Btrfs: Metadata reservation for orphan inodes

reserve metadata space for handling orphan inodes

Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

9 files changed, 365 insertions, 66 deletions

diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index 40510d9351f..6ad63f17eca 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -151,6 +151,7 @@ struct btrfs_inode {
          * of these.
          */
         unsigned ordered_data_close:1;
+        unsigned orphan_meta_reserved:1;
         unsigned dummy_inode:1;
 
         /*
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 504d5daf2f1..fc324f9fcb4 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -1069,7 +1069,6 @@ struct btrfs_root {
         int ref_cows;
         int track_dirty;
         int in_radix;
-        int clean_orphans;
 
         u64 defrag_trans_start;
         struct btrfs_key defrag_progress;
@@ -1083,8 +1082,11 @@ struct btrfs_root {
 
         struct list_head root_list;
 
-        spinlock_t list_lock;
+        spinlock_t orphan_lock;
         struct list_head orphan_list;
+        struct btrfs_block_rsv *orphan_block_rsv;
+        int orphan_item_inserted;
+        int orphan_cleanup_state;
 
         spinlock_t inode_lock;
         /* red-black tree that keeps track of in-memory inodes */
@@ -2080,6 +2082,9 @@ int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans,
                                 int num_items, int *retries);
 void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
                                 struct btrfs_root *root);
+int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans,
+                                  struct inode *inode);
+void btrfs_orphan_release_metadata(struct inode *inode);
 int btrfs_snap_reserve_metadata(struct btrfs_trans_handle *trans,
                                 struct btrfs_pending_snapshot *pending);
 int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes);
@@ -2404,6 +2409,13 @@ int btrfs_update_inode(struct btrfs_trans_handle *trans,
 int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode);
 int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode);
 void btrfs_orphan_cleanup(struct btrfs_root *root);
+void btrfs_orphan_pre_snapshot(struct btrfs_trans_handle *trans,
+                                struct btrfs_pending_snapshot *pending,
+                                u64 *bytes_to_reserve);
+void btrfs_orphan_post_snapshot(struct btrfs_trans_handle *trans,
+                                struct btrfs_pending_snapshot *pending);
+void btrfs_orphan_commit_root(struct btrfs_trans_handle *trans,
+                              struct btrfs_root *root);
 int btrfs_cont_expand(struct inode *inode, loff_t size);
 int btrfs_invalidate_inodes(struct btrfs_root *root);
 void btrfs_add_delayed_iput(struct inode *inode);
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 309d8c08a64..82955b73a96 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -894,7 +894,8 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
         root->ref_cows = 0;
         root->track_dirty = 0;
         root->in_radix = 0;
-        root->clean_orphans = 0;
+        root->orphan_item_inserted = 0;
+        root->orphan_cleanup_state = 0;
 
         root->fs_info = fs_info;
         root->objectid = objectid;
@@ -904,12 +905,13 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
         root->in_sysfs = 0;
         root->inode_tree = RB_ROOT;
         root->block_rsv = NULL;
+        root->orphan_block_rsv = NULL;
 
         INIT_LIST_HEAD(&root->dirty_list);
         INIT_LIST_HEAD(&root->orphan_list);
         INIT_LIST_HEAD(&root->root_list);
         spin_lock_init(&root->node_lock);
-        spin_lock_init(&root->list_lock);
+        spin_lock_init(&root->orphan_lock);
         spin_lock_init(&root->inode_lock);
         spin_lock_init(&root->accounting_lock);
         mutex_init(&root->objectid_mutex);
@@ -1193,19 +1195,23 @@ again:
         if (root)
                 return root;
 
-        ret = btrfs_find_orphan_item(fs_info->tree_root, location->objectid);
-        if (ret == 0)
-                ret = -ENOENT;
-        if (ret < 0)
-                return ERR_PTR(ret);
-
         root = btrfs_read_fs_root_no_radix(fs_info->tree_root, location);
         if (IS_ERR(root))
                 return root;
 
-        WARN_ON(btrfs_root_refs(&root->root_item) == 0);
         set_anon_super(&root->anon_super, NULL);
 
