15 files changed, 474 insertions, 385 deletions

diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index 93b1aa932014..52d7eca8c7bf 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -121,9 +121,6 @@ struct btrfs_inode {
          */
         u64 index_cnt;
 
-        /* the start of block group preferred for allocations. */
-        u64 block_group;
-
         /* the fsync log has some corner cases that mean we have to check
          * directories to see if any unlinks have been done before
          * the directory was logged.  See tree-log.c for all the
diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index b0e18d986e0a..d84089349c82 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -43,8 +43,6 @@ struct btrfs_path *btrfs_alloc_path(void)
 {
         struct btrfs_path *path;
         path = kmem_cache_zalloc(btrfs_path_cachep, GFP_NOFS);
-        if (path)
-                path->reada = 1;
         return path;
 }
 
@@ -1224,6 +1222,7 @@ static void reada_for_search(struct btrfs_root *root,
         u64 search;
         u64 target;
         u64 nread = 0;
+        u64 gen;
         int direction = path->reada;
         struct extent_buffer *eb;
         u32 nr;
@@ -1251,6 +1250,15 @@ static void reada_for_search(struct btrfs_root *root,
         nritems = btrfs_header_nritems(node);
         nr = slot;
         while (1) {
+                if (!node->map_token) {
+                        unsigned long offset = btrfs_node_key_ptr_offset(nr);
+                        map_private_extent_buffer(node, offset,
+                                                  sizeof(struct btrfs_key_ptr),
+                                                  &node->map_token,
+                                                  &node->kaddr,
+                                                  &node->map_start,
+                                                  &node->map_len, KM_USER1);
+                }
                 if (direction < 0) {
                         if (nr == 0)
                                 break;
@@ -1268,14 +1276,23 @@ static void reada_for_search(struct btrfs_root *root,
                 search = btrfs_node_blockptr(node, nr);
                 if ((search <= target && target - search <= 65536) ||
                     (search > target && search - target <= 65536)) {
-                        readahead_tree_block(root, search, blocksize,
-                                     btrfs_node_ptr_generation(node, nr));
+                        gen = btrfs_node_ptr_generation(node, nr);
+                        if (node->map_token) {
+                                unmap_extent_buffer(node, node->map_token,
+                                                    KM_USER1);
+                                node->map_token = NULL;
+                        }
+                        readahead_tree_block(root, search, blocksize, gen);
                         nread += blocksize;
                 }
                 nscan++;
                 if ((nread > 65536 || nscan > 32))
                         break;
         }
+        if (node->map_token) {
+                unmap_extent_buffer(node, node->map_token, KM_USER1);
+                node->map_token = NULL;
+        }
 }
 
 /*
@@ -1648,9 +1665,6 @@ again:
                 }
 cow_done:
                 BUG_ON(!cow && ins_len);
-                if (level != btrfs_header_level(b))
-                        WARN_ON(1);
-                level = btrfs_header_level(b);
 
                 p->nodes[level] = b;
                 if (!p->skip_locking)
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 332323e19dd1..8f98c2005715 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -930,7 +930,6 @@ struct btrfs_fs_info {
          * is required instead of the faster short fsync log commits
          */
         u64 last_trans_log_full_commit;
-        u64 open_ioctl_trans;
         unsigned long mount_opt:20;
         unsigned long compress_type:4;
         u64 max_inline;
@@ -947,7 +946,6 @@ struct btrfs_fs_info {
         struct super_block *sb;
         struct inode *btree_inode;
         struct backing_dev_info bdi;
-        struct mutex trans_mutex;
         struct mutex tree_log_mutex;
         struct mutex transaction_kthread_mutex;
         struct mutex cleaner_mutex;
@@ -968,6 +966,7 @@ struct btrfs_fs_info {
         struct rw_semaphore subvol_sem;
         struct srcu_struct subvol_srcu;
 
+        spinlock_t trans_lock;
         struct list_head trans_list;
         struct list_head hashers;
         struct list_head dead_roots;
@@ -980,6 +979,7 @@ struct btrfs_fs_info {
         atomic_t async_submit_draining;
         atomic_t nr_async_bios;
         atomic_t async_delalloc_pages;
+        atomic_t open_ioctl_trans;
 
         /*
          * this is used by the balancing code to wait for all the pending
@@ -1044,6 +1044,7 @@ struct btrfs_fs_info {
         int closing;
         int log_root_recovering;
         int enospc_unlink;
+        int trans_no_join;
 
         u64 total_pinned;
 
@@ -1065,7 +1066,6 @@ struct btrfs_fs_info {
         struct reloc_control *reloc_ctl;
 
         spinlock_t delalloc_lock;
-        spinlock_t new_trans_lock;
         u64 delalloc_bytes;
 
