17 files changed, 542 insertions, 206 deletions

diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index a8c9693b75a..72677ce2b74 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -66,6 +66,9 @@ struct btrfs_inode {
 	 */
 	struct list_head delalloc_inodes;
 
+	/* the space_info for where this inode's data allocations are done */
+	struct btrfs_space_info *space_info;
+
 	/* full 64 bit generation number, struct vfs_inode doesn't have a big
 	 * enough field for this.
 	 */
@@ -94,6 +97,11 @@ struct btrfs_inode {
 	 */
 	u64 delalloc_bytes;
 
+	/* total number of bytes that may be used for this inode for
+	 * delalloc
+	 */
+	u64 reserved_bytes;
+
 	/*
 	 * the size of the file stored in the metadata on disk.  data=ordered
 	 * means the in-memory i_size might be larger than the size on disk
diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index 551177c0011..37f31b5529a 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -38,19 +38,12 @@ static int balance_node_right(struct btrfs_trans_handle *trans,
 static int del_ptr(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		   struct btrfs_path *path, int level, int slot);
 
-inline void btrfs_init_path(struct btrfs_path *p)
-{
-	memset(p, 0, sizeof(*p));
-}
-
 struct btrfs_path *btrfs_alloc_path(void)
 {
 	struct btrfs_path *path;
-	path = kmem_cache_alloc(btrfs_path_cachep, GFP_NOFS);
-	if (path) {
-		btrfs_init_path(path);
+	path = kmem_cache_zalloc(btrfs_path_cachep, GFP_NOFS);
+	if (path)
 		path->reada = 1;
-	}
 	return path;
 }
 
@@ -69,14 +62,38 @@ noinline void btrfs_set_path_blocking(struct btrfs_path *p)
 
 /*
  * reset all the locked nodes in the patch to spinning locks.
+ *
+ * held is used to keep lockdep happy, when lockdep is enabled
+ * we set held to a blocking lock before we go around and
+ * retake all the spinlocks in the path.  You can safely use NULL
+ * for held
  */
-noinline void btrfs_clear_path_blocking(struct btrfs_path *p)
+noinline void btrfs_clear_path_blocking(struct btrfs_path *p,
+					struct extent_buffer *held)
 {
 	int i;
-	for (i = 0; i < BTRFS_MAX_LEVEL; i++) {
+
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+	/* lockdep really cares that we take all of these spinlocks
+	 * in the right order.  If any of the locks in the path are not
+	 * currently blocking, it is going to complain.  So, make really
+	 * really sure by forcing the path to blocking before we clear
+	 * the path blocking.
+	 */
+	if (held)
+		btrfs_set_lock_blocking(held);
+	btrfs_set_path_blocking(p);
+#endif
+
+	for (i = BTRFS_MAX_LEVEL - 1; i >= 0; i--) {
 		if (p->nodes[i] && p->locks[i])
 			btrfs_clear_lock_blocking(p->nodes[i]);
 	}
+
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+	if (held)
+		btrfs_clear_lock_blocking(held);
+#endif
 }
 
 /* this also releases the path */
@@ -260,7 +277,7 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 	if (*cow_ret == buf)
 		unlock_orig = 1;
 
-	WARN_ON(!btrfs_tree_locked(buf));
+	btrfs_assert_tree_locked(buf);
 
 	if (parent)
 		parent_start = parent->start;
@@ -286,7 +303,7 @@ static noinline int __btrfs_cow_block(struct btrfs_trans_handle *trans,
 						  trans->transid, level, &ins);
 		BUG_ON(ret);
 		cow = btrfs_init_new_buffer(trans, root, prealloc_dest,
-					    buf->len);
+					    buf->len, level);
 	} else {
 		cow = btrfs_alloc_free_block(trans, root, buf->len,
 					     parent_start,
@@ -917,9 +934,9 @@ 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);
 		btrfs_tree_lock(child);
 		btrfs_set_lock_blocking(child);
-		BUG_ON(!child);
 		ret = btrfs_cow_block(trans, root, child, mid, 0, &child, 0);
 		BUG_ON(ret);
 
@@ -1530,8 +1547,7 @@ again:
 			 * for higher level blocks, try not to allocate blocks
 			 * with the block and the parent locks held.
 			 */
-			if (level > 0 && !prealloc_block.objectid &&
-			    btrfs_path_lock_waiting(p, level)) {
+			if (level > 0 && !prealloc_block.objectid) {
 				u32 size = b->len;
 				u64 hint = b->start;
 
@@ -1567,7 +1583,7 @@ cow_done:
 		if (!p->skip_locking)
 			p->locks[level] = 1;
 
