I know this is terrible, but it's temporary until we stop tracking
+	 * the uptodate bits and such for the extent buffers.
+	 */
+	release_extent_buffer(eb, GFP_ATOMIC);
+}
+
+void free_extent_buffer_stale(struct extent_buffer *eb)
+{
+	if (!eb)
 		return;
 
-	WARN_ON(1);
+	spin_lock(&eb->refs_lock);
+	set_bit(EXTENT_BUFFER_STALE, &eb->bflags);
+
+	if (atomic_read(&eb->refs) == 2 && !extent_buffer_under_io(eb) &&
+	    test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
+		atomic_dec(&eb->refs);
+	release_extent_buffer(eb, GFP_NOFS);
 }
 
-int clear_extent_buffer_dirty(struct extent_io_tree *tree,
-			      struct extent_buffer *eb)
+void clear_extent_buffer_dirty(struct extent_buffer *eb)
 {
 	unsigned long i;
 	unsigned long num_pages;
@@ -3812,10 +4298,6 @@ int clear_extent_buffer_dirty(struct extent_io_tree *tree,
 		lock_page(page);
 		WARN_ON(!PagePrivate(page));
 
-		set_page_extent_mapped(page);
-		if (i == 0)
-			set_page_extent_head(page, eb->len);
-
 		clear_page_dirty_for_io(page);
 		spin_lock_irq(&page->mapping->tree_lock);
 		if (!PageDirty(page)) {
@@ -3827,24 +4309,29 @@ int clear_extent_buffer_dirty(struct extent_io_tree *tree,
 		ClearPageError(page);
 		unlock_page(page);
 	}
-	return 0;
+	WARN_ON(atomic_read(&eb->refs) == 0);
 }
 
-int set_extent_buffer_dirty(struct extent_io_tree *tree,
-			     struct extent_buffer *eb)
+int set_extent_buffer_dirty(struct extent_buffer *eb)
 {
 	unsigned long i;
 	unsigned long num_pages;
 	int was_dirty = 0;
 
+	check_buffer_tree_ref(eb);
+
 	was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags);
+
 	num_pages = num_extent_pages(eb->start, eb->len);
+	WARN_ON(atomic_read(&eb->refs) == 0);
+	WARN_ON(!test_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags));
+
 	for (i = 0; i < num_pages; i++)
-		__set_page_dirty_nobuffers(extent_buffer_page(eb, i));
+		set_page_dirty(extent_buffer_page(eb, i));
 	return was_dirty;
 }
 
-static int __eb_straddles_pages(u64 start, u64 len)
+static int range_straddles_pages(u64 start, u64 len)
 {
 	if (len < PAGE_CACHE_SIZE)
 		return 1;
@@ -3855,25 +4342,14 @@ static int __eb_straddles_pages(u64 start, u64 len)
 	return 0;
 }
 
-static int eb_straddles_pages(struct extent_buffer *eb)
-{
-	return __eb_straddles_pages(eb->start, eb->len);
-}
-
-int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
-				struct extent_buffer *eb,
-				struct extent_state **cached_state)
+int clear_extent_buffer_uptodate(struct extent_buffer *eb)
 {
 	unsigned long i;
 	struct page *page;
 	unsigned long num_pages;
 
-	num_pages = num_extent_pages(eb->start, eb->len);
 	clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
-
-	clear_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
-			      cached_state, GFP_NOFS);
-
+	num_pages = num_extent_pages(eb->start, eb->len);
 	for (i = 0; i < num_pages; i++) {
 		page = extent_buffer_page(eb, i);
 		if (page)
@@ -3882,27 +4358,16 @@ int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
 	return 0;
 }
 
-int set_extent_buffer_uptodate(struct extent_io_tree *tree,
-				struct extent_buffer *eb)
+int set_extent_buffer_uptodate(struct extent_buffer *eb)
 {
 	unsigned long i;
 	struct page *page;
 	unsigned long num_pages;
 
+	set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
 	num_pages = num_extent_pages(eb->start, eb->len);
-
-	if (eb_straddles_pages(eb)) {
-		set_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
-				    NULL, GFP_NOFS);
-	}
 	for (i = 0; i < num_pages; i++) {
 		page = extent_buffer_page(eb, i);
-		if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) ||
-		    ((i == num_pages - 1) &&
-		     ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) {
-			check_page_uptodate(tree, page);
-			continue;
-		}
 		SetPageUptodate(page);
 	}
 	return 0;
@@ -3917,7 +4382,7 @@ int extent_range_uptodate(struct extent_io_tree *tree,
 	int uptodate;
 	unsigned long index;
 
-	if (__eb_straddles_pages(start, end - start + 1)) {
+	if (range_straddles_pages(start, end - start + 1)) {
 		ret = test_range_bit(tree, start, end,
 				     EXTENT_UPTODATE, 1, NULL);
 		if (ret)
@@ -3939,35 +4404,9 @@ int extent_range_uptodate(struct extent_io_tree *tree,
 	return pg_uptodate;
 }
 
-int extent_buffer_uptodate(struct extent_io_tree *tree,
-			   struct extent_buffer *eb,
-			   struct extent_state *cached_state)
+int extent_buffer_uptodate(struct extent_buffer *eb)
 {
-	int ret = 0;
-	unsigned long num_pages;
-	unsigned long i;
-	struct page *page;
-	int pg_uptodate = 1;
-
-	if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
-		return 1;
-
-	if (eb_straddles_pages(eb)) {
-		ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1,
-				   EXTENT_UPTODATE, 1, cached_state);
-		if (ret)
-			return ret;
-	}
-
-	num_pages = num_extent_pages(eb->start, eb->len);
-	for (i = 0; i < num_pages; i++) {
-		page = extent_buffer_page(eb, i);
-		if (!PageUptodate(page)) {
-			pg_uptodate = 0;
-			break;
-		}
-	}
-	return pg_uptodate;
+	return test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
 }
 
 int read_extent_buffer_pages(struct extent_io_tree *tree,
@@ -3981,21 +4420,14 @@ int read_extent_buffer_pages(struct extent_io_tree *tree,
 	int ret = 0;
 	int locked_pages = 0;
 	int all_uptodate = 1;
-	int inc_all_pages = 0;
 	unsigned long num_pages;
+	unsigned long num_reads = 0;
 	struct bio *bio = NULL;
 	unsigned long bio_flags = 0;
 
 	if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
 		return 0;
 
-	if (eb_straddles_pages(eb)) {
-		if (test_range_bit(tree, eb->start, eb->start + eb->len - 1,
-				   EXTENT_UPTODATE, 1, NULL)) {
-			return 0;
-		}
-	}
-
 	if (start) {
 		WARN_ON(start < eb->start);
 		start_i = (start >> PAGE_CACHE_SHIFT) -
@@ -4014,8 +4446,10 @@ int read_extent_buffer_pages(struct extent_io_tree *tree,
 			lock_page(page);
 		}
 		locked_pages++;
-		if (!PageUptodate(page))
+		if (!PageUptodate(page)) {
+			num_reads++;
 			all_uptodate = 0;
+		}
 	}
 	if (all_uptodate) {
 		if (start_i == 0)
@@ -4023,20 +4457,12 @@ int read_extent_buffer_pages(struct extent_io_tree *tree,
 		goto unlock_exit;
 	}
 
+	clear_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+	eb->failed_mirror = 0;
+	atomic_set(&eb->io_pages, num_reads);
 	for (i = start_i; i < num_pages; i++) {
 		page = extent_buffer_page(eb, i);
-
-		WARN_ON(!PagePrivate(page));
-
-		set_page_extent_mapped(page);
-		if (i == 0)
-			set_page_extent_head(page, eb->len);
-
-		if (inc_all_pages)
-			page_cache_get(page);
 		if (!PageUptodate(page)) {
-			if (start_i == 0)
-				inc_all_pages = 1;
 			ClearPageError(page);
 			err = __extent_read_full_page(tree, page,
 						      get_extent, &bio,
@@ -4048,8 +4474,11 @@ int read_extent_buffer_pages(struct extent_io_tree *tree,
 		}
 	}
 
-	if (bio)
-		submit_one_bio(READ, bio, mirror_num, bio_flags);
+	if (bio) {
+		err = submit_one_bio(READ, bio, mirror_num, bio_flags);
+		if (err)
+			return err;
+	}
 
 	if (ret || wait != WAIT_COMPLETE)
 		return ret;
@@ -4061,8 +4490,6 @@ int read_extent_buffer_pages(struct extent_io_tree *tree,
 			ret = -EIO;
 	}
 
-	if (!ret)
-		set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
 	return ret;
 
 unlock_exit:
@@ -4304,15 +4731,20 @@ static void copy_pages(struct page *dst_page, struct page *src_page,
 {
 	char *dst_kaddr = page_address(dst_page);
 	char *src_kaddr;
+	int must_memmove = 0;
 
 	if (dst_page != src_page) {
 		src_kaddr = page_address(src_page);
 	} else {
 		src_kaddr = dst_kaddr;
-		BUG_ON(areas_overlap(src_off, dst_off, len));
+		if (areas_overlap(src_off, dst_off, len))
+			must_memmove = 1;
 	}
 
-	memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len);
+	if (must_memmove)
+		memmove(dst_kaddr + dst_off, src_kaddr + src_off, len);
+	else
+		memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len);
 }
 
 void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
@@ -4382,7 +4814,7 @@ void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
 		       "len %lu len %lu\n", dst_offset, len, dst->len);
 		BUG_ON(1);
 	}
-	if (!areas_overlap(src_offset, dst_offset, len)) {
+	if (dst_offset < src_offset) {
 		memcpy_extent_buffer(dst, dst_offset, src_offset, len);
 		return;
 	}
@@ -4408,47 +4840,48 @@ void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
 	}
 }
 
-static inline void btrfs_release_extent_buffer_rcu(struct rcu_head *head)
+int try_release_extent_buffer(struct page *page, gfp_t mask)
 {
-	struct extent_buffer *eb =
-			container_of(head, struct extent_buffer, rcu_head);
-
-	btrfs_release_extent_buffer(eb);
-}
-
-int try_release_extent_buffer(struct extent_io_tree *tree, struct page *page)
-{
-	u64 start = page_offset(page);
 	struct extent_buffer *eb;
-	int ret = 1;
 
-	spin_lock(&tree->buffer_lock);
-	eb = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT);
-	if (!eb) {
-		spin_unlock(&tree->buffer_lock);
-		return ret;
+	/*
+	 * We need to make sure noboody is attaching this page to an eb right
+	 * now.
+	 */
+	spin_lock(&page->mapping->private_lock);
+	if (!PagePrivate(page)) {
+		spin_unlock(&page->mapping->private_lock);
+		return 1;
 	}
 
-	if (test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
-		ret = 0;
-		goto out;
-	}
+	eb = (struct extent_buffer *)page->private;
+	BUG_ON(!eb);
 
 	/*
-	 * set @eb->refs to 0 if it is already 1, and then release the @eb.
-	 * Or go back.
+	 * This is a little awful but should be ok, we need to make sure that
+	 * the eb doesn't disappear out from under us while we're looking at
+	 * this page.
 	 */
-	if (atomic_cmpxchg(&eb->refs, 1, 0) != 1) {
-		ret = 0;
-		goto out;
+	spin_lock(&eb->refs_lock);
+	if (atomic_read(&eb->refs) != 1 || extent_buffer_under_io(eb)) {
+		spin_unlock(&eb->refs_lock);
+		spin_unlock(&page->mapping->private_lock);
+		return 0;
 	}
+	spin_unlock(&page->mapping->private_lock);
 
-	radix_tree_delete(&tree->buffer, start >> PAGE_CACHE_SHIFT);
-out:
-	spin_unlock(&tree->buffer_lock);
+	if ((mask & GFP_NOFS) == GFP_NOFS)
+		mask = GFP_NOFS;
 
-	/* at this point we can safely release the extent buffer */
-	if (atomic_read(&eb->refs) == 0)
-		call_rcu(&eb->rcu_head, btrfs_release_extent_buffer_rcu);
-	return ret;
+	/*
+	 * If tree ref isn't set then we know the ref on this eb is a real ref,
+	 * so just return, this page will likely be freed soon anyway.
+	 */
+	if (!test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags)) {
+		spin_unlock(&eb->refs_lock);
+		return 0;
+	}
+	release_extent_buffer(eb, mask);
+
+	return 1;
 }
diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index cecc3518c12..faf10eb57f7 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -35,6 +35,10 @@
 #define EXTENT_BUFFER_DIRTY 2
 #define EXTENT_BUFFER_CORRUPT 3
 #define EXTENT_BUFFER_READAHEAD 4	/* this got triggered by readahead */
+#define EXTENT_BUFFER_TREE_REF 5
+#define EXTENT_BUFFER_STALE 6
+#define EXTENT_BUFFER_WRITEBACK 7
+#define EXTENT_BUFFER_IOERR 8
 
 /* these are flags for extent_clear_unlock_delalloc */
 #define EXTENT_CLEAR_UNLOCK_PAGE 0x1
@@ -54,6 +58,7 @@
 #define EXTENT_PAGE_PRIVATE_FIRST_PAGE 3
 
 struct extent_state;
+struct btrfs_root;
 
 typedef	int (extent_submit_bio_hook_t)(struct inode *inode, int rw,
 				       struct bio *bio, int mirror_num,
@@ -69,9 +74,7 @@ struct extent_io_ops {
 			      size_t size, struct bio *bio,
 			      unsigned long bio_flags);
 	int (*readpage_io_hook)(struct page *page, u64 start, u64 end);
-	int (*readpage_io_failed_hook)(struct bio *bio, struct page *page,
-				       u64 start, u64 end, int failed_mirror,
-				       struct extent_state *state);
+	int (*readpage_io_failed_hook)(struct page *page, int failed_mirror);
 	int (*writepage_io_failed_hook)(struct bio *bio, struct page *page,
 					u64 start, u64 end,
 				       struct extent_state *state);
@@ -97,6 +100,7 @@ struct extent_io_tree {
 	struct radix_tree_root buffer;
 	struct address_space *mapping;
 	u64 dirty_bytes;
+	int track_uptodate;
 	spinlock_t lock;
 	spinlock_t buffer_lock;
 	struct extent_io_ops *ops;
@@ -119,16 +123,21 @@ struct extent_state {
 	struct list_head leak_list;
 };
 
+#define INLINE_EXTENT_BUFFER_PAGES 16
+#define MAX_INLINE_EXTENT_BUFFER_SIZE (INLINE_EXTENT_BUFFER_PAGES * PAGE_CACHE_SIZE)
 struct extent_buffer {
 	u64 start;
 	unsigned long len;
 	unsigned long map_start;
 	unsigned long map_len;
-	struct page *first_page;
 	unsigned long bflags;
+	struct extent_io_tree *tree;
+	spinlock_t refs_lock;
+	atomic_t refs;
+	atomic_t io_pages;
+	int failed_mirror;
 	struct list_head leak_list;
 	struct rcu_head rcu_head;
-	atomic_t refs;
 	pid_t lock_owner;
 
 	/* count of read lock holders on the extent buffer */
@@ -152,6 +161,9 @@ struct extent_buffer {
 	 * to unlock
 	 */
 	wait_queue_head_t read_lock_wq;
+	wait_queue_head_t lock_wq;
+	struct page *inline_pages[INLINE_EXTENT_BUFFER_PAGES];
+	struct page **pages;
 };
 
 static inline void extent_set_compress_type(unsigned long *bio_flags,
@@ -178,18 +190,17 @@ void extent_io_tree_init(struct extent_io_tree *tree,
 int try_release_extent_mapping(struct extent_map_tree *map,
 			       struct extent_io_tree *tree, struct page *page,
 			       gfp_t mask);
-int try_release_extent_buffer(struct extent_io_tree *tree, struct page *page);
+int try_release_extent_buffer(struct page *page, gfp_t mask);
 int try_release_extent_state(struct extent_map_tree *map,
 			     struct extent_io_tree *tree, struct page *page,
 			     gfp_t mask);
-int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask);
+int lock_extent(struct extent_io_tree *tree, u64 start, u64 end);
 int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
-		     int bits, struct extent_state **cached, gfp_t mask);
-int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask);
+		     int bits, struct extent_state **cached);
+int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end);
 int unlock_extent_cached(struct extent_io_tree *tree, u64 start, u64 end,
 			 struct extent_state **cached, gfp_t mask);
-int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
-		    gfp_t mask);
+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 __init extent_io_init(void);
@@ -210,7 +221,7 @@ int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 		    int bits, gfp_t mask);
 int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
-		   int bits, int exclusive_bits, u64 *failed_start,
+		   int bits, u64 *failed_start,
 		   struct extent_state **cached_state, gfp_t mask);
 int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end,
 			struct extent_state **cached_state, gfp_t mask);
@@ -240,6 +251,8 @@ int extent_writepages(struct extent_io_tree *tree,
 		      struct address_space *mapping,
 		      get_extent_t *get_extent,
 		      struct writeback_control *wbc);
+int btree_write_cache_pages(struct address_space *mapping,
+			    struct writeback_control *wbc);
 int extent_readpages(struct extent_io_tree *tree,
 		     struct address_space *mapping,
 		     struct list_head *pages, unsigned nr_pages,
@@ -251,11 +264,11 @@ int get_state_private(struct extent_io_tree *tree, u64 start, u64 *private);
 void set_page_extent_mapped(struct page *page);
 
 struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
-					  u64 start, unsigned long len,
-					  struct page *page0);
+					  u64 start, unsigned long len);
 struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree,
 					 u64 start, unsigned long len);
 void free_extent_buffer(struct extent_buffer *eb);
+void free_extent_buffer_stale(struct extent_buffer *eb);
 #define WAIT_NONE	0
 #define WAIT_COMPLETE	1
 #define WAIT_PAGE_LOCK	2
@@ -287,19 +300,12 @@ void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
 			   unsigned long src_offset, unsigned long len);
 void memset_extent_buffer(struct extent_buffer *eb, char c,
 			  unsigned long start, unsigned long len);
-int wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits);
-int clear_extent_buffer_dirty(struct extent_io_tree *tree,
-			      struct extent_buffer *eb);
-int set_extent_buffer_dirty(struct extent_io_tree *tree,
-			     struct extent_buffer *eb);
-int set_extent_buffer_uptodate(struct extent_io_tree *tree,
-			       struct extent_buffer *eb);
-int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
-				struct extent_buffer *eb,
-				struct extent_state **cached_state);
-int extent_buffer_uptodate(struct extent_io_tree *tree,
-			   struct extent_buffer *eb,
-			   struct extent_state *cached_state);
+void wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits);
+void clear_extent_buffer_dirty(struct extent_buffer *eb);
+int set_extent_buffer_dirty(struct extent_buffer *eb);
+int set_extent_buffer_uptodate(struct extent_buffer *eb);
+int clear_extent_buffer_uptodate(struct extent_buffer *eb);
+int extent_buffer_uptodate(struct extent_buffer *eb);
 int map_private_extent_buffer(struct extent_buffer *eb, unsigned long offset,
 		      unsigned long min_len, char **map,
 		      unsigned long *map_start,
@@ -320,4 +326,6 @@ int repair_io_failure(struct btrfs_mapping_tree *map_tree, u64 start,
 			u64 length, u64 logical, struct page *page,
 			int mirror_num);
 int end_extent_writepage(struct page *page, int err, u64 start, u64 end);
+int repair_eb_io_failure(struct btrfs_root *root, struct extent_buffer *eb,
+			 int mirror_num);
 #endif
diff --git a/fs/btrfs/file-item.c b/fs/btrfs/file-item.c
index 078b4fd5450..5d158d32023 100644
--- a/fs/btrfs/file-item.c
+++ b/fs/btrfs/file-item.c
@@ -25,10 +25,12 @@
 #include "transaction.h"
 #include "print-tree.h"
 
-#define MAX_CSUM_ITEMS(r, size) ((((BTRFS_LEAF_DATA_SIZE(r) - \
+#define __MAX_CSUM_ITEMS(r, size) ((((BTRFS_LEAF_DATA_SIZE(r) - \
 				   sizeof(struct btrfs_item) * 2) / \
 				  size) - 1))
 
+#define MAX_CSUM_ITEMS(r, size) (min(__MAX_CSUM_ITEMS(r, size), PAGE_CACHE_SIZE))
+
 #define MAX_ORDERED_SUM_BYTES(r) ((PAGE_SIZE - \
 				   sizeof(struct btrfs_ordered_sum)) / \
 				   sizeof(struct btrfs_sector_sum) * \
@@ -59,7 +61,7 @@ int btrfs_insert_file_extent(struct btrfs_trans_handle *trans,
 				      sizeof(*item));
 	if (ret < 0)
 		goto out;
-	BUG_ON(ret);
+	BUG_ON(ret); /* Can't happen */
 	leaf = path->nodes[0];
 	item = btrfs_item_ptr(leaf, path->slots[0],
 			      struct btrfs_file_extent_item);
@@ -284,6 +286,7 @@ int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
 	struct btrfs_ordered_sum *sums;
 	struct btrfs_sector_sum *sector_sum;
 	struct btrfs_csum_item *item;
+	LIST_HEAD(tmplist);
 	unsigned long offset;
 	int ret;
 	size_t size;
@@ -358,7 +361,10 @@ int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
 					MAX_ORDERED_SUM_BYTES(root));
 			sums = kzalloc(btrfs_ordered_sum_size(root, size),
 					GFP_NOFS);
-			BUG_ON(!sums);
+			if (!sums) {
+				ret = -ENOMEM;
+				goto fail;
+			}
 
 			sector_sum = sums->sums;
 			sums->bytenr = start;
@@ -380,12 +386,19 @@ int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
 				offset += csum_size;
 				sector_sum++;
 			}
-			list_add_tail(&sums->list, list);
+			list_add_tail(&sums->list, &tmplist);
 		}
 		path->slots[0]++;
 	}
 	ret = 0;
 fail:
+	while (ret < 0 && !list_empty(&tmplist)) {
+		sums = list_entry(&tmplist, struct btrfs_ordered_sum, list);
+		list_del(&sums->list);
+		kfree(sums);
+	}
+	list_splice_tail(&tmplist, list);
+
 	btrfs_free_path(path);
 	return ret;
 }
@@ -420,7 +433,7 @@ int btrfs_csum_one_bio(struct btrfs_root *root, struct inode *inode,
 		offset = page_offset(bvec->bv_page) + bvec->bv_offset;
 
 	ordered = btrfs_lookup_ordered_extent(inode, offset);
-	BUG_ON(!ordered);
+	BUG_ON(!ordered); /* Logic error */
 	sums->bytenr = ordered->start;
 
 	while (bio_index < bio->bi_vcnt) {
@@ -439,11 +452,11 @@ int btrfs_csum_one_bio(struct btrfs_root *root, struct inode *inode,
 
 			sums = kzalloc(btrfs_ordered_sum_size(root, bytes_left),
 				       GFP_NOFS);
-			BUG_ON(!sums);
+			BUG_ON(!sums); /* -ENOMEM */
 			sector_sum = sums->sums;
 			sums->len = bytes_left;
 			ordered = btrfs_lookup_ordered_extent(inode, offset);
-			BUG_ON(!ordered);
+			BUG_ON(!ordered); /* Logic error */
 			sums->bytenr = ordered->start;
 		}
 
@@ -483,18 +496,17 @@ int btrfs_csum_one_bio(struct btrfs_root *root, struct inode *inode,
  * This calls btrfs_truncate_item with the correct args based on the
  * overlap, and fixes up the key as required.
  */
-static noinline int truncate_one_csum(struct btrfs_trans_handle *trans,
-				      struct btrfs_root *root,
-				      struct btrfs_path *path,
-				      struct btrfs_key *key,
-				      u64 bytenr, u64 len)
+static noinline void truncate_one_csum(struct btrfs_trans_handle *trans,
+				       struct btrfs_root *root,
+				       struct btrfs_path *path,
+				       struct btrfs_key *key,
+				       u64 bytenr, u64 len)
 {
 	struct extent_buffer *leaf;
 	u16 csum_size = btrfs_super_csum_size(root->fs_info->super_copy);
 	u64 csum_end;
 	u64 end_byte = bytenr + len;
 	u32 blocksize_bits = root->fs_info->sb->s_blocksize_bits;
-	int ret;
 
 	leaf = path->nodes[0];
 	csum_end = btrfs_item_size_nr(leaf, path->slots[0]) / csum_size;
@@ -510,7 +522,7 @@ static noinline int truncate_one_csum(struct btrfs_trans_handle *trans,
 		 */
 		u32 new_size = (bytenr - key->offset) >> blocksize_bits;
 		new_size *= csum_size;
-		ret = btrfs_truncate_item(trans, root, path, new_size, 1);
+		btrfs_truncate_item(trans, root, path, new_size, 1);
 	} else if (key->offset >= bytenr && csum_end > end_byte &&
 		   end_byte > key->offset) {
 		/*
@@ -522,15 +534,13 @@ static noinline int truncate_one_csum(struct btrfs_trans_handle *trans,
 		u32 new_size = (csum_end - end_byte) >> blocksize_bits;
 		new_size *= csum_size;
 
-		ret = btrfs_truncate_item(trans, root, path, new_size, 0);
+		btrfs_truncate_item(trans, root, path, new_size, 0);
 
 		key->offset = end_byte;
-		ret = btrfs_set_item_key_safe(trans, root, path, key);
-		BUG_ON(ret);
+		btrfs_set_item_key_safe(trans, root, path, key);
 	} else {
 		BUG();
 	}
-	return 0;
 }
 
 /*
@@ -635,13 +645,14 @@ int btrfs_del_csums(struct btrfs_trans_handle *trans,
 			 * item changed size or key
 			 */
 			ret = btrfs_split_item(trans, root, path, &key, offset);
-			BUG_ON(ret && ret != -EAGAIN);
+			if (ret && ret != -EAGAIN) {
+				btrfs_abort_transaction(trans, root, ret);
+				goto out;
+			}
 
 			key.offset = end_byte - 1;
 		} else {
-			ret = truncate_one_csum(trans, root, path,
-						&key, bytenr, len);
-			BUG_ON(ret);
+			truncate_one_csum(trans, root, path, &key, bytenr, len);
 			if (key.offset < bytenr)
 				break;
 		}
@@ -772,7 +783,7 @@ again:
 		if (diff != csum_size)
 			goto insert;
 
-		ret = btrfs_extend_item(trans, root, path, diff);
+		btrfs_extend_item(trans, root, path, diff);
 		goto csum;
 	}
 
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index e8d06b6b919..d83260d7498 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -452,7 +452,7 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 			split = alloc_extent_map();
 		if (!split2)
 			split2 = alloc_extent_map();
-		BUG_ON(!split || !split2);
+		BUG_ON(!split || !split2); /* -ENOMEM */
 
 		write_lock(&em_tree->lock);
 		em = lookup_extent_mapping(em_tree, start, len);
@@ -494,7 +494,7 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 			split->flags = flags;
 			split->compress_type = em->compress_type;
 			ret = add_extent_mapping(em_tree, split);
-			BUG_ON(ret);
+			BUG_ON(ret); /* Logic error */
 			free_extent_map(split);
 			split = split2;
 			split2 = NULL;
@@ -520,7 +520,7 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 			}
 
 			ret = add_extent_mapping(em_tree, split);
-			BUG_ON(ret);
+			BUG_ON(ret); /* Logic error */
 			free_extent_map(split);
 			split = NULL;
 		}
@@ -679,7 +679,7 @@ next_slot:
 						root->root_key.objectid,
 						new_key.objectid,
 						start - extent_offset, 0);
-				BUG_ON(ret);
+				BUG_ON(ret); /* -ENOMEM */
 				*hint_byte = disk_bytenr;
 			}
 			key.offset = start;
@@ -754,7 +754,7 @@ next_slot:
 						root->root_key.objectid,
 						key.objectid, key.offset -
 						extent_offset, 0);
-				BUG_ON(ret);
+				BUG_ON(ret); /* -ENOMEM */
 				inode_sub_bytes(inode,
 						extent_end - key.offset);
 				*hint_byte = disk_bytenr;
@@ -770,7 +770,10 @@ next_slot:
 
 			ret = btrfs_del_items(trans, root, path, del_slot,
 					      del_nr);
-			BUG_ON(ret);
+			if (ret) {
+				btrfs_abort_transaction(trans, root, ret);
+				goto out;
+			}
 
 			del_nr = 0;
 			del_slot = 0;
@@ -782,11 +785,13 @@ next_slot:
 		BUG_ON(1);
 	}
 
-	if (del_nr > 0) {
+	if (!ret && del_nr > 0) {
 		ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
-		BUG_ON(ret);
+		if (ret)
+			btrfs_abort_transaction(trans, root, ret);
 	}
 
+out:
 	btrfs_free_path(path);
 	return ret;
 }
@@ -944,7 +949,10 @@ again:
 			btrfs_release_path(path);
 			goto again;
 		}
-		BUG_ON(ret < 0);
+		if (ret < 0) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto out;
+		}
 
 		leaf = path->nodes[0];
 		fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
@@ -963,7 +971,7 @@ again:
 		ret = btrfs_inc_extent_ref(trans, root, bytenr, num_bytes, 0,
 					   root->root_key.objectid,
 					   ino, orig_offset, 0);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 
 		if (split == start) {
 			key.offset = start;
@@ -990,7 +998,7 @@ again:
 		ret = btrfs_free_extent(trans, root, bytenr, num_bytes,
 					0, root->root_key.objectid,
 					ino, orig_offset, 0);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 	}
 	other_start = 0;
 	other_end = start;
@@ -1007,7 +1015,7 @@ again:
 		ret = btrfs_free_extent(trans, root, bytenr, num_bytes,
 					0, root->root_key.objectid,
 					ino, orig_offset, 0);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 	}
 	if (del_nr == 0) {
 		fi = btrfs_item_ptr(leaf, path->slots[0],
@@ -1025,7 +1033,10 @@ again:
 		btrfs_mark_buffer_dirty(leaf);
 
 		ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
-		BUG_ON(ret);
+		if (ret < 0) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto out;
+		}
 	}
 out:
 	btrfs_free_path(path);
@@ -1105,8 +1116,7 @@ again:
 	if (start_pos < inode->i_size) {
 		struct btrfs_ordered_extent *ordered;
 		lock_extent_bits(&BTRFS_I(inode)->io_tree,
-				 start_pos, last_pos - 1, 0, &cached_state,
-				 GFP_NOFS);
+				 start_pos, last_pos - 1, 0, &cached_state);
 		ordered = btrfs_lookup_first_ordered_extent(inode,
 							    last_pos - 1);
 		if (ordered &&
@@ -1638,7 +1648,7 @@ static long btrfs_fallocate(struct file *file, int mode,
 		 * transaction
 		 */
 		lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start,
-				 locked_end, 0, &cached_state, GFP_NOFS);
+				 locked_end, 0, &cached_state);
 		ordered = btrfs_lookup_first_ordered_extent(inode,
 							    alloc_end - 1);
 		if (ordered &&
@@ -1667,7 +1677,13 @@ static long btrfs_fallocate(struct file *file, int mode,
 
 		em = btrfs_get_extent(inode, NULL, 0, cur_offset,
 				      alloc_end - cur_offset, 0);
-		BUG_ON(IS_ERR_OR_NULL(em));
+		if (IS_ERR_OR_NULL(em)) {
+			if (!em)
+				ret = -ENOMEM;
+			else
+				ret = PTR_ERR(em);
+			break;
+		}
 		last_byte = min(extent_map_end(em), alloc_end);
 		actual_end = min_t(u64, extent_map_end(em), offset + len);
 		last_byte = (last_byte + mask) & ~mask;
@@ -1737,7 +1753,7 @@ static int find_desired_extent(struct inode *inode, loff_t *offset, int origin)
 		return -ENXIO;
 
 	lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend, 0,
-			 &cached_state, GFP_NOFS);
+			 &cached_state);
 