+        if (btrfs_root_refs(&root->root_item) == 0) {
+                ret = -ENOENT;
+                goto fail;
+        }
+
+        ret = btrfs_find_orphan_item(fs_info->tree_root, location->objectid);
+        if (ret < 0)
+                goto fail;
+        if (ret == 0)
+                root->orphan_item_inserted = 1;
+
         ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
         if (ret)
                 goto fail;
@@ -1214,10 +1220,9 @@ again:
         ret = radix_tree_insert(&fs_info->fs_roots_radix,
                                 (unsigned long)root->root_key.objectid,
                                 root);
-        if (ret == 0) {
+        if (ret == 0)
                 root->in_radix = 1;
-                root->clean_orphans = 1;
-        }
+
         spin_unlock(&fs_info->fs_roots_radix_lock);
         radix_tree_preload_end();
         if (ret) {
@@ -1981,6 +1986,9 @@ struct btrfs_root *open_ctree(struct super_block *sb,
         BUG_ON(ret);
 
         if (!(sb->s_flags & MS_RDONLY)) {
+                ret = btrfs_cleanup_fs_roots(fs_info);
+                BUG_ON(ret);
+
                 ret = btrfs_recover_relocation(tree_root);
                 if (ret < 0) {
                         printk(KERN_WARNING
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index cb814a71800..a713f69f0c7 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -3626,6 +3626,34 @@ void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
         trans->bytes_reserved = 0;
 }
 
+int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans,
+                                  struct inode *inode)
+{
+        struct btrfs_root *root = BTRFS_I(inode)->root;
+        struct btrfs_block_rsv *src_rsv = get_block_rsv(trans, root);
+        struct btrfs_block_rsv *dst_rsv = root->orphan_block_rsv;
+
+        /*
+         * one for deleting orphan item, one for updating inode and
+         * two for calling btrfs_truncate_inode_items.
+         *
+         * btrfs_truncate_inode_items is a delete operation, it frees
+         * more space than it uses in most cases. So two units of
+         * metadata space should be enough for calling it many times.
+         * If all of the metadata space is used, we can commit
+         * transaction and use space it freed.
+         */
+        u64 num_bytes = calc_trans_metadata_size(root, 4);
+        return block_rsv_migrate_bytes(src_rsv, dst_rsv, num_bytes);
+}
+
+void btrfs_orphan_release_metadata(struct inode *inode)
+{
+        struct btrfs_root *root = BTRFS_I(inode)->root;
+        u64 num_bytes = calc_trans_metadata_size(root, 4);
+        btrfs_block_rsv_release(root, root->orphan_block_rsv, num_bytes);
+}
+
 int btrfs_snap_reserve_metadata(struct btrfs_trans_handle *trans,
                                 struct btrfs_pending_snapshot *pending)
 {
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index f9c40b83945..bef69bedf3c 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -1982,32 +1982,196 @@ void btrfs_run_delayed_iputs(struct btrfs_root *root)
 }
 