-		btrfs_clear_path_blocking(p);
+		btrfs_clear_path_blocking(p, NULL);
 
 		/*
 		 * we have a lock on b and as long as we aren't changing
@@ -1606,7 +1622,7 @@ cow_done:
 
 				btrfs_set_path_blocking(p);
 				sret = split_node(trans, root, p, level);
-				btrfs_clear_path_blocking(p);
+				btrfs_clear_path_blocking(p, NULL);
 
 				BUG_ON(sret > 0);
 				if (sret) {
@@ -1626,7 +1642,7 @@ cow_done:
 
 				btrfs_set_path_blocking(p);
 				sret = balance_level(trans, root, p, level);
-				btrfs_clear_path_blocking(p);
+				btrfs_clear_path_blocking(p, NULL);
 
 				if (sret) {
 					ret = sret;
@@ -1689,13 +1705,13 @@ cow_done:
 			if (!p->skip_locking) {
 				int lret;
 
-				btrfs_clear_path_blocking(p);
+				btrfs_clear_path_blocking(p, NULL);
 				lret = btrfs_try_spin_lock(b);
 
 				if (!lret) {
 					btrfs_set_path_blocking(p);
 					btrfs_tree_lock(b);
-					btrfs_clear_path_blocking(p);
+					btrfs_clear_path_blocking(p, b);
 				}
 			}
 		} else {
@@ -1707,7 +1723,7 @@ cow_done:
 				btrfs_set_path_blocking(p);
 				sret = split_leaf(trans, root, key,
 						      p, ins_len, ret == 0);
-				btrfs_clear_path_blocking(p);
+				btrfs_clear_path_blocking(p, NULL);
 
 				BUG_ON(sret > 0);
 				if (sret) {
@@ -2349,7 +2365,7 @@ static int push_leaf_right(struct btrfs_trans_handle *trans, struct btrfs_root
 	if (slot >= btrfs_header_nritems(upper) - 1)
 		return 1;
 
-	WARN_ON(!btrfs_tree_locked(path->nodes[1]));
+	btrfs_assert_tree_locked(path->nodes[1]);
 
 	right = read_node_slot(root, upper, slot + 1);
 	btrfs_tree_lock(right);
@@ -2546,7 +2562,7 @@ static int push_leaf_left(struct btrfs_trans_handle *trans, struct btrfs_root
 	if (right_nritems == 0)
 		return 1;
 
-	WARN_ON(!btrfs_tree_locked(path->nodes[1]));
+	btrfs_assert_tree_locked(path->nodes[1]);
 
 	left = read_node_slot(root, path->nodes[1], slot - 1);
 	btrfs_tree_lock(left);
@@ -3927,7 +3943,6 @@ find_next_key:
 				btrfs_release_path(root, path);
 				goto again;
 			} else {
-				btrfs_clear_path_blocking(path);
 				goto out;
 			}
 		}
@@ -3947,7 +3962,7 @@ find_next_key:
 		path->locks[level - 1] = 1;
 		path->nodes[level - 1] = cur;
 		unlock_up(path, level, 1);
-		btrfs_clear_path_blocking(path);
+		btrfs_clear_path_blocking(path, NULL);
 	}
 out:
 	if (ret == 0)
@@ -4086,7 +4101,7 @@ int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path)
 
 		next = read_node_slot(root, c, slot);
 		if (!path->skip_locking) {
-			WARN_ON(!btrfs_tree_locked(c));
+			btrfs_assert_tree_locked(c);
 			btrfs_tree_lock(next);
 			btrfs_set_lock_blocking(next);
 		}
@@ -4111,7 +4126,7 @@ int btrfs_next_leaf(struct btrfs_root *root, struct btrfs_path *path)
 			reada_for_search(root, path, level, slot, 0);
 		next = read_node_slot(root, next, 0);
 		if (!path->skip_locking) {
-			WARN_ON(!btrfs_tree_locked(path->nodes[level]));
+			btrfs_assert_tree_locked(path->nodes[level]);
 			btrfs_tree_lock(next);
 			btrfs_set_lock_blocking(next);
 		}
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 531db112c8b..5e1d4e30e9d 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -43,11 +43,7 @@ struct btrfs_ordered_sum;
 
 #define BTRFS_ACL_NOT_CACHED    ((void *)-1)
 
-#ifdef CONFIG_LOCKDEP
-# define BTRFS_MAX_LEVEL 7
-#else
-# define BTRFS_MAX_LEVEL 8
-#endif
+#define BTRFS_MAX_LEVEL 8
 