 	/*
 	 * Delalloc is such a pain.  If we have a hole and we have pending
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index b02e379b14c..e88330d3df5 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -230,11 +230,13 @@ int btrfs_truncate_free_space_cache(struct btrfs_root *root,
 
 	if (ret) {
 		trans->block_rsv = rsv;
-		WARN_ON(1);
+		btrfs_abort_transaction(trans, root, ret);
 		return ret;
 	}
 
 	ret = btrfs_update_inode(trans, root, inode);
+	if (ret)
+		btrfs_abort_transaction(trans, root, ret);
 	trans->block_rsv = rsv;
 
 	return ret;
@@ -869,7 +871,7 @@ int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
 	io_ctl_prepare_pages(&io_ctl, inode, 0);
 
 	lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
-			 0, &cached_state, GFP_NOFS);
+			 0, &cached_state);
 
 	node = rb_first(&ctl->free_space_offset);
 	if (!node && cluster) {
@@ -1948,14 +1950,14 @@ again:
 		 */
 		ret = btrfs_add_free_space(block_group, old_start,
 					   offset - old_start);
-		WARN_ON(ret);
+		WARN_ON(ret); /* -ENOMEM */
 		goto out;
 	}
 
 	ret = remove_from_bitmap(ctl, info, &offset, &bytes);
 	if (ret == -EAGAIN)
 		goto again;
-	BUG_ON(ret);
+	BUG_ON(ret); /* logic error */
 out_lock:
 	spin_unlock(&ctl->tree_lock);
 out:
@@ -2346,7 +2348,7 @@ again:
 	rb_erase(&entry->offset_index, &ctl->free_space_offset);
 	ret = tree_insert_offset(&cluster->root, entry->offset,
 				 &entry->offset_index, 1);
-	BUG_ON(ret);
+	BUG_ON(ret); /* -EEXIST; Logic error */
 
 	trace_btrfs_setup_cluster(block_group, cluster,
 				  total_found * block_group->sectorsize, 1);
@@ -2439,7 +2441,7 @@ setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group,
 		ret = tree_insert_offset(&cluster->root, entry->offset,
 					 &entry->offset_index, 0);
 		total_size += entry->bytes;
-		BUG_ON(ret);
+		BUG_ON(ret); /* -EEXIST; Logic error */
 	} while (node && entry != last);
 
 	cluster->max_size = max_extent;
@@ -2830,6 +2832,7 @@ u64 btrfs_find_ino_for_alloc(struct btrfs_root *fs_root)
 		int ret;
 
 		ret = search_bitmap(ctl, entry, &offset, &count);
+		/* Logic error; Should be empty if it can't find anything */
 		BUG_ON(ret);
 
 		ino = offset;
diff --git a/fs/btrfs/inode-item.c b/fs/btrfs/inode-item.c
index baa74f3db69..a13cf1a96c7 100644
--- a/fs/btrfs/inode-item.c
+++ b/fs/btrfs/inode-item.c
@@ -19,6 +19,7 @@
 #include "ctree.h"
 #include "disk-io.h"
 #include "transaction.h"
+#include "print-tree.h"
 
 static int find_name_in_backref(struct btrfs_path *path, const char *name,
 			 int name_len, struct btrfs_inode_ref **ref_ret)
@@ -128,13 +129,14 @@ int btrfs_del_inode_ref(struct btrfs_trans_handle *trans,
 	item_start = btrfs_item_ptr_offset(leaf, path->slots[0]);
 	memmove_extent_buffer(leaf, ptr, ptr + sub_item_len,
 			      item_size - (ptr + sub_item_len - item_start));
-	ret = btrfs_truncate_item(trans, root, path,
+	btrfs_truncate_item(trans, root, path,
 				  item_size - sub_item_len, 1);
 out:
 	btrfs_free_path(path);
 	return ret;
 }
 
+/* Will return 0, -ENOMEM, -EMLINK, or -EEXIST or anything from the CoW path */
 int btrfs_insert_inode_ref(struct btrfs_trans_handle *trans,
 			   struct btrfs_root *root,
 			   const char *name, int name_len,
@@ -165,7 +167,7 @@ int btrfs_insert_inode_ref(struct btrfs_trans_handle *trans,
 			goto out;
 
 		old_size = btrfs_item_size_nr(path->nodes[0], path->slots[0]);
-		ret = btrfs_extend_item(trans, root, path, ins_len);
+		btrfs_extend_item(trans, root, path, ins_len);
 		ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
 				     struct btrfs_inode_ref);
 		ref = (struct btrfs_inode_ref *)((unsigned long)ref + old_size);
diff --git a/fs/btrfs/inode-map.c b/fs/btrfs/inode-map.c
index ee15d88b33d..b1a1c929ba8 100644
--- a/fs/btrfs/inode-map.c
+++ b/fs/btrfs/inode-map.c
@@ -178,7 +178,7 @@ static void start_caching(struct btrfs_root *root)
 
 	tsk = kthread_run(caching_kthread, root, "btrfs-ino-cache-%llu\n",
 			  root->root_key.objectid);
-	BUG_ON(IS_ERR(tsk));
+	BUG_ON(IS_ERR(tsk)); /* -ENOMEM */
 }
 
 int btrfs_find_free_ino(struct btrfs_root *root, u64 *objectid)
@@ -271,7 +271,7 @@ void btrfs_unpin_free_ino(struct btrfs_root *root)
 			break;
 
 		info = rb_entry(n, struct btrfs_free_space, offset_index);
-		BUG_ON(info->bitmap);
+		BUG_ON(info->bitmap); /* Logic error */
 
 		if (info->offset > root->cache_progress)
 			goto free;
@@ -439,17 +439,16 @@ int btrfs_save_ino_cache(struct btrfs_root *root,
 	if (ret)
 		goto out;
 	trace_btrfs_space_reservation(root->fs_info, "ino_cache",
-				      (u64)(unsigned long)trans,
-				      trans->bytes_reserved, 1);
+				      trans->transid, trans->bytes_reserved, 1);
 again:
 	inode = lookup_free_ino_inode(root, path);
-	if (IS_ERR(inode) && PTR_ERR(inode) != -ENOENT) {
+	if (IS_ERR(inode) && (PTR_ERR(inode) != -ENOENT || retry)) {
 		ret = PTR_ERR(inode);
 		goto out_release;
 	}
 
 	if (IS_ERR(inode)) {
-		BUG_ON(retry);
+		BUG_ON(retry); /* Logic error */
 		retry = true;
 
 		ret = create_free_ino_inode(root, trans, path);
@@ -460,12 +459,17 @@ again:
 
 	BTRFS_I(inode)->generation = 0;
 	ret = btrfs_update_inode(trans, root, inode);
-	WARN_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out_put;
+	}
 
 	if (i_size_read(inode) > 0) {
 		ret = btrfs_truncate_free_space_cache(root, trans, path, inode);
-		if (ret)
+		if (ret) {
+			btrfs_abort_transaction(trans, root, ret);
 			goto out_put;
+		}
 	}
 
 	spin_lock(&root->cache_lock);
@@ -502,8 +506,7 @@ out_put:
 	iput(inode);
 out_release:
 	trace_btrfs_space_reservation(root->fs_info, "ino_cache",
-				      (u64)(unsigned long)trans,
-				      trans->bytes_reserved, 0);
+				      trans->transid, trans->bytes_reserved, 0);
 	btrfs_block_rsv_release(root, trans->block_rsv, trans->bytes_reserved);
 out:
 	trans->block_rsv = rsv;
@@ -532,7 +535,7 @@ static int btrfs_find_highest_objectid(struct btrfs_root *root, u64 *objectid)
 	ret = btrfs_search_slot(NULL, root, &search_key, path, 0, 0);
 	if (ret < 0)
 		goto error;
-	BUG_ON(ret == 0);
+	BUG_ON(ret == 0); /* Corruption */
 	if (path->slots[0] > 0) {
 		slot = path->slots[0] - 1;
 		l = path->nodes[0];
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 3a0b5c1f9d3..115bc05e42b 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -150,7 +150,6 @@ static noinline int insert_inline_extent(struct btrfs_trans_handle *trans,
 	inode_add_bytes(inode, size);
 	ret = btrfs_insert_empty_item(trans, root, path, &key,
 				      datasize);
-	BUG_ON(ret);
 	if (ret) {
 		err = ret;
 		goto fail;
@@ -206,9 +205,9 @@ static noinline int insert_inline_extent(struct btrfs_trans_handle *trans,
 	 * could end up racing with unlink.
 	 */
 	BTRFS_I(inode)->disk_i_size = inode->i_size;
-	btrfs_update_inode(trans, root, inode);
+	ret = btrfs_update_inode(trans, root, inode);
 
-	return 0;
+	return ret;
 fail:
 	btrfs_free_path(path);
 	return err;
@@ -250,14 +249,18 @@ static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans,
 
 	ret = btrfs_drop_extents(trans, inode, start, aligned_end,
 				 &hint_byte, 1);
-	BUG_ON(ret);
+	if (ret)
+		return ret;
 
 	if (isize > actual_end)
 		inline_len = min_t(u64, isize, actual_end);
 	ret = insert_inline_extent(trans, root, inode, start,
 				   inline_len, compressed_size,
 				   compress_type, compressed_pages);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		return ret;
+	}
 	btrfs_delalloc_release_metadata(inode, end + 1 - start);
 	btrfs_drop_extent_cache(inode, start, aligned_end - 1, 0);
 	return 0;
@@ -293,7 +296,7 @@ static noinline int add_async_extent(struct async_cow *cow,
 	struct async_extent *async_extent;
 
 	async_extent = kmalloc(sizeof(*async_extent), GFP_NOFS);
-	BUG_ON(!async_extent);
+	BUG_ON(!async_extent); /* -ENOMEM */
 	async_extent->start = start;
 	async_extent->ram_size = ram_size;
 	async_extent->compressed_size = compressed_size;
@@ -344,8 +347,9 @@ static noinline int compress_file_range(struct inode *inode,
 	int will_compress;
 	int compress_type = root->fs_info->compress_type;
 
-	/* if this is a small write inside eof, kick off a defragbot */
-	if (end <= BTRFS_I(inode)->disk_i_size && (end - start + 1) < 16 * 1024)
+	/* if this is a small write inside eof, kick off a defrag */
+	if ((end - start + 1) < 16 * 1024 &&
+	    (start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size))
 		btrfs_add_inode_defrag(NULL, inode);
 
 	actual_end = min_t(u64, isize, end + 1);
@@ -433,7 +437,11 @@ again:
 cont:
 	if (start == 0) {
 		trans = btrfs_join_transaction(root);
-		BUG_ON(IS_ERR(trans));
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+			trans = NULL;
+			goto cleanup_and_out;
+		}
 		trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 
 		/* lets try to make an inline extent */
@@ -450,11 +458,11 @@ cont:
 						    total_compressed,
 						    compress_type, pages);
 		}
-		if (ret == 0) {
+		if (ret <= 0) {
 			/*
-			 * inline extent creation worked, we don't need
-			 * to create any more async work items.  Unlock
-			 * and free up our temp pages.
+			 * inline extent creation worked or returned error,
+			 * we don't need to create any more async work items.
+			 * Unlock and free up our temp pages.
 			 */
 			extent_clear_unlock_delalloc(inode,
 			     &BTRFS_I(inode)->io_tree,
@@ -547,7 +555,7 @@ cleanup_and_bail_uncompressed:
 	}
 
 out:
-	return 0;
+	return ret;
 
 free_pages_out:
 	for (i = 0; i < nr_pages_ret; i++) {
@@ -557,6 +565,20 @@ free_pages_out:
 	kfree(pages);
 
 	goto out;
+
+cleanup_and_out:
+	extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
+				     start, end, NULL,
+				     EXTENT_CLEAR_UNLOCK_PAGE |
+				     EXTENT_CLEAR_DIRTY |
+				     EXTENT_CLEAR_DELALLOC |
+				     EXTENT_SET_WRITEBACK |
+				     EXTENT_END_WRITEBACK);
+	if (!trans || IS_ERR(trans))
+		btrfs_error(root->fs_info, ret, "Failed to join transaction");
+	else
+		btrfs_abort_transaction(trans, root, ret);
+	goto free_pages_out;
 }
 
 /*
@@ -597,7 +619,7 @@ retry:
 
 			lock_extent(io_tree, async_extent->start,
 					 async_extent->start +
-					 async_extent->ram_size - 1, GFP_NOFS);
+					 async_extent->ram_size - 1);
 
 			/* allocate blocks */
 			ret = cow_file_range(inode, async_cow->locked_page,
@@ -606,6 +628,8 @@ retry:
 					     async_extent->ram_size - 1,
 					     &page_started, &nr_written, 0);
 
+			/* JDM XXX */
+
 			/*
 			 * if page_started, cow_file_range inserted an
 			 * inline extent and took care of all the unlocking
@@ -625,18 +649,21 @@ retry:
 		}
 
 		lock_extent(io_tree, async_extent->start,
-			    async_extent->start + async_extent->ram_size - 1,
-			    GFP_NOFS);
+			    async_extent->start + async_extent->ram_size - 1);
 
 		trans = btrfs_join_transaction(root);
-		BUG_ON(IS_ERR(trans));
-		trans->block_rsv = &root->fs_info->delalloc_block_rsv;
-		ret = btrfs_reserve_extent(trans, root,
+		if (IS_ERR(trans)) {
+			ret = PTR_ERR(trans);
+		} else {
+			trans->block_rsv = &root->fs_info->delalloc_block_rsv;
+			ret = btrfs_reserve_extent(trans, root,
 					   async_extent->compressed_size,
 					   async_extent->compressed_size,
-					   0, alloc_hint,
-					   (u64)-1, &ins, 1);
-		btrfs_end_transaction(trans, root);
+					   0, alloc_hint, &ins, 1);
+			if (ret)
+				btrfs_abort_transaction(trans, root, ret);
+			btrfs_end_transaction(trans, root);
+		}
 
 		if (ret) {
 			int i;
@@ -649,8 +676,10 @@ retry:
 			async_extent->pages = NULL;
 			unlock_extent(io_tree, async_extent->start,
 				      async_extent->start +
-				      async_extent->ram_size - 1, GFP_NOFS);
-			goto retry;
+				      async_extent->ram_size - 1);
+			if (ret == -ENOSPC)
+				goto retry;
+			goto out_free; /* JDM: Requeue? */
 		}
 
 		/*
@@ -662,7 +691,7 @@ retry:
 					async_extent->ram_size - 1, 0);
 
 		em = alloc_extent_map();
-		BUG_ON(!em);
+		BUG_ON(!em); /* -ENOMEM */
 		em->start = async_extent->start;
 		em->len = async_extent->ram_size;
 		em->orig_start = em->start;
@@ -694,7 +723,7 @@ retry:
 						ins.offset,
 						BTRFS_ORDERED_COMPRESSED,
 						async_extent->compress_type);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 
 		/*
 		 * clear dirty, set writeback and unlock the pages.
@@ -716,13 +745,17 @@ retry:
 				    ins.offset, async_extent->pages,
 				    async_extent->nr_pages);
 
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 		alloc_hint = ins.objectid + ins.offset;
 		kfree(async_extent);
 		cond_resched();
 	}
-
-	return 0;
+	ret = 0;
+out:
+	return ret;
+out_free:
+	kfree(async_extent);
+	goto out;
 }
 
 static u64 get_extent_allocation_hint(struct inode *inode, u64 start,
@@ -791,7 +824,18 @@ static noinline int cow_file_range(struct inode *inode,
 
 	BUG_ON(btrfs_is_free_space_inode(root, inode));
 	trans = btrfs_join_transaction(root);
-	BUG_ON(IS_ERR(trans));
+	if (IS_ERR(trans)) {
+		extent_clear_unlock_delalloc(inode,
+			     &BTRFS_I(inode)->io_tree,
+			     start, end, NULL,
+			     EXTENT_CLEAR_UNLOCK_PAGE |
+			     EXTENT_CLEAR_UNLOCK |
+			     EXTENT_CLEAR_DELALLOC |
+			     EXTENT_CLEAR_DIRTY |
+			     EXTENT_SET_WRITEBACK |
+			     EXTENT_END_WRITEBACK);
+		return PTR_ERR(trans);
+	}
 	trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 
 	num_bytes = (end - start + blocksize) & ~(blocksize - 1);
@@ -800,7 +844,8 @@ static noinline int cow_file_range(struct inode *inode,
 	ret = 0;
 
 	/* if this is a small write inside eof, kick off defrag */
-	if (end <= BTRFS_I(inode)->disk_i_size && num_bytes < 64 * 1024)
+	if (num_bytes < 64 * 1024 &&
+	    (start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size))
 		btrfs_add_inode_defrag(trans, inode);
 
 	if (start == 0) {
@@ -821,8 +866,10 @@ static noinline int cow_file_range(struct inode *inode,
 			*nr_written = *nr_written +
 			     (end - start + PAGE_CACHE_SIZE) / PAGE_CACHE_SIZE;
 			*page_started = 1;
-			ret = 0;
 			goto out;
+		} else if (ret < 0) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto out_unlock;
 		}
 	}
 
@@ -838,11 +885,14 @@ static noinline int cow_file_range(struct inode *inode,
 		cur_alloc_size = disk_num_bytes;
 		ret = btrfs_reserve_extent(trans, root, cur_alloc_size,
 					   root->sectorsize, 0, alloc_hint,
-					   (u64)-1, &ins, 1);
-		BUG_ON(ret);
+					   &ins, 1);
+		if (ret < 0) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto out_unlock;
+		}
 
 		em = alloc_extent_map();
-		BUG_ON(!em);
+		BUG_ON(!em); /* -ENOMEM */
 		em->start = start;
 		em->orig_start = em->start;
 		ram_size = ins.offset;
@@ -868,13 +918,16 @@ static noinline int cow_file_range(struct inode *inode,
 		cur_alloc_size = ins.offset;
 		ret = btrfs_add_ordered_extent(inode, start, ins.objectid,
 					       ram_size, cur_alloc_size, 0);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 
 		if (root->root_key.objectid ==
 		    BTRFS_DATA_RELOC_TREE_OBJECTID) {
 			ret = btrfs_reloc_clone_csums(inode, start,
 						      cur_alloc_size);
-			BUG_ON(ret);
+			if (ret) {
+				btrfs_abort_transaction(trans, root, ret);
+				goto out_unlock;
+			}
 		}
 
 		if (disk_num_bytes < cur_alloc_size)
@@ -899,11 +952,23 @@ static noinline int cow_file_range(struct inode *inode,
 		alloc_hint = ins.objectid + ins.offset;
 		start += cur_alloc_size;
 	}
-out:
 	ret = 0;
+out:
 	btrfs_end_transaction(trans, root);
 
 	return ret;
+out_unlock:
+	extent_clear_unlock_delalloc(inode,
+		     &BTRFS_I(inode)->io_tree,
+		     start, end, NULL,
+		     EXTENT_CLEAR_UNLOCK_PAGE |
+		     EXTENT_CLEAR_UNLOCK |
+		     EXTENT_CLEAR_DELALLOC |
+		     EXTENT_CLEAR_DIRTY |
+		     EXTENT_SET_WRITEBACK |
+		     EXTENT_END_WRITEBACK);
+
+	goto out;
 }
 
 /*
@@ -969,7 +1034,7 @@ static int cow_file_range_async(struct inode *inode, struct page *locked_page,
 			 1, 0, NULL, GFP_NOFS);
 	while (start < end) {
 		async_cow = kmalloc(sizeof(*async_cow), GFP_NOFS);
-		BUG_ON(!async_cow);
+		BUG_ON(!async_cow); /* -ENOMEM */
 		async_cow->inode = inode;
 		async_cow->root = root;
 		async_cow->locked_page = locked_page;
@@ -1060,7 +1125,7 @@ static noinline int run_delalloc_nocow(struct inode *inode,
 	u64 disk_bytenr;
 	u64 num_bytes;
 	int extent_type;
-	int ret;
+	int ret, err;
 	int type;
 	int nocow;
 	int check_prev = 1;
@@ -1078,7 +1143,11 @@ static noinline int run_delalloc_nocow(struct inode *inode,
 	else
 		trans = btrfs_join_transaction(root);
 
-	BUG_ON(IS_ERR(trans));
+	if (IS_ERR(trans)) {
+		btrfs_free_path(path);
+		return PTR_ERR(trans);
+	}
+
 	trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 
 	cow_start = (u64)-1;
@@ -1086,7 +1155,10 @@ static noinline int run_delalloc_nocow(struct inode *inode,
 	while (1) {
 		ret = btrfs_lookup_file_extent(trans, root, path, ino,
 					       cur_offset, 0);
-		BUG_ON(ret < 0);
+		if (ret < 0) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto error;
+		}
 		if (ret > 0 && path->slots[0] > 0 && check_prev) {
 			leaf = path->nodes[0];
 			btrfs_item_key_to_cpu(leaf, &found_key,
@@ -1100,8 +1172,10 @@ next_slot:
 		leaf = path->nodes[0];
 		if (path->slots[0] >= btrfs_header_nritems(leaf)) {
 			ret = btrfs_next_leaf(root, path);
-			if (ret < 0)
-				BUG_ON(1);
+			if (ret < 0) {
+				btrfs_abort_transaction(trans, root, ret);
+				goto error;
+			}
 			if (ret > 0)
 				break;
 			leaf = path->nodes[0];
@@ -1189,7 +1263,10 @@ out_check:
 			ret = cow_file_range(inode, locked_page, cow_start,
 					found_key.offset - 1, page_started,
 					nr_written, 1);
-			BUG_ON(ret);
+			if (ret) {
+				btrfs_abort_transaction(trans, root, ret);
+				goto error;
+			}
 			cow_start = (u64)-1;
 		}
 
@@ -1198,7 +1275,7 @@ out_check:
 			struct extent_map_tree *em_tree;
 			em_tree = &BTRFS_I(inode)->extent_tree;
 			em = alloc_extent_map();
-			BUG_ON(!em);
+			BUG_ON(!em); /* -ENOMEM */
 			em->start = cur_offset;
 			em->orig_start = em->start;
 			em->len = num_bytes;
@@ -1224,13 +1301,16 @@ out_check:
 
 		ret = btrfs_add_ordered_extent(inode, cur_offset, disk_bytenr,
 					       num_bytes, num_bytes, type);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 
 		if (root->root_key.objectid ==
 		    BTRFS_DATA_RELOC_TREE_OBJECTID) {
 			ret = btrfs_reloc_clone_csums(inode, cur_offset,
 						      num_bytes);
-			BUG_ON(ret);
+			if (ret) {
+				btrfs_abort_transaction(trans, root, ret);
+				goto error;
+			}
 		}
 
 		extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
@@ -1249,18 +1329,23 @@ out_check:
 	if (cow_start != (u64)-1) {
 		ret = cow_file_range(inode, locked_page, cow_start, end,
 				     page_started, nr_written, 1);
-		BUG_ON(ret);
+		if (ret) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto error;
+		}
 	}
 
+error:
 	if (nolock) {
-		ret = btrfs_end_transaction_nolock(trans, root);
-		BUG_ON(ret);
+		err = btrfs_end_transaction_nolock(trans, root);
 	} else {
-		ret = btrfs_end_transaction(trans, root);
-		BUG_ON(ret);
+		err = btrfs_end_transaction(trans, root);
 	}
+	if (!ret)
+		ret = err;
+
 	btrfs_free_path(path);
-	return 0;
+	return ret;
 }
 
 /*
@@ -1425,10 +1510,11 @@ int btrfs_merge_bio_hook(struct page *page, unsigned long offset,
 	map_length = length;
 	ret = btrfs_map_block(map_tree, READ, logical,
 			      &map_length, NULL, 0);
-
+	/* Will always return 0 or 1 with map_multi == NULL */
+	BUG_ON(ret < 0);
 	if (map_length < length + size)
 		return 1;
-	return ret;
+	return 0;
 }
 
 /*
@@ -1448,7 +1534,7 @@ static int __btrfs_submit_bio_start(struct inode *inode, int rw,
 	int ret = 0;
 
 	ret = btrfs_csum_one_bio(root, inode, bio, 0, 0);
-	BUG_ON(ret);
+	BUG_ON(ret); /* -ENOMEM */
 	return 0;
 }
 
@@ -1479,14 +1565,16 @@ static int btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	int ret = 0;
 	int skip_sum;
+	int metadata = 0;
 
 	skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
 
 	if (btrfs_is_free_space_inode(root, inode))
-		ret = btrfs_bio_wq_end_io(root->fs_info, bio, 2);
-	else
-		ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
-	BUG_ON(ret);
+		metadata = 2;
+
+	ret = btrfs_bio_wq_end_io(root->fs_info, bio, metadata);
+	if (ret)
+		return ret;
 
 	if (!(rw & REQ_WRITE)) {
 		if (bio_flags & EXTENT_BIO_COMPRESSED) {
@@ -1571,7 +1659,7 @@ again:
 	page_end = page_offset(page) + PAGE_CACHE_SIZE - 1;
 
 	lock_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end, 0,
-			 &cached_state, GFP_NOFS);
+			 &cached_state);
 
 	/* already ordered? We're done */
 	if (PagePrivate2(page))
@@ -1675,13 +1763,15 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
 	 */
 	ret = btrfs_drop_extents(trans, inode, file_pos, file_pos + num_bytes,
 				 &hint, 0);
-	BUG_ON(ret);
+	if (ret)
+		goto out;
 
 	ins.objectid = btrfs_ino(inode);
 	ins.offset = file_pos;
 	ins.type = BTRFS_EXTENT_DATA_KEY;
 	ret = btrfs_insert_empty_item(trans, root, path, &ins, sizeof(*fi));
-	BUG_ON(ret);
+	if (ret)
+		goto out;
 	leaf = path->nodes[0];
 	fi = btrfs_item_ptr(leaf, path->slots[0],
 			    struct btrfs_file_extent_item);
@@ -1709,10 +1799,10 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
 	ret = btrfs_alloc_reserved_file_extent(trans, root,
 					root->root_key.objectid,
 					btrfs_ino(inode), file_pos, &ins);
-	BUG_ON(ret);
+out:
 	btrfs_free_path(path);
 
-	return 0;
+	return ret;
 }
 
 /*
@@ -1740,35 +1830,41 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 					     end - start + 1);
 	if (!ret)
 		return 0;
-	BUG_ON(!ordered_extent);
+	BUG_ON(!ordered_extent); /* Logic error */
 
 	nolock = btrfs_is_free_space_inode(root, inode);
 
 	if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
-		BUG_ON(!list_empty(&ordered_extent->list));
+		BUG_ON(!list_empty(&ordered_extent->list)); /* Logic error */
 		ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent);
 		if (!ret) {
 			if (nolock)
 				trans = btrfs_join_transaction_nolock(root);
 			else
 				trans = btrfs_join_transaction(root);
-			BUG_ON(IS_ERR(trans));
+			if (IS_ERR(trans))
+				return PTR_ERR(trans);
 			trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 			ret = btrfs_update_inode_fallback(trans, root, inode);
-			BUG_ON(ret);
+			if (ret) /* -ENOMEM or corruption */
+				btrfs_abort_transaction(trans, root, ret);
 		}
 		goto out;
 	}
 
 	lock_extent_bits(io_tree, ordered_extent->file_offset,
 			 ordered_extent->file_offset + ordered_extent->len - 1,
-			 0, &cached_state, GFP_NOFS);
+			 0, &cached_state);
 
 	if (nolock)
 		trans = btrfs_join_transaction_nolock(root);
 	else
 		trans = btrfs_join_transaction(root);
-	BUG_ON(IS_ERR(trans));
+	if (IS_ERR(trans)) {
+		ret = PTR_ERR(trans);
+		trans = NULL;
+		goto out_unlock;
+	}
 	trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 
 	if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags))
@@ -1779,7 +1875,6 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 						ordered_extent->file_offset,
 						ordered_extent->file_offset +
 						ordered_extent->len);
-		BUG_ON(ret);
 	} else {
 		BUG_ON(root == root->fs_info->tree_root);
 		ret = insert_reserved_file_extent(trans, inode,
@@ -1793,11 +1888,14 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 		unpin_extent_cache(&BTRFS_I(inode)->extent_tree,
 				   ordered_extent->file_offset,
 				   ordered_extent->len);
-		BUG_ON(ret);
 	}
 	unlock_extent_cached(io_tree, ordered_extent->file_offset,
 			     ordered_extent->file_offset +
 			     ordered_extent->len - 1, &cached_state, GFP_NOFS);
+	if (ret < 0) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out;
+	}
 
 	add_pending_csums(trans, inode, ordered_extent->file_offset,
 			  &ordered_extent->list);
@@ -1805,7 +1903,10 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 	ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent);
 	if (!ret || !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) {
 		ret = btrfs_update_inode_fallback(trans, root, inode);
-		BUG_ON(ret);
+		if (ret) { /* -ENOMEM or corruption */
+			btrfs_abort_transaction(trans, root, ret);
+			goto out;
+		}
 	}
 	ret = 0;
 out:
@@ -1824,6 +1925,11 @@ out:
 	btrfs_put_ordered_extent(ordered_extent);
 
 	return 0;
+out_unlock:
+	unlock_extent_cached(io_tree, ordered_extent->file_offset,
+			     ordered_extent->file_offset +
+			     ordered_extent->len - 1, &cached_state, GFP_NOFS);
+	goto out;
 }
 
 static int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end,
@@ -1905,6 +2011,8 @@ struct delayed_iput {
 	struct inode *inode;
 };
 
+/* JDM: If this is fs-wide, why can't we add a pointer to
+ * btrfs_inode instead and avoid the allocation? */
 void btrfs_add_delayed_iput(struct inode *inode)
 {
 	struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
@@ -2051,20 +2159,27 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode)
 	/* grab metadata reservation from transaction handle */
 	if (reserve) {
 		ret = btrfs_orphan_reserve_metadata(trans, inode);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOSPC in reservation; Logic error? JDM */
 	}
 
 	/* insert an orphan item to track this unlinked/truncated file */
 	if (insert >= 1) {
 		ret = btrfs_insert_orphan_item(trans, root, btrfs_ino(inode));
-		BUG_ON(ret && ret != -EEXIST);
+		if (ret && ret != -EEXIST) {
+			btrfs_abort_transaction(trans, root, ret);
+			return ret;
+		}
+		ret = 0;
 	}
 