         /* data_alloc_cluster is only used in ssd mode */
@@ -2238,6 +2238,9 @@ int btrfs_block_rsv_migrate(struct btrfs_block_rsv *src_rsv,
 void btrfs_block_rsv_release(struct btrfs_root *root,
                              struct btrfs_block_rsv *block_rsv,
                              u64 num_bytes);
+int btrfs_truncate_reserve_metadata(struct btrfs_trans_handle *trans,
+                                    struct btrfs_root *root,
+                                    struct btrfs_block_rsv *rsv);
 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,
@@ -2512,8 +2515,7 @@ int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end,
 int btrfs_writepages(struct address_space *mapping,
                      struct writeback_control *wbc);
 int btrfs_create_subvol_root(struct btrfs_trans_handle *trans,
-                             struct btrfs_root *new_root,
-                             u64 new_dirid, u64 alloc_hint);
+                             struct btrfs_root *new_root, u64 new_dirid);
 int btrfs_merge_bio_hook(struct page *page, unsigned long offset,
                          size_t size, struct bio *bio, unsigned long bio_flags);
 
diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c
index 01e29503a54b..b46d94d1dea8 100644
--- a/fs/btrfs/delayed-inode.c
+++ b/fs/btrfs/delayed-inode.c
@@ -1129,7 +1129,7 @@ static void btrfs_async_run_delayed_node_done(struct btrfs_work *work)
         delayed_node = async_node->delayed_node;
         root = delayed_node->root;
 
-        trans = btrfs_join_transaction(root, 0);
+        trans = btrfs_join_transaction(root);
         if (IS_ERR(trans))
                 goto free_path;
 
@@ -1572,8 +1572,7 @@ static void fill_stack_inode_item(struct btrfs_trans_handle *trans,
         btrfs_set_stack_inode_transid(inode_item, trans->transid);
         btrfs_set_stack_inode_rdev(inode_item, inode->i_rdev);
         btrfs_set_stack_inode_flags(inode_item, BTRFS_I(inode)->flags);
-        btrfs_set_stack_inode_block_group(inode_item,
-                                          BTRFS_I(inode)->block_group);
+        btrfs_set_stack_inode_block_group(inode_item, 0);
 
         btrfs_set_stack_timespec_sec(btrfs_inode_atime(inode_item),
                                      inode->i_atime.tv_sec);
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 98b6a71decba..a203d363184d 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1505,24 +1505,24 @@ static int transaction_kthread(void *arg)
                 vfs_check_frozen(root->fs_info->sb, SB_FREEZE_WRITE);
                 mutex_lock(&root->fs_info->transaction_kthread_mutex);
 
-                spin_lock(&root->fs_info->new_trans_lock);
+                spin_lock(&root->fs_info->trans_lock);
                 cur = root->fs_info->running_transaction;
                 if (!cur) {
-                        spin_unlock(&root->fs_info->new_trans_lock);
+                        spin_unlock(&root->fs_info->trans_lock);
                         goto sleep;
                 }
 
                 now = get_seconds();
                 if (!cur->blocked &&
                     (now < cur->start_time || now - cur->start_time < 30)) {
-                        spin_unlock(&root->fs_info->new_trans_lock);
+                        spin_unlock(&root->fs_info->trans_lock);
                         delay = HZ * 5;
                         goto sleep;
                 }
                 transid = cur->transid;
-                spin_unlock(&root->fs_info->new_trans_lock);
+                spin_unlock(&root->fs_info->trans_lock);
 
-                trans = btrfs_join_transaction(root, 1);
+                trans = btrfs_join_transaction(root);
                 BUG_ON(IS_ERR(trans));
                 if (transid == trans->transid) {
                         ret = btrfs_commit_transaction(trans, root);
@@ -1613,7 +1613,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
         INIT_LIST_HEAD(&fs_info->ordered_operations);
         INIT_LIST_HEAD(&fs_info->caching_block_groups);
         spin_lock_init(&fs_info->delalloc_lock);
-        spin_lock_init(&fs_info->new_trans_lock);
+        spin_lock_init(&fs_info->trans_lock);
         spin_lock_init(&fs_info->ref_cache_lock);
         spin_lock_init(&fs_info->fs_roots_radix_lock);
         spin_lock_init(&fs_info->delayed_iput_lock);
@@ -1645,6 +1645,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
         fs_info->max_inline = 8192 * 1024;
         fs_info->metadata_ratio = 0;
         fs_info->defrag_inodes = RB_ROOT;
+        fs_info->trans_no_join = 0;
 
         fs_info->thread_pool_size = min_t(unsigned long,
                                           num_online_cpus() + 2, 8);
@@ -1709,7 +1710,6 @@ struct btrfs_root *open_ctree(struct super_block *sb,
         fs_info->do_barriers = 1;
 
 
-        mutex_init(&fs_info->trans_mutex);
         mutex_init(&fs_info->ordered_operations_mutex);
         mutex_init(&fs_info->tree_log_mutex);
         mutex_init(&fs_info->chunk_mutex);
@@ -2479,13 +2479,13 @@ int btrfs_commit_super(struct btrfs_root *root)
         down_write(&root->fs_info->cleanup_work_sem);
         up_write(&root->fs_info->cleanup_work_sem);
 