 /* holds pointers to all of the tree roots */
 #define BTRFS_ROOT_TREE_OBJECTID 1ULL
@@ -600,13 +596,27 @@ struct btrfs_block_group_item {
 
 struct btrfs_space_info {
 	u64 flags;
-	u64 total_bytes;
-	u64 bytes_used;
-	u64 bytes_pinned;
-	u64 bytes_reserved;
-	u64 bytes_readonly;
-	int full;
-	int force_alloc;
+
+	u64 total_bytes;	/* total bytes in the space */
+	u64 bytes_used;		/* total bytes used on disk */
+	u64 bytes_pinned;	/* total bytes pinned, will be freed when the
+				   transaction finishes */
+	u64 bytes_reserved;	/* total bytes the allocator has reserved for
+				   current allocations */
+	u64 bytes_readonly;	/* total bytes that are read only */
+
+	/* delalloc accounting */
+	u64 bytes_delalloc;	/* number of bytes reserved for allocation,
+				   this space is not necessarily reserved yet
+				   by the allocator */
+	u64 bytes_may_use;	/* number of bytes that may be used for
+				   delalloc */
+
+	int full;		/* indicates that we cannot allocate any more
+				   chunks for this space */
+	int force_alloc;	/* set if we need to force a chunk alloc for
+				   this space */
+
 	struct list_head list;
 
 	/* for block groups in our same type */
@@ -774,7 +784,14 @@ struct btrfs_fs_info {
 	struct list_head dirty_cowonly_roots;
 
 	struct btrfs_fs_devices *fs_devices;
+
+	/*
+	 * the space_info list is almost entirely read only.  It only changes
+	 * when we add a new raid type to the FS, and that happens
+	 * very rarely.  RCU is used to protect it.
+	 */
 	struct list_head space_info;
+
 	spinlock_t delalloc_lock;
 	spinlock_t new_trans_lock;
 	u64 delalloc_bytes;
@@ -1715,7 +1732,8 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
 					     u64 empty_size);
 struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans,
 					    struct btrfs_root *root,
-					    u64 bytenr, u32 blocksize);
+					    u64 bytenr, u32 blocksize,
+					    int level);
 int btrfs_alloc_extent(struct btrfs_trans_handle *trans,
 		       struct btrfs_root *root,
 		       u64 num_bytes, u64 parent, u64 min_bytes,
@@ -1785,6 +1803,18 @@ int btrfs_add_dead_reloc_root(struct btrfs_root *root);
 int btrfs_cleanup_reloc_trees(struct btrfs_root *root);
 int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len);
 u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags);
+void btrfs_set_inode_space_info(struct btrfs_root *root, struct inode *ionde);
+void btrfs_clear_space_info_full(struct btrfs_fs_info *info);
+
+int btrfs_check_metadata_free_space(struct btrfs_root *root);
+int btrfs_check_data_free_space(struct btrfs_root *root, struct inode *inode,
+				u64 bytes);
+void btrfs_free_reserved_data_space(struct btrfs_root *root,
+				    struct inode *inode, u64 bytes);
+void btrfs_delalloc_reserve_space(struct btrfs_root *root, struct inode *inode,
+				 u64 bytes);
+void btrfs_delalloc_free_space(struct btrfs_root *root, struct inode *inode,
+			      u64 bytes);
 /* ctree.c */
 int btrfs_previous_item(struct btrfs_root *root,
 			struct btrfs_path *path, u64 min_objectid,
@@ -1834,9 +1864,7 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans,
 void btrfs_release_path(struct btrfs_root *root, struct btrfs_path *p);
 struct btrfs_path *btrfs_alloc_path(void);
 void btrfs_free_path(struct btrfs_path *p);
-void btrfs_init_path(struct btrfs_path *p);
 void btrfs_set_path_blocking(struct btrfs_path *p);
-void btrfs_clear_path_blocking(struct btrfs_path *p);
 void btrfs_unlock_up_safe(struct btrfs_path *p, int level);
 
 int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
@@ -2032,8 +2060,6 @@ int btrfs_merge_bio_hook(struct page *page, unsigned long offset,
 unsigned long btrfs_force_ra(struct address_space *mapping,
 			      struct file_ra_state *ra, struct file *file,
 			      pgoff_t offset, pgoff_t last_index);
-int btrfs_check_free_space(struct btrfs_root *root, u64 num_required,
-			   int for_del);
 int btrfs_page_mkwrite(struct vm_area_struct *vma, struct page *page);
 int btrfs_readpage(struct file *file, struct page *page);
 void btrfs_delete_inode(struct inode *inode);
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 5aebddd7119..3e18175248e 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -75,6 +75,40 @@ struct async_submit_bio {
 	struct btrfs_work work;
 };
 