 /*
+ * calculate extra metadata reservation when snapshotting a subvolume
+ * contains orphan files.
+ */
+void btrfs_orphan_pre_snapshot(struct btrfs_trans_handle *trans,
+                                struct btrfs_pending_snapshot *pending,
+                                u64 *bytes_to_reserve)
+{
+        struct btrfs_root *root;
+        struct btrfs_block_rsv *block_rsv;
+        u64 num_bytes;
+        int index;
+
+        root = pending->root;
+        if (!root->orphan_block_rsv || list_empty(&root->orphan_list))
+                return;
+
+        block_rsv = root->orphan_block_rsv;
+
+        /* orphan block reservation for the snapshot */
+        num_bytes = block_rsv->size;
+
+        /*
+         * after the snapshot is created, COWing tree blocks may use more
+         * space than it frees. So we should make sure there is enough
+         * reserved space.
+         */
+        index = trans->transid & 0x1;
+        if (block_rsv->reserved + block_rsv->freed[index] < block_rsv->size) {
+                num_bytes += block_rsv->size -
+                             (block_rsv->reserved + block_rsv->freed[index]);
+        }
+
+        *bytes_to_reserve += num_bytes;
+}
+
+void btrfs_orphan_post_snapshot(struct btrfs_trans_handle *trans,
+                                struct btrfs_pending_snapshot *pending)
+{
+        struct btrfs_root *root = pending->root;
+        struct btrfs_root *snap = pending->snap;
+        struct btrfs_block_rsv *block_rsv;
+        u64 num_bytes;
+        int index;
+        int ret;
+
+        if (!root->orphan_block_rsv || list_empty(&root->orphan_list))
+                return;
+
+        /* refill source subvolume's orphan block reservation */
+        block_rsv = root->orphan_block_rsv;
+        index = trans->transid & 0x1;
+        if (block_rsv->reserved + block_rsv->freed[index] < block_rsv->size) {
+                num_bytes = block_rsv->size -
+                            (block_rsv->reserved + block_rsv->freed[index]);
+                ret = btrfs_block_rsv_migrate(&pending->block_rsv,
+                                              root->orphan_block_rsv,
+                                              num_bytes);
+                BUG_ON(ret);
+        }
+
+        /* setup orphan block reservation for the snapshot */
+        block_rsv = btrfs_alloc_block_rsv(snap);
+        BUG_ON(!block_rsv);
+
+        btrfs_add_durable_block_rsv(root->fs_info, block_rsv);
+        snap->orphan_block_rsv = block_rsv;
+
+        num_bytes = root->orphan_block_rsv->size;
+        ret = btrfs_block_rsv_migrate(&pending->block_rsv,
+                                      block_rsv, num_bytes);
+        BUG_ON(ret);
+
+#if 0
+        /* insert orphan item for the snapshot */
+        WARN_ON(!root->orphan_item_inserted);
+        ret = btrfs_insert_orphan_item(trans, root->fs_info->tree_root,
+                                       snap->root_key.objectid);
+        BUG_ON(ret);
+        snap->orphan_item_inserted = 1;
+#endif
+}
+
+enum btrfs_orphan_cleanup_state {
+        ORPHAN_CLEANUP_STARTED  = 1,
+        ORPHAN_CLEANUP_DONE     = 2,
+};
+
+/*
+ * This is called in transaction commmit time. If there are no orphan
+ * files in the subvolume, it removes orphan item and frees block_rsv
+ * structure.
+ */
+void btrfs_orphan_commit_root(struct btrfs_trans_handle *trans,
+                              struct btrfs_root *root)
+{
+        int ret;
+
+        if (!list_empty(&root->orphan_list) ||
+            root->orphan_cleanup_state != ORPHAN_CLEANUP_DONE)
+                return;
+
+        if (root->orphan_item_inserted &&
+            btrfs_root_refs(&root->root_item) > 0) {
+                ret = btrfs_del_orphan_item(trans, root->fs_info->tree_root,
+                                            root->root_key.objectid);
+                BUG_ON(ret);
+                root->orphan_item_inserted = 0;
+        }
+
+        if (root->orphan_block_rsv) {
+                WARN_ON(root->orphan_block_rsv->size > 0);
+                btrfs_free_block_rsv(root, root->orphan_block_rsv);
+                root->orphan_block_rsv = NULL;
+        }
+}
+
+/*
  * This creates an orphan entry for the given inode in case something goes
  * wrong in the middle of an unlink/truncate.
+ *
+ * NOTE: caller of this function should reserve 5 units of metadata for
+ *       this function.
  */
 int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode)
 {
         struct btrfs_root *root = BTRFS_I(inode)->root;
-        int ret = 0;
+        struct btrfs_block_rsv *block_rsv = NULL;
+        int reserve = 0;
+        int insert = 0;
+        int ret;
 
-        spin_lock(&root->list_lock);
+        if (!root->orphan_block_rsv) {
+                block_rsv = btrfs_alloc_block_rsv(root);
+                BUG_ON(!block_rsv);
+        }
 