-        trans = btrfs_join_transaction(root, 1);
+        trans = btrfs_join_transaction(root);
         if (IS_ERR(trans))
                 return PTR_ERR(trans);
         ret = btrfs_commit_transaction(trans, root);
         BUG_ON(ret);
         /* run commit again to drop the original snapshot */
-        trans = btrfs_join_transaction(root, 1);
+        trans = btrfs_join_transaction(root);
         if (IS_ERR(trans))
                 return PTR_ERR(trans);
         btrfs_commit_transaction(trans, root);
@@ -3024,10 +3024,13 @@ static int btrfs_cleanup_transaction(struct btrfs_root *root)
 
         WARN_ON(1);
 
-        mutex_lock(&root->fs_info->trans_mutex);
         mutex_lock(&root->fs_info->transaction_kthread_mutex);
 
+        spin_lock(&root->fs_info->trans_lock);
         list_splice_init(&root->fs_info->trans_list, &list);
+        root->fs_info->trans_no_join = 1;
+        spin_unlock(&root->fs_info->trans_lock);
+
         while (!list_empty(&list)) {
                 t = list_entry(list.next, struct btrfs_transaction, list);
                 if (!t)
@@ -3052,23 +3055,18 @@ static int btrfs_cleanup_transaction(struct btrfs_root *root)
                 t->blocked = 0;
                 if (waitqueue_active(&root->fs_info->transaction_wait))
                         wake_up(&root->fs_info->transaction_wait);
-                mutex_unlock(&root->fs_info->trans_mutex);
 
-                mutex_lock(&root->fs_info->trans_mutex);
                 t->commit_done = 1;
                 if (waitqueue_active(&t->commit_wait))
                         wake_up(&t->commit_wait);
-                mutex_unlock(&root->fs_info->trans_mutex);
-
-                mutex_lock(&root->fs_info->trans_mutex);
 
                 btrfs_destroy_pending_snapshots(t);
 
                 btrfs_destroy_delalloc_inodes(root);
 
-                spin_lock(&root->fs_info->new_trans_lock);
+                spin_lock(&root->fs_info->trans_lock);
                 root->fs_info->running_transaction = NULL;
-                spin_unlock(&root->fs_info->new_trans_lock);
+                spin_unlock(&root->fs_info->trans_lock);
 
                 btrfs_destroy_marked_extents(root, &t->dirty_pages,
                                              EXTENT_DIRTY);
@@ -3082,8 +3080,10 @@ static int btrfs_cleanup_transaction(struct btrfs_root *root)
                 kmem_cache_free(btrfs_transaction_cachep, t);
         }
 
+        spin_lock(&root->fs_info->trans_lock);
+        root->fs_info->trans_no_join = 0;
+        spin_unlock(&root->fs_info->trans_lock);
         mutex_unlock(&root->fs_info->transaction_kthread_mutex);
-        mutex_unlock(&root->fs_info->trans_mutex);
 
         return 0;
 }
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 169bd62ce776..c9173a7827b0 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -348,7 +348,7 @@ static int caching_kthread(void *data)
          */
         path->skip_locking = 1;
         path->search_commit_root = 1;
-        path->reada = 2;
+        path->reada = 1;
 
         key.objectid = last;
         key.offset = 0;
@@ -379,15 +379,18 @@ again:
                         if (ret)
                                 break;
 
-                        caching_ctl->progress = last;
-                        btrfs_release_path(path);
-                        up_read(&fs_info->extent_commit_sem);
-                        mutex_unlock(&caching_ctl->mutex);
-                        if (btrfs_transaction_in_commit(fs_info))
-                                schedule_timeout(1);
-                        else
+                        if (need_resched() ||
+                            btrfs_next_leaf(extent_root, path)) {
+                                caching_ctl->progress = last;
+                                btrfs_release_path(path);
+                                up_read(&fs_info->extent_commit_sem);
+                                mutex_unlock(&caching_ctl->mutex);
                                 cond_resched();
-                        goto again;
+                                goto again;
+                        }
+                        leaf = path->nodes[0];
+                        nritems = btrfs_header_nritems(leaf);
+                        continue;
                 }
 
                 if (key.objectid < block_group->key.objectid) {
@@ -3065,7 +3068,7 @@ again:
                         spin_unlock(&data_sinfo->lock);
 alloc:
                         alloc_target = btrfs_get_alloc_profile(root, 1);
-                        trans = btrfs_join_transaction(root, 1);
+                        trans = btrfs_join_transaction(root);
                         if (IS_ERR(trans))
                                 return PTR_ERR(trans);
 
@@ -3091,9 +3094,10 @@ alloc:
 