 	/* insert an orphan item to track subvolume contains orphan files */
 	if (insert >= 2) {
 		ret = btrfs_insert_orphan_item(trans, root->fs_info->tree_root,
 					       root->root_key.objectid);
-		BUG_ON(ret);
+		if (ret && ret != -EEXIST) {
+			btrfs_abort_transaction(trans, root, ret);
+			return ret;
+		}
 	}
 	return 0;
 }
@@ -2094,7 +2209,7 @@ int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode)
 
 	if (trans && delete_item) {
 		ret = btrfs_del_orphan_item(trans, root, btrfs_ino(inode));
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM or corruption (JDM: Recheck) */
 	}
 
 	if (release_rsv)
@@ -2228,7 +2343,7 @@ int btrfs_orphan_cleanup(struct btrfs_root *root)
 			}
 			ret = btrfs_del_orphan_item(trans, root,
 						    found_key.objectid);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM or corruption (JDM: Recheck) */
 			btrfs_end_transaction(trans, root);
 			continue;
 		}
@@ -2610,16 +2725,22 @@ static int __btrfs_unlink_inode(struct btrfs_trans_handle *trans,
 		printk(KERN_INFO "btrfs failed to delete reference to %.*s, "
 		       "inode %llu parent %llu\n", name_len, name,
 		       (unsigned long long)ino, (unsigned long long)dir_ino);
+		btrfs_abort_transaction(trans, root, ret);
 		goto err;
 	}
 
 	ret = btrfs_delete_delayed_dir_index(trans, root, dir, index);
-	if (ret)
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
 		goto err;
+	}
 
 	ret = btrfs_del_inode_ref_in_log(trans, root, name, name_len,
 					 inode, dir_ino);
-	BUG_ON(ret != 0 && ret != -ENOENT);
+	if (ret != 0 && ret != -ENOENT) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto err;
+	}
 
 	ret = btrfs_del_dir_entries_in_log(trans, root, name, name_len,
 					   dir, index);
@@ -2777,7 +2898,7 @@ static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir,
 			err = ret;
 			goto out;
 		}
-		BUG_ON(ret == 0);
+		BUG_ON(ret == 0); /* Corruption */
 		if (check_path_shared(root, path))
 			goto out;
 		btrfs_release_path(path);
@@ -2810,7 +2931,7 @@ static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir,
 		err = PTR_ERR(ref);
 		goto out;
 	}
-	BUG_ON(!ref);
+	BUG_ON(!ref); /* Logic error */
 	if (check_path_shared(root, path))
 		goto out;
 	index = btrfs_inode_ref_index(path->nodes[0], ref);
@@ -2917,23 +3038,42 @@ int btrfs_unlink_subvol(struct btrfs_trans_handle *trans,
 
 	di = btrfs_lookup_dir_item(trans, root, path, dir_ino,
 				   name, name_len, -1);
-	BUG_ON(IS_ERR_OR_NULL(di));
+	if (IS_ERR_OR_NULL(di)) {
+		if (!di)
+			ret = -ENOENT;
+		else
+			ret = PTR_ERR(di);
+		goto out;
+	}
 
 	leaf = path->nodes[0];
 	btrfs_dir_item_key_to_cpu(leaf, di, &key);
 	WARN_ON(key.type != BTRFS_ROOT_ITEM_KEY || key.objectid != objectid);
 	ret = btrfs_delete_one_dir_name(trans, root, path, di);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out;
+	}
 	btrfs_release_path(path);
 
 	ret = btrfs_del_root_ref(trans, root->fs_info->tree_root,
 				 objectid, root->root_key.objectid,
 				 dir_ino, &index, name, name_len);
 	if (ret < 0) {
-		BUG_ON(ret != -ENOENT);
+		if (ret != -ENOENT) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto out;
+		}
 		di = btrfs_search_dir_index_item(root, path, dir_ino,
 						 name, name_len);
-		BUG_ON(IS_ERR_OR_NULL(di));
+		if (IS_ERR_OR_NULL(di)) {
+			if (!di)
+				ret = -ENOENT;
+			else
+				ret = PTR_ERR(di);
+			btrfs_abort_transaction(trans, root, ret);
+			goto out;
+		}
 
 		leaf = path->nodes[0];
 		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
@@ -2943,15 +3083,19 @@ int btrfs_unlink_subvol(struct btrfs_trans_handle *trans,
 	btrfs_release_path(path);
 
 	ret = btrfs_delete_delayed_dir_index(trans, root, dir, index);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out;
+	}
 
 	btrfs_i_size_write(dir, dir->i_size - name_len * 2);
 	dir->i_mtime = dir->i_ctime = CURRENT_TIME;
 	ret = btrfs_update_inode(trans, root, dir);
-	BUG_ON(ret);
-
+	if (ret)
+		btrfs_abort_transaction(trans, root, ret);
+out:
 	btrfs_free_path(path);
-	return 0;
+	return ret;
 }
 
 static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
@@ -3161,8 +3305,8 @@ search_again:
 				}
 				size =
 				    btrfs_file_extent_calc_inline_size(size);
-				ret = btrfs_truncate_item(trans, root, path,
-							  size, 1);
+				btrfs_truncate_item(trans, root, path,
+						    size, 1);
 			} else if (root->ref_cows) {
 				inode_sub_bytes(inode, item_end + 1 -
 						found_key.offset);
@@ -3210,7 +3354,11 @@ delete:
 				ret = btrfs_del_items(trans, root, path,
 						pending_del_slot,
 						pending_del_nr);
-				BUG_ON(ret);
+				if (ret) {
+					btrfs_abort_transaction(trans,
+								root, ret);
+					goto error;
+				}
 				pending_del_nr = 0;
 			}
 			btrfs_release_path(path);
@@ -3223,8 +3371,10 @@ out:
 	if (pending_del_nr) {
 		ret = btrfs_del_items(trans, root, path, pending_del_slot,
 				      pending_del_nr);
-		BUG_ON(ret);
+		if (ret)
+			btrfs_abort_transaction(trans, root, ret);
 	}
+error:
 	btrfs_free_path(path);
 	return err;
 }
@@ -3282,8 +3432,7 @@ again:
 	}
 	wait_on_page_writeback(page);
 
-	lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state,
-			 GFP_NOFS);
+	lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state);
 	set_page_extent_mapped(page);
 
 	ordered = btrfs_lookup_ordered_extent(inode, page_start);
@@ -3359,7 +3508,7 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 		btrfs_wait_ordered_range(inode, hole_start,
 					 block_end - hole_start);
 		lock_extent_bits(io_tree, hole_start, block_end - 1, 0,
-				 &cached_state, GFP_NOFS);
+				 &cached_state);
 		ordered = btrfs_lookup_ordered_extent(inode, hole_start);
 		if (!ordered)
 			break;
@@ -3372,7 +3521,10 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 	while (1) {
 		em = btrfs_get_extent(inode, NULL, 0, cur_offset,
 				block_end - cur_offset, 0);
-		BUG_ON(IS_ERR_OR_NULL(em));
+		if (IS_ERR(em)) {
+			err = PTR_ERR(em);
+			break;
+		}
 		last_byte = min(extent_map_end(em), block_end);
 		last_byte = (last_byte + mask) & ~mask;
 		if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
@@ -3389,7 +3541,7 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 						 cur_offset + hole_size,
 						 &hint_byte, 1);
 			if (err) {
-				btrfs_update_inode(trans, root, inode);
+				btrfs_abort_transaction(trans, root, err);
 				btrfs_end_transaction(trans, root);
 				break;
 			}
@@ -3399,7 +3551,7 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 					0, hole_size, 0, hole_size,
 					0, 0, 0);
 			if (err) {
-				btrfs_update_inode(trans, root, inode);
+				btrfs_abort_transaction(trans, root, err);
 				btrfs_end_transaction(trans, root);
 				break;
 			}
@@ -3779,7 +3931,7 @@ static void inode_tree_del(struct inode *inode)
 	}
 }
 
-int btrfs_invalidate_inodes(struct btrfs_root *root)
+void btrfs_invalidate_inodes(struct btrfs_root *root)
 {
 	struct rb_node *node;
 	struct rb_node *prev;
@@ -3839,7 +3991,6 @@ again:
 		node = rb_next(node);
 	}
 	spin_unlock(&root->inode_lock);
-	return 0;
 }
 
 static int btrfs_init_locked_inode(struct inode *inode, void *p)
@@ -4581,18 +4732,26 @@ int btrfs_add_link(struct btrfs_trans_handle *trans,
 					     parent_ino, index);
 	}
 
-	if (ret == 0) {
-		ret = btrfs_insert_dir_item(trans, root, name, name_len,
-					    parent_inode, &key,
-					    btrfs_inode_type(inode), index);
-		if (ret)
-			goto fail_dir_item;
+	/* Nothing to clean up yet */
+	if (ret)
+		return ret;
 
-		btrfs_i_size_write(parent_inode, parent_inode->i_size +
-				   name_len * 2);
-		parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME;
-		ret = btrfs_update_inode(trans, root, parent_inode);
+	ret = btrfs_insert_dir_item(trans, root, name, name_len,
+				    parent_inode, &key,
+				    btrfs_inode_type(inode), index);
+	if (ret == -EEXIST)
+		goto fail_dir_item;
+	else if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		return ret;
 	}
+
+	btrfs_i_size_write(parent_inode, parent_inode->i_size +
+			   name_len * 2);
+	parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME;
+	ret = btrfs_update_inode(trans, root, parent_inode);
+	if (ret)
+		btrfs_abort_transaction(trans, root, ret);
 	return ret;
 
 fail_dir_item:
@@ -4806,7 +4965,8 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
 	} else {
 		struct dentry *parent = dentry->d_parent;
 		err = btrfs_update_inode(trans, root, inode);
-		BUG_ON(err);
+		if (err)
+			goto fail;
 		d_instantiate(dentry, inode);
 		btrfs_log_new_name(trans, inode, NULL, parent);
 	}
@@ -5137,7 +5297,7 @@ again:
 				ret = uncompress_inline(path, inode, page,
 							pg_offset,
 							extent_offset, item);
-				BUG_ON(ret);
+				BUG_ON(ret); /* -ENOMEM */
 			} else {
 				map = kmap(page);
 				read_extent_buffer(leaf, map + pg_offset, ptr,
@@ -5252,6 +5412,7 @@ out:
 		free_extent_map(em);
 		return ERR_PTR(err);
 	}
+	BUG_ON(!em); /* Error is always set */
 	return em;
 }
 
@@ -5414,7 +5575,7 @@ static struct extent_map *btrfs_new_extent_direct(struct inode *inode,
 
 	alloc_hint = get_extent_allocation_hint(inode, start, len);
 	ret = btrfs_reserve_extent(trans, root, len, root->sectorsize, 0,
-				   alloc_hint, (u64)-1, &ins, 1);
+				   alloc_hint, &ins, 1);
 	if (ret) {
 		em = ERR_PTR(ret);
 		goto out;
@@ -5602,7 +5763,7 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 		free_extent_map(em);
 		/* DIO will do one hole at a time, so just unlock a sector */
 		unlock_extent(&BTRFS_I(inode)->io_tree, start,
-			      start + root->sectorsize - 1, GFP_NOFS);
+			      start + root->sectorsize - 1);
 		return 0;
 	}
 
@@ -5743,7 +5904,7 @@ static void btrfs_endio_direct_read(struct bio *bio, int err)
 	} while (bvec <= bvec_end);
 
 	unlock_extent(&BTRFS_I(inode)->io_tree, dip->logical_offset,
-		      dip->logical_offset + dip->bytes - 1, GFP_NOFS);
+		      dip->logical_offset + dip->bytes - 1);
 	bio->bi_private = dip->private;
 
 	kfree(dip->csums);
@@ -5794,7 +5955,7 @@ again:
 
 	lock_extent_bits(&BTRFS_I(inode)->io_tree, ordered->file_offset,
 			 ordered->file_offset + ordered->len - 1, 0,
-			 &cached_state, GFP_NOFS);
+			 &cached_state);
 
 	if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags)) {
 		ret = btrfs_mark_extent_written(trans, inode,
@@ -5868,7 +6029,7 @@ static int __btrfs_submit_bio_start_direct_io(struct inode *inode, int rw,
 	int ret;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	ret = btrfs_csum_one_bio(root, inode, bio, offset, 1);
-	BUG_ON(ret);
+	BUG_ON(ret); /* -ENOMEM */
 	return 0;
 }
 
@@ -6209,7 +6370,7 @@ static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb,
 
 	while (1) {
 		lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-				 0, &cached_state, GFP_NOFS);
+				 0, &cached_state);
 		/*
 		 * We're concerned with the entire range that we're going to be
 		 * doing DIO to, so we need to make sure theres no ordered
@@ -6233,7 +6394,7 @@ static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb,
 	if (writing) {
 		write_bits = EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING;
 		ret = set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-				     EXTENT_DELALLOC, 0, NULL, &cached_state,
+				     EXTENT_DELALLOC, NULL, &cached_state,
 				     GFP_NOFS);
 		if (ret) {
 			clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart,
@@ -6363,8 +6524,7 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset)
 		btrfs_releasepage(page, GFP_NOFS);
 		return;
 	}
-	lock_extent_bits(tree, page_start, page_end, 0, &cached_state,
-			 GFP_NOFS);
+	lock_extent_bits(tree, page_start, page_end, 0, &cached_state);
 	ordered = btrfs_lookup_ordered_extent(page->mapping->host,
 					   page_offset(page));
 	if (ordered) {
@@ -6386,8 +6546,7 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset)
 		}
 		btrfs_put_ordered_extent(ordered);
 		cached_state = NULL;
-		lock_extent_bits(tree, page_start, page_end, 0, &cached_state,
-				 GFP_NOFS);
+		lock_extent_bits(tree, page_start, page_end, 0, &cached_state);
 	}
 	clear_extent_bit(tree, page_start, page_end,
 		 EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
@@ -6462,8 +6621,7 @@ again:
 	}
 	wait_on_page_writeback(page);
 
-	lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state,
-			 GFP_NOFS);
+	lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state);
 	set_page_extent_mapped(page);
 
 	/*
@@ -6737,10 +6895,9 @@ int btrfs_create_subvol_root(struct btrfs_trans_handle *trans,
 	btrfs_i_size_write(inode, 0);
 
 	err = btrfs_update_inode(trans, new_root, inode);
-	BUG_ON(err);
 
 	iput(inode);
-	return 0;
+	return err;
 }
 
 struct inode *btrfs_alloc_inode(struct super_block *sb)
@@ -6783,6 +6940,8 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
 	extent_map_tree_init(&ei->extent_tree);
 	extent_io_tree_init(&ei->io_tree, &inode->i_data);
 	extent_io_tree_init(&ei->io_failure_tree, &inode->i_data);
+	ei->io_tree.track_uptodate = 1;
+	ei->io_failure_tree.track_uptodate = 1;
 	mutex_init(&ei->log_mutex);
 	mutex_init(&ei->delalloc_mutex);
 	btrfs_ordered_inode_tree_init(&ei->ordered_tree);
@@ -7072,7 +7231,10 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 		if (!ret)
 			ret = btrfs_update_inode(trans, root, old_inode);
 	}
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out_fail;
+	}
 
 	if (new_inode) {
 		new_inode->i_ctime = CURRENT_TIME;
@@ -7090,11 +7252,14 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 						 new_dentry->d_name.name,
 						 new_dentry->d_name.len);
 		}
-		BUG_ON(ret);
-		if (new_inode->i_nlink == 0) {
+		if (!ret && new_inode->i_nlink == 0) {
 			ret = btrfs_orphan_add(trans, new_dentry->d_inode);
 			BUG_ON(ret);
 		}
+		if (ret) {
+			btrfs_abort_transaction(trans, root, ret);
+			goto out_fail;
+		}
 	}
 
 	fixup_inode_flags(new_dir, old_inode);
@@ -7102,7 +7267,10 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	ret = btrfs_add_link(trans, new_dir, old_inode,
 			     new_dentry->d_name.name,
 			     new_dentry->d_name.len, 0, index);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto out_fail;
+	}
 
 	if (old_ino != BTRFS_FIRST_FREE_OBJECTID) {
 		struct dentry *parent = new_dentry->d_parent;
@@ -7315,7 +7483,7 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 		}
 
 		ret = btrfs_reserve_extent(trans, root, num_bytes, min_size,
-					   0, *alloc_hint, (u64)-1, &ins, 1);
+					   0, *alloc_hint, &ins, 1);
 		if (ret) {
 			if (own_trans)
 				btrfs_end_transaction(trans, root);
@@ -7327,7 +7495,12 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 						  ins.offset, ins.offset,
 						  ins.offset, 0, 0, 0,
 						  BTRFS_FILE_EXTENT_PREALLOC);
-		BUG_ON(ret);
+		if (ret) {
+			btrfs_abort_transaction(trans, root, ret);
+			if (own_trans)
+				btrfs_end_transaction(trans, root);
+			break;
+		}
 		btrfs_drop_extent_cache(inode, cur_offset,
 					cur_offset + ins.offset -1, 0);
 
@@ -7349,7 +7522,13 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 		}
 
 		ret = btrfs_update_inode(trans, root, inode);
-		BUG_ON(ret);
+
+		if (ret) {
+			btrfs_abort_transaction(trans, root, ret);
+			if (own_trans)
+				btrfs_end_transaction(trans, root);
+			break;
+		}
 
 		if (own_trans)
 			btrfs_end_transaction(trans, root);
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index d8b54715c2d..18cc23d164a 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -425,22 +425,37 @@ static noinline int create_subvol(struct btrfs_root *root,
 
 	key.offset = (u64)-1;
 	new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
-	BUG_ON(IS_ERR(new_root));
+	if (IS_ERR(new_root)) {
+		btrfs_abort_transaction(trans, root, PTR_ERR(new_root));
+		ret = PTR_ERR(new_root);
+		goto fail;
+	}
 
 	btrfs_record_root_in_trans(trans, new_root);
 
 	ret = btrfs_create_subvol_root(trans, new_root, new_dirid);
+	if (ret) {
+		/* We potentially lose an unused inode item here */
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
+	}
+
 	/*
 	 * insert the directory item
 	 */
 	ret = btrfs_set_inode_index(dir, &index);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
+	}
 
 	ret = btrfs_insert_dir_item(trans, root,
 				    name, namelen, dir, &key,
 				    BTRFS_FT_DIR, index);
-	if (ret)
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
 		goto fail;
+	}
 
 	btrfs_i_size_write(dir, dir->i_size + namelen * 2);
 	ret = btrfs_update_inode(trans, root, dir);
@@ -769,6 +784,31 @@ none:
 	return -ENOENT;
 }
 
+/*
+ * Validaty check of prev em and next em:
+ * 1) no prev/next em
+ * 2) prev/next em is an hole/inline extent
+ */
+static int check_adjacent_extents(struct inode *inode, struct extent_map *em)
+{
+	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+	struct extent_map *prev = NULL, *next = NULL;
+	int ret = 0;
+
+	read_lock(&em_tree->lock);
+	prev = lookup_extent_mapping(em_tree, em->start - 1, (u64)-1);
+	next = lookup_extent_mapping(em_tree, em->start + em->len, (u64)-1);
+	read_unlock(&em_tree->lock);
+
+	if ((!prev || prev->block_start >= EXTENT_MAP_LAST_BYTE) &&
+	    (!next || next->block_start >= EXTENT_MAP_LAST_BYTE))
+		ret = 1;
+	free_extent_map(prev);
+	free_extent_map(next);
+
+	return ret;
+}
+
 static int should_defrag_range(struct inode *inode, u64 start, u64 len,
 			       int thresh, u64 *last_len, u64 *skip,
 			       u64 *defrag_end)
@@ -797,17 +837,25 @@ static int should_defrag_range(struct inode *inode, u64 start, u64 len,
 
 	if (!em) {
 		/* get the big lock and read metadata off disk */
-		lock_extent(io_tree, start, start + len - 1, GFP_NOFS);
+		lock_extent(io_tree, start, start + len - 1);
 		em = btrfs_get_extent(inode, NULL, 0, start, len, 0);
-		unlock_extent(io_tree, start, start + len - 1, GFP_NOFS);
+		unlock_extent(io_tree, start, start + len - 1);
 
 		if (IS_ERR(em))
 			return 0;
 	}
 
 	/* this will cover holes, and inline extents */
-	if (em->block_start >= EXTENT_MAP_LAST_BYTE)
+	if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
 		ret = 0;
+		goto out;
+	}
+
+	/* If we have nothing to merge with us, just skip. */
+	if (check_adjacent_extents(inode, em)) {
+		ret = 0;
+		goto out;
+	}
 
 	/*
 	 * we hit a real extent, if it is big don't bother defragging it again
@@ -815,6 +863,7 @@ static int should_defrag_range(struct inode *inode, u64 start, u64 len,
 	if ((*last_len == 0 || *last_len >= thresh) && em->len >= thresh)
 		ret = 0;
 
+out:
 	/*
 	 * last_len ends up being a counter of how many bytes we've defragged.
 	 * every time we choose not to defrag an extent, we reset *last_len
@@ -856,6 +905,7 @@ static int cluster_pages_for_defrag(struct inode *inode,
 	u64 isize = i_size_read(inode);
 	u64 page_start;
 	u64 page_end;
+	u64 page_cnt;
 	int ret;
 	int i;
 	int i_done;
@@ -864,19 +914,21 @@ static int cluster_pages_for_defrag(struct inode *inode,
 	struct extent_io_tree *tree;
 	gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
 
-	if (isize == 0)
-		return 0;
 	file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
+	if (!isize || start_index > file_end)
+		return 0;
+
+	page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1);
 
 	ret = btrfs_delalloc_reserve_space(inode,
-					   num_pages << PAGE_CACHE_SHIFT);
+					   page_cnt << PAGE_CACHE_SHIFT);
 	if (ret)
 		return ret;
 	i_done = 0;
 	tree = &BTRFS_I(inode)->io_tree;
 
 	/* step one, lock all the pages */
-	for (i = 0; i < num_pages; i++) {
+	for (i = 0; i < page_cnt; i++) {
 		struct page *page;
 again:
 		page = find_or_create_page(inode->i_mapping,
@@ -887,10 +939,10 @@ again:
 		page_start = page_offset(page);
 		page_end = page_start + PAGE_CACHE_SIZE - 1;
 		while (1) {
-			lock_extent(tree, page_start, page_end, GFP_NOFS);
+			lock_extent(tree, page_start, page_end);
 			ordered = btrfs_lookup_ordered_extent(inode,
 							      page_start);
-			unlock_extent(tree, page_start, page_end, GFP_NOFS);
+			unlock_extent(tree, page_start, page_end);
 			if (!ordered)
 				break;
 
@@ -898,6 +950,15 @@ again:
 			btrfs_start_ordered_extent(inode, ordered, 1);
 			btrfs_put_ordered_extent(ordered);
 			lock_page(page);
+			/*
+			 * we unlocked the page above, so we need check if
+			 * it was released or not.
+			 */
+			if (page->mapping != inode->i_mapping) {
+				unlock_page(page);
+				page_cache_release(page);
+				goto again;
+			}
 		}
 
 		if (!PageUptodate(page)) {
@@ -911,15 +972,6 @@ again:
 			}
 		}
 
-		isize = i_size_read(inode);
-		file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
-		if (!isize || page->index > file_end) {
-			/* whoops, we blew past eof, skip this page */
-			unlock_page(page);
-			page_cache_release(page);
-			break;
-		}
-
 		if (page->mapping != inode->i_mapping) {
 			unlock_page(page);
 			page_cache_release(page);
@@ -946,19 +998,18 @@ again:
 	page_end = page_offset(pages[i_done - 1]) + PAGE_CACHE_SIZE;
 
 	lock_extent_bits(&BTRFS_I(inode)->io_tree,
-			 page_start, page_end - 1, 0, &cached_state,
-			 GFP_NOFS);
+			 page_start, page_end - 1, 0, &cached_state);
 	clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
 			  page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
 			  EXTENT_DO_ACCOUNTING, 0, 0, &cached_state,
 			  GFP_NOFS);
 
-	if (i_done != num_pages) {
+	if (i_done != page_cnt) {
 		spin_lock(&BTRFS_I(inode)->lock);
 		BTRFS_I(inode)->outstanding_extents++;
 		spin_unlock(&BTRFS_I(inode)->lock);
 		btrfs_delalloc_release_space(inode,
-				     (num_pages - i_done) << PAGE_CACHE_SHIFT);
+				     (page_cnt - i_done) << PAGE_CACHE_SHIFT);
 	}
 
 
@@ -983,7 +1034,7 @@ out:
 		unlock_page(pages[i]);
 		page_cache_release(pages[i]);
 	}
-	btrfs_delalloc_release_space(inode, num_pages << PAGE_CACHE_SHIFT);
+	btrfs_delalloc_release_space(inode, page_cnt << PAGE_CACHE_SHIFT);
 	return ret;
 
 }
@@ -1089,12 +1140,9 @@ int btrfs_defrag_file(struct inode *inode, struct file *file,
 		if (!(inode->i_sb->s_flags & MS_ACTIVE))
 			break;
 
-		if (!newer_than &&
-		    !should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT,
-					PAGE_CACHE_SIZE,
-					extent_thresh,
-					&last_len, &skip,
-					&defrag_end)) {
+		if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT,
+					 PAGE_CACHE_SIZE, extent_thresh,
+					 &last_len, &skip, &defrag_end)) {
 			unsigned long next;
 			/*
 			 * the should_defrag function tells us how much to skip
@@ -1123,17 +1171,24 @@ int btrfs_defrag_file(struct inode *inode, struct file *file,
 			ra_index += max_cluster;
 		}
 
+		mutex_lock(&inode->i_mutex);
 		ret = cluster_pages_for_defrag(inode, pages, i, cluster);
-		if (ret < 0)
+		if (ret < 0) {
+			mutex_unlock(&inode->i_mutex);
 			goto out_ra;
+		}
 
 		defrag_count += ret;
 		balance_dirty_pages_ratelimited_nr(inode->i_mapping, ret);
+		mutex_unlock(&inode->i_mutex);
 
 		if (newer_than) {
 			if (newer_off == (u64)-1)
 				break;
 
+			if (ret > 0)
+				i += ret;
+
 			newer_off = max(newer_off + 1,
 					(u64)i << PAGE_CACHE_SHIFT);
 
@@ -1966,7 +2021,11 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file,
 				dest->root_key.objectid,
 				dentry->d_name.name,
 				dentry->d_name.len);
-	BUG_ON(ret);
+	if (ret) {
+		err = ret;
+		btrfs_abort_transaction(trans, root, ret);
+		goto out_end_trans;
+	}
 
 	btrfs_record_root_in_trans(trans, dest);
 
@@ -1979,11 +2038,16 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file,
 		ret = btrfs_insert_orphan_item(trans,
 					root->fs_info->tree_root,
 					dest->root_key.objectid);
-		BUG_ON(ret);
+		if (ret) {
+			btrfs_abort_transaction(trans, root, ret);
+			err = ret;
+			goto out_end_trans;
+		}
 	}
-
+out_end_trans:
 	ret = btrfs_end_transaction(trans, root);
-	BUG_ON(ret);
+	if (ret && !err)
+		err = ret;
 	inode->i_flags |= S_DEAD;
 out_up_write:
 	up_write(&root->fs_info->subvol_sem);
@@ -2326,13 +2390,13 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 	   another, and lock file content */
 	while (1) {
 		struct btrfs_ordered_extent *ordered;
-		lock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
+		lock_extent(&BTRFS_I(src)->io_tree, off, off+len);
 		ordered = btrfs_lookup_first_ordered_extent(src, off+len);
 		if (!ordered &&
 		    !test_range_bit(&BTRFS_I(src)->io_tree, off, off+len,
 				   EXTENT_DELALLOC, 0, NULL))
 			break;
-		unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
+		unlock_extent(&BTRFS_I(src)->io_tree, off, off+len);
 		if (ordered)
 			btrfs_put_ordered_extent(ordered);
 		btrfs_wait_ordered_range(src, off, len);
@@ -2447,11 +2511,21 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 							 new_key.offset,
 							 new_key.offset + datal,
 							 &hint_byte, 1);
-				BUG_ON(ret);
+				if (ret) {
+					btrfs_abort_transaction(trans, root,
+								ret);
+					btrfs_end_transaction(trans, root);
+					goto out;
+				}
 
 				ret = btrfs_insert_empty_item(trans, root, path,
 							      &new_key, size);
-				BUG_ON(ret);
+				if (ret) {
+					btrfs_abort_transaction(trans, root,
+								ret);
+					btrfs_end_transaction(trans, root);
+					goto out;
+				}
 
 				leaf = path->nodes[0];
 				slot = path->slots[0];
@@ -2478,7 +2552,15 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 							btrfs_ino(inode),
 							new_key.offset - datao,
 							0);
-					BUG_ON(ret);
+					if (ret) {
+						btrfs_abort_transaction(trans,
+									root,
+									ret);
+						btrfs_end_transaction(trans,
+								      root);
+						goto out;
+
+					}
 				}
 			} else if (type == BTRFS_FILE_EXTENT_INLINE) {
 				u64 skip = 0;
@@ -2503,11 +2585,21 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 							 new_key.offset,
 							 new_key.offset + datal,
 							 &hint_byte, 1);
-				BUG_ON(ret);
+				if (ret) {
+					btrfs_abort_transaction(trans, root,
+								ret);
+					btrfs_end_transaction(trans, root);
+					goto out;
+				}
 
 				ret = btrfs_insert_empty_item(trans, root, path,
 							      &new_key, size);
-				BUG_ON(ret);
+				if (ret) {
+					btrfs_abort_transaction(trans, root,
+								ret);
+					btrfs_end_transaction(trans, root);
+					goto out;
+				}
 
 				if (skip) {
 					u32 start =
@@ -2541,8 +2633,12 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 				btrfs_i_size_write(inode, endoff);
 
 			ret = btrfs_update_inode(trans, root, inode);
-			BUG_ON(ret);
-			btrfs_end_transaction(trans, root);
+			if (ret) {
+				btrfs_abort_transaction(trans, root, ret);
+				btrfs_end_transaction(trans, root);
+				goto out;
+			}
+			ret = btrfs_end_transaction(trans, root);
 		}
 next:
 		btrfs_release_path(path);
@@ -2551,7 +2647,7 @@ next:
 	ret = 0;
 out:
 	btrfs_release_path(path);
-	unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS);
+	unlock_extent(&BTRFS_I(src)->io_tree, off, off+len);
 out_unlock:
 	mutex_unlock(&src->i_mutex);
 	mutex_unlock(&inode->i_mutex);
@@ -3066,8 +3162,8 @@ static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root,
 		goto out;
 
 	extent_item_pos = loi->logical - key.objectid;
-	ret = iterate_extent_inodes(root->fs_info, path, key.objectid,
-					extent_item_pos, build_ino_list,
+	ret = iterate_extent_inodes(root->fs_info, key.objectid,
+					extent_item_pos, 0, build_ino_list,
 					inodes);
 
 	if (ret < 0)
diff --git a/fs/btrfs/locking.c b/fs/btrfs/locking.c
index 5e178d8f716..272f911203f 100644
--- a/fs/btrfs/locking.c
+++ b/fs/btrfs/locking.c
@@ -208,7 +208,7 @@ void btrfs_tree_read_unlock_blocking(struct extent_buffer *eb)
  * take a spinning write lock.  This will wait for both
  * blocking readers or writers
  */
-int btrfs_tree_lock(struct extent_buffer *eb)
+void btrfs_tree_lock(struct extent_buffer *eb)
 {
 again:
 	wait_event(eb->read_lock_wq, atomic_read(&eb->blocking_readers) == 0);
@@ -230,13 +230,12 @@ again:
 	atomic_inc(&eb->spinning_writers);
 	atomic_inc(&eb->write_locks);
 	eb->lock_owner = current->pid;
-	return 0;
 }
 
 /*
  * drop a spinning or a blocking write lock.
  */
-int btrfs_tree_unlock(struct extent_buffer *eb)
+void btrfs_tree_unlock(struct extent_buffer *eb)
 {
 	int blockers = atomic_read(&eb->blocking_writers);
 
@@ -255,7 +254,6 @@ int btrfs_tree_unlock(struct extent_buffer *eb)
 		atomic_dec(&eb->spinning_writers);
 		write_unlock(&eb->lock);
 	}
-	return 0;
 }
 
 void btrfs_assert_tree_locked(struct extent_buffer *eb)
diff --git a/fs/btrfs/locking.h b/fs/btrfs/locking.h
index 17247ddb81a..ca52681e5f4 100644
--- a/fs/btrfs/locking.h
+++ b/fs/btrfs/locking.h
@@ -24,8 +24,8 @@
 #define BTRFS_WRITE_LOCK_BLOCKING 3
 #define BTRFS_READ_LOCK_BLOCKING 4
 
-int btrfs_tree_lock(struct extent_buffer *eb);
-int btrfs_tree_unlock(struct extent_buffer *eb);
+void btrfs_tree_lock(struct extent_buffer *eb);
+void btrfs_tree_unlock(struct extent_buffer *eb);
 int btrfs_try_spin_lock(struct extent_buffer *eb);
 
 void btrfs_tree_read_lock(struct extent_buffer *eb);
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index a1c94042530..bbf6d0d9aeb 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -59,6 +59,14 @@ static struct rb_node *tree_insert(struct rb_root *root, u64 file_offset,
 	return NULL;
 }
 
+static void ordered_data_tree_panic(struct inode *inode, int errno,
+					       u64 offset)
+{
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	btrfs_panic(fs_info, errno, "Inconsistency in ordered tree at offset "
+		    "%llu\n", (unsigned long long)offset);
+}
+
 /*
  * look for a given offset in the tree, and if it can't be found return the
  * first lesser offset
@@ -207,7 +215,8 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
 	spin_lock(&tree->lock);
 	node = tree_insert(&tree->tree, file_offset,
 			   &entry->rb_node);
-	BUG_ON(node);
+	if (node)
+		ordered_data_tree_panic(inode, -EEXIST, file_offset);
 	spin_unlock(&tree->lock);
 
 	spin_lock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
@@ -215,7 +224,6 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
 		      &BTRFS_I(inode)->root->fs_info->ordered_extents);
 	spin_unlock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
 
-	BUG_ON(node);
 	return 0;
 }
 
@@ -249,9 +257,9 @@ int btrfs_add_ordered_extent_compress(struct inode *inode, u64 file_offset,
  * when an ordered extent is finished.  If the list covers more than one
  * ordered extent, it is split across multiples.
  */
-int btrfs_add_ordered_sum(struct inode *inode,
-			  struct btrfs_ordered_extent *entry,
-			  struct btrfs_ordered_sum *sum)
+void btrfs_add_ordered_sum(struct inode *inode,
+			   struct btrfs_ordered_extent *entry,
+			   struct btrfs_ordered_sum *sum)
 {
 	struct btrfs_ordered_inode_tree *tree;
 
@@ -259,7 +267,6 @@ int btrfs_add_ordered_sum(struct inode *inode,
 	spin_lock(&tree->lock);
 	list_add_tail(&sum->list, &entry->list);
 	spin_unlock(&tree->lock);
-	return 0;
 }
 
 /*
@@ -384,7 +391,7 @@ out:
  * used to drop a reference on an ordered extent.  This will free
  * the extent if the last reference is dropped
  */
-int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
+void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
 {
 	struct list_head *cur;
 	struct btrfs_ordered_sum *sum;
@@ -400,7 +407,6 @@ int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
 		}
 		kfree(entry);
 	}
-	return 0;
 }
 