-        /* already on the orphan list, we're good */
-        if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
-                spin_unlock(&root->list_lock);
-                return 0;
+        spin_lock(&root->orphan_lock);
+        if (!root->orphan_block_rsv) {
+                root->orphan_block_rsv = block_rsv;
+        } else if (block_rsv) {
+                btrfs_free_block_rsv(root, block_rsv);
+                block_rsv = NULL;
         }
 
-        list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list);
+        if (list_empty(&BTRFS_I(inode)->i_orphan)) {
+                list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list);
+#if 0
+                /*
+                 * For proper ENOSPC handling, we should do orphan
+                 * cleanup when mounting. But this introduces backward
+                 * compatibility issue.
+                 */
+                if (!xchg(&root->orphan_item_inserted, 1))
+                        insert = 2;
+                else
+                        insert = 1;
+#endif
+                insert = 1;
+        } else {
+                WARN_ON(!BTRFS_I(inode)->orphan_meta_reserved);
+        }
 
-        spin_unlock(&root->list_lock);
+        if (!BTRFS_I(inode)->orphan_meta_reserved) {
+                BTRFS_I(inode)->orphan_meta_reserved = 1;
+                reserve = 1;
+        }
+        spin_unlock(&root->orphan_lock);
 
-        /*
-         * insert an orphan item to track this unlinked/truncated file
-         */
-        ret = btrfs_insert_orphan_item(trans, root, inode->i_ino);
+        if (block_rsv)
+                btrfs_add_durable_block_rsv(root->fs_info, block_rsv);
 
-        return ret;
+        /* grab metadata reservation from transaction handle */
+        if (reserve) {
+                ret = btrfs_orphan_reserve_metadata(trans, inode);
+                BUG_ON(ret);
+        }
+
+        /* insert an orphan item to track this unlinked/truncated file */
+        if (insert >= 1) {
+                ret = btrfs_insert_orphan_item(trans, root, inode->i_ino);
+                BUG_ON(ret);
+        }
+
+        /* 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);
+        }
+        return 0;
 }
 
 /*
@@ -2017,26 +2181,31 @@ int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode)
 int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode)
 {
         struct btrfs_root *root = BTRFS_I(inode)->root;
+        int delete_item = 0;
+        int release_rsv = 0;
         int ret = 0;
 
-        spin_lock(&root->list_lock);
-
-        if (list_empty(&BTRFS_I(inode)->i_orphan)) {
-                spin_unlock(&root->list_lock);
-                return 0;
+        spin_lock(&root->orphan_lock);
+        if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
+                list_del_init(&BTRFS_I(inode)->i_orphan);
+                delete_item = 1;
         }
 
-        list_del_init(&BTRFS_I(inode)->i_orphan);
-        if (!trans) {
-                spin_unlock(&root->list_lock);
-                return 0;
+        if (BTRFS_I(inode)->orphan_meta_reserved) {
+                BTRFS_I(inode)->orphan_meta_reserved = 0;
+                release_rsv = 1;
         }
+        spin_unlock(&root->orphan_lock);
 
-        spin_unlock(&root->list_lock);
+        if (trans && delete_item) {
+                ret = btrfs_del_orphan_item(trans, root, inode->i_ino);
+                BUG_ON(ret);
+        }
 
-        ret = btrfs_del_orphan_item(trans, root, inode->i_ino);
+        if (release_rsv)
+                btrfs_orphan_release_metadata(inode);
 
-        return ret;
+        return 0;
 }
 
 /*
@@ -2053,7 +2222,7 @@ void btrfs_orphan_cleanup(struct btrfs_root *root)
         struct inode *inode;
         int ret = 0, nr_unlink = 0, nr_truncate = 0;
 
-        if (!xchg(&root->clean_orphans, 0))
+        if (cmpxchg(&root->orphan_cleanup_state, 0, ORPHAN_CLEANUP_STARTED))
                 return;
 
         path = btrfs_alloc_path();
@@ -2106,16 +2275,15 @@ void btrfs_orphan_cleanup(struct btrfs_root *root)
                 found_key.type = BTRFS_INODE_ITEM_KEY;
                 found_key.offset = 0;
                 inode = btrfs_iget(root->fs_info->sb, &found_key, root, NULL);
-                if (IS_ERR(inode))
-                        break;
+                BUG_ON(IS_ERR(inode));
 