                 /* commit the current transaction and try again */
 commit_trans:
-                if (!committed && !root->fs_info->open_ioctl_trans) {
+                if (!committed &&
+                    !atomic_read(&root->fs_info->open_ioctl_trans)) {
                         committed = 1;
-                        trans = btrfs_join_transaction(root, 1);
+                        trans = btrfs_join_transaction(root);
                         if (IS_ERR(trans))
                                 return PTR_ERR(trans);
                         ret = btrfs_commit_transaction(trans, root);
@@ -3472,7 +3476,7 @@ again:
                 goto out;
 
         ret = -ENOSPC;
-        trans = btrfs_join_transaction(root, 1);
+        trans = btrfs_join_transaction(root);
         if (IS_ERR(trans))
                 goto out;
         ret = btrfs_commit_transaction(trans, root);
@@ -3699,7 +3703,7 @@ int btrfs_block_rsv_check(struct btrfs_trans_handle *trans,
                 if (trans)
                         return -EAGAIN;
 
-                trans = btrfs_join_transaction(root, 1);
+                trans = btrfs_join_transaction(root);
                 BUG_ON(IS_ERR(trans));
                 ret = btrfs_commit_transaction(trans, root);
                 return 0;
@@ -3837,6 +3841,37 @@ static void release_global_block_rsv(struct btrfs_fs_info *fs_info)
         WARN_ON(fs_info->chunk_block_rsv.reserved > 0);
 }
 
+int btrfs_truncate_reserve_metadata(struct btrfs_trans_handle *trans,
+                                    struct btrfs_root *root,
+                                    struct btrfs_block_rsv *rsv)
+{
+        struct btrfs_block_rsv *trans_rsv = &root->fs_info->trans_block_rsv;
+        u64 num_bytes;
+        int ret;
+
+        /*
+         * Truncate should be freeing data, but give us 2 items just in case it
+         * needs to use some space.  We may want to be smarter about this in the
+         * future.
+         */
+        num_bytes = btrfs_calc_trans_metadata_size(root, 2);
+
+        /* We already have enough bytes, just return */
+        if (rsv->reserved >= num_bytes)
+                return 0;
+
+        num_bytes -= rsv->reserved;
+
+        /*
+         * You should have reserved enough space before hand to do this, so this
+         * should not fail.
+         */
+        ret = block_rsv_migrate_bytes(trans_rsv, rsv, num_bytes);
+        BUG_ON(ret);
+
+        return 0;
+}
+
 int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans,
                                  struct btrfs_root *root,
                                  int num_items)
@@ -3877,23 +3912,18 @@ int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans,
         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.
+         * We need to hold space in order to delete our orphan item once we've
+         * added it, so this takes the reservation so we can release it later
+         * when we are truly done with the orphan item.
          */
-        u64 num_bytes = btrfs_calc_trans_metadata_size(root, 4);
+        u64 num_bytes = btrfs_calc_trans_metadata_size(root, 1);
         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 = btrfs_calc_trans_metadata_size(root, 4);
+        u64 num_bytes = btrfs_calc_trans_metadata_size(root, 1);
         btrfs_block_rsv_release(root, root->orphan_block_rsv, num_bytes);
 }
 
@@ -4987,6 +5017,15 @@ have_block_group:
                 if (unlikely(block_group->ro))
                         goto loop;
 
+                spin_lock(&block_group->free_space_ctl->tree_lock);
+                if (cached &&
+                    block_group->free_space_ctl->free_space <
+                    num_bytes + empty_size) {
+                        spin_unlock(&block_group->free_space_ctl->tree_lock);
+                        goto loop;
+                }
+                spin_unlock(&block_group->free_space_ctl->tree_lock);
+
                 /*
                  * Ok we want to try and use the cluster allocator, so lets look
                  * there, unless we are on LOOP_NO_EMPTY_SIZE, since we will
@@ -5150,6 +5189,7 @@ checks:
                         btrfs_add_free_space(block_group, offset,
                                              search_start - offset);
                 BUG_ON(offset > search_start);
+                btrfs_put_block_group(block_group);
                 break;
 loop:
                 failed_cluster_refill = false;
@@ -5242,14 +5282,7 @@ loop:
                 ret = -ENOSPC;
         } else if (!ins->objectid) {
                 ret = -ENOSPC;
-        }
-
-        /* we found what we needed */
-        if (ins->objectid) {
-                if (!(data & BTRFS_BLOCK_GROUP_DATA))
-                        trans->block_group = block_group->key.objectid;
-
-                btrfs_put_block_group(block_group);
+        } else if (ins->objectid) {
                 ret = 0;
         }
 
@@ -6526,7 +6559,7 @@ int btrfs_set_block_group_ro(struct btrfs_root *root,
 
         BUG_ON(cache->ro);
 