 /*
@@ -408,8 +414,8 @@ int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
  * and you must wake_up entry->wait.  You must hold the tree lock
  * while you call this function.
  */
-static int __btrfs_remove_ordered_extent(struct inode *inode,
-				struct btrfs_ordered_extent *entry)
+static void __btrfs_remove_ordered_extent(struct inode *inode,
+					  struct btrfs_ordered_extent *entry)
 {
 	struct btrfs_ordered_inode_tree *tree;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
@@ -436,35 +442,30 @@ static int __btrfs_remove_ordered_extent(struct inode *inode,
 		list_del_init(&BTRFS_I(inode)->ordered_operations);
 	}
 	spin_unlock(&root->fs_info->ordered_extent_lock);
-
-	return 0;
 }
 
 /*
  * remove an ordered extent from the tree.  No references are dropped
  * but any waiters are woken.
  */
-int btrfs_remove_ordered_extent(struct inode *inode,
-				struct btrfs_ordered_extent *entry)
+void btrfs_remove_ordered_extent(struct inode *inode,
+				 struct btrfs_ordered_extent *entry)
 {
 	struct btrfs_ordered_inode_tree *tree;
-	int ret;
 
 	tree = &BTRFS_I(inode)->ordered_tree;
 	spin_lock(&tree->lock);
-	ret = __btrfs_remove_ordered_extent(inode, entry);
+	__btrfs_remove_ordered_extent(inode, entry);
 	spin_unlock(&tree->lock);
 	wake_up(&entry->wait);
-
-	return ret;
 }
 
 /*
  * wait for all the ordered extents in a root.  This is done when balancing
  * space between drives.
  */
-int btrfs_wait_ordered_extents(struct btrfs_root *root,
-			       int nocow_only, int delay_iput)
+void btrfs_wait_ordered_extents(struct btrfs_root *root,
+				int nocow_only, int delay_iput)
 {
 	struct list_head splice;
 	struct list_head *cur;
@@ -512,7 +513,6 @@ int btrfs_wait_ordered_extents(struct btrfs_root *root,
 		spin_lock(&root->fs_info->ordered_extent_lock);
 	}
 	spin_unlock(&root->fs_info->ordered_extent_lock);
-	return 0;
 }
 
 /*
@@ -525,7 +525,7 @@ int btrfs_wait_ordered_extents(struct btrfs_root *root,
  * extra check to make sure the ordered operation list really is empty
  * before we return
  */
-int btrfs_run_ordered_operations(struct btrfs_root *root, int wait)
+void btrfs_run_ordered_operations(struct btrfs_root *root, int wait)
 {
 	struct btrfs_inode *btrfs_inode;
 	struct inode *inode;
@@ -573,8 +573,6 @@ again:
 
 	spin_unlock(&root->fs_info->ordered_extent_lock);
 	mutex_unlock(&root->fs_info->ordered_operations_mutex);
-
-	return 0;
 }
 
 /*
@@ -609,7 +607,7 @@ void btrfs_start_ordered_extent(struct inode *inode,
 /*
  * Used to wait on ordered extents across a large range of bytes.
  */
-int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len)
+void btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len)
 {
 	u64 end;
 	u64 orig_end;
@@ -664,7 +662,6 @@ again:
 		schedule_timeout(1);
 		goto again;
 	}
-	return 0;
 }
 
 /*
@@ -948,9 +945,8 @@ out:
  * If trans is not null, we'll do a friendly check for a transaction that
  * is already flushing things and force the IO down ourselves.
  */
-int btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root,
-				struct inode *inode)
+void btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
+				 struct btrfs_root *root, struct inode *inode)
 {
 	u64 last_mod;
 
@@ -961,7 +957,7 @@ int btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
 	 * commit, we can safely return without doing anything
 	 */
 	if (last_mod < root->fs_info->last_trans_committed)
-		return 0;
+		return;
 
 	/*
 	 * the transaction is already committing.  Just start the IO and
@@ -969,7 +965,7 @@ int btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
 	 */
 	if (trans && root->fs_info->running_transaction->blocked) {
 		btrfs_wait_ordered_range(inode, 0, (u64)-1);
-		return 0;
+		return;
 	}
 
 	spin_lock(&root->fs_info->ordered_extent_lock);
@@ -978,6 +974,4 @@ int btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
 			      &root->fs_info->ordered_operations);
 	}
 	spin_unlock(&root->fs_info->ordered_extent_lock);
-
-	return 0;
 }
diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h
index ff1f69aa188..c355ad4dc1a 100644
--- a/fs/btrfs/ordered-data.h
+++ b/fs/btrfs/ordered-data.h
@@ -138,8 +138,8 @@ btrfs_ordered_inode_tree_init(struct btrfs_ordered_inode_tree *t)
 	t->last = NULL;
 }
 
-int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry);
-int btrfs_remove_ordered_extent(struct inode *inode,
+void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry);
+void btrfs_remove_ordered_extent(struct inode *inode,
 				struct btrfs_ordered_extent *entry);
 int btrfs_dec_test_ordered_pending(struct inode *inode,
 				   struct btrfs_ordered_extent **cached,
@@ -154,14 +154,14 @@ int btrfs_add_ordered_extent_dio(struct inode *inode, u64 file_offset,
 int btrfs_add_ordered_extent_compress(struct inode *inode, u64 file_offset,
 				      u64 start, u64 len, u64 disk_len,
 				      int type, int compress_type);
-int btrfs_add_ordered_sum(struct inode *inode,
-			  struct btrfs_ordered_extent *entry,
-			  struct btrfs_ordered_sum *sum);
+void btrfs_add_ordered_sum(struct inode *inode,
+			   struct btrfs_ordered_extent *entry,
+			   struct btrfs_ordered_sum *sum);
 struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode,
 							 u64 file_offset);
 void btrfs_start_ordered_extent(struct inode *inode,
 				struct btrfs_ordered_extent *entry, int wait);
-int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len);
+void btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len);
 struct btrfs_ordered_extent *
 btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset);
 struct btrfs_ordered_extent *btrfs_lookup_ordered_range(struct inode *inode,
@@ -170,10 +170,10 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_range(struct inode *inode,
 int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
 				struct btrfs_ordered_extent *ordered);
 int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, u32 *sum);
-int btrfs_run_ordered_operations(struct btrfs_root *root, int wait);
-int btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root,
-				struct inode *inode);
-int btrfs_wait_ordered_extents(struct btrfs_root *root,
-			       int nocow_only, int delay_iput);
+void btrfs_run_ordered_operations(struct btrfs_root *root, int wait);
+void btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
+				 struct btrfs_root *root,
+				 struct inode *inode);
+void btrfs_wait_ordered_extents(struct btrfs_root *root,
+				int nocow_only, int delay_iput);
 #endif
diff --git a/fs/btrfs/orphan.c b/fs/btrfs/orphan.c
index f8be250963a..24cad1695af 100644
--- a/fs/btrfs/orphan.c
+++ b/fs/btrfs/orphan.c
@@ -58,7 +58,7 @@ int btrfs_del_orphan_item(struct btrfs_trans_handle *trans,
 	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
 	if (ret < 0)
 		goto out;
-	if (ret) {
+	if (ret) { /* JDM: Really? */
 		ret = -ENOENT;
 		goto out;
 	}
diff --git a/fs/btrfs/reada.c b/fs/btrfs/reada.c
index 22db04550f6..dc5d33146fd 100644
--- a/fs/btrfs/reada.c
+++ b/fs/btrfs/reada.c
@@ -54,7 +54,6 @@
  * than the 2 started one after another.
  */
 
-#define MAX_MIRRORS 2
 #define MAX_IN_FLIGHT 6
 
 struct reada_extctl {
@@ -71,7 +70,7 @@ struct reada_extent {
 	struct list_head	extctl;
 	struct kref		refcnt;
 	spinlock_t		lock;
-	struct reada_zone	*zones[MAX_MIRRORS];
+	struct reada_zone	*zones[BTRFS_MAX_MIRRORS];
 	int			nzones;
 	struct btrfs_device	*scheduled_for;
 };
@@ -84,7 +83,8 @@ struct reada_zone {
 	spinlock_t		lock;
 	int			locked;
 	struct btrfs_device	*device;
-	struct btrfs_device	*devs[MAX_MIRRORS]; /* full list, incl self */
+	struct btrfs_device	*devs[BTRFS_MAX_MIRRORS]; /* full list, incl
+							   * self */
 	int			ndevs;
 	struct kref		refcnt;
 };
@@ -365,9 +365,9 @@ again:
 	if (ret || !bbio || length < blocksize)
 		goto error;
 
-	if (bbio->num_stripes > MAX_MIRRORS) {
+	if (bbio->num_stripes > BTRFS_MAX_MIRRORS) {
 		printk(KERN_ERR "btrfs readahead: more than %d copies not "
-				"supported", MAX_MIRRORS);
+				"supported", BTRFS_MAX_MIRRORS);
 		goto error;
 	}
 
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 8c1aae2c845..017281dbb2a 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -326,6 +326,19 @@ static struct rb_node *tree_search(struct rb_root *root, u64 bytenr)
 	return NULL;
 }
 
+void backref_tree_panic(struct rb_node *rb_node, int errno,
+					  u64 bytenr)
+{
+
+	struct btrfs_fs_info *fs_info = NULL;
+	struct backref_node *bnode = rb_entry(rb_node, struct backref_node,
+					      rb_node);
+	if (bnode->root)
+		fs_info = bnode->root->fs_info;
+	btrfs_panic(fs_info, errno, "Inconsistency in backref cache "
+		    "found at offset %llu\n", (unsigned long long)bytenr);
+}
+
 /*
  * walk up backref nodes until reach node presents tree root
  */
@@ -452,7 +465,8 @@ static void update_backref_node(struct backref_cache *cache,
 	rb_erase(&node->rb_node, &cache->rb_root);
 	node->bytenr = bytenr;
 	rb_node = tree_insert(&cache->rb_root, node->bytenr, &node->rb_node);
-	BUG_ON(rb_node);
+	if (rb_node)
+		backref_tree_panic(rb_node, -EEXIST, bytenr);
 }
 
 /*
@@ -999,7 +1013,8 @@ next:
 	if (!cowonly) {
 		rb_node = tree_insert(&cache->rb_root, node->bytenr,
 				      &node->rb_node);
-		BUG_ON(rb_node);
+		if (rb_node)
+			backref_tree_panic(rb_node, -EEXIST, node->bytenr);
 		list_add_tail(&node->lower, &cache->leaves);
 	}
 
@@ -1034,7 +1049,9 @@ next:
 		if (!cowonly) {
 			rb_node = tree_insert(&cache->rb_root, upper->bytenr,
 					      &upper->rb_node);
-			BUG_ON(rb_node);
+			if (rb_node)
+				backref_tree_panic(rb_node, -EEXIST,
+						   upper->bytenr);
 		}
 
 		list_add_tail(&edge->list[UPPER], &upper->lower);
@@ -1180,7 +1197,8 @@ static int clone_backref_node(struct btrfs_trans_handle *trans,
 
 	rb_node = tree_insert(&cache->rb_root, new_node->bytenr,
 			      &new_node->rb_node);
-	BUG_ON(rb_node);
+	if (rb_node)
+		backref_tree_panic(rb_node, -EEXIST, new_node->bytenr);
 
 	if (!new_node->lowest) {
 		list_for_each_entry(new_edge, &new_node->lower, list[UPPER]) {
@@ -1203,14 +1221,15 @@ fail:
 /*
  * helper to add 'address of tree root -> reloc tree' mapping
  */
-static int __add_reloc_root(struct btrfs_root *root)
+static int __must_check __add_reloc_root(struct btrfs_root *root)
 {
 	struct rb_node *rb_node;
 	struct mapping_node *node;
 	struct reloc_control *rc = root->fs_info->reloc_ctl;
 
 	node = kmalloc(sizeof(*node), GFP_NOFS);
-	BUG_ON(!node);
+	if (!node)
+		return -ENOMEM;
 
 	node->bytenr = root->node->start;
 	node->data = root;
@@ -1219,7 +1238,12 @@ static int __add_reloc_root(struct btrfs_root *root)
 	rb_node = tree_insert(&rc->reloc_root_tree.rb_root,
 			      node->bytenr, &node->rb_node);
 	spin_unlock(&rc->reloc_root_tree.lock);
-	BUG_ON(rb_node);
+	if (rb_node) {
+		kfree(node);
+		btrfs_panic(root->fs_info, -EEXIST, "Duplicate root found "
+			    "for start=%llu while inserting into relocation "
+			    "tree\n");
+	}
 
 	list_add_tail(&root->root_list, &rc->reloc_roots);
 	return 0;
@@ -1252,7 +1276,8 @@ static int __update_reloc_root(struct btrfs_root *root, int del)
 		rb_node = tree_insert(&rc->reloc_root_tree.rb_root,
 				      node->bytenr, &node->rb_node);
 		spin_unlock(&rc->reloc_root_tree.lock);
-		BUG_ON(rb_node);
+		if (rb_node)
+			backref_tree_panic(rb_node, -EEXIST, node->bytenr);
 	} else {
 		list_del_init(&root->root_list);
 		kfree(node);
@@ -1334,6 +1359,7 @@ int btrfs_init_reloc_root(struct btrfs_trans_handle *trans,
 	struct btrfs_root *reloc_root;
 	struct reloc_control *rc = root->fs_info->reloc_ctl;
 	int clear_rsv = 0;
+	int ret;
 
 	if (root->reloc_root) {
 		reloc_root = root->reloc_root;
@@ -1353,7 +1379,8 @@ int btrfs_init_reloc_root(struct btrfs_trans_handle *trans,
 	if (clear_rsv)
 		trans->block_rsv = NULL;
 
-	__add_reloc_root(reloc_root);
+	ret = __add_reloc_root(reloc_root);
+	BUG_ON(ret < 0);
 	root->reloc_root = reloc_root;
 	return 0;
 }
@@ -1577,15 +1604,14 @@ int replace_file_extents(struct btrfs_trans_handle *trans,
 				WARN_ON(!IS_ALIGNED(end, root->sectorsize));
 				end--;
 				ret = try_lock_extent(&BTRFS_I(inode)->io_tree,
-						      key.offset, end,
-						      GFP_NOFS);
+						      key.offset, end);
 				if (!ret)
 					continue;
 
 				btrfs_drop_extent_cache(inode, key.offset, end,
 							1);
 				unlock_extent(&BTRFS_I(inode)->io_tree,
-					      key.offset, end, GFP_NOFS);
+					      key.offset, end);
 			}
 		}
 
@@ -1956,9 +1982,9 @@ static int invalidate_extent_cache(struct btrfs_root *root,
 		}
 
 		/* the lock_extent waits for readpage to complete */
-		lock_extent(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS);
+		lock_extent(&BTRFS_I(inode)->io_tree, start, end);
 		btrfs_drop_extent_cache(inode, start, end, 1);
-		unlock_extent(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS);
+		unlock_extent(&BTRFS_I(inode)->io_tree, start, end);
 	}
 	return 0;
 }
@@ -2246,7 +2272,8 @@ again:
 		} else {
 			list_del_init(&reloc_root->root_list);
 		}
-		btrfs_drop_snapshot(reloc_root, rc->block_rsv, 0, 1);
+		ret = btrfs_drop_snapshot(reloc_root, rc->block_rsv, 0, 1);
+		BUG_ON(ret < 0);
 	}
 
 	if (found) {
@@ -2862,12 +2889,12 @@ int prealloc_file_extent_cluster(struct inode *inode,
 		else
 			end = cluster->end - offset;
 
-		lock_extent(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS);
+		lock_extent(&BTRFS_I(inode)->io_tree, start, end);
 		num_bytes = end + 1 - start;
 		ret = btrfs_prealloc_file_range(inode, 0, start,
 						num_bytes, num_bytes,
 						end + 1, &alloc_hint);
-		unlock_extent(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS);
+		unlock_extent(&BTRFS_I(inode)->io_tree, start, end);
 		if (ret)
 			break;
 		nr++;
@@ -2899,7 +2926,7 @@ int setup_extent_mapping(struct inode *inode, u64 start, u64 end,
 	em->bdev = root->fs_info->fs_devices->latest_bdev;
 	set_bit(EXTENT_FLAG_PINNED, &em->flags);
 
-	lock_extent(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS);
+	lock_extent(&BTRFS_I(inode)->io_tree, start, end);
 	while (1) {
 		write_lock(&em_tree->lock);
 		ret = add_extent_mapping(em_tree, em);
@@ -2910,7 +2937,7 @@ int setup_extent_mapping(struct inode *inode, u64 start, u64 end,
 		}
 		btrfs_drop_extent_cache(inode, start, end, 0);
 	}
-	unlock_extent(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS);
+	unlock_extent(&BTRFS_I(inode)->io_tree, start, end);
 	return ret;
 }
 
@@ -2990,8 +3017,7 @@ static int relocate_file_extent_cluster(struct inode *inode,
 		page_start = (u64)page->index << PAGE_CACHE_SHIFT;
 		page_end = page_start + PAGE_CACHE_SIZE - 1;
 
-		lock_extent(&BTRFS_I(inode)->io_tree,
-			    page_start, page_end, GFP_NOFS);
+		lock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end);
 
 		set_page_extent_mapped(page);
 
@@ -3007,7 +3033,7 @@ static int relocate_file_extent_cluster(struct inode *inode,
 		set_page_dirty(page);
 
 		unlock_extent(&BTRFS_I(inode)->io_tree,
-			      page_start, page_end, GFP_NOFS);
+			      page_start, page_end);
 		unlock_page(page);
 		page_cache_release(page);
 
@@ -3154,7 +3180,8 @@ static int add_tree_block(struct reloc_control *rc,
 	block->key_ready = 0;
 
 	rb_node = tree_insert(blocks, block->bytenr, &block->rb_node);
-	BUG_ON(rb_node);
+	if (rb_node)
+		backref_tree_panic(rb_node, -EEXIST, block->bytenr);
 
 	return 0;
 }
@@ -3426,7 +3453,9 @@ static int find_data_references(struct reloc_control *rc,
 			block->key_ready = 1;
 			rb_node = tree_insert(blocks, block->bytenr,
 					      &block->rb_node);
-			BUG_ON(rb_node);
+			if (rb_node)
+				backref_tree_panic(rb_node, -EEXIST,
+						   block->bytenr);
 		}
 		if (counted)
 			added = 1;
@@ -4073,10 +4102,11 @@ out:
 static noinline_for_stack int mark_garbage_root(struct btrfs_root *root)
 {
 	struct btrfs_trans_handle *trans;
-	int ret;
+	int ret, err;
 
 	trans = btrfs_start_transaction(root->fs_info->tree_root, 0);
-	BUG_ON(IS_ERR(trans));
+	if (IS_ERR(trans))
+		return PTR_ERR(trans);
 
 	memset(&root->root_item.drop_progress, 0,
 		sizeof(root->root_item.drop_progress));
@@ -4084,11 +4114,11 @@ static noinline_for_stack int mark_garbage_root(struct btrfs_root *root)
 	btrfs_set_root_refs(&root->root_item, 0);
 	ret = btrfs_update_root(trans, root->fs_info->tree_root,
 				&root->root_key, &root->root_item);
-	BUG_ON(ret);
 
-	ret = btrfs_end_transaction(trans, root->fs_info->tree_root);
-	BUG_ON(ret);
-	return 0;
+	err = btrfs_end_transaction(trans, root->fs_info->tree_root);
+	if (err)
+		return err;
+	return ret;
 }
 
 /*
@@ -4156,7 +4186,11 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 					err = ret;
 					goto out;
 				}
-				mark_garbage_root(reloc_root);
+				ret = mark_garbage_root(reloc_root);
+				if (ret < 0) {
+					err = ret;
+					goto out;
+				}
 			}
 		}
 
@@ -4202,13 +4236,19 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 
 		fs_root = read_fs_root(root->fs_info,
 				       reloc_root->root_key.offset);
-		BUG_ON(IS_ERR(fs_root));
+		if (IS_ERR(fs_root)) {
+			err = PTR_ERR(fs_root);
+			goto out_free;
+		}
 
-		__add_reloc_root(reloc_root);
+		err = __add_reloc_root(reloc_root);
+		BUG_ON(err < 0); /* -ENOMEM or logic error */
 		fs_root->reloc_root = reloc_root;
 	}
 
-	btrfs_commit_transaction(trans, rc->extent_root);
+	err = btrfs_commit_transaction(trans, rc->extent_root);
+	if (err)
+		goto out_free;
 
 	merge_reloc_roots(rc);
 
@@ -4218,7 +4258,7 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 	if (IS_ERR(trans))
 		err = PTR_ERR(trans);
 	else
-		btrfs_commit_transaction(trans, rc->extent_root);
+		err = btrfs_commit_transaction(trans, rc->extent_root);
 out_free:
 	kfree(rc);
 out:
@@ -4267,6 +4307,8 @@ int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len)
 	disk_bytenr = file_pos + BTRFS_I(inode)->index_cnt;
 	ret = btrfs_lookup_csums_range(root->fs_info->csum_root, disk_bytenr,
 				       disk_bytenr + len - 1, &list, 0);
+	if (ret)
+		goto out;
 
 	while (!list_empty(&list)) {
 		sums = list_entry(list.next, struct btrfs_ordered_sum, list);
@@ -4284,6 +4326,7 @@ int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len)
 
 		btrfs_add_ordered_sum(inode, ordered, sums);
 	}
+out:
 	btrfs_put_ordered_extent(ordered);
 	return ret;
 }
@@ -4380,7 +4423,7 @@ void btrfs_reloc_pre_snapshot(struct btrfs_trans_handle *trans,
  * called after snapshot is created. migrate block reservation
  * and create reloc root for the newly created snapshot
  */
-void btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
+int btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
 			       struct btrfs_pending_snapshot *pending)
 {
 	struct btrfs_root *root = pending->root;
@@ -4390,7 +4433,7 @@ void btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
 	int ret;
 
 	if (!root->reloc_root)
-		return;
+		return 0;
 
 	rc = root->fs_info->reloc_ctl;
 	rc->merging_rsv_size += rc->nodes_relocated;
@@ -4399,18 +4442,21 @@ void btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
 		ret = btrfs_block_rsv_migrate(&pending->block_rsv,
 					      rc->block_rsv,
 					      rc->nodes_relocated);
-		BUG_ON(ret);
+		if (ret)
+			return ret;
 	}
 
 	new_root = pending->snap;
 	reloc_root = create_reloc_root(trans, root->reloc_root,
 				       new_root->root_key.objectid);
+	if (IS_ERR(reloc_root))
+		return PTR_ERR(reloc_root);
 
-	__add_reloc_root(reloc_root);
+	ret = __add_reloc_root(reloc_root);
+	BUG_ON(ret < 0);
 	new_root->reloc_root = reloc_root;
 
-	if (rc->create_reloc_tree) {
+	if (rc->create_reloc_tree)
 		ret = clone_backref_node(trans, rc, root, reloc_root);
-		BUG_ON(ret);
-	}
+	return ret;
 }
diff --git a/fs/btrfs/root-tree.c b/fs/btrfs/root-tree.c
index f4099904565..24fb8ce4e07 100644
--- a/fs/btrfs/root-tree.c
+++ b/fs/btrfs/root-tree.c
@@ -93,10 +93,14 @@ int btrfs_update_root(struct btrfs_trans_handle *trans, struct btrfs_root
 	unsigned long ptr;
 
 	path = btrfs_alloc_path();
-	BUG_ON(!path);
+	if (!path)
+		return -ENOMEM;
+
 	ret = btrfs_search_slot(trans, root, key, path, 0, 1);
-	if (ret < 0)
+	if (ret < 0) {
+		btrfs_abort_transaction(trans, root, ret);
 		goto out;
+	}
 
 	if (ret != 0) {
 		btrfs_print_leaf(root, path->nodes[0]);
@@ -116,13 +120,10 @@ out:
 	return ret;
 }
 
-int btrfs_insert_root(struct btrfs_trans_handle *trans, struct btrfs_root
-		      *root, struct btrfs_key *key, struct btrfs_root_item
-		      *item)
+int btrfs_insert_root(struct btrfs_trans_handle *trans, struct btrfs_root *root,
+		      struct btrfs_key *key, struct btrfs_root_item *item)
 {
-	int ret;
-	ret = btrfs_insert_item(trans, root, key, item, sizeof(*item));
-	return ret;
+	return btrfs_insert_item(trans, root, key, item, sizeof(*item));
 }
 
 /*
@@ -384,6 +385,8 @@ int btrfs_find_root_ref(struct btrfs_root *tree_root,
  *
  * For a back ref the root_id is the id of the subvol or snapshot and
  * ref_id is the id of the tree referencing it.
+ *
+ * Will return 0, -ENOMEM, or anything from the CoW path
  */
 int btrfs_add_root_ref(struct btrfs_trans_handle *trans,
 		       struct btrfs_root *tree_root,
@@ -407,7 +410,11 @@ int btrfs_add_root_ref(struct btrfs_trans_handle *trans,
 again:
 	ret = btrfs_insert_empty_item(trans, tree_root, path, &key,
 				      sizeof(*ref) + name_len);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, tree_root, ret);
+		btrfs_free_path(path);
+		return ret;
+	}
 
 	leaf = path->nodes[0];
 	ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_root_ref);
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 390e7102b0f..90acc82046c 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -36,37 +36,30 @@
  * Future enhancements:
  *  - In case an unrepairable extent is encountered, track which files are
  *    affected and report them
- *  - In case of a read error on files with nodatasum, map the file and read
- *    the extent to trigger a writeback of the good copy
  *  - track and record media errors, throw out bad devices
  *  - add a mode to also read unallocated space
  */
 
-struct scrub_bio;
-struct scrub_page;
+struct scrub_block;
 struct scrub_dev;
-static void scrub_bio_end_io(struct bio *bio, int err);
-static void scrub_checksum(struct btrfs_work *work);
-static int scrub_checksum_data(struct scrub_dev *sdev,
-			       struct scrub_page *spag, void *buffer);
-static int scrub_checksum_tree_block(struct scrub_dev *sdev,
-				     struct scrub_page *spag, u64 logical,
-				     void *buffer);
-static int scrub_checksum_super(struct scrub_bio *sbio, void *buffer);
-static int scrub_fixup_check(struct scrub_bio *sbio, int ix);
-static void scrub_fixup_end_io(struct bio *bio, int err);
-static int scrub_fixup_io(int rw, struct block_device *bdev, sector_t sector,
-			  struct page *page);
-static void scrub_fixup(struct scrub_bio *sbio, int ix);
 
 #define SCRUB_PAGES_PER_BIO	16	/* 64k per bio */
 #define SCRUB_BIOS_PER_DEV	16	/* 1 MB per device in flight */
+#define SCRUB_MAX_PAGES_PER_BLOCK	16	/* 64k per node/leaf/sector */
 
 struct scrub_page {
+	struct scrub_block	*sblock;
+	struct page		*page;
+	struct block_device	*bdev;
 	u64			flags;  /* extent flags */
 	u64			generation;
-	int			mirror_num;
-	int			have_csum;
+	u64			logical;
+	u64			physical;
+	struct {
+		unsigned int	mirror_num:8;
+		unsigned int	have_csum:1;
+		unsigned int	io_error:1;
+	};
 	u8			csum[BTRFS_CSUM_SIZE];
 };
 
@@ -77,12 +70,25 @@ struct scrub_bio {
 	int			err;
 	u64			logical;
 	u64			physical;
-	struct scrub_page	spag[SCRUB_PAGES_PER_BIO];
-	u64			count;
+	struct scrub_page	*pagev[SCRUB_PAGES_PER_BIO];
+	int			page_count;
 	int			next_free;
 	struct btrfs_work	work;
 };
 
+struct scrub_block {
+	struct scrub_page	pagev[SCRUB_MAX_PAGES_PER_BLOCK];
+	int			page_count;
+	atomic_t		outstanding_pages;
+	atomic_t		ref_count; /* free mem on transition to zero */
+	struct scrub_dev	*sdev;
+	struct {
+		unsigned int	header_error:1;
+		unsigned int	checksum_error:1;
+		unsigned int	no_io_error_seen:1;
+	};
+};
+
 struct scrub_dev {
 	struct scrub_bio	*bios[SCRUB_BIOS_PER_DEV];
 	struct btrfs_device	*dev;
@@ -96,6 +102,10 @@ struct scrub_dev {
 	struct list_head	csum_list;
 	atomic_t		cancel_req;
 	int			readonly;
+	int			pages_per_bio; /* <= SCRUB_PAGES_PER_BIO */
+	u32			sectorsize;
+	u32			nodesize;
+	u32			leafsize;
 	/*
 	 * statistics
 	 */
@@ -124,6 +134,43 @@ struct scrub_warning {
 	int			scratch_bufsize;
 };
 
+
+static int scrub_handle_errored_block(struct scrub_block *sblock_to_check);
+static int scrub_setup_recheck_block(struct scrub_dev *sdev,
+				     struct btrfs_mapping_tree *map_tree,
+				     u64 length, u64 logical,
+				     struct scrub_block *sblock);
+static int scrub_recheck_block(struct btrfs_fs_info *fs_info,
+			       struct scrub_block *sblock, int is_metadata,
+			       int have_csum, u8 *csum, u64 generation,
+			       u16 csum_size);
+static void scrub_recheck_block_checksum(struct btrfs_fs_info *fs_info,
+					 struct scrub_block *sblock,
+					 int is_metadata, int have_csum,
+					 const u8 *csum, u64 generation,
+					 u16 csum_size);
+static void scrub_complete_bio_end_io(struct bio *bio, int err);
+static int scrub_repair_block_from_good_copy(struct scrub_block *sblock_bad,
+					     struct scrub_block *sblock_good,
+					     int force_write);
+static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad,
+					    struct scrub_block *sblock_good,
+					    int page_num, int force_write);
+static int scrub_checksum_data(struct scrub_block *sblock);
+static int scrub_checksum_tree_block(struct scrub_block *sblock);
+static int scrub_checksum_super(struct scrub_block *sblock);
+static void scrub_block_get(struct scrub_block *sblock);
+static void scrub_block_put(struct scrub_block *sblock);
+static int scrub_add_page_to_bio(struct scrub_dev *sdev,
+				 struct scrub_page *spage);
+static int scrub_pages(struct scrub_dev *sdev, u64 logical, u64 len,
+		       u64 physical, u64 flags, u64 gen, int mirror_num,
+		       u8 *csum, int force);
+static void scrub_bio_end_io(struct bio *bio, int err);
+static void scrub_bio_end_io_worker(struct btrfs_work *work);
+static void scrub_block_complete(struct scrub_block *sblock);
+
+
 static void scrub_free_csums(struct scrub_dev *sdev)
 {
 	while (!list_empty(&sdev->csum_list)) {
@@ -135,23 +182,6 @@ static void scrub_free_csums(struct scrub_dev *sdev)
 	}
 }
 
-static void scrub_free_bio(struct bio *bio)
-{
-	int i;
-	struct page *last_page = NULL;
-
-	if (!bio)
-		return;
-
-	for (i = 0; i < bio->bi_vcnt; ++i) {
-		if (bio->bi_io_vec[i].bv_page == last_page)
-			continue;
-		last_page = bio->bi_io_vec[i].bv_page;
-		__free_page(last_page);
-	}
-	bio_put(bio);
-}
-
 static noinline_for_stack void scrub_free_dev(struct scrub_dev *sdev)
 {
 	int i;
@@ -159,13 +189,23 @@ static noinline_for_stack void scrub_free_dev(struct scrub_dev *sdev)
 	if (!sdev)
 		return;
 
+	/* this can happen when scrub is cancelled */
+	if (sdev->curr != -1) {
+		struct scrub_bio *sbio = sdev->bios[sdev->curr];
+
+		for (i = 0; i < sbio->page_count; i++) {
+			BUG_ON(!sbio->pagev[i]);
+			BUG_ON(!sbio->pagev[i]->page);
+			scrub_block_put(sbio->pagev[i]->sblock);
+		}
+		bio_put(sbio->bio);
+	}
+
 	for (i = 0; i < SCRUB_BIOS_PER_DEV; ++i) {
 		struct scrub_bio *sbio = sdev->bios[i];
 
 		if (!sbio)
 			break;
-
-		scrub_free_bio(sbio->bio);
 		kfree(sbio);
 	}
 
@@ -179,11 +219,16 @@ struct scrub_dev *scrub_setup_dev(struct btrfs_device *dev)
 	struct scrub_dev *sdev;
 	int		i;
 	struct btrfs_fs_info *fs_info = dev->dev_root->fs_info;
+	int pages_per_bio;
 