                 /*
                  * add this inode to the orphan list so btrfs_orphan_del does
                  * the proper thing when we hit it
                  */
-                spin_lock(&root->list_lock);
+                spin_lock(&root->orphan_lock);
                 list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list);
-                spin_unlock(&root->list_lock);
+                spin_unlock(&root->orphan_lock);
 
                 /*
                  * if this is a bad inode, means we actually succeeded in
@@ -2142,13 +2310,23 @@ void btrfs_orphan_cleanup(struct btrfs_root *root)
                 /* this will do delete_inode and everything for us */
                 iput(inode);
         }
+        btrfs_free_path(path);
+
+        root->orphan_cleanup_state = ORPHAN_CLEANUP_DONE;
+
+        if (root->orphan_block_rsv)
+                btrfs_block_rsv_release(root, root->orphan_block_rsv,
+                                        (u64)-1);
+
+        if (root->orphan_block_rsv || root->orphan_item_inserted) {
+                trans = btrfs_join_transaction(root, 1);
+                btrfs_end_transaction(trans, root);
+        }
 
         if (nr_unlink)
                 printk(KERN_INFO "btrfs: unlinked %d orphans\n", nr_unlink);
         if (nr_truncate)
                 printk(KERN_INFO "btrfs: truncated %d orphans\n", nr_truncate);
-
-        btrfs_free_path(path);
 }
 
 /*
@@ -3181,6 +3359,7 @@ out:
         if (pending_del_nr) {
                 ret = btrfs_del_items(trans, root, path, pending_del_slot,
                                       pending_del_nr);
+                BUG_ON(ret);
         }
         btrfs_free_path(path);
         return err;
@@ -3386,7 +3565,10 @@ static int btrfs_setattr_size(struct inode *inode, struct iattr *attr)
                 }
         }
 
-        trans = btrfs_start_transaction(root, 1);
+        trans = btrfs_start_transaction(root, 5);
+        if (IS_ERR(trans))
+                return PTR_ERR(trans);
+
         btrfs_set_trans_block_group(trans, inode);
 
         ret = btrfs_orphan_add(trans, inode);
@@ -3406,8 +3588,11 @@ static int btrfs_setattr_size(struct inode *inode, struct iattr *attr)
                 i_size_write(inode, attr->ia_size);
                 btrfs_ordered_update_i_size(inode, inode->i_size, NULL);
 
-                trans = btrfs_start_transaction(root, 1);
+                trans = btrfs_start_transaction(root, 0);
+                BUG_ON(IS_ERR(trans));
                 btrfs_set_trans_block_group(trans, inode);
+                trans->block_rsv = root->orphan_block_rsv;
+                BUG_ON(!trans->block_rsv);
 
                 ret = btrfs_update_inode(trans, root, inode);
                 BUG_ON(ret);
@@ -3487,10 +3672,21 @@ void btrfs_delete_inode(struct inode *inode)
         btrfs_i_size_write(inode, 0);
 
         while (1) {
-                trans = btrfs_start_transaction(root, 1);
+                trans = btrfs_start_transaction(root, 0);
+                BUG_ON(IS_ERR(trans));
                 btrfs_set_trans_block_group(trans, inode);
-                ret = btrfs_truncate_inode_items(trans, root, inode, 0, 0);
+                trans->block_rsv = root->orphan_block_rsv;
+
+                ret = btrfs_block_rsv_check(trans, root,
+                                            root->orphan_block_rsv, 0, 5);
+                if (ret) {
+                        BUG_ON(ret != -EAGAIN);
+                        ret = btrfs_commit_transaction(trans, root);
+                        BUG_ON(ret);
+                        continue;
+                }
 
+                ret = btrfs_truncate_inode_items(trans, root, inode, 0, 0);
                 if (ret != -EAGAIN)
                         break;
 