+/* These are used to set the lockdep class on the extent buffer locks.
+ * The class is set by the readpage_end_io_hook after the buffer has
+ * passed csum validation but before the pages are unlocked.
+ *
+ * The lockdep class is also set by btrfs_init_new_buffer on freshly
+ * allocated blocks.
+ *
+ * The class is based on the level in the tree block, which allows lockdep
+ * to know that lower nodes nest inside the locks of higher nodes.
+ *
+ * We also add a check to make sure the highest level of the tree is
+ * the same as our lockdep setup here.  If BTRFS_MAX_LEVEL changes, this
+ * code needs update as well.
+ */
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+# if BTRFS_MAX_LEVEL != 8
+#  error
+# endif
+static struct lock_class_key btrfs_eb_class[BTRFS_MAX_LEVEL + 1];
+static const char *btrfs_eb_name[BTRFS_MAX_LEVEL + 1] = {
+	/* leaf */
+	"btrfs-extent-00",
+	"btrfs-extent-01",
+	"btrfs-extent-02",
+	"btrfs-extent-03",
+	"btrfs-extent-04",
+	"btrfs-extent-05",
+	"btrfs-extent-06",
+	"btrfs-extent-07",
+	/* highest possible level */
+	"btrfs-extent-08",
+};
+#endif
+
 /*
  * extents on the btree inode are pretty simple, there's one extent
  * that covers the entire device
@@ -347,6 +381,15 @@ static int check_tree_block_fsid(struct btrfs_root *root,
 	return ret;
 }
 
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+void btrfs_set_buffer_lockdep_class(struct extent_buffer *eb, int level)
+{
+	lockdep_set_class_and_name(&eb->lock,
+			   &btrfs_eb_class[level],
+			   btrfs_eb_name[level]);
+}
+#endif
+
 static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 			       struct extent_state *state)
 {
@@ -392,6 +435,8 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 	}
 	found_level = btrfs_header_level(eb);
 
+	btrfs_set_buffer_lockdep_class(eb, found_level);
+
 	ret = csum_tree_block(root, eb, 1);
 	if (ret)
 		ret = -EIO;
@@ -812,7 +857,7 @@ int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 	struct inode *btree_inode = root->fs_info->btree_inode;
 	if (btrfs_header_generation(buf) ==
 	    root->fs_info->running_transaction->transid) {
-		WARN_ON(!btrfs_tree_locked(buf));
+		btrfs_assert_tree_locked(buf);
 
 		/* ugh, clear_extent_buffer_dirty can be expensive */
 		btrfs_set_lock_blocking(buf);
@@ -1777,7 +1822,6 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	ret = find_and_setup_root(tree_root, fs_info,
 				  BTRFS_DEV_TREE_OBJECTID, dev_root);
 	dev_root->track_dirty = 1;
-
 	if (ret)
 		goto fail_extent_root;
 
@@ -2317,7 +2361,7 @@ void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
 
 	btrfs_set_lock_blocking(buf);
 
-	WARN_ON(!btrfs_tree_locked(buf));
+	btrfs_assert_tree_locked(buf);
 	if (transid != root->fs_info->generation) {
 		printk(KERN_CRIT "btrfs transid mismatch buffer %llu, "
 		       "found %llu running %llu\n",
diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h
index 494a56eb298..95029db227b 100644
--- a/fs/btrfs/disk-io.h
+++ b/fs/btrfs/disk-io.h
@@ -101,4 +101,14 @@ int btrfs_init_log_root_tree(struct btrfs_trans_handle *trans,
 int btrfs_add_log_tree(struct btrfs_trans_handle *trans,
 		       struct btrfs_root *root);
 int btree_lock_page_hook(struct page *page);
+
+
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+void btrfs_set_buffer_lockdep_class(struct extent_buffer *eb, int level);
+#else
+static inline void btrfs_set_buffer_lockdep_class(struct extent_buffer *eb,
+						 int level)
+{
+}
+#endif
 #endif
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 7527523c2d2..fefe83ad205 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -20,6 +20,7 @@
 #include <linux/writeback.h>
 #include <linux/blkdev.h>
 #include <linux/sort.h>
+#include <linux/rcupdate.h>
 #include "compat.h"
 #include "hash.h"
 #include "crc32c.h"
@@ -60,6 +61,10 @@ static int update_block_group(struct btrfs_trans_handle *trans,
 			      u64 bytenr, u64 num_bytes, int alloc,
 			      int mark_free);
 