+	pages_per_bio = min_t(int, SCRUB_PAGES_PER_BIO,
+			      bio_get_nr_vecs(dev->bdev));
 	sdev = kzalloc(sizeof(*sdev), GFP_NOFS);
 	if (!sdev)
 		goto nomem;
 	sdev->dev = dev;
+	sdev->pages_per_bio = pages_per_bio;
+	sdev->curr = -1;
 	for (i = 0; i < SCRUB_BIOS_PER_DEV; ++i) {
 		struct scrub_bio *sbio;
 
@@ -194,8 +239,8 @@ struct scrub_dev *scrub_setup_dev(struct btrfs_device *dev)
 
 		sbio->index = i;
 		sbio->sdev = sdev;
-		sbio->count = 0;
-		sbio->work.func = scrub_checksum;
+		sbio->page_count = 0;
+		sbio->work.func = scrub_bio_end_io_worker;
 
 		if (i != SCRUB_BIOS_PER_DEV-1)
 			sdev->bios[i]->next_free = i + 1;
@@ -203,7 +248,9 @@ struct scrub_dev *scrub_setup_dev(struct btrfs_device *dev)
 			sdev->bios[i]->next_free = -1;
 	}
 	sdev->first_free = 0;
-	sdev->curr = -1;
+	sdev->nodesize = dev->dev_root->nodesize;
+	sdev->leafsize = dev->dev_root->leafsize;
+	sdev->sectorsize = dev->dev_root->sectorsize;
 	atomic_set(&sdev->in_flight, 0);
 	atomic_set(&sdev->fixup_cnt, 0);
 	atomic_set(&sdev->cancel_req, 0);
@@ -294,10 +341,9 @@ err:
 	return 0;
 }
 
-static void scrub_print_warning(const char *errstr, struct scrub_bio *sbio,
-				int ix)
+static void scrub_print_warning(const char *errstr, struct scrub_block *sblock)
 {
-	struct btrfs_device *dev = sbio->sdev->dev;
+	struct btrfs_device *dev = sblock->sdev->dev;
 	struct btrfs_fs_info *fs_info = dev->dev_root->fs_info;
 	struct btrfs_path *path;
 	struct btrfs_key found_key;
@@ -316,8 +362,9 @@ static void scrub_print_warning(const char *errstr, struct scrub_bio *sbio,
 
 	swarn.scratch_buf = kmalloc(bufsize, GFP_NOFS);
 	swarn.msg_buf = kmalloc(bufsize, GFP_NOFS);
-	swarn.sector = (sbio->physical + ix * PAGE_SIZE) >> 9;
-	swarn.logical = sbio->logical + ix * PAGE_SIZE;
+	BUG_ON(sblock->page_count < 1);
+	swarn.sector = (sblock->pagev[0].physical) >> 9;
+	swarn.logical = sblock->pagev[0].logical;
 	swarn.errstr = errstr;
 	swarn.dev = dev;
 	swarn.msg_bufsize = bufsize;
@@ -342,7 +389,8 @@ static void scrub_print_warning(const char *errstr, struct scrub_bio *sbio,
 		do {
 			ret = tree_backref_for_extent(&ptr, eb, ei, item_size,
 							&ref_root, &ref_level);
-			printk(KERN_WARNING "%s at logical %llu on dev %s, "
+			printk(KERN_WARNING
+				"btrfs: %s at logical %llu on dev %s, "
 				"sector %llu: metadata %s (level %d) in tree "
 				"%llu\n", errstr, swarn.logical, dev->name,
 				(unsigned long long)swarn.sector,
@@ -352,8 +400,8 @@ static void scrub_print_warning(const char *errstr, struct scrub_bio *sbio,
 		} while (ret != 1);
 	} else {
 		swarn.path = path;
-		iterate_extent_inodes(fs_info, path, found_key.objectid,
-					extent_item_pos,
+		iterate_extent_inodes(fs_info, found_key.objectid,
+					extent_item_pos, 1,
 					scrub_print_warning_inode, &swarn);
 	}
 
@@ -531,9 +579,9 @@ out:
 		spin_lock(&sdev->stat_lock);
 		++sdev->stat.uncorrectable_errors;
 		spin_unlock(&sdev->stat_lock);
-		printk_ratelimited(KERN_ERR "btrfs: unable to fixup "
-					"(nodatasum) error at logical %llu\n",
-					fixup->logical);
+		printk_ratelimited(KERN_ERR
+			"btrfs: unable to fixup (nodatasum) error at logical %llu on dev %s\n",
+			(unsigned long long)fixup->logical, sdev->dev->name);
 	}
 
 	btrfs_free_path(path);
@@ -550,91 +598,168 @@ out:
 }
 
 /*
- * scrub_recheck_error gets called when either verification of the page
- * failed or the bio failed to read, e.g. with EIO. In the latter case,
- * recheck_error gets called for every page in the bio, even though only
- * one may be bad
+ * scrub_handle_errored_block gets called when either verification of the
+ * pages failed or the bio failed to read, e.g. with EIO. In the latter
+ * case, this function handles all pages in the bio, even though only one
+ * may be bad.
+ * The goal of this function is to repair the errored block by using the
+ * contents of one of the mirrors.
  */
-static int scrub_recheck_error(struct scrub_bio *sbio, int ix)
+static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 {
-	struct scrub_dev *sdev = sbio->sdev;
-	u64 sector = (sbio->physical + ix * PAGE_SIZE) >> 9;
+	struct scrub_dev *sdev = sblock_to_check->sdev;
+	struct btrfs_fs_info *fs_info;
+	u64 length;
+	u64 logical;
+	u64 generation;
+	unsigned int failed_mirror_index;
+	unsigned int is_metadata;
+	unsigned int have_csum;
+	u8 *csum;
+	struct scrub_block *sblocks_for_recheck; /* holds one for each mirror */
+	struct scrub_block *sblock_bad;
+	int ret;
+	int mirror_index;
+	int page_num;
+	int success;
 	static DEFINE_RATELIMIT_STATE(_rs, DEFAULT_RATELIMIT_INTERVAL,
-					DEFAULT_RATELIMIT_BURST);
+				      DEFAULT_RATELIMIT_BURST);
+
+	BUG_ON(sblock_to_check->page_count < 1);
+	fs_info = sdev->dev->dev_root->fs_info;
+	length = sblock_to_check->page_count * PAGE_SIZE;
+	logical = sblock_to_check->pagev[0].logical;
+	generation = sblock_to_check->pagev[0].generation;
+	BUG_ON(sblock_to_check->pagev[0].mirror_num < 1);
+	failed_mirror_index = sblock_to_check->pagev[0].mirror_num - 1;
+	is_metadata = !(sblock_to_check->pagev[0].flags &
+			BTRFS_EXTENT_FLAG_DATA);
+	have_csum = sblock_to_check->pagev[0].have_csum;
+	csum = sblock_to_check->pagev[0].csum;
 
-	if (sbio->err) {
-		if (scrub_fixup_io(READ, sbio->sdev->dev->bdev, sector,
-				   sbio->bio->bi_io_vec[ix].bv_page) == 0) {
-			if (scrub_fixup_check(sbio, ix) == 0)
-				return 0;
-		}
-		if (__ratelimit(&_rs))
-			scrub_print_warning("i/o error", sbio, ix);
-	} else {
-		if (__ratelimit(&_rs))
-			scrub_print_warning("checksum error", sbio, ix);
+	/*
+	 * read all mirrors one after the other. This includes to
+	 * re-read the extent or metadata block that failed (that was
+	 * the cause that this fixup code is called) another time,
+	 * page by page this time in order to know which pages
+	 * caused I/O errors and which ones are good (for all mirrors).
+	 * It is the goal to handle the situation when more than one
+	 * mirror contains I/O errors, but the errors do not
+	 * overlap, i.e. the data can be repaired by selecting the
+	 * pages from those mirrors without I/O error on the
+	 * particular pages. One example (with blocks >= 2 * PAGE_SIZE)
+	 * would be that mirror #1 has an I/O error on the first page,
+	 * the second page is good, and mirror #2 has an I/O error on
+	 * the second page, but the first page is good.
+	 * Then the first page of the first mirror can be repaired by
+	 * taking the first page of the second mirror, and the
+	 * second page of the second mirror can be repaired by
+	 * copying the contents of the 2nd page of the 1st mirror.
+	 * One more note: if the pages of one mirror contain I/O
+	 * errors, the checksum cannot be verified. In order to get
+	 * the best data for repairing, the first attempt is to find
+	 * a mirror without I/O errors and with a validated checksum.
+	 * Only if this is not possible, the pages are picked from
+	 * mirrors with I/O errors without considering the checksum.
+	 * If the latter is the case, at the end, the checksum of the
+	 * repaired area is verified in order to correctly maintain
+	 * the statistics.
+	 */
+
+	sblocks_for_recheck = kzalloc(BTRFS_MAX_MIRRORS *
+				     sizeof(*sblocks_for_recheck),
+				     GFP_NOFS);
+	if (!sblocks_for_recheck) {
+		spin_lock(&sdev->stat_lock);
+		sdev->stat.malloc_errors++;
+		sdev->stat.read_errors++;
+		sdev->stat.uncorrectable_errors++;
+		spin_unlock(&sdev->stat_lock);
+		goto out;
 	}
 
-	spin_lock(&sdev->stat_lock);
-	++sdev->stat.read_errors;
-	spin_unlock(&sdev->stat_lock);
+	/* setup the context, map the logical blocks and alloc the pages */
+	ret = scrub_setup_recheck_block(sdev, &fs_info->mapping_tree, length,
+					logical, sblocks_for_recheck);
+	if (ret) {
+		spin_lock(&sdev->stat_lock);
+		sdev->stat.read_errors++;
+		sdev->stat.uncorrectable_errors++;
+		spin_unlock(&sdev->stat_lock);
+		goto out;
+	}
+	BUG_ON(failed_mirror_index >= BTRFS_MAX_MIRRORS);
+	sblock_bad = sblocks_for_recheck + failed_mirror_index;
 
-	scrub_fixup(sbio, ix);
-	return 1;
-}
+	/* build and submit the bios for the failed mirror, check checksums */
+	ret = scrub_recheck_block(fs_info, sblock_bad, is_metadata, have_csum,
+				  csum, generation, sdev->csum_size);
+	if (ret) {
+		spin_lock(&sdev->stat_lock);
+		sdev->stat.read_errors++;
+		sdev->stat.uncorrectable_errors++;
+		spin_unlock(&sdev->stat_lock);
+		goto out;
+	}
 
-static int scrub_fixup_check(struct scrub_bio *sbio, int ix)
-{
-	int ret = 1;
-	struct page *page;
-	void *buffer;
-	u64 flags = sbio->spag[ix].flags;
+	if (!sblock_bad->header_error && !sblock_bad->checksum_error &&
+	    sblock_bad->no_io_error_seen) {
+		/*
+		 * the error disappeared after reading page by page, or
+		 * the area was part of a huge bio and other parts of the
+		 * bio caused I/O errors, or the block layer merged several
+		 * read requests into one and the error is caused by a
+		 * different bio (usually one of the two latter cases is
+		 * the cause)
+		 */
+		spin_lock(&sdev->stat_lock);
+		sdev->stat.unverified_errors++;
+		spin_unlock(&sdev->stat_lock);
 
-	page = sbio->bio->bi_io_vec[ix].bv_page;
-	buffer = kmap_atomic(page);
-	if (flags & BTRFS_EXTENT_FLAG_DATA) {
-		ret = scrub_checksum_data(sbio->sdev,
-					  sbio->spag + ix, buffer);
-	} else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
-		ret = scrub_checksum_tree_block(sbio->sdev,
-						sbio->spag + ix,
-						sbio->logical + ix * PAGE_SIZE,
-						buffer);
-	} else {
-		WARN_ON(1);
+		goto out;
 	}
-	kunmap_atomic(buffer);
 
-	return ret;
-}
+	if (!sblock_bad->no_io_error_seen) {
+		spin_lock(&sdev->stat_lock);
+		sdev->stat.read_errors++;
+		spin_unlock(&sdev->stat_lock);
+		if (__ratelimit(&_rs))
+			scrub_print_warning("i/o error", sblock_to_check);
+	} else if (sblock_bad->checksum_error) {
+		spin_lock(&sdev->stat_lock);
+		sdev->stat.csum_errors++;
+		spin_unlock(&sdev->stat_lock);
+		if (__ratelimit(&_rs))
+			scrub_print_warning("checksum error", sblock_to_check);
+	} else if (sblock_bad->header_error) {
+		spin_lock(&sdev->stat_lock);
+		sdev->stat.verify_errors++;
+		spin_unlock(&sdev->stat_lock);
+		if (__ratelimit(&_rs))
+			scrub_print_warning("checksum/header error",
+					    sblock_to_check);
+	}
 
-static void scrub_fixup_end_io(struct bio *bio, int err)
-{
-	complete((struct completion *)bio->bi_private);
-}
+	if (sdev->readonly)
+		goto did_not_correct_error;
 
-static void scrub_fixup(struct scrub_bio *sbio, int ix)
-{
-	struct scrub_dev *sdev = sbio->sdev;
-	struct btrfs_fs_info *fs_info = sdev->dev->dev_root->fs_info;
-	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
-	struct btrfs_bio *bbio = NULL;
-	struct scrub_fixup_nodatasum *fixup;
-	u64 logical = sbio->logical + ix * PAGE_SIZE;
-	u64 length;
-	int i;
-	int ret;
-	DECLARE_COMPLETION_ONSTACK(complete);
-
-	if ((sbio->spag[ix].flags & BTRFS_EXTENT_FLAG_DATA) &&
-	    (sbio->spag[ix].have_csum == 0)) {
-		fixup = kzalloc(sizeof(*fixup), GFP_NOFS);
-		if (!fixup)
-			goto uncorrectable;
-		fixup->sdev = sdev;
-		fixup->logical = logical;
-		fixup->root = fs_info->extent_root;
-		fixup->mirror_num = sbio->spag[ix].mirror_num;
+	if (!is_metadata && !have_csum) {
+		struct scrub_fixup_nodatasum *fixup_nodatasum;
+
+		/*
+		 * !is_metadata and !have_csum, this means that the data
+		 * might not be COW'ed, that it might be modified
+		 * concurrently. The general strategy to work on the
+		 * commit root does not help in the case when COW is not
+		 * used.
+		 */
+		fixup_nodatasum = kzalloc(sizeof(*fixup_nodatasum), GFP_NOFS);
+		if (!fixup_nodatasum)
+			goto did_not_correct_error;
+		fixup_nodatasum->sdev = sdev;
+		fixup_nodatasum->logical = logical;
+		fixup_nodatasum->root = fs_info->extent_root;
+		fixup_nodatasum->mirror_num = failed_mirror_index + 1;
 		/*
 		 * increment scrubs_running to prevent cancel requests from
 		 * completing as long as a fixup worker is running. we must also
@@ -649,235 +774,528 @@ static void scrub_fixup(struct scrub_bio *sbio, int ix)
 		atomic_inc(&fs_info->scrubs_paused);
 		mutex_unlock(&fs_info->scrub_lock);
 		atomic_inc(&sdev->fixup_cnt);
-		fixup->work.func = scrub_fixup_nodatasum;
-		btrfs_queue_worker(&fs_info->scrub_workers, &fixup->work);
-		return;
+		fixup_nodatasum->work.func = scrub_fixup_nodatasum;
+		btrfs_queue_worker(&fs_info->scrub_workers,
+				   &fixup_nodatasum->work);
+		goto out;
 	}
 
-	length = PAGE_SIZE;
-	ret = btrfs_map_block(map_tree, REQ_WRITE, logical, &length,
-			      &bbio, 0);
-	if (ret || !bbio || length < PAGE_SIZE) {
-		printk(KERN_ERR
-		       "scrub_fixup: btrfs_map_block failed us for %llu\n",
-		       (unsigned long long)logical);
-		WARN_ON(1);
-		kfree(bbio);
-		return;
+	/*
+	 * now build and submit the bios for the other mirrors, check
+	 * checksums
+	 */
+	for (mirror_index = 0;
+	     mirror_index < BTRFS_MAX_MIRRORS &&
+	     sblocks_for_recheck[mirror_index].page_count > 0;
+	     mirror_index++) {
+		if (mirror_index == failed_mirror_index)
+			continue;
+
+		/* build and submit the bios, check checksums */
+		ret = scrub_recheck_block(fs_info,
+					  sblocks_for_recheck + mirror_index,
+					  is_metadata, have_csum, csum,
+					  generation, sdev->csum_size);
+		if (ret)
+			goto did_not_correct_error;
 	}
 
-	if (bbio->num_stripes == 1)
-		/* there aren't any replicas */
-		goto uncorrectable;
+	/*
+	 * first try to pick the mirror which is completely without I/O
+	 * errors and also does not have a checksum error.
+	 * If one is found, and if a checksum is present, the full block
+	 * that is known to contain an error is rewritten. Afterwards
+	 * the block is known to be corrected.
+	 * If a mirror is found which is completely correct, and no
+	 * checksum is present, only those pages are rewritten that had
+	 * an I/O error in the block to be repaired, since it cannot be
+	 * determined, which copy of the other pages is better (and it
+	 * could happen otherwise that a correct page would be
+	 * overwritten by a bad one).
+	 */
+	for (mirror_index = 0;
+	     mirror_index < BTRFS_MAX_MIRRORS &&
+	     sblocks_for_recheck[mirror_index].page_count > 0;
+	     mirror_index++) {
+		struct scrub_block *sblock_other = sblocks_for_recheck +
+						   mirror_index;
+
+		if (!sblock_other->header_error &&
+		    !sblock_other->checksum_error &&
+		    sblock_other->no_io_error_seen) {
+			int force_write = is_metadata || have_csum;
+
+			ret = scrub_repair_block_from_good_copy(sblock_bad,
+								sblock_other,
+								force_write);
+			if (0 == ret)
+				goto corrected_error;
+		}
+	}
 
 	/*
-	 * first find a good copy
+	 * in case of I/O errors in the area that is supposed to be
+	 * repaired, continue by picking good copies of those pages.
+	 * Select the good pages from mirrors to rewrite bad pages from
+	 * the area to fix. Afterwards verify the checksum of the block
+	 * that is supposed to be repaired. This verification step is
+	 * only done for the purpose of statistic counting and for the
+	 * final scrub report, whether errors remain.
+	 * A perfect algorithm could make use of the checksum and try
+	 * all possible combinations of pages from the different mirrors
+	 * until the checksum verification succeeds. For example, when
+	 * the 2nd page of mirror #1 faces I/O errors, and the 2nd page
+	 * of mirror #2 is readable but the final checksum test fails,
+	 * then the 2nd page of mirror #3 could be tried, whether now
+	 * the final checksum succeedes. But this would be a rare
+	 * exception and is therefore not implemented. At least it is
+	 * avoided that the good copy is overwritten.
+	 * A more useful improvement would be to pick the sectors
+	 * without I/O error based on sector sizes (512 bytes on legacy
+	 * disks) instead of on PAGE_SIZE. Then maybe 512 byte of one
+	 * mirror could be repaired by taking 512 byte of a different
+	 * mirror, even if other 512 byte sectors in the same PAGE_SIZE
+	 * area are unreadable.
 	 */
-	for (i = 0; i < bbio->num_stripes; ++i) {
-		if (i + 1 == sbio->spag[ix].mirror_num)
-			continue;
 
-		if (scrub_fixup_io(READ, bbio->stripes[i].dev->bdev,
-				   bbio->stripes[i].physical >> 9,
-				   sbio->bio->bi_io_vec[ix].bv_page)) {
-			/* I/O-error, this is not a good copy */
+	/* can only fix I/O errors from here on */
+	if (sblock_bad->no_io_error_seen)
+		goto did_not_correct_error;
+
+	success = 1;
+	for (page_num = 0; page_num < sblock_bad->page_count; page_num++) {
+		struct scrub_page *page_bad = sblock_bad->pagev + page_num;
+
+		if (!page_bad->io_error)
 			continue;
+
+		for (mirror_index = 0;
+		     mirror_index < BTRFS_MAX_MIRRORS &&
+		     sblocks_for_recheck[mirror_index].page_count > 0;
+		     mirror_index++) {
+			struct scrub_block *sblock_other = sblocks_for_recheck +
+							   mirror_index;
+			struct scrub_page *page_other = sblock_other->pagev +
+							page_num;
+
+			if (!page_other->io_error) {
+				ret = scrub_repair_page_from_good_copy(
+					sblock_bad, sblock_other, page_num, 0);
+				if (0 == ret) {
+					page_bad->io_error = 0;
+					break; /* succeeded for this page */
+				}
+			}
 		}
 
-		if (scrub_fixup_check(sbio, ix) == 0)
-			break;
+		if (page_bad->io_error) {
+			/* did not find a mirror to copy the page from */
+			success = 0;
+		}
 	}
-	if (i == bbio->num_stripes)
-		goto uncorrectable;
 
-	if (!sdev->readonly) {
-		/*
-		 * bi_io_vec[ix].bv_page now contains good data, write it back
-		 */
-		if (scrub_fixup_io(WRITE, sdev->dev->bdev,
-				   (sbio->physical + ix * PAGE_SIZE) >> 9,
-				   sbio->bio->bi_io_vec[ix].bv_page)) {
-			/* I/O-error, writeback failed, give up */
-			goto uncorrectable;
+	if (success) {
+		if (is_metadata || have_csum) {
+			/*
+			 * need to verify the checksum now that all
+			 * sectors on disk are repaired (the write
+			 * request for data to be repaired is on its way).
+			 * Just be lazy and use scrub_recheck_block()
+			 * which re-reads the data before the checksum
+			 * is verified, but most likely the data comes out
+			 * of the page cache.
+			 */
+			ret = scrub_recheck_block(fs_info, sblock_bad,
+						  is_metadata, have_csum, csum,
+						  generation, sdev->csum_size);
+			if (!ret && !sblock_bad->header_error &&
+			    !sblock_bad->checksum_error &&
+			    sblock_bad->no_io_error_seen)
+				goto corrected_error;
+			else
+				goto did_not_correct_error;
+		} else {
+corrected_error:
+			spin_lock(&sdev->stat_lock);
+			sdev->stat.corrected_errors++;
+			spin_unlock(&sdev->stat_lock);
+			printk_ratelimited(KERN_ERR
+				"btrfs: fixed up error at logical %llu on dev %s\n",
+				(unsigned long long)logical, sdev->dev->name);
 		}
+	} else {
+did_not_correct_error:
+		spin_lock(&sdev->stat_lock);
+		sdev->stat.uncorrectable_errors++;
+		spin_unlock(&sdev->stat_lock);
+		printk_ratelimited(KERN_ERR
+			"btrfs: unable to fixup (regular) error at logical %llu on dev %s\n",
+			(unsigned long long)logical, sdev->dev->name);
 	}
 
-	kfree(bbio);
-	spin_lock(&sdev->stat_lock);
-	++sdev->stat.corrected_errors;
-	spin_unlock(&sdev->stat_lock);
+out:
+	if (sblocks_for_recheck) {
+		for (mirror_index = 0; mirror_index < BTRFS_MAX_MIRRORS;
+		     mirror_index++) {
+			struct scrub_block *sblock = sblocks_for_recheck +
+						     mirror_index;
+			int page_index;
+
+			for (page_index = 0; page_index < SCRUB_PAGES_PER_BIO;
+			     page_index++)
+				if (sblock->pagev[page_index].page)
+					__free_page(
+						sblock->pagev[page_index].page);
+		}
+		kfree(sblocks_for_recheck);
+	}
 
-	printk_ratelimited(KERN_ERR "btrfs: fixed up error at logical %llu\n",
-			       (unsigned long long)logical);
-	return;
+	return 0;
+}
 
-uncorrectable:
-	kfree(bbio);
-	spin_lock(&sdev->stat_lock);
-	++sdev->stat.uncorrectable_errors;
-	spin_unlock(&sdev->stat_lock);
+static int scrub_setup_recheck_block(struct scrub_dev *sdev,
+				     struct btrfs_mapping_tree *map_tree,
+				     u64 length, u64 logical,
+				     struct scrub_block *sblocks_for_recheck)
+{
+	int page_index;
+	int mirror_index;
+	int ret;
+
+	/*
+	 * note: the three members sdev, ref_count and outstanding_pages
+	 * are not used (and not set) in the blocks that are used for
+	 * the recheck procedure
+	 */
+
+	page_index = 0;
+	while (length > 0) {
+		u64 sublen = min_t(u64, length, PAGE_SIZE);
+		u64 mapped_length = sublen;
+		struct btrfs_bio *bbio = NULL;
+
+		/*
+		 * with a length of PAGE_SIZE, each returned stripe
+		 * represents one mirror
+		 */
+		ret = btrfs_map_block(map_tree, WRITE, logical, &mapped_length,
+				      &bbio, 0);
+		if (ret || !bbio || mapped_length < sublen) {
+			kfree(bbio);
+			return -EIO;
+		}
 
-	printk_ratelimited(KERN_ERR "btrfs: unable to fixup (regular) error at "
-				"logical %llu\n", (unsigned long long)logical);
+		BUG_ON(page_index >= SCRUB_PAGES_PER_BIO);
+		for (mirror_index = 0; mirror_index < (int)bbio->num_stripes;
+		     mirror_index++) {
+			struct scrub_block *sblock;
+			struct scrub_page *page;
+
+			if (mirror_index >= BTRFS_MAX_MIRRORS)
+				continue;
+
+			sblock = sblocks_for_recheck + mirror_index;
+			page = sblock->pagev + page_index;
+			page->logical = logical;
+			page->physical = bbio->stripes[mirror_index].physical;
+			page->bdev = bbio->stripes[mirror_index].dev->bdev;
+			page->mirror_num = mirror_index + 1;
+			page->page = alloc_page(GFP_NOFS);
+			if (!page->page) {
+				spin_lock(&sdev->stat_lock);
+				sdev->stat.malloc_errors++;
+				spin_unlock(&sdev->stat_lock);
+				return -ENOMEM;
+			}
+			sblock->page_count++;
+		}
+		kfree(bbio);
+		length -= sublen;
+		logical += sublen;
+		page_index++;
+	}
+
+	return 0;
 }
 
-static int scrub_fixup_io(int rw, struct block_device *bdev, sector_t sector,
-			 struct page *page)
+/*
+ * this function will check the on disk data for checksum errors, header
+ * errors and read I/O errors. If any I/O errors happen, the exact pages
+ * which are errored are marked as being bad. The goal is to enable scrub
+ * to take those pages that are not errored from all the mirrors so that
+ * the pages that are errored in the just handled mirror can be repaired.
+ */
+static int scrub_recheck_block(struct btrfs_fs_info *fs_info,
+			       struct scrub_block *sblock, int is_metadata,
+			       int have_csum, u8 *csum, u64 generation,
+			       u16 csum_size)
 {
-	struct bio *bio = NULL;
-	int ret;
-	DECLARE_COMPLETION_ONSTACK(complete);
+	int page_num;
+
+	sblock->no_io_error_seen = 1;
+	sblock->header_error = 0;
+	sblock->checksum_error = 0;
+
+	for (page_num = 0; page_num < sblock->page_count; page_num++) {
+		struct bio *bio;
+		int ret;
+		struct scrub_page *page = sblock->pagev + page_num;
+		DECLARE_COMPLETION_ONSTACK(complete);
+
+		BUG_ON(!page->page);
+		bio = bio_alloc(GFP_NOFS, 1);
+		bio->bi_bdev = page->bdev;
+		bio->bi_sector = page->physical >> 9;
+		bio->bi_end_io = scrub_complete_bio_end_io;
+		bio->bi_private = &complete;
+
+		ret = bio_add_page(bio, page->page, PAGE_SIZE, 0);
+		if (PAGE_SIZE != ret) {
+			bio_put(bio);
+			return -EIO;
+		}
+		btrfsic_submit_bio(READ, bio);
 
-	bio = bio_alloc(GFP_NOFS, 1);
-	bio->bi_bdev = bdev;
-	bio->bi_sector = sector;
-	bio_add_page(bio, page, PAGE_SIZE, 0);
-	bio->bi_end_io = scrub_fixup_end_io;
-	bio->bi_private = &complete;
-	btrfsic_submit_bio(rw, bio);
+		/* this will also unplug the queue */
+		wait_for_completion(&complete);
 
-	/* this will also unplug the queue */
-	wait_for_completion(&complete);
+		page->io_error = !test_bit(BIO_UPTODATE, &bio->bi_flags);
+		if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
+			sblock->no_io_error_seen = 0;
+		bio_put(bio);
+	}
 
-	ret = !test_bit(BIO_UPTODATE, &bio->bi_flags);
-	bio_put(bio);
-	return ret;
+	if (sblock->no_io_error_seen)
+		scrub_recheck_block_checksum(fs_info, sblock, is_metadata,
+					     have_csum, csum, generation,
+					     csum_size);
+
+	return 0;
 }
 
-static void scrub_bio_end_io(struct bio *bio, int err)
+static void scrub_recheck_block_checksum(struct btrfs_fs_info *fs_info,
+					 struct scrub_block *sblock,
+					 int is_metadata, int have_csum,
+					 const u8 *csum, u64 generation,
+					 u16 csum_size)
 {
-	struct scrub_bio *sbio = bio->bi_private;
-	struct scrub_dev *sdev = sbio->sdev;
-	struct btrfs_fs_info *fs_info = sdev->dev->dev_root->fs_info;
+	int page_num;
+	u8 calculated_csum[BTRFS_CSUM_SIZE];
+	u32 crc = ~(u32)0;
+	struct btrfs_root *root = fs_info->extent_root;
+	void *mapped_buffer;
+
+	BUG_ON(!sblock->pagev[0].page);
+	if (is_metadata) {
+		struct btrfs_header *h;
+
+		mapped_buffer = kmap_atomic(sblock->pagev[0].page);
+		h = (struct btrfs_header *)mapped_buffer;
+
+		if (sblock->pagev[0].logical != le64_to_cpu(h->bytenr) ||
+		    generation != le64_to_cpu(h->generation) ||
+		    memcmp(h->fsid, fs_info->fsid, BTRFS_UUID_SIZE) ||
+		    memcmp(h->chunk_tree_uuid, fs_info->chunk_tree_uuid,
+			   BTRFS_UUID_SIZE))
+			sblock->header_error = 1;
+		csum = h->csum;
+	} else {
+		if (!have_csum)
+			return;
 
-	sbio->err = err;
-	sbio->bio = bio;
+		mapped_buffer = kmap_atomic(sblock->pagev[0].page);
+	}
 
-	btrfs_queue_worker(&fs_info->scrub_workers, &sbio->work);
+	for (page_num = 0;;) {
+		if (page_num == 0 && is_metadata)
+			crc = btrfs_csum_data(root,
+				((u8 *)mapped_buffer) + BTRFS_CSUM_SIZE,
+				crc, PAGE_SIZE - BTRFS_CSUM_SIZE);
+		else
+			crc = btrfs_csum_data(root, mapped_buffer, crc,
+					      PAGE_SIZE);
+
+		kunmap_atomic(mapped_buffer);
+		page_num++;
+		if (page_num >= sblock->page_count)
+			break;
+		BUG_ON(!sblock->pagev[page_num].page);
+
+		mapped_buffer = kmap_atomic(sblock->pagev[page_num].page);
+	}
+
+	btrfs_csum_final(crc, calculated_csum);
+	if (memcmp(calculated_csum, csum, csum_size))
+		sblock->checksum_error = 1;
 }
 
-static void scrub_checksum(struct btrfs_work *work)
+static void scrub_complete_bio_end_io(struct bio *bio, int err)
 {
-	struct scrub_bio *sbio = container_of(work, struct scrub_bio, work);
-	struct scrub_dev *sdev = sbio->sdev;
-	struct page *page;
-	void *buffer;
-	int i;
-	u64 flags;
-	u64 logical;
-	int ret;
+	complete((struct completion *)bio->bi_private);
+}
 
-	if (sbio->err) {
-		ret = 0;
-		for (i = 0; i < sbio->count; ++i)
-			ret |= scrub_recheck_error(sbio, i);
-		if (!ret) {
-			spin_lock(&sdev->stat_lock);
-			++sdev->stat.unverified_errors;
-			spin_unlock(&sdev->stat_lock);
-		}
+static int scrub_repair_block_from_good_copy(struct scrub_block *sblock_bad,
+					     struct scrub_block *sblock_good,
+					     int force_write)
+{
+	int page_num;
+	int ret = 0;
 