@@ -3498,6 +3694,7 @@ void btrfs_delete_inode(struct inode *inode)
                 btrfs_end_transaction(trans, root);
                 trans = NULL;
                 btrfs_btree_balance_dirty(root, nr);
+
         }
 
         if (ret == 0) {
@@ -5247,8 +5444,10 @@ static void btrfs_truncate(struct inode *inode)
         btrfs_wait_ordered_range(inode, inode->i_size & (~mask), (u64)-1);
         btrfs_ordered_update_i_size(inode, inode->i_size, NULL);
 
-        trans = btrfs_start_transaction(root, 1);
+        trans = btrfs_start_transaction(root, 0);
+        BUG_ON(IS_ERR(trans));
         btrfs_set_trans_block_group(trans, inode);
+        trans->block_rsv = root->orphan_block_rsv;
 
         /*
          * setattr is responsible for setting the ordered_data_close flag,
@@ -5271,6 +5470,23 @@ static void btrfs_truncate(struct inode *inode)
                 btrfs_add_ordered_operation(trans, root, inode);
 
         while (1) {
+                if (!trans) {
+                        trans = btrfs_start_transaction(root, 0);
+                        BUG_ON(IS_ERR(trans));
+                        btrfs_set_trans_block_group(trans, inode);
+                        trans->block_rsv = root->orphan_block_rsv;
+                }
+
+                ret = btrfs_block_rsv_check(trans, root,
+                                            root->orphan_block_rsv, 0, 5);
+                if (ret) {
+                        BUG_ON(ret != -EAGAIN);
+                        ret = btrfs_commit_transaction(trans, root);
+                        BUG_ON(ret);
+                        trans = NULL;
+                        continue;
+                }
+
                 ret = btrfs_truncate_inode_items(trans, root, inode,
                                                  inode->i_size,
                                                  BTRFS_EXTENT_DATA_KEY);
@@ -5282,10 +5498,8 @@ static void btrfs_truncate(struct inode *inode)
 
                 nr = trans->blocks_used;
                 btrfs_end_transaction(trans, root);
+                trans = NULL;
                 btrfs_btree_balance_dirty(root, nr);
-
-                trans = btrfs_start_transaction(root, 1);
-                btrfs_set_trans_block_group(trans, inode);
         }
 
         if (ret == 0 && inode->i_nlink > 0) {
@@ -5371,6 +5585,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
         ei->reserved_extents = 0;
 
         ei->ordered_data_close = 0;
+        ei->orphan_meta_reserved = 0;
         ei->dummy_inode = 0;
         ei->force_compress = 0;
 
@@ -5417,13 +5632,13 @@ void btrfs_destroy_inode(struct inode *inode)
                 spin_unlock(&root->fs_info->ordered_extent_lock);
         }
 
-        spin_lock(&root->list_lock);
+        spin_lock(&root->orphan_lock);
         if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
                 printk(KERN_INFO "BTRFS: inode %lu still on the orphan list\n",
                        inode->i_ino);
                 list_del_init(&BTRFS_I(inode)->i_orphan);
         }
-        spin_unlock(&root->list_lock);
+        spin_unlock(&root->orphan_lock);
 
         while (1) {
                 ordered = btrfs_lookup_first_ordered_extent(inode, (u64)-1);
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index a0686658a3f..4cdb98cf26d 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -1297,10 +1297,12 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file,
         dest->root_item.drop_level = 0;
         btrfs_set_root_refs(&dest->root_item, 0);
 
-        ret = btrfs_insert_orphan_item(trans,
-                                root->fs_info->tree_root,
-                                dest->root_key.objectid);
-        BUG_ON(ret);
+        if (!xchg(&dest->orphan_item_inserted, 1)) {
+                ret = btrfs_insert_orphan_item(trans,
+                                        root->fs_info->tree_root,
+                                        dest->root_key.objectid);
+                BUG_ON(ret);
+        }
 
         ret = btrfs_commit_transaction(trans, root);
         BUG_ON(ret);
diff --git a/fs/btrfs/root-tree.c b/fs/btrfs/root-tree.c
index 67fa2d29d66..b91ccd97264 100644
--- a/fs/btrfs/root-tree.c
+++ b/fs/btrfs/root-tree.c
@@ -259,6 +259,8 @@ int btrfs_find_orphan_roots(struct btrfs_root *tree_root)
         struct extent_buffer *leaf;
         struct btrfs_path *path;
         struct btrfs_key key;
+        struct btrfs_key root_key;
+        struct btrfs_root *root;
         int err = 0;
         int ret;
 