+static int do_chunk_alloc(struct btrfs_trans_handle *trans,
+			  struct btrfs_root *extent_root, u64 alloc_bytes,
+			  u64 flags, int force);
+
 static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
 {
 	return (cache->flags & bits) == bits;
@@ -326,13 +331,33 @@ static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info,
 {
 	struct list_head *head = &info->space_info;
 	struct btrfs_space_info *found;
-	list_for_each_entry(found, head, list) {
-		if (found->flags == flags)
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(found, head, list) {
+		if (found->flags == flags) {
+			rcu_read_unlock();
 			return found;
+		}
 	}
+	rcu_read_unlock();
 	return NULL;
 }
 
+/*
+ * after adding space to the filesystem, we need to clear the full flags
+ * on all the space infos.
+ */
+void btrfs_clear_space_info_full(struct btrfs_fs_info *info)
+{
+	struct list_head *head = &info->space_info;
+	struct btrfs_space_info *found;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(found, head, list)
+		found->full = 0;
+	rcu_read_unlock();
+}
+
 static u64 div_factor(u64 num, int factor)
 {
 	if (factor == 10)
@@ -1323,8 +1348,25 @@ int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 int btrfs_extent_post_op(struct btrfs_trans_handle *trans,
 			 struct btrfs_root *root)
 {
-	finish_current_insert(trans, root->fs_info->extent_root, 1);
-	del_pending_extents(trans, root->fs_info->extent_root, 1);
+	u64 start;
+	u64 end;
+	int ret;
+
+	while(1) {
+		finish_current_insert(trans, root->fs_info->extent_root, 1);
+		del_pending_extents(trans, root->fs_info->extent_root, 1);
+
+		/* is there more work to do? */
+		ret = find_first_extent_bit(&root->fs_info->pending_del,
+					    0, &start, &end, EXTENT_WRITEBACK);
+		if (!ret)
+			continue;
+		ret = find_first_extent_bit(&root->fs_info->extent_ins,
+					    0, &start, &end, EXTENT_WRITEBACK);
+		if (!ret)
+			continue;
+		break;
+	}
 	return 0;
 }
 
@@ -1882,7 +1924,6 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags,
 	if (!found)
 		return -ENOMEM;
 
-	list_add(&found->list, &info->space_info);
 	INIT_LIST_HEAD(&found->block_groups);
 	init_rwsem(&found->groups_sem);
 	spin_lock_init(&found->lock);
@@ -1892,9 +1933,11 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags,
 	found->bytes_pinned = 0;
 	found->bytes_reserved = 0;
 	found->bytes_readonly = 0;
+	found->bytes_delalloc = 0;
 	found->full = 0;
 	found->force_alloc = 0;
 	*space_info = found;
+	list_add_rcu(&found->list, &info->space_info);
 	return 0;
 }
 
@@ -1955,6 +1998,233 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags)
 	return flags;
 }
 