-        trans = btrfs_join_transaction(root, 1);
+        trans = btrfs_join_transaction(root);
         BUG_ON(IS_ERR(trans));
 
         alloc_flags = update_block_group_flags(root, cache->flags);
@@ -6882,6 +6915,7 @@ int btrfs_read_block_groups(struct btrfs_root *root)
         path = btrfs_alloc_path();
         if (!path)
                 return -ENOMEM;
+        path->reada = 1;
 
         cache_gen = btrfs_super_cache_generation(&root->fs_info->super_copy);
         if (cache_gen != 0 &&
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 0e0fe0f6ec75..b181a94a7170 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -1474,7 +1474,7 @@ u64 count_range_bits(struct extent_io_tree *tree,
                         if (total_bytes >= max_bytes)
                                 break;
                         if (!found) {
-                                *start = state->start;
+                                *start = max(cur_start, state->start);
                                 found = 1;
                         }
                         last = state->end;
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index c6a22d783c35..e3a1b0c2394c 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1480,14 +1480,12 @@ int btrfs_sync_file(struct file *file, int datasync)
          * the current transaction, we can bail out now without any
          * syncing
          */
-        mutex_lock(&root->fs_info->trans_mutex);
+        smp_mb();
         if (BTRFS_I(inode)->last_trans <=
             root->fs_info->last_trans_committed) {
                 BTRFS_I(inode)->last_trans = 0;
-                mutex_unlock(&root->fs_info->trans_mutex);
                 goto out;
         }
-        mutex_unlock(&root->fs_info->trans_mutex);
 
         /*
          * ok we haven't committed the transaction yet, lets do a commit
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 70d45795d758..dd38d4c3a599 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -402,7 +402,14 @@ int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
                                 spin_lock(&ctl->tree_lock);
                                 ret = link_free_space(ctl, e);
                                 spin_unlock(&ctl->tree_lock);
-                                BUG_ON(ret);
+                                if (ret) {
+                                        printk(KERN_ERR "Duplicate entries in "
+                                               "free space cache, dumping\n");
+                                        kunmap(page);
+                                        unlock_page(page);
+                                        page_cache_release(page);
+                                        goto free_cache;
+                                }
                         } else {
                                 e->bitmap = kzalloc(PAGE_CACHE_SIZE, GFP_NOFS);
                                 if (!e->bitmap) {
@@ -419,6 +426,14 @@ int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
                                 ctl->op->recalc_thresholds(ctl);
                                 spin_unlock(&ctl->tree_lock);
                                 list_add_tail(&e->list, &bitmaps);
+                                if (ret) {
+                                        printk(KERN_ERR "Duplicate entries in "
+                                               "free space cache, dumping\n");
+                                        kunmap(page);
+                                        unlock_page(page);
+                                        page_cache_release(page);
+                                        goto free_cache;
+                                }
                         }
 
                         num_entries--;
@@ -963,10 +978,16 @@ static int tree_insert_offset(struct rb_root *root, u64 offset,
                          * logically.
                          */
                         if (bitmap) {
-                                WARN_ON(info->bitmap);
+                                if (info->bitmap) {
+                                        WARN_ON_ONCE(1);
+                                        return -EEXIST;
+                                }
                                 p = &(*p)->rb_right;
                         } else {
-                                WARN_ON(!info->bitmap);
+                                if (!info->bitmap) {
+                                        WARN_ON_ONCE(1);
+                                        return -EEXIST;
+                                }
                                 p = &(*p)->rb_left;
                         }
                 }
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index bb51bb1fa44f..a83e44bf3206 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -138,7 +138,6 @@ static noinline int insert_inline_extent(struct btrfs_trans_handle *trans,
                 return -ENOMEM;
 
         path->leave_spinning = 1;
-        btrfs_set_trans_block_group(trans, inode);
 
         key.objectid = btrfs_ino(inode);
         key.offset = start;
@@ -426,9 +425,8 @@ again:
                 }
         }
         if (start == 0) {
-                trans = btrfs_join_transaction(root, 1);
+                trans = btrfs_join_transaction(root);
                 BUG_ON(IS_ERR(trans));
-                btrfs_set_trans_block_group(trans, inode);
                 trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 
                 /* lets try to make an inline extent */
@@ -623,8 +621,9 @@ retry:
                             async_extent->start + async_extent->ram_size - 1,
                             GFP_NOFS);
 
-                trans = btrfs_join_transaction(root, 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,
                                            async_extent->compressed_size,
                                            async_extent->compressed_size,
@@ -793,9 +792,8 @@ static noinline int cow_file_range(struct inode *inode,
         int ret = 0;
 
         BUG_ON(is_free_space_inode(root, inode));
-        trans = btrfs_join_transaction(root, 1);
+        trans = btrfs_join_transaction(root);
         BUG_ON(IS_ERR(trans));
-        btrfs_set_trans_block_group(trans, inode);
         trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 
         num_bytes = (end - start + blocksize) & ~(blocksize - 1);
@@ -1077,10 +1075,12 @@ static noinline int run_delalloc_nocow(struct inode *inode,
         nolock = is_free_space_inode(root, inode);
 
         if (nolock)
-                trans = btrfs_join_transaction_nolock(root, 1);
+                trans = btrfs_join_transaction_nolock(root);
         else
-                trans = btrfs_join_transaction(root, 1);
+                trans = btrfs_join_transaction(root);
+
         BUG_ON(IS_ERR(trans));
+        trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 
         cow_start = (u64)-1;
         cur_offset = start;
@@ -1519,8 +1519,6 @@ static noinline int add_pending_csums(struct btrfs_trans_handle *trans,
 {
         struct btrfs_ordered_sum *sum;
 