-		sbio->bio->bi_flags &= ~(BIO_POOL_MASK - 1);
-		sbio->bio->bi_flags |= 1 << BIO_UPTODATE;
-		sbio->bio->bi_phys_segments = 0;
-		sbio->bio->bi_idx = 0;
+	for (page_num = 0; page_num < sblock_bad->page_count; page_num++) {
+		int ret_sub;
 
-		for (i = 0; i < sbio->count; i++) {
-			struct bio_vec *bi;
-			bi = &sbio->bio->bi_io_vec[i];
-			bi->bv_offset = 0;
-			bi->bv_len = PAGE_SIZE;
-		}
-		goto out;
+		ret_sub = scrub_repair_page_from_good_copy(sblock_bad,
+							   sblock_good,
+							   page_num,
+							   force_write);
+		if (ret_sub)
+			ret = ret_sub;
 	}
-	for (i = 0; i < sbio->count; ++i) {
-		page = sbio->bio->bi_io_vec[i].bv_page;
-		buffer = kmap_atomic(page);
-		flags = sbio->spag[i].flags;
-		logical = sbio->logical + i * PAGE_SIZE;
-		ret = 0;
-		if (flags & BTRFS_EXTENT_FLAG_DATA) {
-			ret = scrub_checksum_data(sdev, sbio->spag + i, buffer);
-		} else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
-			ret = scrub_checksum_tree_block(sdev, sbio->spag + i,
-							logical, buffer);
-		} else if (flags & BTRFS_EXTENT_FLAG_SUPER) {
-			BUG_ON(i);
-			(void)scrub_checksum_super(sbio, buffer);
-		} else {
-			WARN_ON(1);
-		}
-		kunmap_atomic(buffer);
-		if (ret) {
-			ret = scrub_recheck_error(sbio, i);
-			if (!ret) {
-				spin_lock(&sdev->stat_lock);
-				++sdev->stat.unverified_errors;
-				spin_unlock(&sdev->stat_lock);
-			}
+
+	return ret;
+}
+
+static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad,
+					    struct scrub_block *sblock_good,
+					    int page_num, int force_write)
+{
+	struct scrub_page *page_bad = sblock_bad->pagev + page_num;
+	struct scrub_page *page_good = sblock_good->pagev + page_num;
+
+	BUG_ON(sblock_bad->pagev[page_num].page == NULL);
+	BUG_ON(sblock_good->pagev[page_num].page == NULL);
+	if (force_write || sblock_bad->header_error ||
+	    sblock_bad->checksum_error || page_bad->io_error) {
+		struct bio *bio;
+		int ret;
+		DECLARE_COMPLETION_ONSTACK(complete);
+
+		bio = bio_alloc(GFP_NOFS, 1);
+		bio->bi_bdev = page_bad->bdev;
+		bio->bi_sector = page_bad->physical >> 9;
+		bio->bi_end_io = scrub_complete_bio_end_io;
+		bio->bi_private = &complete;
+
+		ret = bio_add_page(bio, page_good->page, PAGE_SIZE, 0);
+		if (PAGE_SIZE != ret) {
+			bio_put(bio);
+			return -EIO;
 		}
+		btrfsic_submit_bio(WRITE, bio);
+
+		/* this will also unplug the queue */
+		wait_for_completion(&complete);
+		bio_put(bio);
 	}
 
-out:
-	scrub_free_bio(sbio->bio);
-	sbio->bio = NULL;
-	spin_lock(&sdev->list_lock);
-	sbio->next_free = sdev->first_free;
-	sdev->first_free = sbio->index;
-	spin_unlock(&sdev->list_lock);
-	atomic_dec(&sdev->in_flight);
-	wake_up(&sdev->list_wait);
+	return 0;
+}
+
+static void scrub_checksum(struct scrub_block *sblock)
+{
+	u64 flags;
+	int ret;
+
+	BUG_ON(sblock->page_count < 1);
+	flags = sblock->pagev[0].flags;
+	ret = 0;
+	if (flags & BTRFS_EXTENT_FLAG_DATA)
+		ret = scrub_checksum_data(sblock);
+	else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
+		ret = scrub_checksum_tree_block(sblock);
+	else if (flags & BTRFS_EXTENT_FLAG_SUPER)
+		(void)scrub_checksum_super(sblock);
+	else
+		WARN_ON(1);
+	if (ret)
+		scrub_handle_errored_block(sblock);
 }
 
-static int scrub_checksum_data(struct scrub_dev *sdev,
-			       struct scrub_page *spag, void *buffer)
+static int scrub_checksum_data(struct scrub_block *sblock)
 {
+	struct scrub_dev *sdev = sblock->sdev;
 	u8 csum[BTRFS_CSUM_SIZE];
+	u8 *on_disk_csum;
+	struct page *page;
+	void *buffer;
 	u32 crc = ~(u32)0;
 	int fail = 0;
 	struct btrfs_root *root = sdev->dev->dev_root;
+	u64 len;
+	int index;
 
-	if (!spag->have_csum)
+	BUG_ON(sblock->page_count < 1);
+	if (!sblock->pagev[0].have_csum)
 		return 0;
 
-	crc = btrfs_csum_data(root, buffer, crc, PAGE_SIZE);
+	on_disk_csum = sblock->pagev[0].csum;
+	page = sblock->pagev[0].page;
+	buffer = kmap_atomic(page);
+
+	len = sdev->sectorsize;
+	index = 0;
+	for (;;) {
+		u64 l = min_t(u64, len, PAGE_SIZE);
+
+		crc = btrfs_csum_data(root, buffer, crc, l);
+		kunmap_atomic(buffer);
+		len -= l;
+		if (len == 0)
+			break;
+		index++;
+		BUG_ON(index >= sblock->page_count);
+		BUG_ON(!sblock->pagev[index].page);
+		page = sblock->pagev[index].page;
+		buffer = kmap_atomic(page);
+	}
+
 	btrfs_csum_final(crc, csum);
-	if (memcmp(csum, spag->csum, sdev->csum_size))
+	if (memcmp(csum, on_disk_csum, sdev->csum_size))
 		fail = 1;
 
-	spin_lock(&sdev->stat_lock);
-	++sdev->stat.data_extents_scrubbed;
-	sdev->stat.data_bytes_scrubbed += PAGE_SIZE;
-	if (fail)
+	if (fail) {
+		spin_lock(&sdev->stat_lock);
 		++sdev->stat.csum_errors;
-	spin_unlock(&sdev->stat_lock);
+		spin_unlock(&sdev->stat_lock);
+	}
 
 	return fail;
 }
 
-static int scrub_checksum_tree_block(struct scrub_dev *sdev,
-				     struct scrub_page *spag, u64 logical,
-				     void *buffer)
+static int scrub_checksum_tree_block(struct scrub_block *sblock)
 {
+	struct scrub_dev *sdev = sblock->sdev;
 	struct btrfs_header *h;
 	struct btrfs_root *root = sdev->dev->dev_root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	u8 csum[BTRFS_CSUM_SIZE];
+	u8 calculated_csum[BTRFS_CSUM_SIZE];
+	u8 on_disk_csum[BTRFS_CSUM_SIZE];
+	struct page *page;
+	void *mapped_buffer;
+	u64 mapped_size;
+	void *p;
 	u32 crc = ~(u32)0;
 	int fail = 0;
 	int crc_fail = 0;
+	u64 len;
+	int index;
+
+	BUG_ON(sblock->page_count < 1);
+	page = sblock->pagev[0].page;
+	mapped_buffer = kmap_atomic(page);
+	h = (struct btrfs_header *)mapped_buffer;
+	memcpy(on_disk_csum, h->csum, sdev->csum_size);
 
 	/*
 	 * we don't use the getter functions here, as we
 	 * a) don't have an extent buffer and
 	 * b) the page is already kmapped
 	 */
-	h = (struct btrfs_header *)buffer;
 
-	if (logical != le64_to_cpu(h->bytenr))
+	if (sblock->pagev[0].logical != le64_to_cpu(h->bytenr))
 		++fail;
 
-	if (spag->generation != le64_to_cpu(h->generation))
+	if (sblock->pagev[0].generation != le64_to_cpu(h->generation))
 		++fail;
 
 	if (memcmp(h->fsid, fs_info->fsid, BTRFS_UUID_SIZE))
@@ -887,51 +1305,99 @@ static int scrub_checksum_tree_block(struct scrub_dev *sdev,
 		   BTRFS_UUID_SIZE))
 		++fail;
 
-	crc = btrfs_csum_data(root, buffer + BTRFS_CSUM_SIZE, crc,
-			      PAGE_SIZE - BTRFS_CSUM_SIZE);
-	btrfs_csum_final(crc, csum);
-	if (memcmp(csum, h->csum, sdev->csum_size))
+	BUG_ON(sdev->nodesize != sdev->leafsize);
+	len = sdev->nodesize - BTRFS_CSUM_SIZE;
+	mapped_size = PAGE_SIZE - BTRFS_CSUM_SIZE;
+	p = ((u8 *)mapped_buffer) + BTRFS_CSUM_SIZE;
+	index = 0;
+	for (;;) {
+		u64 l = min_t(u64, len, mapped_size);
+
+		crc = btrfs_csum_data(root, p, crc, l);
+		kunmap_atomic(mapped_buffer);
+		len -= l;
+		if (len == 0)
+			break;
+		index++;
+		BUG_ON(index >= sblock->page_count);
+		BUG_ON(!sblock->pagev[index].page);
+		page = sblock->pagev[index].page;
+		mapped_buffer = kmap_atomic(page);
+		mapped_size = PAGE_SIZE;
+		p = mapped_buffer;
+	}
+
+	btrfs_csum_final(crc, calculated_csum);
+	if (memcmp(calculated_csum, on_disk_csum, sdev->csum_size))
 		++crc_fail;
 
-	spin_lock(&sdev->stat_lock);
-	++sdev->stat.tree_extents_scrubbed;
-	sdev->stat.tree_bytes_scrubbed += PAGE_SIZE;
-	if (crc_fail)
-		++sdev->stat.csum_errors;
-	if (fail)
-		++sdev->stat.verify_errors;
-	spin_unlock(&sdev->stat_lock);
+	if (crc_fail || fail) {
+		spin_lock(&sdev->stat_lock);
+		if (crc_fail)
+			++sdev->stat.csum_errors;
+		if (fail)
+			++sdev->stat.verify_errors;
+		spin_unlock(&sdev->stat_lock);
+	}
 
 	return fail || crc_fail;
 }
 
-static int scrub_checksum_super(struct scrub_bio *sbio, void *buffer)
+static int scrub_checksum_super(struct scrub_block *sblock)
 {
 	struct btrfs_super_block *s;
-	u64 logical;
-	struct scrub_dev *sdev = sbio->sdev;
+	struct scrub_dev *sdev = sblock->sdev;
 	struct btrfs_root *root = sdev->dev->dev_root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
-	u8 csum[BTRFS_CSUM_SIZE];
+	u8 calculated_csum[BTRFS_CSUM_SIZE];
+	u8 on_disk_csum[BTRFS_CSUM_SIZE];
+	struct page *page;
+	void *mapped_buffer;
+	u64 mapped_size;
+	void *p;
 	u32 crc = ~(u32)0;
 	int fail = 0;
+	u64 len;
+	int index;
 
-	s = (struct btrfs_super_block *)buffer;
-	logical = sbio->logical;
+	BUG_ON(sblock->page_count < 1);
+	page = sblock->pagev[0].page;
+	mapped_buffer = kmap_atomic(page);
+	s = (struct btrfs_super_block *)mapped_buffer;
+	memcpy(on_disk_csum, s->csum, sdev->csum_size);
 
-	if (logical != le64_to_cpu(s->bytenr))
+	if (sblock->pagev[0].logical != le64_to_cpu(s->bytenr))
 		++fail;
 
-	if (sbio->spag[0].generation != le64_to_cpu(s->generation))
+	if (sblock->pagev[0].generation != le64_to_cpu(s->generation))
 		++fail;
 
 	if (memcmp(s->fsid, fs_info->fsid, BTRFS_UUID_SIZE))
 		++fail;
 
-	crc = btrfs_csum_data(root, buffer + BTRFS_CSUM_SIZE, crc,
-			      PAGE_SIZE - BTRFS_CSUM_SIZE);
-	btrfs_csum_final(crc, csum);
-	if (memcmp(csum, s->csum, sbio->sdev->csum_size))
+	len = BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE;
+	mapped_size = PAGE_SIZE - BTRFS_CSUM_SIZE;
+	p = ((u8 *)mapped_buffer) + BTRFS_CSUM_SIZE;
+	index = 0;
+	for (;;) {
+		u64 l = min_t(u64, len, mapped_size);
+
+		crc = btrfs_csum_data(root, p, crc, l);
+		kunmap_atomic(mapped_buffer);
+		len -= l;
+		if (len == 0)
+			break;
+		index++;
+		BUG_ON(index >= sblock->page_count);
+		BUG_ON(!sblock->pagev[index].page);
+		page = sblock->pagev[index].page;
+		mapped_buffer = kmap_atomic(page);
+		mapped_size = PAGE_SIZE;
+		p = mapped_buffer;
+	}
+
+	btrfs_csum_final(crc, calculated_csum);
+	if (memcmp(calculated_csum, on_disk_csum, sdev->csum_size))
 		++fail;
 
 	if (fail) {
@@ -948,29 +1414,42 @@ static int scrub_checksum_super(struct scrub_bio *sbio, void *buffer)
 	return fail;
 }
 
-static int scrub_submit(struct scrub_dev *sdev)
+static void scrub_block_get(struct scrub_block *sblock)
+{
+	atomic_inc(&sblock->ref_count);
+}
+
+static void scrub_block_put(struct scrub_block *sblock)
+{
+	if (atomic_dec_and_test(&sblock->ref_count)) {
+		int i;
+
+		for (i = 0; i < sblock->page_count; i++)
+			if (sblock->pagev[i].page)
+				__free_page(sblock->pagev[i].page);
+		kfree(sblock);
+	}
+}
+
+static void scrub_submit(struct scrub_dev *sdev)
 {
 	struct scrub_bio *sbio;
 
 	if (sdev->curr == -1)
-		return 0;
+		return;
 
 	sbio = sdev->bios[sdev->curr];
-	sbio->err = 0;
 	sdev->curr = -1;
 	atomic_inc(&sdev->in_flight);
 
 	btrfsic_submit_bio(READ, sbio->bio);
-
-	return 0;
 }
 
-static int scrub_page(struct scrub_dev *sdev, u64 logical, u64 len,
-		      u64 physical, u64 flags, u64 gen, int mirror_num,
-		      u8 *csum, int force)
+static int scrub_add_page_to_bio(struct scrub_dev *sdev,
+				 struct scrub_page *spage)
 {
+	struct scrub_block *sblock = spage->sblock;
 	struct scrub_bio *sbio;
-	struct page *page;
 	int ret;
 
 again:
@@ -983,7 +1462,7 @@ again:
 		if (sdev->curr != -1) {
 			sdev->first_free = sdev->bios[sdev->curr]->next_free;
 			sdev->bios[sdev->curr]->next_free = -1;
-			sdev->bios[sdev->curr]->count = 0;
+			sdev->bios[sdev->curr]->page_count = 0;
 			spin_unlock(&sdev->list_lock);
 		} else {
 			spin_unlock(&sdev->list_lock);
@@ -991,62 +1470,200 @@ again:
 		}
 	}
 	sbio = sdev->bios[sdev->curr];
-	if (sbio->count == 0) {
+	if (sbio->page_count == 0) {
 		struct bio *bio;
 
-		sbio->physical = physical;
-		sbio->logical = logical;
-		bio = bio_alloc(GFP_NOFS, SCRUB_PAGES_PER_BIO);
-		if (!bio)
-			return -ENOMEM;
+		sbio->physical = spage->physical;
+		sbio->logical = spage->logical;
+		bio = sbio->bio;
+		if (!bio) {
+			bio = bio_alloc(GFP_NOFS, sdev->pages_per_bio);
+			if (!bio)
+				return -ENOMEM;
+			sbio->bio = bio;
+		}
 
 		bio->bi_private = sbio;
 		bio->bi_end_io = scrub_bio_end_io;
 		bio->bi_bdev = sdev->dev->bdev;
-		bio->bi_sector = sbio->physical >> 9;
+		bio->bi_sector = spage->physical >> 9;
 		sbio->err = 0;
-		sbio->bio = bio;
-	} else if (sbio->physical + sbio->count * PAGE_SIZE != physical ||
-		   sbio->logical + sbio->count * PAGE_SIZE != logical) {
-		ret = scrub_submit(sdev);
-		if (ret)
-			return ret;
+	} else if (sbio->physical + sbio->page_count * PAGE_SIZE !=
+		   spage->physical ||
+		   sbio->logical + sbio->page_count * PAGE_SIZE !=
+		   spage->logical) {
+		scrub_submit(sdev);
 		goto again;
 	}
-	sbio->spag[sbio->count].flags = flags;
-	sbio->spag[sbio->count].generation = gen;
-	sbio->spag[sbio->count].have_csum = 0;
-	sbio->spag[sbio->count].mirror_num = mirror_num;
-
-	page = alloc_page(GFP_NOFS);
-	if (!page)
-		return -ENOMEM;
 
-	ret = bio_add_page(sbio->bio, page, PAGE_SIZE, 0);
-	if (!ret) {
-		__free_page(page);
-		ret = scrub_submit(sdev);
-		if (ret)
-			return ret;
+	sbio->pagev[sbio->page_count] = spage;
+	ret = bio_add_page(sbio->bio, spage->page, PAGE_SIZE, 0);
+	if (ret != PAGE_SIZE) {
+		if (sbio->page_count < 1) {
+			bio_put(sbio->bio);
+			sbio->bio = NULL;
+			return -EIO;
+		}
+		scrub_submit(sdev);
 		goto again;
 	}
 
-	if (csum) {
-		sbio->spag[sbio->count].have_csum = 1;
-		memcpy(sbio->spag[sbio->count].csum, csum, sdev->csum_size);
+	scrub_block_get(sblock); /* one for the added page */
+	atomic_inc(&sblock->outstanding_pages);
+	sbio->page_count++;
+	if (sbio->page_count == sdev->pages_per_bio)
+		scrub_submit(sdev);
+
+	return 0;
+}
+
+static int scrub_pages(struct scrub_dev *sdev, u64 logical, u64 len,
+		       u64 physical, u64 flags, u64 gen, int mirror_num,
+		       u8 *csum, int force)
+{
+	struct scrub_block *sblock;
+	int index;
+
+	sblock = kzalloc(sizeof(*sblock), GFP_NOFS);
+	if (!sblock) {
+		spin_lock(&sdev->stat_lock);
+		sdev->stat.malloc_errors++;
+		spin_unlock(&sdev->stat_lock);
+		return -ENOMEM;
 	}
-	++sbio->count;
-	if (sbio->count == SCRUB_PAGES_PER_BIO || force) {
+
+	/* one ref inside this function, plus one for each page later on */
+	atomic_set(&sblock->ref_count, 1);
+	sblock->sdev = sdev;
+	sblock->no_io_error_seen = 1;
+
+	for (index = 0; len > 0; index++) {
+		struct scrub_page *spage = sblock->pagev + index;
+		u64 l = min_t(u64, len, PAGE_SIZE);
+
+		BUG_ON(index >= SCRUB_MAX_PAGES_PER_BLOCK);
+		spage->page = alloc_page(GFP_NOFS);
+		if (!spage->page) {
+			spin_lock(&sdev->stat_lock);
+			sdev->stat.malloc_errors++;
+			spin_unlock(&sdev->stat_lock);
+			while (index > 0) {
+				index--;
+				__free_page(sblock->pagev[index].page);
+			}
+			kfree(sblock);
+			return -ENOMEM;
+		}
+		spage->sblock = sblock;
+		spage->bdev = sdev->dev->bdev;
+		spage->flags = flags;
+		spage->generation = gen;
+		spage->logical = logical;
+		spage->physical = physical;
+		spage->mirror_num = mirror_num;
+		if (csum) {
+			spage->have_csum = 1;
+			memcpy(spage->csum, csum, sdev->csum_size);
+		} else {
+			spage->have_csum = 0;
+		}
+		sblock->page_count++;
+		len -= l;
+		logical += l;
+		physical += l;
+	}
+
+	BUG_ON(sblock->page_count == 0);
+	for (index = 0; index < sblock->page_count; index++) {
+		struct scrub_page *spage = sblock->pagev + index;
 		int ret;
 
-		ret = scrub_submit(sdev);
-		if (ret)
+		ret = scrub_add_page_to_bio(sdev, spage);
+		if (ret) {
+			scrub_block_put(sblock);
 			return ret;
+		}
 	}
 
+	if (force)
+		scrub_submit(sdev);
+
+	/* last one frees, either here or in bio completion for last page */
+	scrub_block_put(sblock);
 	return 0;
 }
 
+static void scrub_bio_end_io(struct bio *bio, int err)
+{
+	struct scrub_bio *sbio = bio->bi_private;
+	struct scrub_dev *sdev = sbio->sdev;
+	struct btrfs_fs_info *fs_info = sdev->dev->dev_root->fs_info;
+
+	sbio->err = err;
+	sbio->bio = bio;
+
+	btrfs_queue_worker(&fs_info->scrub_workers, &sbio->work);
+}
+
+static void scrub_bio_end_io_worker(struct btrfs_work *work)
+{
+	struct scrub_bio *sbio = container_of(work, struct scrub_bio, work);
+	struct scrub_dev *sdev = sbio->sdev;
+	int i;
+
+	BUG_ON(sbio->page_count > SCRUB_PAGES_PER_BIO);
+	if (sbio->err) {
+		for (i = 0; i < sbio->page_count; i++) {
+			struct scrub_page *spage = sbio->pagev[i];
+
+			spage->io_error = 1;
+			spage->sblock->no_io_error_seen = 0;
+		}
+	}
+
+	/* now complete the scrub_block items that have all pages completed */
+	for (i = 0; i < sbio->page_count; i++) {
+		struct scrub_page *spage = sbio->pagev[i];
+		struct scrub_block *sblock = spage->sblock;
+
+		if (atomic_dec_and_test(&sblock->outstanding_pages))
+			scrub_block_complete(sblock);
+		scrub_block_put(sblock);
+	}
+
+	if (sbio->err) {
+		/* what is this good for??? */
+		sbio->bio->bi_flags &= ~(BIO_POOL_MASK - 1);
+		sbio->bio->bi_flags |= 1 << BIO_UPTODATE;
+		sbio->bio->bi_phys_segments = 0;
+		sbio->bio->bi_idx = 0;
+
+		for (i = 0; i < sbio->page_count; i++) {
+			struct bio_vec *bi;
+			bi = &sbio->bio->bi_io_vec[i];
+			bi->bv_offset = 0;
+			bi->bv_len = PAGE_SIZE;
+		}
+	}
+
+	bio_put(sbio->bio);
+	sbio->bio = NULL;
+	spin_lock(&sdev->list_lock);
+	sbio->next_free = sdev->first_free;
+	sdev->first_free = sbio->index;
+	spin_unlock(&sdev->list_lock);
+	atomic_dec(&sdev->in_flight);
+	wake_up(&sdev->list_wait);
+}
+
+static void scrub_block_complete(struct scrub_block *sblock)
+{
+	if (!sblock->no_io_error_seen)
+		scrub_handle_errored_block(sblock);
+	else
+		scrub_checksum(sblock);
+}
+
 static int scrub_find_csum(struct scrub_dev *sdev, u64 logical, u64 len,
 			   u8 *csum)
 {
@@ -1054,7 +1671,6 @@ static int scrub_find_csum(struct scrub_dev *sdev, u64 logical, u64 len,
 	int ret = 0;
 	unsigned long i;
 	unsigned long num_sectors;
-	u32 sectorsize = sdev->dev->dev_root->sectorsize;
 
 	while (!list_empty(&sdev->csum_list)) {
 		sum = list_first_entry(&sdev->csum_list,
@@ -1072,7 +1688,7 @@ static int scrub_find_csum(struct scrub_dev *sdev, u64 logical, u64 len,
 	if (!sum)
 		return 0;
 
-	num_sectors = sum->len / sectorsize;
+	num_sectors = sum->len / sdev->sectorsize;
 	for (i = 0; i < num_sectors; ++i) {
 		if (sum->sums[i].bytenr == logical) {
 			memcpy(csum, &sum->sums[i].sum, sdev->csum_size);
@@ -1093,9 +1709,28 @@ static int scrub_extent(struct scrub_dev *sdev, u64 logical, u64 len,
 {
 	int ret;
 	u8 csum[BTRFS_CSUM_SIZE];
+	u32 blocksize;
+
+	if (flags & BTRFS_EXTENT_FLAG_DATA) {
+		blocksize = sdev->sectorsize;
+		spin_lock(&sdev->stat_lock);
+		sdev->stat.data_extents_scrubbed++;
+		sdev->stat.data_bytes_scrubbed += len;
+		spin_unlock(&sdev->stat_lock);
+	} else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
+		BUG_ON(sdev->nodesize != sdev->leafsize);
+		blocksize = sdev->nodesize;
+		spin_lock(&sdev->stat_lock);
+		sdev->stat.tree_extents_scrubbed++;
+		sdev->stat.tree_bytes_scrubbed += len;
+		spin_unlock(&sdev->stat_lock);
+	} else {
+		blocksize = sdev->sectorsize;
+		BUG_ON(1);
+	}
 
 	while (len) {
-		u64 l = min_t(u64, len, PAGE_SIZE);
+		u64 l = min_t(u64, len, blocksize);
 		int have_csum = 0;
 
 		if (flags & BTRFS_EXTENT_FLAG_DATA) {
@@ -1104,8 +1739,8 @@ static int scrub_extent(struct scrub_dev *sdev, u64 logical, u64 len,
 			if (have_csum == 0)
 				++sdev->stat.no_csum;
 		}
-		ret = scrub_page(sdev, logical, l, physical, flags, gen,
-				 mirror_num, have_csum ? csum : NULL, 0);
+		ret = scrub_pages(sdev, logical, l, physical, flags, gen,
+				  mirror_num, have_csum ? csum : NULL, 0);
 		if (ret)
 			return ret;
 		len -= l;
@@ -1170,6 +1805,11 @@ static noinline_for_stack int scrub_stripe(struct scrub_dev *sdev,
 	if (!path)
 		return -ENOMEM;
 
+	/*
+	 * work on commit root. The related disk blocks are static as
+	 * long as COW is applied. This means, it is save to rewrite
+	 * them to repair disk errors without any race conditions
+	 */
 	path->search_commit_root = 1;
 	path->skip_locking = 1;
 
@@ -1516,15 +2156,18 @@ static noinline_for_stack int scrub_supers(struct scrub_dev *sdev)
 	struct btrfs_device *device = sdev->dev;
 	struct btrfs_root *root = device->dev_root;
 
+	if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR)
+		return -EIO;
+
 	gen = root->fs_info->last_trans_committed;
 
 	for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
 		bytenr = btrfs_sb_offset(i);
-		if (bytenr + BTRFS_SUPER_INFO_SIZE >= device->total_bytes)
+		if (bytenr + BTRFS_SUPER_INFO_SIZE > device->total_bytes)
 			break;
 
-		ret = scrub_page(sdev, bytenr, PAGE_SIZE, bytenr,
-				 BTRFS_EXTENT_FLAG_SUPER, gen, i, NULL, 1);
+		ret = scrub_pages(sdev, bytenr, BTRFS_SUPER_INFO_SIZE, bytenr,
+				     BTRFS_EXTENT_FLAG_SUPER, gen, i, NULL, 1);
 		if (ret)
 			return ret;
 	}
@@ -1583,10 +2226,30 @@ int btrfs_scrub_dev(struct btrfs_root *root, u64 devid, u64 start, u64 end,
 	/*
 	 * check some assumptions
 	 */
-	if (root->sectorsize != PAGE_SIZE ||
-	    root->sectorsize != root->leafsize ||
-	    root->sectorsize != root->nodesize) {
-		printk(KERN_ERR "btrfs_scrub: size assumptions fail\n");
+	if (root->nodesize != root->leafsize) {
+		printk(KERN_ERR
+		       "btrfs_scrub: size assumption nodesize == leafsize (%d == %d) fails\n",
+		       root->nodesize, root->leafsize);
+		return -EINVAL;
+	}
+
+	if (root->nodesize > BTRFS_STRIPE_LEN) {
+		/*
+		 * in this case scrub is unable to calculate the checksum
+		 * the way scrub is implemented. Do not handle this
+		 * situation at all because it won't ever happen.
+		 */
+		printk(KERN_ERR
+		       "btrfs_scrub: size assumption nodesize <= BTRFS_STRIPE_LEN (%d <= %d) fails\n",
+		       root->nodesize, BTRFS_STRIPE_LEN);
+		return -EINVAL;
+	}
+
+	if (root->sectorsize != PAGE_SIZE) {
+		/* not supported for data w/o checksums */
+		printk(KERN_ERR
+		       "btrfs_scrub: size assumption sectorsize != PAGE_SIZE (%d != %lld) fails\n",
+		       root->sectorsize, (unsigned long long)PAGE_SIZE);
 		return -EINVAL;
 	}
 
@@ -1656,7 +2319,7 @@ int btrfs_scrub_dev(struct btrfs_root *root, u64 devid, u64 start, u64 end,
 	return ret;
 }
 
-int btrfs_scrub_pause(struct btrfs_root *root)
+void btrfs_scrub_pause(struct btrfs_root *root)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
@@ -1671,34 +2334,28 @@ int btrfs_scrub_pause(struct btrfs_root *root)
 		mutex_lock(&fs_info->scrub_lock);
 	}
 	mutex_unlock(&fs_info->scrub_lock);
-
-	return 0;
 }
 
-int btrfs_scrub_continue(struct btrfs_root *root)
+void btrfs_scrub_continue(struct btrfs_root *root)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
 	atomic_dec(&fs_info->scrub_pause_req);
 	wake_up(&fs_info->scrub_pause_wait);
-	return 0;
 }
 
-int btrfs_scrub_pause_super(struct btrfs_root *root)
+void btrfs_scrub_pause_super(struct btrfs_root *root)
 {
 	down_write(&root->fs_info->scrub_super_lock);
-	return 0;
 }
 
-int btrfs_scrub_continue_super(struct btrfs_root *root)
+void btrfs_scrub_continue_super(struct btrfs_root *root)
 {
 	up_write(&root->fs_info->scrub_super_lock);
-	return 0;
 }
 
-int btrfs_scrub_cancel(struct btrfs_root *root)
+int __btrfs_scrub_cancel(struct btrfs_fs_info *fs_info)
 {
-	struct btrfs_fs_info *fs_info = root->fs_info;
 
 	mutex_lock(&fs_info->scrub_lock);
 	if (!atomic_read(&fs_info->scrubs_running)) {
@@ -1719,6 +2376,11 @@ int btrfs_scrub_cancel(struct btrfs_root *root)
 	return 0;
 }
 