+static u64 btrfs_get_alloc_profile(struct btrfs_root *root, u64 data)
+{
+	struct btrfs_fs_info *info = root->fs_info;
+	u64 alloc_profile;
+
+	if (data) {
+		alloc_profile = info->avail_data_alloc_bits &
+			info->data_alloc_profile;
+		data = BTRFS_BLOCK_GROUP_DATA | alloc_profile;
+	} else if (root == root->fs_info->chunk_root) {
+		alloc_profile = info->avail_system_alloc_bits &
+			info->system_alloc_profile;
+		data = BTRFS_BLOCK_GROUP_SYSTEM | alloc_profile;
+	} else {
+		alloc_profile = info->avail_metadata_alloc_bits &
+			info->metadata_alloc_profile;
+		data = BTRFS_BLOCK_GROUP_METADATA | alloc_profile;
+	}
+
+	return btrfs_reduce_alloc_profile(root, data);
+}
+
+void btrfs_set_inode_space_info(struct btrfs_root *root, struct inode *inode)
+{
+	u64 alloc_target;
+
+	alloc_target = btrfs_get_alloc_profile(root, 1);
+	BTRFS_I(inode)->space_info = __find_space_info(root->fs_info,
+						       alloc_target);
+}
+
+/*
+ * for now this just makes sure we have at least 5% of our metadata space free
+ * for use.
+ */
+int btrfs_check_metadata_free_space(struct btrfs_root *root)
+{
+	struct btrfs_fs_info *info = root->fs_info;
+	struct btrfs_space_info *meta_sinfo;
+	u64 alloc_target, thresh;
+	int committed = 0, ret;
+
+	/* get the space info for where the metadata will live */
+	alloc_target = btrfs_get_alloc_profile(root, 0);
+	meta_sinfo = __find_space_info(info, alloc_target);
+
+again:
+	spin_lock(&meta_sinfo->lock);
+	if (!meta_sinfo->full)
+		thresh = meta_sinfo->total_bytes * 80;
+	else
+		thresh = meta_sinfo->total_bytes * 95;
+
+	do_div(thresh, 100);
+
+	if (meta_sinfo->bytes_used + meta_sinfo->bytes_reserved +
+	    meta_sinfo->bytes_pinned + meta_sinfo->bytes_readonly > thresh) {
+		struct btrfs_trans_handle *trans;
+		if (!meta_sinfo->full) {
+			meta_sinfo->force_alloc = 1;
+			spin_unlock(&meta_sinfo->lock);
+
+			trans = btrfs_start_transaction(root, 1);
+			if (!trans)
+				return -ENOMEM;
+
+			ret = do_chunk_alloc(trans, root->fs_info->extent_root,
+					     2 * 1024 * 1024, alloc_target, 0);
+			btrfs_end_transaction(trans, root);
+			goto again;
+		}
+		spin_unlock(&meta_sinfo->lock);
+
+		if (!committed) {
+			committed = 1;
+			trans = btrfs_join_transaction(root, 1);
+			if (!trans)
+				return -ENOMEM;
+			ret = btrfs_commit_transaction(trans, root);
+			if (ret)
+				return ret;
+			goto again;
+		}
+		return -ENOSPC;
+	}
+	spin_unlock(&meta_sinfo->lock);
+
+	return 0;
+}
+
+/*
+ * This will check the space that the inode allocates from to make sure we have
+ * enough space for bytes.
+ */
+int btrfs_check_data_free_space(struct btrfs_root *root, struct inode *inode,
+				u64 bytes)
+{
+	struct btrfs_space_info *data_sinfo;
+	int ret = 0, committed = 0;
+
+	/* make sure bytes are sectorsize aligned */
+	bytes = (bytes + root->sectorsize - 1) & ~((u64)root->sectorsize - 1);
+
+	data_sinfo = BTRFS_I(inode)->space_info;
+again:
+	/* make sure we have enough space to handle the data first */
+	spin_lock(&data_sinfo->lock);
+	if (data_sinfo->total_bytes - data_sinfo->bytes_used -
+	    data_sinfo->bytes_delalloc - data_sinfo->bytes_reserved -
+	    data_sinfo->bytes_pinned - data_sinfo->bytes_readonly -
+	    data_sinfo->bytes_may_use < bytes) {
+		struct btrfs_trans_handle *trans;
+
+		/*
+		 * if we don't have enough free bytes in this space then we need
+		 * to alloc a new chunk.
+		 */
+		if (!data_sinfo->full) {
+			u64 alloc_target;
+
+			data_sinfo->force_alloc = 1;
+			spin_unlock(&data_sinfo->lock);
+
+			alloc_target = btrfs_get_alloc_profile(root, 1);
+			trans = btrfs_start_transaction(root, 1);
+			if (!trans)
+				return -ENOMEM;
+
+			ret = do_chunk_alloc(trans, root->fs_info->extent_root,
+					     bytes + 2 * 1024 * 1024,
+					     alloc_target, 0);
+			btrfs_end_transaction(trans, root);
+			if (ret)
+				return ret;
+			goto again;
+		}
+		spin_unlock(&data_sinfo->lock);
+
+		/* commit the current transaction and try again */
+		if (!committed) {
+			committed = 1;
+			trans = btrfs_join_transaction(root, 1);
+			if (!trans)
+				return -ENOMEM;
+			ret = btrfs_commit_transaction(trans, root);
+			if (ret)
+				return ret;
+			goto again;
+		}
+
+		printk(KERN_ERR "no space left, need %llu, %llu delalloc bytes"
+		       ", %llu bytes_used, %llu bytes_reserved, "
+		       "%llu bytes_pinned, %llu bytes_readonly, %llu may use"
+		       "%llu total\n", bytes, data_sinfo->bytes_delalloc,
+		       data_sinfo->bytes_used, data_sinfo->bytes_reserved,
+		       data_sinfo->bytes_pinned, data_sinfo->bytes_readonly,
+		       data_sinfo->bytes_may_use, data_sinfo->total_bytes);
+		return -ENOSPC;
+	}
+	data_sinfo->bytes_may_use += bytes;
+	BTRFS_I(inode)->reserved_bytes += bytes;
+	spin_unlock(&data_sinfo->lock);
+
+	return btrfs_check_metadata_free_space(root);
+}
+
+/*
+ * if there was an error for whatever reason after calling
+ * btrfs_check_data_free_space, call this so we can cleanup the counters.
+ */
+void btrfs_free_reserved_data_space(struct btrfs_root *root,
+				    struct inode *inode, u64 bytes)
+{
+	struct btrfs_space_info *data_sinfo;
+
+	/* make sure bytes are sectorsize aligned */
+	bytes = (bytes + root->sectorsize - 1) & ~((u64)root->sectorsize - 1);
+
+	data_sinfo = BTRFS_I(inode)->space_info;
+	spin_lock(&data_sinfo->lock);
+	data_sinfo->bytes_may_use -= bytes;
+	BTRFS_I(inode)->reserved_bytes -= bytes;
+	spin_unlock(&data_sinfo->lock);
+}
+
+/* called when we are adding a delalloc extent to the inode's io_tree */
+void btrfs_delalloc_reserve_space(struct btrfs_root *root, struct inode *inode,
+				  u64 bytes)
+{
+	struct btrfs_space_info *data_sinfo;
+
+	/* get the space info for where this inode will be storing its data */
+	data_sinfo = BTRFS_I(inode)->space_info;
+
+	/* make sure we have enough space to handle the data first */
+	spin_lock(&data_sinfo->lock);
+	data_sinfo->bytes_delalloc += bytes;
+
+	/*
+	 * we are adding a delalloc extent without calling
+	 * btrfs_check_data_free_space first.  This happens on a weird
+	 * writepage condition, but shouldn't hurt our accounting
+	 */
+	if (unlikely(bytes > BTRFS_I(inode)->reserved_bytes)) {
+		data_sinfo->bytes_may_use -= BTRFS_I(inode)->reserved_bytes;
+		BTRFS_I(inode)->reserved_bytes = 0;
+	} else {
+		data_sinfo->bytes_may_use -= bytes;
+		BTRFS_I(inode)->reserved_bytes -= bytes;
+	}
+
+	spin_unlock(&data_sinfo->lock);
+}
+
+/* called when we are clearing an delalloc extent from the inode's io_tree */
+void btrfs_delalloc_free_space(struct btrfs_root *root, struct inode *inode,
+			      u64 bytes)
+{
+	struct btrfs_space_info *info;
+
+	info = BTRFS_I(inode)->space_info;
+
+	spin_lock(&info->lock);
+	info->bytes_delalloc -= bytes;
+	spin_unlock(&info->lock);
+}
+
 static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *extent_root, u64 alloc_bytes,
 			  u64 flags, int force)
@@ -2211,13 +2481,12 @@ static int finish_current_insert(struct btrfs_trans_handle *trans,
 	u64 end;
 	u64 priv;
 	u64 search = 0;
-	u64 skipped = 0;
 	struct btrfs_fs_info *info = extent_root->fs_info;
 	struct btrfs_path *path;
 	struct pending_extent_op *extent_op, *tmp;
 	struct list_head insert_list, update_list;
 	int ret;
-	int num_inserts = 0, max_inserts;
+	int num_inserts = 0, max_inserts, restart = 0;
 