@@ -270,6 +272,9 @@ int btrfs_find_orphan_roots(struct btrfs_root *tree_root)
         key.type = BTRFS_ORPHAN_ITEM_KEY;
         key.offset = 0;
 
+        root_key.type = BTRFS_ROOT_ITEM_KEY;
+        root_key.offset = (u64)-1;
+
         while (1) {
                 ret = btrfs_search_slot(NULL, tree_root, &key, path, 0, 0);
                 if (ret < 0) {
@@ -294,13 +299,25 @@ int btrfs_find_orphan_roots(struct btrfs_root *tree_root)
                     key.type != BTRFS_ORPHAN_ITEM_KEY)
                         break;
 
-                ret = btrfs_find_dead_roots(tree_root, key.offset);
-                if (ret) {
+                root_key.objectid = key.offset;
+                key.offset++;
+
+                root = btrfs_read_fs_root_no_name(tree_root->fs_info,
+                                                  &root_key);
+                if (!IS_ERR(root))
+                        continue;
+
+                ret = PTR_ERR(root);
+                if (ret != -ENOENT) {
                         err = ret;
                         break;
                 }
 
-                key.offset++;
+                ret = btrfs_find_dead_roots(tree_root, root_key.objectid);
+                if (ret) {
+                        err = ret;
+                        break;
+                }
         }
 
         btrfs_free_path(path);
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 38d91c7e8db..574285c8cbd 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -694,11 +694,11 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
                 if (btrfs_super_log_root(&root->fs_info->super_copy) != 0)
                         return -EINVAL;
 
-                /* recover relocation */
-                ret = btrfs_recover_relocation(root);
+                ret = btrfs_cleanup_fs_roots(root->fs_info);
                 WARN_ON(ret);
 
-                ret = btrfs_cleanup_fs_roots(root->fs_info);
+                /* recover relocation */
+                ret = btrfs_recover_relocation(root);
                 WARN_ON(ret);
 
                 sb->s_flags &= ~MS_RDONLY;
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 6217bb6d516..cfe7f588ef0 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -676,6 +676,7 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans,
 
                         btrfs_free_log(trans, root);
                         btrfs_update_reloc_root(trans, root);
+                        btrfs_orphan_commit_root(trans, root);
 
                         if (root->commit_root != root->node) {
                                 switch_commit_root(root);
@@ -835,6 +836,8 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
         struct extent_buffer *tmp;
         struct extent_buffer *old;
         int ret;
+        int retries = 0;
+        u64 to_reserve = 0;
         u64 index = 0;
         u64 objectid;
 
@@ -850,6 +853,17 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
                 goto fail;
         }
 
+        btrfs_orphan_pre_snapshot(trans, pending, &to_reserve);
+
+        if (to_reserve > 0) {
+                ret = btrfs_block_rsv_add(trans, root, &pending->block_rsv,
+                                          to_reserve, &retries);
+                if (ret) {
+                        pending->error = ret;
+                        goto fail;
+                }
+        }
+
         key.objectid = objectid;
         key.offset = (u64)-1;
         key.type = BTRFS_ROOT_ITEM_KEY;
@@ -909,6 +923,8 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
         key.offset = (u64)-1;
         pending->snap = btrfs_read_fs_root_no_name(root->fs_info, &key);
         BUG_ON(IS_ERR(pending->snap));
+
+        btrfs_orphan_post_snapshot(trans, pending);
 fail:
         kfree(new_root_item);
         btrfs_block_rsv_release(root, &pending->block_rsv, (u64)-1);