+int btrfs_scrub_cancel(struct btrfs_root *root)
+{
+	return __btrfs_scrub_cancel(root->fs_info);
+}
+
 int btrfs_scrub_cancel_dev(struct btrfs_root *root, struct btrfs_device *dev)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -1741,6 +2403,7 @@ int btrfs_scrub_cancel_dev(struct btrfs_root *root, struct btrfs_device *dev)
 
 	return 0;
 }
+
 int btrfs_scrub_cancel_devid(struct btrfs_root *root, u64 devid)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
diff --git a/fs/btrfs/struct-funcs.c b/fs/btrfs/struct-funcs.c
index bc1f6ad1844..c6ffa581241 100644
--- a/fs/btrfs/struct-funcs.c
+++ b/fs/btrfs/struct-funcs.c
@@ -44,8 +44,9 @@
 #define BTRFS_SETGET_FUNCS(name, type, member, bits)			\
 u##bits btrfs_##name(struct extent_buffer *eb, type *s);		\
 void btrfs_set_##name(struct extent_buffer *eb, type *s, u##bits val);	\
-u##bits btrfs_##name(struct extent_buffer *eb,				\
-				   type *s)				\
+void btrfs_set_token_##name(struct extent_buffer *eb, type *s, u##bits val, struct btrfs_map_token *token);	\
+u##bits btrfs_token_##name(struct extent_buffer *eb,				\
+			   type *s, struct btrfs_map_token *token)	\
 {									\
 	unsigned long part_offset = (unsigned long)s;			\
 	unsigned long offset = part_offset + offsetof(type, member);	\
@@ -54,9 +55,18 @@ u##bits btrfs_##name(struct extent_buffer *eb,				\
 	char *kaddr;						\
 	unsigned long map_start;				\
 	unsigned long map_len;					\
+	unsigned long mem_len = sizeof(((type *)0)->member);	\
 	u##bits res;						\
+	if (token && token->kaddr && token->offset <= offset &&	\
+	    token->eb == eb &&					\
+	   (token->offset + PAGE_CACHE_SIZE >= offset + mem_len)) { \
+		kaddr = token->kaddr;				\
+		p = (type *)(kaddr + part_offset - token->offset);  \
+		res = le##bits##_to_cpu(p->member);		\
+		return res;					\
+	}							\
 	err = map_private_extent_buffer(eb, offset,		\
-			sizeof(((type *)0)->member),		\
+			mem_len,				\
 			&kaddr, &map_start, &map_len);		\
 	if (err) {						\
 		__le##bits leres;				\
@@ -65,10 +75,15 @@ u##bits btrfs_##name(struct extent_buffer *eb,				\
 	}							\
 	p = (type *)(kaddr + part_offset - map_start);		\
 	res = le##bits##_to_cpu(p->member);			\
+	if (token) {						\
+		token->kaddr = kaddr;				\
+		token->offset = map_start;			\
+		token->eb = eb;					\
+	}							\
 	return res;						\
 }									\
-void btrfs_set_##name(struct extent_buffer *eb,				\
-				    type *s, u##bits val)		\
+void btrfs_set_token_##name(struct extent_buffer *eb,				\
+			    type *s, u##bits val, struct btrfs_map_token *token)		\
 {									\
 	unsigned long part_offset = (unsigned long)s;			\
 	unsigned long offset = part_offset + offsetof(type, member);	\
@@ -77,8 +92,17 @@ void btrfs_set_##name(struct extent_buffer *eb,				\
 	char *kaddr;						\
 	unsigned long map_start;				\
 	unsigned long map_len;					\
+	unsigned long mem_len = sizeof(((type *)0)->member);	\
+	if (token && token->kaddr && token->offset <= offset &&	\
+	    token->eb == eb &&					\
+	   (token->offset + PAGE_CACHE_SIZE >= offset + mem_len)) { \
+		kaddr = token->kaddr;				\
+		p = (type *)(kaddr + part_offset - token->offset);  \
+		p->member = cpu_to_le##bits(val);		\
+		return;						\
+	}							\
 	err = map_private_extent_buffer(eb, offset,		\
-			sizeof(((type *)0)->member),		\
+			mem_len,				\
 			&kaddr, &map_start, &map_len);		\
 	if (err) {						\
 		__le##bits val2;				\
@@ -88,7 +112,22 @@ void btrfs_set_##name(struct extent_buffer *eb,				\
 	}							\
 	p = (type *)(kaddr + part_offset - map_start);		\
 	p->member = cpu_to_le##bits(val);			\
-}
+	if (token) {						\
+		token->kaddr = kaddr;				\
+		token->offset = map_start;			\
+		token->eb = eb;					\
+	}							\
+}								\
+void btrfs_set_##name(struct extent_buffer *eb,			\
+		      type *s, u##bits val)			\
+{								\
+	btrfs_set_token_##name(eb, s, val, NULL);		\
+}								\
+u##bits btrfs_##name(struct extent_buffer *eb,			\
+		      type *s)					\
+{								\
+	return btrfs_token_##name(eb, s, NULL);			\
+}								\
 
 #include "ctree.h"
 
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 81df3fec6a6..8d5d380f7bd 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -76,6 +76,9 @@ static const char *btrfs_decode_error(struct btrfs_fs_info *fs_info, int errno,
 	case -EROFS:
 		errstr = "Readonly filesystem";
 		break;
+	case -EEXIST:
+		errstr = "Object already exists";
+		break;
 	default:
 		if (nbuf) {
 			if (snprintf(nbuf, 16, "error %d", -errno) >= 0)
@@ -116,6 +119,8 @@ static void btrfs_handle_error(struct btrfs_fs_info *fs_info)
 	if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) {
 		sb->s_flags |= MS_RDONLY;
 		printk(KERN_INFO "btrfs is forced readonly\n");
+		__btrfs_scrub_cancel(fs_info);
+//		WARN_ON(1);
 	}
 }
 
@@ -124,25 +129,132 @@ static void btrfs_handle_error(struct btrfs_fs_info *fs_info)
  * invokes the approciate error response.
  */
 void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
-		     unsigned int line, int errno)
+		       unsigned int line, int errno, const char *fmt, ...)
 {
 	struct super_block *sb = fs_info->sb;
 	char nbuf[16];
 	const char *errstr;
+	va_list args;
+	va_start(args, fmt);
 
 	/*
 	 * Special case: if the error is EROFS, and we're already
 	 * under MS_RDONLY, then it is safe here.
 	 */
 	if (errno == -EROFS && (sb->s_flags & MS_RDONLY))
+  		return;
+
+  	errstr = btrfs_decode_error(fs_info, errno, nbuf);
+	if (fmt) {
+		struct va_format vaf = {
+			.fmt = fmt,
+			.va = &args,
+		};
+
+		printk(KERN_CRIT "BTRFS error (device %s) in %s:%d: %s (%pV)\n",
+			sb->s_id, function, line, errstr, &vaf);
+	} else {
+		printk(KERN_CRIT "BTRFS error (device %s) in %s:%d: %s\n",
+			sb->s_id, function, line, errstr);
+	}
+
+	/* Don't go through full error handling during mount */
+	if (sb->s_flags & MS_BORN) {
+		save_error_info(fs_info);
+		btrfs_handle_error(fs_info);
+	}
+	va_end(args);
+}
+
+const char *logtypes[] = {
+	"emergency",
+	"alert",
+	"critical",
+	"error",
+	"warning",
+	"notice",
+	"info",
+	"debug",
+};
+
+void btrfs_printk(struct btrfs_fs_info *fs_info, const char *fmt, ...)
+{
+	struct super_block *sb = fs_info->sb;
+	char lvl[4];
+	struct va_format vaf;
+	va_list args;
+	const char *type = logtypes[4];
+
+	va_start(args, fmt);
+
+	if (fmt[0] == '<' && isdigit(fmt[1]) && fmt[2] == '>') {
+		strncpy(lvl, fmt, 3);
+		fmt += 3;
+		type = logtypes[fmt[1] - '0'];
+	} else
+		*lvl = '\0';
+
+	vaf.fmt = fmt;
+	vaf.va = &args;
+	printk("%sBTRFS %s (device %s): %pV", lvl, type, sb->s_id, &vaf);
+}
+
+/*
+ * We only mark the transaction aborted and then set the file system read-only.
+ * This will prevent new transactions from starting or trying to join this
+ * one.
+ *
+ * This means that error recovery at the call site is limited to freeing
+ * any local memory allocations and passing the error code up without
+ * further cleanup. The transaction should complete as it normally would
+ * in the call path but will return -EIO.
+ *
+ * We'll complete the cleanup in btrfs_end_transaction and
+ * btrfs_commit_transaction.
+ */
+void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
+			       struct btrfs_root *root, const char *function,
+			       unsigned int line, int errno)
+{
+	WARN_ONCE(1, KERN_DEBUG "btrfs: Transaction aborted");
+	trans->aborted = errno;
+	/* Nothing used. The other threads that have joined this
+	 * transaction may be able to continue. */
+	if (!trans->blocks_used) {
+		btrfs_printk(root->fs_info, "Aborting unused transaction.\n");
 		return;
+	}
+	trans->transaction->aborted = errno;
+	__btrfs_std_error(root->fs_info, function, line, errno, NULL);
+}
+/*
+ * __btrfs_panic decodes unexpected, fatal errors from the caller,
+ * issues an alert, and either panics or BUGs, depending on mount options.
+ */
+void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function,
+		   unsigned int line, int errno, const char *fmt, ...)
+{
+	char nbuf[16];
+	char *s_id = "<unknown>";
+	const char *errstr;
+	struct va_format vaf = { .fmt = fmt };
+	va_list args;
 
-	errstr = btrfs_decode_error(fs_info, errno, nbuf);
-	printk(KERN_CRIT "BTRFS error (device %s) in %s:%d: %s\n",
-		sb->s_id, function, line, errstr);
-	save_error_info(fs_info);
+	if (fs_info)
+		s_id = fs_info->sb->s_id;
 
-	btrfs_handle_error(fs_info);
+	va_start(args, fmt);
+	vaf.va = &args;
+
+	errstr = btrfs_decode_error(fs_info, errno, nbuf);
+	if (fs_info->mount_opt & BTRFS_MOUNT_PANIC_ON_FATAL_ERROR)
+		panic(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (%s)\n",
+			s_id, function, line, &vaf, errstr);
+
+	printk(KERN_CRIT "BTRFS panic (device %s) in %s:%d: %pV (%s)\n",
+	       s_id, function, line, &vaf, errstr);
+	va_end(args);
+	/* Caller calls BUG() */
 }
 
 static void btrfs_put_super(struct super_block *sb)
@@ -166,7 +278,7 @@ enum {
 	Opt_enospc_debug, Opt_subvolrootid, Opt_defrag, Opt_inode_cache,
 	Opt_no_space_cache, Opt_recovery, Opt_skip_balance,
 	Opt_check_integrity, Opt_check_integrity_including_extent_data,
-	Opt_check_integrity_print_mask,
+	Opt_check_integrity_print_mask, Opt_fatal_errors,
 	Opt_err,
 };
 
@@ -206,12 +318,14 @@ static match_table_t tokens = {
 	{Opt_check_integrity, "check_int"},
 	{Opt_check_integrity_including_extent_data, "check_int_data"},
 	{Opt_check_integrity_print_mask, "check_int_print_mask=%d"},
+	{Opt_fatal_errors, "fatal_errors=%s"},
 	{Opt_err, NULL},
 };
 
 /*
  * Regular mount options parser.  Everything that is needed only when
  * reading in a new superblock is parsed here.
+ * XXX JDM: This needs to be cleaned up for remount.
  */
 int btrfs_parse_options(struct btrfs_root *root, char *options)
 {
@@ -438,6 +552,18 @@ int btrfs_parse_options(struct btrfs_root *root, char *options)
 			ret = -EINVAL;
 			goto out;
 #endif
+		case Opt_fatal_errors:
+			if (strcmp(args[0].from, "panic") == 0)
+				btrfs_set_opt(info->mount_opt,
+					      PANIC_ON_FATAL_ERROR);
+			else if (strcmp(args[0].from, "bug") == 0)
+				btrfs_clear_opt(info->mount_opt,
+					      PANIC_ON_FATAL_ERROR);
+			else {
+				ret = -EINVAL;
+				goto out;
+			}
+			break;
 		case Opt_err:
 			printk(KERN_INFO "btrfs: unrecognized mount option "
 			       "'%s'\n", p);
@@ -762,6 +888,8 @@ static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
 		seq_puts(seq, ",inode_cache");
 	if (btrfs_test_opt(root, SKIP_BALANCE))
 		seq_puts(seq, ",skip_balance");
+	if (btrfs_test_opt(root, PANIC_ON_FATAL_ERROR))
+		seq_puts(seq, ",fatal_errors=panic");
 	return 0;
 }
 
@@ -995,11 +1123,20 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
 	struct btrfs_root *root = fs_info->tree_root;
+	unsigned old_flags = sb->s_flags;
+	unsigned long old_opts = fs_info->mount_opt;
+	unsigned long old_compress_type = fs_info->compress_type;
+	u64 old_max_inline = fs_info->max_inline;
+	u64 old_alloc_start = fs_info->alloc_start;
+	int old_thread_pool_size = fs_info->thread_pool_size;
+	unsigned int old_metadata_ratio = fs_info->metadata_ratio;
 	int ret;
 
 	ret = btrfs_parse_options(root, data);
-	if (ret)
-		return -EINVAL;
+	if (ret) {
+		ret = -EINVAL;
+		goto restore;
+	}
 
 	if ((*flags & MS_RDONLY) == (sb->s_flags & MS_RDONLY))
 		return 0;
@@ -1007,26 +1144,44 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 	if (*flags & MS_RDONLY) {
 		sb->s_flags |= MS_RDONLY;
 
-		ret =  btrfs_commit_super(root);
-		WARN_ON(ret);
+		ret = btrfs_commit_super(root);
+		if (ret)
+			goto restore;
 	} else {
 		if (fs_info->fs_devices->rw_devices == 0)
-			return -EACCES;
+			ret = -EACCES;
+			goto restore;
 
 		if (btrfs_super_log_root(fs_info->super_copy) != 0)
-			return -EINVAL;
+			ret = -EINVAL;
+			goto restore;
 
 		ret = btrfs_cleanup_fs_roots(fs_info);
-		WARN_ON(ret);
+		if (ret)
+			goto restore;
 
 		/* recover relocation */
 		ret = btrfs_recover_relocation(root);
-		WARN_ON(ret);
+		if (ret)
+			goto restore;
 
 		sb->s_flags &= ~MS_RDONLY;
 	}
 
 	return 0;
+
+restore:
+	/* We've hit an error - don't reset MS_RDONLY */
+	if (sb->s_flags & MS_RDONLY)
+		old_flags |= MS_RDONLY;
+	sb->s_flags = old_flags;
+	fs_info->mount_opt = old_opts;
+	fs_info->compress_type = old_compress_type;
+	fs_info->max_inline = old_max_inline;
+	fs_info->alloc_start = old_alloc_start;
+	fs_info->thread_pool_size = old_thread_pool_size;
+	fs_info->metadata_ratio = old_metadata_ratio;
+	return ret;
 }
 
 /* Used to sort the devices by max_avail(descending sort) */
@@ -1356,9 +1511,7 @@ static int __init init_btrfs_fs(void)
 	if (err)
 		return err;
 
-	err = btrfs_init_compress();
-	if (err)
-		goto free_sysfs;
+	btrfs_init_compress();
 
 	err = btrfs_init_cachep();
 	if (err)
@@ -1384,6 +1537,8 @@ static int __init init_btrfs_fs(void)
 	if (err)
 		goto unregister_ioctl;
 
+	btrfs_init_lockdep();
+
 	printk(KERN_INFO "%s loaded\n", BTRFS_BUILD_VERSION);
 	return 0;
 
@@ -1399,7 +1554,6 @@ free_cachep:
 	btrfs_destroy_cachep();
 free_compress:
 	btrfs_exit_compress();
-free_sysfs:
 	btrfs_exit_sysfs();
 	return err;
 }
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 04b77e3ceb7..8da29e8e4de 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -31,7 +31,7 @@
 
 #define BTRFS_ROOT_TRANS_TAG 0
 
-static noinline void put_transaction(struct btrfs_transaction *transaction)
+void put_transaction(struct btrfs_transaction *transaction)
 {
 	WARN_ON(atomic_read(&transaction->use_count) == 0);
 	if (atomic_dec_and_test(&transaction->use_count)) {
@@ -58,6 +58,12 @@ static noinline int join_transaction(struct btrfs_root *root, int nofail)
 
 	spin_lock(&root->fs_info->trans_lock);
 loop:
+	/* The file system has been taken offline. No new transactions. */
+	if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) {
+		spin_unlock(&root->fs_info->trans_lock);
+		return -EROFS;
+	}
+
 	if (root->fs_info->trans_no_join) {
 		if (!nofail) {
 			spin_unlock(&root->fs_info->trans_lock);
@@ -67,6 +73,8 @@ loop:
 
 	cur_trans = root->fs_info->running_transaction;
 	if (cur_trans) {
+		if (cur_trans->aborted)
+			return cur_trans->aborted;
 		atomic_inc(&cur_trans->use_count);
 		atomic_inc(&cur_trans->num_writers);
 		cur_trans->num_joined++;
@@ -123,6 +131,7 @@ loop:
 	root->fs_info->generation++;
 	cur_trans->transid = root->fs_info->generation;
 	root->fs_info->running_transaction = cur_trans;
+	cur_trans->aborted = 0;
 	spin_unlock(&root->fs_info->trans_lock);
 
 	return 0;
@@ -318,6 +327,7 @@ again:
 	h->use_count = 1;
 	h->block_rsv = NULL;
 	h->orig_rsv = NULL;
+	h->aborted = 0;
 
 	smp_mb();
 	if (cur_trans->blocked && may_wait_transaction(root, type)) {
@@ -327,8 +337,7 @@ again:
 
 	if (num_bytes) {
 		trace_btrfs_space_reservation(root->fs_info, "transaction",
-					      (u64)(unsigned long)h,
-					      num_bytes, 1);
+					      h->transid, num_bytes, 1);
 		h->block_rsv = &root->fs_info->trans_block_rsv;
 		h->bytes_reserved = num_bytes;
 	}
@@ -440,6 +449,7 @@ int btrfs_should_end_transaction(struct btrfs_trans_handle *trans,
 	struct btrfs_transaction *cur_trans = trans->transaction;
 	struct btrfs_block_rsv *rsv = trans->block_rsv;
 	int updates;
+	int err;
 
 	smp_mb();
 	if (cur_trans->blocked || cur_trans->delayed_refs.flushing)
@@ -453,8 +463,11 @@ int btrfs_should_end_transaction(struct btrfs_trans_handle *trans,
 
 	updates = trans->delayed_ref_updates;
 	trans->delayed_ref_updates = 0;
-	if (updates)
-		btrfs_run_delayed_refs(trans, root, updates);
+	if (updates) {
+		err = btrfs_run_delayed_refs(trans, root, updates);
+		if (err) /* Error code will also eval true */
+			return err;
+	}
 
 	trans->block_rsv = rsv;
 
@@ -525,6 +538,11 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 	if (throttle)
 		btrfs_run_delayed_iputs(root);
 
+	if (trans->aborted ||
+	    root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) {
+		return -EIO;
+	}
+
 	return 0;
 }
 
@@ -690,11 +708,13 @@ static int update_cowonly_root(struct btrfs_trans_handle *trans,
 		ret = btrfs_update_root(trans, tree_root,
 					&root->root_key,
 					&root->root_item);
-		BUG_ON(ret);
+		if (ret)
+			return ret;
 
 		old_root_used = btrfs_root_used(&root->root_item);
 		ret = btrfs_write_dirty_block_groups(trans, root);
-		BUG_ON(ret);
+		if (ret)
+			return ret;
 	}
 
 	if (root != root->fs_info->extent_root)
@@ -705,6 +725,10 @@ static int update_cowonly_root(struct btrfs_trans_handle *trans,
 
 /*
  * update all the cowonly tree roots on disk
+ *
+ * The error handling in this function may not be obvious. Any of the
+ * failures will cause the file system to go offline. We still need
+ * to clean up the delayed refs.
  */
 static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans,
 					 struct btrfs_root *root)
@@ -715,22 +739,30 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans,
 	int ret;
 
 	ret = btrfs_run_delayed_refs(trans, root, (unsigned long)-1);
-	BUG_ON(ret);
+	if (ret)
+		return ret;
 
 	eb = btrfs_lock_root_node(fs_info->tree_root);
-	btrfs_cow_block(trans, fs_info->tree_root, eb, NULL, 0, &eb);
+	ret = btrfs_cow_block(trans, fs_info->tree_root, eb, NULL,
+			      0, &eb);
 	btrfs_tree_unlock(eb);
 	free_extent_buffer(eb);
 
+	if (ret)
+		return ret;
+
 	ret = btrfs_run_delayed_refs(trans, root, (unsigned long)-1);
-	BUG_ON(ret);
+	if (ret)
+		return ret;
 
 	while (!list_empty(&fs_info->dirty_cowonly_roots)) {
 		next = fs_info->dirty_cowonly_roots.next;
 		list_del_init(next);
 		root = list_entry(next, struct btrfs_root, dirty_list);
 
-		update_cowonly_root(trans, root);
+		ret = update_cowonly_root(trans, root);
+		if (ret)
+			return ret;
 	}
 
 	down_write(&fs_info->extent_commit_sem);
@@ -874,7 +906,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 
 	new_root_item = kmalloc(sizeof(*new_root_item), GFP_NOFS);
 	if (!new_root_item) {
-		pending->error = -ENOMEM;
+		ret = pending->error = -ENOMEM;
 		goto fail;
 	}
 
@@ -911,21 +943,24 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	 * insert the directory item
 	 */
 	ret = btrfs_set_inode_index(parent_inode, &index);
-	BUG_ON(ret);
+	BUG_ON(ret); /* -ENOMEM */
 	ret = btrfs_insert_dir_item(trans, parent_root,
 				dentry->d_name.name, dentry->d_name.len,
 				parent_inode, &key,
 				BTRFS_FT_DIR, index);
-	if (ret) {
+	if (ret == -EEXIST) {
 		pending->error = -EEXIST;
 		dput(parent);
 		goto fail;
+	} else if (ret) {
+		goto abort_trans_dput;
 	}
 
 	btrfs_i_size_write(parent_inode, parent_inode->i_size +
 					 dentry->d_name.len * 2);
 	ret = btrfs_update_inode(trans, parent_root, parent_inode);
-	BUG_ON(ret);
+	if (ret)
+		goto abort_trans_dput;
 
 	/*
 	 * pull in the delayed directory update
@@ -934,7 +969,10 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	 * snapshot
 	 */
 	ret = btrfs_run_delayed_items(trans, root);
-	BUG_ON(ret);
+	if (ret) { /* Transaction aborted */
+		dput(parent);
+		goto fail;
+	}
 
 	record_root_in_trans(trans, root);
 	btrfs_set_root_last_snapshot(&root->root_item, trans->transid);
@@ -949,12 +987,21 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	btrfs_set_root_flags(new_root_item, root_flags);
 
 	old = btrfs_lock_root_node(root);
-	btrfs_cow_block(trans, root, old, NULL, 0, &old);
+	ret = btrfs_cow_block(trans, root, old, NULL, 0, &old);
+	if (ret) {
+		btrfs_tree_unlock(old);
+		free_extent_buffer(old);
+		goto abort_trans_dput;
+	}
+
 	btrfs_set_lock_blocking(old);
 
-	btrfs_copy_root(trans, root, old, &tmp, objectid);
+	ret = btrfs_copy_root(trans, root, old, &tmp, objectid);
+	/* clean up in any case */
 	btrfs_tree_unlock(old);
 	free_extent_buffer(old);
+	if (ret)
+		goto abort_trans_dput;
 
 	/* see comments in should_cow_block() */
 	root->force_cow = 1;
@@ -966,7 +1013,8 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	ret = btrfs_insert_root(trans, tree_root, &key, new_root_item);
 	btrfs_tree_unlock(tmp);
 	free_extent_buffer(tmp);
-	BUG_ON(ret);
+	if (ret)
+		goto abort_trans_dput;
 
 	/*
 	 * insert root back/forward references
@@ -975,19 +1023,32 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 				 parent_root->root_key.objectid,
 				 btrfs_ino(parent_inode), index,
 				 dentry->d_name.name, dentry->d_name.len);
-	BUG_ON(ret);
 	dput(parent);
+	if (ret)
+		goto fail;
 
 	key.offset = (u64)-1;
 	pending->snap = btrfs_read_fs_root_no_name(root->fs_info, &key);
-	BUG_ON(IS_ERR(pending->snap));
+	if (IS_ERR(pending->snap)) {
+		ret = PTR_ERR(pending->snap);
+		goto abort_trans;
+	}
 
-	btrfs_reloc_post_snapshot(trans, pending);
+	ret = btrfs_reloc_post_snapshot(trans, pending);
+	if (ret)
+		goto abort_trans;
+	ret = 0;
 fail:
 	kfree(new_root_item);
 	trans->block_rsv = rsv;
 	btrfs_block_rsv_release(root, &pending->block_rsv, (u64)-1);
-	return 0;
+	return ret;
+
+abort_trans_dput:
+	dput(parent);
+abort_trans:
+	btrfs_abort_transaction(trans, root, ret);
+	goto fail;
 }
 
 /*
@@ -1124,6 +1185,33 @@ int btrfs_commit_transaction_async(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
+
+static void cleanup_transaction(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root)
+{
+	struct btrfs_transaction *cur_trans = trans->transaction;
+
+	WARN_ON(trans->use_count > 1);
+
+	spin_lock(&root->fs_info->trans_lock);
+	list_del_init(&cur_trans->list);
+	spin_unlock(&root->fs_info->trans_lock);
+
+	btrfs_cleanup_one_transaction(trans->transaction, root);
+
+	put_transaction(cur_trans);
+	put_transaction(cur_trans);
+
+	trace_btrfs_transaction_commit(root);
+
+	btrfs_scrub_continue(root);
+
+	if (current->journal_info == trans)
+		current->journal_info = NULL;
+
+	kmem_cache_free(btrfs_trans_handle_cachep, trans);
+}
+
 /*
  * btrfs_transaction state sequence:
  *    in_commit = 0, blocked = 0  (initial)
@@ -1135,10 +1223,10 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root)
 {
 	unsigned long joined = 0;
-	struct btrfs_transaction *cur_trans;
+	struct btrfs_transaction *cur_trans = trans->transaction;
 	struct btrfs_transaction *prev_trans = NULL;
 	DEFINE_WAIT(wait);
-	int ret;
+	int ret = -EIO;
 	int should_grow = 0;
 	unsigned long now = get_seconds();
 	int flush_on_commit = btrfs_test_opt(root, FLUSHONCOMMIT);
@@ -1148,13 +1236,18 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	btrfs_trans_release_metadata(trans, root);
 	trans->block_rsv = NULL;
 
+	if (cur_trans->aborted)
+		goto cleanup_transaction;
+
 	/* make a pass through all the delayed refs we have so far
 	 * any runnings procs may add more while we are here
 	 */
 	ret = btrfs_run_delayed_refs(trans, root, 0);
-	BUG_ON(ret);
+	if (ret)
+		goto cleanup_transaction;
 
 	cur_trans = trans->transaction;
+
 	/*
 	 * set the flushing flag so procs in this transaction have to
 	 * start sending their work down.
@@ -1162,19 +1255,20 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	cur_trans->delayed_refs.flushing = 1;
 
 	ret = btrfs_run_delayed_refs(trans, root, 0);
-	BUG_ON(ret);
+	if (ret)
+		goto cleanup_transaction;
 
 	spin_lock(&cur_trans->commit_lock);
 	if (cur_trans->in_commit) {
 		spin_unlock(&cur_trans->commit_lock);
 		atomic_inc(&cur_trans->use_count);
-		btrfs_end_transaction(trans, root);
+		ret = btrfs_end_transaction(trans, root);
 
 		wait_for_commit(root, cur_trans);
 
 		put_transaction(cur_trans);
 
-		return 0;
+		return ret;
 	}
 
 	trans->transaction->in_commit = 1;
@@ -1214,12 +1308,12 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 
 		if (flush_on_commit || snap_pending) {
 			btrfs_start_delalloc_inodes(root, 1);
-			ret = btrfs_wait_ordered_extents(root, 0, 1);
-			BUG_ON(ret);
+			btrfs_wait_ordered_extents(root, 0, 1);
 		}
 
 		ret = btrfs_run_delayed_items(trans, root);
-		BUG_ON(ret);
+		if (ret)
+			goto cleanup_transaction;
 
 		/*
 		 * rename don't use btrfs_join_transaction, so, once we
@@ -1261,13 +1355,22 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	mutex_lock(&root->fs_info->reloc_mutex);
 
 	ret = btrfs_run_delayed_items(trans, root);
-	BUG_ON(ret);
+	if (ret) {
+		mutex_unlock(&root->fs_info->reloc_mutex);
+		goto cleanup_transaction;
+	}
 
 	ret = create_pending_snapshots(trans, root->fs_info);
-	BUG_ON(ret);
+	if (ret) {
+		mutex_unlock(&root->fs_info->reloc_mutex);
+		goto cleanup_transaction;
+	}
 
 	ret = btrfs_run_delayed_refs(trans, root, (unsigned long)-1);
-	BUG_ON(ret);
+	if (ret) {
+		mutex_unlock(&root->fs_info->reloc_mutex);
+		goto cleanup_transaction;
+	}
 
 	/*
 	 * make sure none of the code above managed to slip in a
@@ -1294,7 +1397,10 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	mutex_lock(&root->fs_info->tree_log_mutex);
 
 	ret = commit_fs_roots(trans, root);
-	BUG_ON(ret);
+	if (ret) {
+		mutex_unlock(&root->fs_info->tree_log_mutex);
+		goto cleanup_transaction;
+	}
 
 	/* commit_fs_roots gets rid of all the tree log roots, it is now
 	 * safe to free the root of tree log roots
@@ -1302,7 +1408,10 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	btrfs_free_log_root_tree(trans, root->fs_info);
 
 	ret = commit_cowonly_roots(trans, root);
-	BUG_ON(ret);
+	if (ret) {
+		mutex_unlock(&root->fs_info->tree_log_mutex);
+		goto cleanup_transaction;
+	}
 
 	btrfs_prepare_extent_commit(trans, root);
 
@@ -1336,8 +1445,18 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	wake_up(&root->fs_info->transaction_wait);
 
 	ret = btrfs_write_and_wait_transaction(trans, root);
-	BUG_ON(ret);
-	write_ctree_super(trans, root, 0);
+	if (ret) {
+		btrfs_error(root->fs_info, ret,
+			    "Error while writing out transaction.");
+		mutex_unlock(&root->fs_info->tree_log_mutex);
+		goto cleanup_transaction;
+	}
+
+	ret = write_ctree_super(trans, root, 0);
+	if (ret) {
+		mutex_unlock(&root->fs_info->tree_log_mutex);
+		goto cleanup_transaction;
+	}
 
 	/*
 	 * the super is written, we can safely allow the tree-loggers
@@ -1373,6 +1492,15 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 		btrfs_run_delayed_iputs(root);
 
 	return ret;
+
+cleanup_transaction:
+	btrfs_printk(root->fs_info, "Skipping commit of aborted transaction.\n");
+//	WARN_ON(1);
+	if (current->journal_info == trans)
+		current->journal_info = NULL;
+	cleanup_transaction(trans, root);
+
+	return ret;
 }
 
 /*
@@ -1388,6 +1516,8 @@ int btrfs_clean_old_snapshots(struct btrfs_root *root)
 	spin_unlock(&fs_info->trans_lock);
 
 	while (!list_empty(&list)) {
+		int ret;
+
 		root = list_entry(list.next, struct btrfs_root, root_list);
 		list_del(&root->root_list);
 
@@ -1395,9 +1525,10 @@ int btrfs_clean_old_snapshots(struct btrfs_root *root)
 
 		if (btrfs_header_backref_rev(root->node) <
 		    BTRFS_MIXED_BACKREF_REV)
-			btrfs_drop_snapshot(root, NULL, 0, 0);
+			ret = btrfs_drop_snapshot(root, NULL, 0, 0);
 		else
-			btrfs_drop_snapshot(root, NULL, 1, 0);
+			ret =btrfs_drop_snapshot(root, NULL, 1, 0);
+		BUG_ON(ret < 0);
 	}
 	return 0;
 }
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index 02564e6230a..fe27379e368 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -43,6 +43,7 @@ struct btrfs_transaction {
 	wait_queue_head_t commit_wait;
 	struct list_head pending_snapshots;
 	struct btrfs_delayed_ref_root delayed_refs;
+	int aborted;
 };
 
 struct btrfs_trans_handle {
@@ -55,6 +56,7 @@ struct btrfs_trans_handle {
 	struct btrfs_transaction *transaction;
 	struct btrfs_block_rsv *block_rsv;
 	struct btrfs_block_rsv *orig_rsv;
+	int aborted;
 };
 
 struct btrfs_pending_snapshot {
@@ -114,4 +116,5 @@ int btrfs_wait_marked_extents(struct btrfs_root *root,
 				struct extent_io_tree *dirty_pages, int mark);
 int btrfs_transaction_blocked(struct btrfs_fs_info *info);
 int btrfs_transaction_in_commit(struct btrfs_fs_info *info);
+void put_transaction(struct btrfs_transaction *transaction);
 #endif
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index 966cc74f5d6..d017283ae6f 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -212,14 +212,13 @@ int btrfs_pin_log_trans(struct btrfs_root *root)
  * indicate we're done making changes to the log tree
  * and wake up anyone waiting to do a sync
  */
-int btrfs_end_log_trans(struct btrfs_root *root)
+void btrfs_end_log_trans(struct btrfs_root *root)
 {
 	if (atomic_dec_and_test(&root->log_writers)) {
 		smp_mb();
 		if (waitqueue_active(&root->log_writer_wait))
 			wake_up(&root->log_writer_wait);
 	}
-	return 0;
 }
 