 	path = btrfs_alloc_path();
 	INIT_LIST_HEAD(&insert_list);
@@ -2233,19 +2502,19 @@ again:
 		ret = find_first_extent_bit(&info->extent_ins, search, &start,
 					    &end, EXTENT_WRITEBACK);
 		if (ret) {
-			if (skipped && all && !num_inserts &&
+			if (restart && !num_inserts &&
 			    list_empty(&update_list)) {
-				skipped = 0;
+				restart = 0;
 				search = 0;
 				continue;
 			}
-			mutex_unlock(&info->extent_ins_mutex);
 			break;
 		}
 
 		ret = try_lock_extent(&info->extent_ins, start, end, GFP_NOFS);
 		if (!ret) {
-			skipped = 1;
+			if (all)
+				restart = 1;
 			search = end + 1;
 			if (need_resched()) {
 				mutex_unlock(&info->extent_ins_mutex);
@@ -2264,7 +2533,7 @@ again:
 			list_add_tail(&extent_op->list, &insert_list);
 			search = end + 1;
 			if (num_inserts == max_inserts) {
-				mutex_unlock(&info->extent_ins_mutex);
+				restart = 1;
 				break;
 			}
 		} else if (extent_op->type == PENDING_BACKREF_UPDATE) {
@@ -2280,7 +2549,6 @@ again:
 	 * somebody marked this thing for deletion then just unlock it and be
 	 * done, the free_extents will handle it
 	 */
-	mutex_lock(&info->extent_ins_mutex);
 	list_for_each_entry_safe(extent_op, tmp, &update_list, list) {
 		clear_extent_bits(&info->extent_ins, extent_op->bytenr,
 				  extent_op->bytenr + extent_op->num_bytes - 1,
@@ -2302,6 +2570,10 @@ again:
 	if (!list_empty(&update_list)) {
 		ret = update_backrefs(trans, extent_root, path, &update_list);
 		BUG_ON(ret);
+
+		/* we may have COW'ed new blocks, so lets start over */
+		if (all)
+			restart = 1;
 	}
 