-        btrfs_set_trans_block_group(trans, inode);
-
         list_for_each_entry(sum, list, list) {
                 btrfs_csum_file_blocks(trans,
                        BTRFS_I(inode)->root->fs_info->csum_root, sum);
@@ -1735,11 +1733,10 @@ 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) {
                         if (nolock)
-                                trans = btrfs_join_transaction_nolock(root, 1);
+                                trans = btrfs_join_transaction_nolock(root);
                         else
-                                trans = btrfs_join_transaction(root, 1);
+                                trans = btrfs_join_transaction(root);
                         BUG_ON(IS_ERR(trans));
-                        btrfs_set_trans_block_group(trans, inode);
                         trans->block_rsv = &root->fs_info->delalloc_block_rsv;
                         ret = btrfs_update_inode(trans, root, inode);
                         BUG_ON(ret);
@@ -1752,11 +1749,10 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
                          0, &cached_state, GFP_NOFS);
 
         if (nolock)
-                trans = btrfs_join_transaction_nolock(root, 1);
+                trans = btrfs_join_transaction_nolock(root);
         else
-                trans = btrfs_join_transaction(root, 1);
+                trans = btrfs_join_transaction(root);
         BUG_ON(IS_ERR(trans));
-        btrfs_set_trans_block_group(trans, inode);
         trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 
         if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags))
@@ -2431,7 +2427,7 @@ int btrfs_orphan_cleanup(struct btrfs_root *root)
                                         (u64)-1);
 
         if (root->orphan_block_rsv || root->orphan_item_inserted) {
-                trans = btrfs_join_transaction(root, 1);
+                trans = btrfs_join_transaction(root);
                 if (!IS_ERR(trans))
                         btrfs_end_transaction(trans, root);
         }
@@ -2511,12 +2507,12 @@ static void btrfs_read_locked_inode(struct inode *inode)
         struct btrfs_root *root = BTRFS_I(inode)->root;
         struct btrfs_key location;
         int maybe_acls;
-        u64 alloc_group_block;
         u32 rdev;
         int ret;
 
         path = btrfs_alloc_path();
         BUG_ON(!path);
+        path->leave_spinning = 1;
         memcpy(&location, &BTRFS_I(inode)->location, sizeof(location));
 
         ret = btrfs_lookup_inode(NULL, root, path, &location, 0);
@@ -2526,6 +2522,12 @@ static void btrfs_read_locked_inode(struct inode *inode)
         leaf = path->nodes[0];
         inode_item = btrfs_item_ptr(leaf, path->slots[0],
                                     struct btrfs_inode_item);
+        if (!leaf->map_token)
+                map_private_extent_buffer(leaf, (unsigned long)inode_item,
+                                          sizeof(struct btrfs_inode_item),
+                                          &leaf->map_token, &leaf->kaddr,
+                                          &leaf->map_start, &leaf->map_len,
+                                          KM_USER1);
 
         inode->i_mode = btrfs_inode_mode(leaf, inode_item);
         inode->i_nlink = btrfs_inode_nlink(leaf, inode_item);
@@ -2555,8 +2557,6 @@ static void btrfs_read_locked_inode(struct inode *inode)
         BTRFS_I(inode)->index_cnt = (u64)-1;
         BTRFS_I(inode)->flags = btrfs_inode_flags(leaf, inode_item);
 
-        alloc_group_block = btrfs_inode_block_group(leaf, inode_item);
-
         /*
          * try to precache a NULL acl entry for files that don't have
          * any xattrs or acls
@@ -2566,8 +2566,11 @@ static void btrfs_read_locked_inode(struct inode *inode)
         if (!maybe_acls)
                 cache_no_acl(inode);
 
-        BTRFS_I(inode)->block_group = btrfs_find_block_group(root, 0,
-                                                alloc_group_block, 0);
+        if (leaf->map_token) {
+                unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
+                leaf->map_token = NULL;
+        }
+
         btrfs_free_path(path);
         inode_item = NULL;
 
@@ -2647,7 +2650,7 @@ static void fill_inode_item(struct btrfs_trans_handle *trans,
         btrfs_set_inode_transid(leaf, item, trans->transid);
         btrfs_set_inode_rdev(leaf, item, inode->i_rdev);
         btrfs_set_inode_flags(leaf, item, BTRFS_I(inode)->flags);
-        btrfs_set_inode_block_group(leaf, item, BTRFS_I(inode)->block_group);
+        btrfs_set_inode_block_group(leaf, item, 0);
 
         if (leaf->map_token) {
                 unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
@@ -3004,8 +3007,6 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
         if (IS_ERR(trans))
                 return PTR_ERR(trans);
 