 
@@ -378,12 +377,11 @@ insert:
 		u32 found_size;
 		found_size = btrfs_item_size_nr(path->nodes[0],
 						path->slots[0]);
-		if (found_size > item_size) {
+		if (found_size > item_size)
 			btrfs_truncate_item(trans, root, path, item_size, 1);
-		} else if (found_size < item_size) {
-			ret = btrfs_extend_item(trans, root, path,
-						item_size - found_size);
-		}
+		else if (found_size < item_size)
+			btrfs_extend_item(trans, root, path,
+					  item_size - found_size);
 	} else if (ret) {
 		return ret;
 	}
@@ -1763,7 +1761,7 @@ static noinline int walk_down_log_tree(struct btrfs_trans_handle *trans,
 					BTRFS_TREE_LOG_OBJECTID);
 				ret = btrfs_free_and_pin_reserved_extent(root,
 							 bytenr, blocksize);
-				BUG_ON(ret);
+				BUG_ON(ret); /* -ENOMEM or logic errors */
 			}
 			free_extent_buffer(next);
 			continue;
@@ -1871,20 +1869,26 @@ static int walk_log_tree(struct btrfs_trans_handle *trans,
 		wret = walk_down_log_tree(trans, log, path, &level, wc);
 		if (wret > 0)
 			break;
-		if (wret < 0)
+		if (wret < 0) {
 			ret = wret;
+			goto out;
+		}
 
 		wret = walk_up_log_tree(trans, log, path, &level, wc);
 		if (wret > 0)
 			break;
-		if (wret < 0)
+		if (wret < 0) {
 			ret = wret;
+			goto out;
+		}
 	}
 
 	/* was the root node processed? if not, catch it here */
 	if (path->nodes[orig_level]) {
-		wc->process_func(log, path->nodes[orig_level], wc,
+		ret = wc->process_func(log, path->nodes[orig_level], wc,
 			 btrfs_header_generation(path->nodes[orig_level]));
+		if (ret)
+			goto out;
 		if (wc->free) {
 			struct extent_buffer *next;
 
@@ -1900,10 +1904,11 @@ static int walk_log_tree(struct btrfs_trans_handle *trans,
 				BTRFS_TREE_LOG_OBJECTID);
 			ret = btrfs_free_and_pin_reserved_extent(log, next->start,
 							 next->len);
-			BUG_ON(ret);
+			BUG_ON(ret); /* -ENOMEM or logic errors */
 		}
 	}
 
+out:
 	for (i = 0; i <= orig_level; i++) {
 		if (path->nodes[i]) {
 			free_extent_buffer(path->nodes[i]);
@@ -1963,8 +1968,8 @@ static int wait_log_commit(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
-static int wait_for_writer(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root)
+static void wait_for_writer(struct btrfs_trans_handle *trans,
+			    struct btrfs_root *root)
 {
 	DEFINE_WAIT(wait);
 	while (root->fs_info->last_trans_log_full_commit !=
@@ -1978,7 +1983,6 @@ static int wait_for_writer(struct btrfs_trans_handle *trans,
 		mutex_lock(&root->log_mutex);
 		finish_wait(&root->log_writer_wait, &wait);
 	}
-	return 0;
 }
 
 /*
@@ -2046,7 +2050,11 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	 * wait for them until later.
 	 */
 	ret = btrfs_write_marked_extents(log, &log->dirty_log_pages, mark);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		mutex_unlock(&root->log_mutex);
+		goto out;
+	}
 
 	btrfs_set_root_node(&log->root_item, log->node);
 
@@ -2077,7 +2085,11 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	}
 
 	if (ret) {
-		BUG_ON(ret != -ENOSPC);
+		if (ret != -ENOSPC) {
+			btrfs_abort_transaction(trans, root, ret);
+			mutex_unlock(&log_root_tree->log_mutex);
+			goto out;
+		}
 		root->fs_info->last_trans_log_full_commit = trans->transid;
 		btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
 		mutex_unlock(&log_root_tree->log_mutex);
@@ -2117,7 +2129,11 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	ret = btrfs_write_and_wait_marked_extents(log_root_tree,
 				&log_root_tree->dirty_log_pages,
 				EXTENT_DIRTY | EXTENT_NEW);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		mutex_unlock(&log_root_tree->log_mutex);
+		goto out_wake_log_root;
+	}
 	btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
 
 	btrfs_set_super_log_root(root->fs_info->super_for_commit,
@@ -2326,7 +2342,9 @@ out_unlock:
 	if (ret == -ENOSPC) {
 		root->fs_info->last_trans_log_full_commit = trans->transid;
 		ret = 0;
-	}
+	} else if (ret < 0)
+		btrfs_abort_transaction(trans, root, ret);
+
 	btrfs_end_log_trans(root);
 
 	return err;
@@ -2357,7 +2375,8 @@ int btrfs_del_inode_ref_in_log(struct btrfs_trans_handle *trans,
 	if (ret == -ENOSPC) {
 		root->fs_info->last_trans_log_full_commit = trans->transid;
 		ret = 0;
-	}
+	} else if (ret < 0 && ret != -ENOENT)
+		btrfs_abort_transaction(trans, root, ret);
 	btrfs_end_log_trans(root);
 
 	return ret;
@@ -3169,13 +3188,20 @@ int btrfs_recover_log_trees(struct btrfs_root *log_root_tree)
 	fs_info->log_root_recovering = 1;
 
 	trans = btrfs_start_transaction(fs_info->tree_root, 0);
-	BUG_ON(IS_ERR(trans));
+	if (IS_ERR(trans)) {
+		ret = PTR_ERR(trans);
+		goto error;
+	}
 
 	wc.trans = trans;
 	wc.pin = 1;
 
 	ret = walk_log_tree(trans, log_root_tree, &wc);
-	BUG_ON(ret);
+	if (ret) {
+		btrfs_error(fs_info, ret, "Failed to pin buffers while "
+			    "recovering log root tree.");
+		goto error;
+	}
 
 again:
 	key.objectid = BTRFS_TREE_LOG_OBJECTID;
@@ -3184,8 +3210,12 @@ again:
 
 	while (1) {
 		ret = btrfs_search_slot(NULL, log_root_tree, &key, path, 0, 0);
-		if (ret < 0)
-			break;
+
+		if (ret < 0) {
+			btrfs_error(fs_info, ret,
+				    "Couldn't find tree log root.");
+			goto error;
+		}
 		if (ret > 0) {
 			if (path->slots[0] == 0)
 				break;
@@ -3199,14 +3229,24 @@ again:
 
 		log = btrfs_read_fs_root_no_radix(log_root_tree,
 						  &found_key);
-		BUG_ON(IS_ERR(log));
+		if (IS_ERR(log)) {
+			ret = PTR_ERR(log);
+			btrfs_error(fs_info, ret,
+				    "Couldn't read tree log root.");
+			goto error;
+		}
 
 		tmp_key.objectid = found_key.offset;
 		tmp_key.type = BTRFS_ROOT_ITEM_KEY;
 		tmp_key.offset = (u64)-1;
 
 		wc.replay_dest = btrfs_read_fs_root_no_name(fs_info, &tmp_key);
-		BUG_ON(IS_ERR_OR_NULL(wc.replay_dest));
+		if (IS_ERR(wc.replay_dest)) {
+			ret = PTR_ERR(wc.replay_dest);
+			btrfs_error(fs_info, ret, "Couldn't read target root "
+				    "for tree log recovery.");
+			goto error;
+		}
 
 		wc.replay_dest->log_root = log;
 		btrfs_record_root_in_trans(trans, wc.replay_dest);
@@ -3254,6 +3294,10 @@ again:
 
 	kfree(log_root_tree);
 	return 0;
+
+error:
+	btrfs_free_path(path);
+	return ret;
 }
 
 /*
diff --git a/fs/btrfs/tree-log.h b/fs/btrfs/tree-log.h
index 2270ac58d74..862ac813f6b 100644
--- a/fs/btrfs/tree-log.h
+++ b/fs/btrfs/tree-log.h
@@ -38,7 +38,7 @@ int btrfs_del_inode_ref_in_log(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root,
 			       const char *name, int name_len,
 			       struct inode *inode, u64 dirid);
-int btrfs_end_log_trans(struct btrfs_root *root);
+void btrfs_end_log_trans(struct btrfs_root *root);
 int btrfs_pin_log_trans(struct btrfs_root *root);
 int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 		    struct btrfs_root *root, struct inode *inode,
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index ef41f285a47..a872b48be0a 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -67,7 +67,7 @@ static void free_fs_devices(struct btrfs_fs_devices *fs_devices)
 	kfree(fs_devices);
 }
 
-int btrfs_cleanup_fs_uuids(void)
+void btrfs_cleanup_fs_uuids(void)
 {
 	struct btrfs_fs_devices *fs_devices;
 
@@ -77,7 +77,6 @@ int btrfs_cleanup_fs_uuids(void)
 		list_del(&fs_devices->list);
 		free_fs_devices(fs_devices);
 	}
-	return 0;
 }
 
 static noinline struct btrfs_device *__find_device(struct list_head *head,
@@ -130,7 +129,7 @@ static void requeue_list(struct btrfs_pending_bios *pending_bios,
  * the list if the block device is congested.  This way, multiple devices
  * can make progress from a single worker thread.
  */
-static noinline int run_scheduled_bios(struct btrfs_device *device)
+static noinline void run_scheduled_bios(struct btrfs_device *device)
 {
 	struct bio *pending;
 	struct backing_dev_info *bdi;
@@ -316,7 +315,6 @@ loop_lock:
 
 done:
 	blk_finish_plug(&plug);
-	return 0;
 }
 
 static void pending_bios_fn(struct btrfs_work *work)
@@ -455,7 +453,7 @@ error:
 	return ERR_PTR(-ENOMEM);
 }
 
-int btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices)
+void btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices)
 {
 	struct btrfs_device *device, *next;
 
@@ -503,7 +501,6 @@ again:
 	fs_devices->latest_trans = latest_transid;
 
 	mutex_unlock(&uuid_mutex);
-	return 0;
 }
 
 static void __free_device(struct work_struct *work)
@@ -552,10 +549,10 @@ static int __btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
 			fs_devices->num_can_discard--;
 
 		new_device = kmalloc(sizeof(*new_device), GFP_NOFS);
-		BUG_ON(!new_device);
+		BUG_ON(!new_device); /* -ENOMEM */
 		memcpy(new_device, device, sizeof(*new_device));
 		new_device->name = kstrdup(device->name, GFP_NOFS);
-		BUG_ON(device->name && !new_device->name);
+		BUG_ON(device->name && !new_device->name); /* -ENOMEM */
 		new_device->bdev = NULL;
 		new_device->writeable = 0;
 		new_device->in_fs_metadata = 0;
@@ -625,6 +622,8 @@ static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
 			printk(KERN_INFO "open %s failed\n", device->name);
 			goto error;
 		}
+		filemap_write_and_wait(bdev->bd_inode->i_mapping);
+		invalidate_bdev(bdev);
 		set_blocksize(bdev, 4096);
 
 		bh = btrfs_read_dev_super(bdev);
@@ -1039,8 +1038,10 @@ again:
 		leaf = path->nodes[0];
 		extent = btrfs_item_ptr(leaf, path->slots[0],
 					struct btrfs_dev_extent);
+	} else {
+		btrfs_error(root->fs_info, ret, "Slot search failed");
+		goto out;
 	}
-	BUG_ON(ret);
 
 	if (device->bytes_used > 0) {
 		u64 len = btrfs_dev_extent_length(leaf, extent);
@@ -1050,7 +1051,10 @@ again:
 		spin_unlock(&root->fs_info->free_chunk_lock);
 	}
 	ret = btrfs_del_item(trans, root, path);
-
+	if (ret) {
+		btrfs_error(root->fs_info, ret,
+			    "Failed to remove dev extent item");
+	}
 out:
 	btrfs_free_path(path);
 	return ret;
@@ -1078,7 +1082,8 @@ int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
 	key.type = BTRFS_DEV_EXTENT_KEY;
 	ret = btrfs_insert_empty_item(trans, root, path, &key,
 				      sizeof(*extent));
-	BUG_ON(ret);
+	if (ret)
+		goto out;
 
 	leaf = path->nodes[0];
 	extent = btrfs_item_ptr(leaf, path->slots[0],
@@ -1093,6 +1098,7 @@ int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
 
 	btrfs_set_dev_extent_length(leaf, extent, num_bytes);
 	btrfs_mark_buffer_dirty(leaf);
+out:
 	btrfs_free_path(path);
 	return ret;
 }
@@ -1118,7 +1124,7 @@ static noinline int find_next_chunk(struct btrfs_root *root,
 	if (ret < 0)
 		goto error;
 
-	BUG_ON(ret == 0);
+	BUG_ON(ret == 0); /* Corruption */
 
 	ret = btrfs_previous_item(root, path, 0, BTRFS_CHUNK_ITEM_KEY);
 	if (ret) {
@@ -1162,7 +1168,7 @@ static noinline int find_next_devid(struct btrfs_root *root, u64 *objectid)
 	if (ret < 0)
 		goto error;
 
-	BUG_ON(ret == 0);
+	BUG_ON(ret == 0); /* Corruption */
 
 	ret = btrfs_previous_item(root, path, BTRFS_DEV_ITEMS_OBJECTID,
 				  BTRFS_DEV_ITEM_KEY);
@@ -1350,6 +1356,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
 		}
 
 		set_blocksize(bdev, 4096);
+		invalidate_bdev(bdev);
 		bh = btrfs_read_dev_super(bdev);
 		if (!bh) {
 			ret = -EINVAL;
@@ -1596,7 +1603,7 @@ next_slot:
 				   (unsigned long)btrfs_device_fsid(dev_item),
 				   BTRFS_UUID_SIZE);
 		device = btrfs_find_device(root, devid, dev_uuid, fs_uuid);
-		BUG_ON(!device);
+		BUG_ON(!device); /* Logic error */
 
 		if (device->fs_devices->seeding) {
 			btrfs_set_device_generation(leaf, dev_item,
@@ -1706,7 +1713,7 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 	if (seeding_dev) {
 		sb->s_flags &= ~MS_RDONLY;
 		ret = btrfs_prepare_sprout(root);
-		BUG_ON(ret);
+		BUG_ON(ret); /* -ENOMEM */
 	}
 
 	device->fs_devices = root->fs_info->fs_devices;
@@ -1744,11 +1751,15 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 
 	if (seeding_dev) {
 		ret = init_first_rw_device(trans, root, device);
-		BUG_ON(ret);
+		if (ret)
+			goto error_trans;
 		ret = btrfs_finish_sprout(trans, root);
-		BUG_ON(ret);
+		if (ret)
+			goto error_trans;
 	} else {
 		ret = btrfs_add_device(trans, root, device);
+		if (ret)
+			goto error_trans;
 	}
 
 	/*
@@ -1758,17 +1769,31 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 	btrfs_clear_space_info_full(root->fs_info);
 
 	unlock_chunks(root);
-	btrfs_commit_transaction(trans, root);
+	ret = btrfs_commit_transaction(trans, root);
 
 	if (seeding_dev) {
 		mutex_unlock(&uuid_mutex);
 		up_write(&sb->s_umount);
 
+		if (ret) /* transaction commit */
+			return ret;
+
 		ret = btrfs_relocate_sys_chunks(root);
-		BUG_ON(ret);
+		if (ret < 0)
+			btrfs_error(root->fs_info, ret,
+				    "Failed to relocate sys chunks after "
+				    "device initialization. This can be fixed "
+				    "using the \"btrfs balance\" command.");
 	}
 
 	return ret;
+
+error_trans:
+	unlock_chunks(root);
+	btrfs_abort_transaction(trans, root, ret);
+	btrfs_end_transaction(trans, root);
+	kfree(device->name);
+	kfree(device);
 error:
 	blkdev_put(bdev, FMODE_EXCL);
 	if (seeding_dev) {
@@ -1876,10 +1901,20 @@ static int btrfs_free_chunk(struct btrfs_trans_handle *trans,
 	key.type = BTRFS_CHUNK_ITEM_KEY;
 
 	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
-	BUG_ON(ret);
+	if (ret < 0)
+		goto out;
+	else if (ret > 0) { /* Logic error or corruption */
+		btrfs_error(root->fs_info, -ENOENT,
+			    "Failed lookup while freeing chunk.");
+		ret = -ENOENT;
+		goto out;
+	}
 
 	ret = btrfs_del_item(trans, root, path);
-
+	if (ret < 0)
+		btrfs_error(root->fs_info, ret,
+			    "Failed to delete chunk item.");
+out:
 	btrfs_free_path(path);
 	return ret;
 }
@@ -2041,7 +2076,7 @@ again:
 		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
 		if (ret < 0)
 			goto error;
-		BUG_ON(ret == 0);
+		BUG_ON(ret == 0); /* Corruption */
 
 		ret = btrfs_previous_item(chunk_root, path, key.objectid,
 					  key.type);
@@ -2250,15 +2285,13 @@ static void unset_balance_control(struct btrfs_fs_info *fs_info)
  * Balance filters.  Return 1 if chunk should be filtered out
  * (should not be balanced).
  */
-static int chunk_profiles_filter(u64 chunk_profile,
+static int chunk_profiles_filter(u64 chunk_type,
 				 struct btrfs_balance_args *bargs)
 {
-	chunk_profile &= BTRFS_BLOCK_GROUP_PROFILE_MASK;
-
-	if (chunk_profile == 0)
-		chunk_profile = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
+	chunk_type = chunk_to_extended(chunk_type) &
+				BTRFS_EXTENDED_PROFILE_MASK;
 
-	if (bargs->profiles & chunk_profile)
+	if (bargs->profiles & chunk_type)
 		return 0;
 
 	return 1;
@@ -2365,18 +2398,16 @@ static int chunk_vrange_filter(struct extent_buffer *leaf,
 	return 1;
 }
 
-static int chunk_soft_convert_filter(u64 chunk_profile,
+static int chunk_soft_convert_filter(u64 chunk_type,
 				     struct btrfs_balance_args *bargs)
 {
 	if (!(bargs->flags & BTRFS_BALANCE_ARGS_CONVERT))
 		return 0;
 
-	chunk_profile &= BTRFS_BLOCK_GROUP_PROFILE_MASK;
+	chunk_type = chunk_to_extended(chunk_type) &
+				BTRFS_EXTENDED_PROFILE_MASK;
 
-	if (chunk_profile == 0)
-		chunk_profile = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
-
-	if (bargs->target & chunk_profile)
+	if (bargs->target == chunk_type)
 		return 1;
 
 	return 0;
@@ -2602,6 +2633,30 @@ error:
 	return ret;
 }
 
+/**
+ * alloc_profile_is_valid - see if a given profile is valid and reduced
+ * @flags: profile to validate
+ * @extended: if true @flags is treated as an extended profile
+ */
+static int alloc_profile_is_valid(u64 flags, int extended)
+{
+	u64 mask = (extended ? BTRFS_EXTENDED_PROFILE_MASK :
+			       BTRFS_BLOCK_GROUP_PROFILE_MASK);
+
+	flags &= ~BTRFS_BLOCK_GROUP_TYPE_MASK;
+
+	/* 1) check that all other bits are zeroed */
+	if (flags & ~mask)
+		return 0;
+
+	/* 2) see if profile is reduced */
+	if (flags == 0)
+		return !extended; /* "0" is valid for usual profiles */
+
+	/* true if exactly one bit set */
+	return (flags & (flags - 1)) == 0;
+}
+
 static inline int balance_need_close(struct btrfs_fs_info *fs_info)
 {
 	/* cancel requested || normal exit path */
@@ -2630,6 +2685,7 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 {
 	struct btrfs_fs_info *fs_info = bctl->fs_info;
 	u64 allowed;
+	int mixed = 0;
 	int ret;
 
 	if (btrfs_fs_closing(fs_info) ||
@@ -2639,13 +2695,16 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		goto out;
 	}
 
+	allowed = btrfs_super_incompat_flags(fs_info->super_copy);
+	if (allowed & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS)
+		mixed = 1;
+
 	/*
 	 * In case of mixed groups both data and meta should be picked,
 	 * and identical options should be given for both of them.
 	 */
-	allowed = btrfs_super_incompat_flags(fs_info->super_copy);
-	if ((allowed & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) &&
-	    (bctl->flags & (BTRFS_BALANCE_DATA | BTRFS_BALANCE_METADATA))) {
+	allowed = BTRFS_BALANCE_DATA | BTRFS_BALANCE_METADATA;
+	if (mixed && (bctl->flags & allowed)) {
 		if (!(bctl->flags & BTRFS_BALANCE_DATA) ||
 		    !(bctl->flags & BTRFS_BALANCE_METADATA) ||
 		    memcmp(&bctl->data, &bctl->meta, sizeof(bctl->data))) {
@@ -2656,14 +2715,6 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		}
 	}
 
-	/*
-	 * Profile changing sanity checks.  Skip them if a simple
-	 * balance is requested.
-	 */
-	if (!((bctl->data.flags | bctl->sys.flags | bctl->meta.flags) &
-	      BTRFS_BALANCE_ARGS_CONVERT))
-		goto do_balance;
-
 	allowed = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
 	if (fs_info->fs_devices->num_devices == 1)
 		allowed |= BTRFS_BLOCK_GROUP_DUP;
@@ -2673,24 +2724,27 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1 |
 				BTRFS_BLOCK_GROUP_RAID10);
 
-	if (!profile_is_valid(bctl->data.target, 1) ||
-	    bctl->data.target & ~allowed) {
+	if ((bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
+	    (!alloc_profile_is_valid(bctl->data.target, 1) ||
+	     (bctl->data.target & ~allowed))) {
 		printk(KERN_ERR "btrfs: unable to start balance with target "
 		       "data profile %llu\n",
 		       (unsigned long long)bctl->data.target);
 		ret = -EINVAL;
 		goto out;
 	}
-	if (!profile_is_valid(bctl->meta.target, 1) ||
-	    bctl->meta.target & ~allowed) {
+	if ((bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
+	    (!alloc_profile_is_valid(bctl->meta.target, 1) ||
+	     (bctl->meta.target & ~allowed))) {
 		printk(KERN_ERR "btrfs: unable to start balance with target "
 		       "metadata profile %llu\n",
 		       (unsigned long long)bctl->meta.target);
 		ret = -EINVAL;
 		goto out;
 	}
-	if (!profile_is_valid(bctl->sys.target, 1) ||
-	    bctl->sys.target & ~allowed) {
+	if ((bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
+	    (!alloc_profile_is_valid(bctl->sys.target, 1) ||
+	     (bctl->sys.target & ~allowed))) {
 		printk(KERN_ERR "btrfs: unable to start balance with target "
 		       "system profile %llu\n",
 		       (unsigned long long)bctl->sys.target);
@@ -2698,7 +2752,9 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		goto out;
 	}
 
-	if (bctl->data.target & BTRFS_BLOCK_GROUP_DUP) {
+	/* allow dup'ed data chunks only in mixed mode */
+	if (!mixed && (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
+	    (bctl->data.target & BTRFS_BLOCK_GROUP_DUP)) {
 		printk(KERN_ERR "btrfs: dup for data is not allowed\n");
 		ret = -EINVAL;
 		goto out;
@@ -2724,7 +2780,6 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		}
 	}
 
-do_balance:
 	ret = insert_balance_item(fs_info->tree_root, bctl);
 	if (ret && ret != -EEXIST)
 		goto out;
@@ -2967,7 +3022,7 @@ again:
 	key.offset = (u64)-1;
 	key.type = BTRFS_DEV_EXTENT_KEY;
 
-	while (1) {
+	do {
 		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 		if (ret < 0)
 			goto done;
@@ -3009,8 +3064,7 @@ again:
 			goto done;
 		if (ret == -ENOSPC)
 			failed++;
-		key.offset -= 1;
-	}
+	} while (key.offset-- > 0);
 
 	if (failed && !retried) {
 		failed = 0;
@@ -3128,11 +3182,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 	int i;
 	int j;
 
-	if ((type & BTRFS_BLOCK_GROUP_RAID1) &&
-	    (type & BTRFS_BLOCK_GROUP_DUP)) {
-		WARN_ON(1);
-		type &= ~BTRFS_BLOCK_GROUP_DUP;
-	}
+	BUG_ON(!alloc_profile_is_valid(type, 0));
 
 	if (list_empty(&fs_devices->alloc_list))
 		return -ENOSPC;
@@ -3328,13 +3378,15 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 	write_lock(&em_tree->lock);
 	ret = add_extent_mapping(em_tree, em);
 	write_unlock(&em_tree->lock);
-	BUG_ON(ret);
 	free_extent_map(em);
+	if (ret)
+		goto error;
 
 	ret = btrfs_make_block_group(trans, extent_root, 0, type,
 				     BTRFS_FIRST_CHUNK_TREE_OBJECTID,
 				     start, num_bytes);
-	BUG_ON(ret);
+	if (ret)
+		goto error;
 
 	for (i = 0; i < map->num_stripes; ++i) {
 		struct btrfs_device *device;
@@ -3347,7 +3399,10 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 				info->chunk_root->root_key.objectid,
 				BTRFS_FIRST_CHUNK_TREE_OBJECTID,
 				start, dev_offset, stripe_size);
-		BUG_ON(ret);
+		if (ret) {
+			btrfs_abort_transaction(trans, extent_root, ret);
+			goto error;
+		}
 	}
 
 	kfree(devices_info);
@@ -3383,7 +3438,8 @@ static int __finish_chunk_alloc(struct btrfs_trans_handle *trans,
 		device = map->stripes[index].dev;
 		device->bytes_used += stripe_size;
 		ret = btrfs_update_device(trans, device);
-		BUG_ON(ret);
+		if (ret)
+			goto out_free;
 		index++;
 	}
 
@@ -3420,16 +3476,19 @@ static int __finish_chunk_alloc(struct btrfs_trans_handle *trans,
 	key.offset = chunk_offset;
 
 	ret = btrfs_insert_item(trans, chunk_root, &key, chunk, item_size);
-	BUG_ON(ret);
 
-	if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
+	if (ret == 0 && map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
+		/*
+		 * TODO: Cleanup of inserted chunk root in case of
+		 * failure.
+		 */
 		ret = btrfs_add_system_chunk(chunk_root, &key, chunk,
 					     item_size);
-		BUG_ON(ret);
 	}
 
+out_free:
 	kfree(chunk);
-	return 0;
+	return ret;
 }
 
 /*
@@ -3461,7 +3520,8 @@ int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 
 	ret = __finish_chunk_alloc(trans, extent_root, map, chunk_offset,
 				   chunk_size, stripe_size);
-	BUG_ON(ret);
+	if (ret)
+		return ret;
 	return 0;
 }
 
@@ -3493,7 +3553,8 @@ static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
 
 	ret = __btrfs_alloc_chunk(trans, extent_root, &map, &chunk_size,
 				  &stripe_size, chunk_offset, alloc_profile);
-	BUG_ON(ret);
+	if (ret)
+		return ret;
 
 	sys_chunk_offset = chunk_offset + chunk_size;
 
@@ -3504,10 +3565,12 @@ static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
 	ret = __btrfs_alloc_chunk(trans, extent_root, &sys_map,
 				  &sys_chunk_size, &sys_stripe_size,
 				  sys_chunk_offset, alloc_profile);
-	BUG_ON(ret);
+	if (ret)
+		goto abort;
 
 	ret = btrfs_add_device(trans, fs_info->chunk_root, device);
-	BUG_ON(ret);
+	if (ret)
+		goto abort;
 
 	/*
 	 * Modifying chunk tree needs allocating new blocks from both
@@ -3517,13 +3580,20 @@ static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
 	 */
 	ret = __finish_chunk_alloc(trans, extent_root, map, chunk_offset,
 				   chunk_size, stripe_size);
-	BUG_ON(ret);
+	if (ret)
+		goto abort;
 
 	ret = __finish_chunk_alloc(trans, extent_root, sys_map,
 				   sys_chunk_offset, sys_chunk_size,
 				   sys_stripe_size);
-	BUG_ON(ret);
+	if (ret)
+		goto abort;
+
 	return 0;
+
+abort:
+	btrfs_abort_transaction(trans, root, ret);
+	return ret;
 }
 
 int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset)
@@ -3874,7 +3944,7 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
 		do_div(length, map->num_stripes);
 
 	buf = kzalloc(sizeof(u64) * map->num_stripes, GFP_NOFS);
-	BUG_ON(!buf);
+	BUG_ON(!buf); /* -ENOMEM */
 
 	for (i = 0; i < map->num_stripes; i++) {
 		if (devid && map->stripes[i].dev->devid != devid)
@@ -3967,7 +4037,7 @@ struct async_sched {
  * This will add one bio to the pending list for a device and make sure
  * the work struct is scheduled.
  */
-static noinline int schedule_bio(struct btrfs_root *root,
+static noinline void schedule_bio(struct btrfs_root *root,
 				 struct btrfs_device *device,
 				 int rw, struct bio *bio)
 {
@@ -3979,7 +4049,7 @@ static noinline int schedule_bio(struct btrfs_root *root,
 		bio_get(bio);
 		btrfsic_submit_bio(rw, bio);
 		bio_put(bio);
-		return 0;
+		return;
 	}
 
 	/*
@@ -4013,7 +4083,6 @@ static noinline int schedule_bio(struct btrfs_root *root,
 	if (should_queue)
 		btrfs_queue_worker(&root->fs_info->submit_workers,
 				   &device->work);
-	return 0;
 }
 
 int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
@@ -4036,7 +4105,8 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
 
 	ret = btrfs_map_block(map_tree, rw, logical, &map_length, &bbio,
 			      mirror_num);
-	BUG_ON(ret);
+	if (ret) /* -ENOMEM */
+		return ret;
 
 	total_devs = bbio->num_stripes;
 	if (map_length < length) {
@@ -4055,7 +4125,7 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
 	while (dev_nr < total_devs) {
 		if (dev_nr < total_devs - 1) {
 			bio = bio_clone(first_bio, GFP_NOFS);
-			BUG_ON(!bio);
+			BUG_ON(!bio); /* -ENOMEM */
 		} else {
 			bio = first_bio;
 		}
@@ -4209,13 +4279,13 @@ static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
 	write_lock(&map_tree->map_tree.lock);
 	ret = add_extent_mapping(&map_tree->map_tree, em);
 	write_unlock(&map_tree->map_tree.lock);
-	BUG_ON(ret);
+	BUG_ON(ret); /* Tree corruption */
 	free_extent_map(em);
 
 	return 0;
 }
 
-static int fill_device_from_item(struct extent_buffer *leaf,
+static void fill_device_from_item(struct extent_buffer *leaf,
 				 struct btrfs_dev_item *dev_item,
 				 struct btrfs_device *device)
 {
@@ -4232,8 +4302,6 @@ static int fill_device_from_item(struct extent_buffer *leaf,
 
 	ptr = (unsigned long)btrfs_device_uuid(dev_item);
 	read_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
-
-	return 0;
 }
 
 static int open_seed_devices(struct btrfs_root *root, u8 *fsid)
@@ -4384,7 +4452,7 @@ int btrfs_read_sys_array(struct btrfs_root *root)
 	 * to silence the warning eg. on PowerPC 64.
 	 */
 	if (PAGE_CACHE_SIZE > BTRFS_SUPER_INFO_SIZE)
-		SetPageUptodate(sb->first_page);
+		SetPageUptodate(sb->pages[0]);
 
 	write_extent_buffer(sb, super_copy, 0, BTRFS_SUPER_INFO_SIZE);
 	array_size = btrfs_super_sys_array_size(super_copy);
diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h
index 19ac95048b8..bb6b03f97aa 100644
--- a/fs/btrfs/volumes.h
+++ b/fs/btrfs/volumes.h
@@ -260,12 +260,12 @@ int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
 int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
 			  struct btrfs_fs_devices **fs_devices_ret);
 int btrfs_close_devices(struct btrfs_fs_devices *fs_devices);
-int btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices);
+void btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices);
 int btrfs_add_device(struct btrfs_trans_handle *trans,
 		     struct btrfs_root *root,
 		     struct btrfs_device *device);
 int btrfs_rm_device(struct btrfs_root *root, char *device_path);
-int btrfs_cleanup_fs_uuids(void);
+void btrfs_cleanup_fs_uuids(void);
 int btrfs_num_copies(struct btrfs_mapping_tree *map_tree, u64 logical, u64 len);
 int btrfs_grow_device(struct btrfs_trans_handle *trans,
 		      struct btrfs_device *device, u64 new_size);