 	/*
@@ -2309,9 +2581,9 @@ again:
 	 * need to make sure everything is cleaned then reset everything and
 	 * go back to the beginning
 	 */
-	if (!num_inserts && all && skipped) {
+	if (!num_inserts && restart) {
 		search = 0;
-		skipped = 0;
+		restart = 0;
 		INIT_LIST_HEAD(&update_list);
 		INIT_LIST_HEAD(&insert_list);
 		goto again;
@@ -2368,27 +2640,19 @@ again:
 	BUG_ON(ret);
 
 	/*
-	 * if we broke out of the loop in order to insert stuff because we hit
-	 * the maximum number of inserts at a time we can handle, then loop
-	 * back and pick up where we left off
+	 * if restart is set for whatever reason we need to go back and start
+	 * searching through the pending list again.
+	 *
+	 * We just inserted some extents, which could have resulted in new
+	 * blocks being allocated, which would result in new blocks needing
+	 * updates, so if all is set we _must_ restart to get the updated
+	 * blocks.
 	 */
-	if (num_inserts == max_inserts) {
-		INIT_LIST_HEAD(&insert_list);
-		INIT_LIST_HEAD(&update_list);
-		num_inserts = 0;
-		goto again;
-	}
-
-	/*
-	 * again, if we need to make absolutely sure there are no more pending
-	 * extent operations left and we know that we skipped some, go back to
-	 * the beginning and do it all again
-	 */
-	if (all && skipped) {
+	if (restart || all) {
 		INIT_LIST_HEAD(&insert_list);
 		INIT_LIST_HEAD(&update_list);
 		search = 0;
-		skipped = 0;
+		restart = 0;
 		num_inserts = 0;
 		goto again;
 	}
@@ -2709,6 +2973,8 @@ again:
 		goto again;
 	}
 
+	if (!err)
+		finish_current_insert(trans, extent_root, 0);
 	return err;
 }
 
@@ -2859,7 +3125,8 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 
 	if (data & BTRFS_BLOCK_GROUP_METADATA) {
 		last_ptr = &root->fs_info->last_alloc;
-		empty_cluster = 64 * 1024;
+		if (!btrfs_test_opt(root, SSD))
+			empty_cluster = 64 * 1024;
 	}
 
 	if ((data & BTRFS_BLOCK_GROUP_DATA) && btrfs_test_opt(root, SSD))
@@ -3091,6 +3358,10 @@ static void dump_space_info(struct btrfs_space_info *info, u64 bytes)
 	       (unsigned long long)(info->total_bytes - info->bytes_used -
 				    info->bytes_pinned - info->bytes_reserved),
 	       (info->full) ? "" : "not ");
+	printk(KERN_INFO "space_info total=%llu, pinned=%llu, delalloc=%llu,"
+	       " may_use=%llu, used=%llu\n", info->total_bytes,
+	       info->bytes_pinned, info->bytes_delalloc, info->bytes_may_use,
+	       info->bytes_used);
 
 	down_read(&info->groups_sem);
 	list_for_each_entry(cache, &info->block_groups, list) {
@@ -3117,24 +3388,10 @@ static int __btrfs_reserve_extent(struct btrfs_trans_handle *trans,
 {
 	int ret;
 	u64 search_start = 0;
-	u64 alloc_profile;
 	struct btrfs_fs_info *info = root->fs_info;
 
-	if (data) {
-		alloc_profile = info->avail_data_alloc_bits &
-			info->data_alloc_profile;
-		data = BTRFS_BLOCK_GROUP_DATA | alloc_profile;
-	} else if (root == root->fs_info->chunk_root) {
-		alloc_profile = info->avail_system_alloc_bits &
-			info->system_alloc_profile;
-		data = BTRFS_BLOCK_GROUP_SYSTEM | alloc_profile;
-	} else {
-		alloc_profile = info->avail_metadata_alloc_bits &
-			info->metadata_alloc_profile;
-		data = BTRFS_BLOCK_GROUP_METADATA | alloc_profile;
-	}
+	data = btrfs_get_alloc_profile(root, data);
 a