-        btrfs_set_trans_block_group(trans, dir);
-
         btrfs_record_unlink_dir(trans, dir, dentry->d_inode, 0);
 
         ret = btrfs_unlink_inode(trans, root, dir, dentry->d_inode,
@@ -3094,8 +3095,6 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
         if (IS_ERR(trans))
                 return PTR_ERR(trans);
 
-        btrfs_set_trans_block_group(trans, dir);
-
         if (unlikely(btrfs_ino(inode) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) {
                 err = btrfs_unlink_subvol(trans, root, dir,
                                           BTRFS_I(inode)->location.objectid,
@@ -3514,7 +3513,6 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
                                 err = PTR_ERR(trans);
                                 break;
                         }
-                        btrfs_set_trans_block_group(trans, inode);
 
                         err = btrfs_drop_extents(trans, inode, cur_offset,
                                                  cur_offset + hole_size,
@@ -3650,7 +3648,6 @@ void btrfs_evict_inode(struct inode *inode)
         while (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;
 
                 ret = btrfs_block_rsv_check(trans, root,
@@ -4133,7 +4130,8 @@ static int btrfs_real_readdir(struct file *filp, void *dirent,
         path = btrfs_alloc_path();
         if (!path)
                 return -ENOMEM;
-        path->reada = 2;
+
+        path->reada = 1;
 
         if (key_type == BTRFS_DIR_INDEX_KEY) {
                 INIT_LIST_HEAD(&ins_list);
@@ -4274,12 +4272,11 @@ int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc)
 
         if (wbc->sync_mode == WB_SYNC_ALL) {
                 if (nolock)
-                        trans = btrfs_join_transaction_nolock(root, 1);
+                        trans = btrfs_join_transaction_nolock(root);
                 else
-                        trans = btrfs_join_transaction(root, 1);
+                        trans = btrfs_join_transaction(root);
                 if (IS_ERR(trans))
                         return PTR_ERR(trans);
-                btrfs_set_trans_block_group(trans, inode);
                 if (nolock)
                         ret = btrfs_end_transaction_nolock(trans, root);
                 else
@@ -4303,9 +4300,8 @@ void btrfs_dirty_inode(struct inode *inode)
         if (BTRFS_I(inode)->dummy_inode)
                 return;
 
-        trans = btrfs_join_transaction(root, 1);
+        trans = btrfs_join_transaction(root);
         BUG_ON(IS_ERR(trans));
-        btrfs_set_trans_block_group(trans, inode);
 
         ret = btrfs_update_inode(trans, root, inode);
         if (ret && ret == -ENOSPC) {
@@ -4319,7 +4315,6 @@ void btrfs_dirty_inode(struct inode *inode)
                                        PTR_ERR(trans));
                         return;
                 }
-                btrfs_set_trans_block_group(trans, inode);
 
                 ret = btrfs_update_inode(trans, root, inode);
                 if (ret) {
@@ -4418,8 +4413,8 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
                                      struct btrfs_root *root,
                                      struct inode *dir,
                                      const char *name, int name_len,
-                                     u64 ref_objectid, u64 objectid,
-                                     u64 alloc_hint, int mode, u64 *index)
+                                     u64 ref_objectid, u64 objectid, int mode,
+                                     u64 *index)
 {
         struct inode *inode;
         struct btrfs_inode_item *inode_item;
@@ -4472,8 +4467,6 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
                 owner = 0;
         else
                 owner = 1;
-        BTRFS_I(inode)->block_group =
-                        btrfs_find_block_group(root, 0, alloc_hint, owner);
 
         key[0].objectid = objectid;
         btrfs_set_key_type(&key[0], BTRFS_INODE_ITEM_KEY);
@@ -4629,15 +4622,13 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry,
         if (IS_ERR(trans))
                 return PTR_ERR(trans);
 
-        btrfs_set_trans_block_group(trans, dir);
-
         err = btrfs_find_free_ino(root, &objectid);
         if (err)
                 goto out_unlock;
 
         inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
                                 dentry->d_name.len, btrfs_ino(dir), objectid,
-                                BTRFS_I(dir)->block_group, mode, &index);
+                                mode, &index);
         if (IS_ERR(inode)) {
                 err = PTR_ERR(inode);
                 goto out_unlock;
@@ -4649,7 +4640,6 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry,
                 goto out_unlock;
         }
 
-        btrfs_set_trans_block_group(trans, inode);
         err = btrfs_add_nondir(trans, dir, dentry, inode, 0, index);
         if (err)
                 drop_inode = 1;
@@ -4658,8 +4648,6 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry,
                 init_special_inode(inode, inode->i_mode, rdev);
                 btrfs_update_inode(trans, root, inode);
         }
-        btrfs_update_inode_block_group(trans, inode);
-        btrfs_update_inode_block_group(trans, dir);
 out_unlock:
         nr = trans->blocks_used;
         btrfs_end_transaction_throttle(trans, root);
@@ -4692,15 +4680,13 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry,
         if (IS_ERR(trans))
                 return PTR_ERR(trans);
 
-        btrfs_set_trans_block_group(trans, dir);
-
         err = btrfs_find_free_ino(root, &objectid);
         if (err)
                 goto out_unlock;
 
         inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
                                 dentry->d_name.len, btrfs_ino(dir), objectid,
-                                BTRFS_I(dir)->block_group, mode, &index);
+                                mode, &index);
         if (IS_ERR(inode)) {
                 err = PTR_ERR(inode);
                 goto out_unlock;
@@ -4712,7 +4698,6 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry,
                 goto out_unlock;
         }
 
-        btrfs_set_trans_block_group(trans, inode);
         err = btrfs_add_nondir(trans, dir, dentry, inode, 0, index);
         if (err)
                 drop_inode = 1;
@@ -4723,8 +4708,6 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry,
                 inode->i_op = &btrfs_file_inode_operations;
                 BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
         }
-        btrfs_update_inode_block_group(trans, inode);
-        btrfs_update_inode_block_group(trans, dir);
 out_unlock:
         nr = trans->blocks_used;
         btrfs_end_transaction_throttle(trans, root);
@@ -4771,8 +4754,6 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
 
         btrfs_inc_nlink(inode);
         inode->i_ctime = CURRENT_TIME;
-
-        btrfs_set_trans_block_group(trans, dir);
         ihold(inode);
 
         err = btrfs_add_nondir(trans, dir, dentry, inode, 1, index);
@@ -4781,7 +4762,6 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
                 drop_inode = 1;
         } else {
                 struct dentry *parent = dget_parent(dentry);
-                btrfs_update_inode_block_group(trans, dir);
                 err = btrfs_update_inode(trans, root, inode);
                 BUG_ON(err);
                 btrfs_log_new_name(trans, inode, NULL, parent);
@@ -4818,7 +4798,6 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
         trans = btrfs_start_transaction(root, 5);
         if (IS_ERR(trans))
                 return PTR_ERR(trans);
-        btrfs_set_trans_block_group(trans, dir);
 
         err = btrfs_find_free_ino(root, &objectid);
         if (err)
@@ -4826,8 +4805,7 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
 
         inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
                                 dentry->d_name.len, btrfs_ino(dir), objectid,
-                                BTRFS_I(dir)->block_group, S_IFDIR | mode,
-                                &index);
+                                S_IFDIR | mode, &index);
         if (IS_ERR(inode)) {
                 err = PTR_ERR(inode);
                 goto out_fail;
@@ -4841,7 +4819,6 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
 
         inode->i_op = &btrfs_dir_inode_operations;
         inode->i_fop = &btrfs_dir_file_operations;
-        btrfs_set_trans_block_group(trans, inode);
 
         btrfs_i_size_write(inode, 0);
         err = btrfs_update_inode(trans, root, inode);
@@ -4855,8 +4832,6 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
 
         d_instantiate(dentry, inode);
         drop_on_err = 0;
-        btrfs_update_inode_block_group(trans, inode);
-        btrfs_update_inode_block_group(trans, dir);
 
 out_fail:
         nr = trans->blocks_used;
@@ -4989,7 +4964,15 @@ again:
 
         if (!path) {
                 path = btrfs_alloc_path();
-                BUG_ON(!path);
+                if (!path) {
+                        err = -ENOMEM;
+                        goto out;
+                }
+                /*
+                 * Chances are we'll be called again, so go ahead and do
+                 * readahead
+                 */
+                path->reada = 1;
         }
 
         ret = btrfs_lookup_file_extent(trans, root, path,
@@ -5130,8 +5113,10 @@ again:
                                 kunmap(page);
                                 free_extent_map(em);
                                 em = NULL;
+
                                 btrfs_release_path(path);
-                                trans = btrfs_join_transaction(root, 1);
+                                trans = btrfs_join_transaction(root);
+
                                 if (IS_ERR(trans))
                                         return ERR_CAST(trans);
                                 goto again;
@@ -5375,7 +5360,7 @@ static struct extent_map *btrfs_new_extent_direct(struct inode *inode,
                 btrfs_drop_extent_cache(inode, start, start + len - 1, 0);
         }
 
-        trans = btrfs_join_transaction(root, 0);
+        trans = btrfs_join_transaction(root);
         if (IS_ERR(trans))
                 return ERR_CAST(trans);
 
@@ -5611,7 +5596,7 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
                  * to make sure the current transaction stays open
                  * while we look for nocow cross refs
                  */
-                trans = btrfs_join_transaction(root, 0);
+                trans = btrfs_join_transaction(root);
                 if (IS_ERR(trans))
                         goto must_cow;
 
@@ -5750,7 +5735,7 @@ again:
 
         BUG_ON(!ordered);
 
-        trans = btrfs_join_transaction(root, 1);
+        trans = btrfs_join_transaction(root);
         if (IS_ERR(trans)) {
                 err = -ENOMEM;
                 goto out;
@@ -6500,6 +6485,7 @@ out:
 static int btrfs_truncate(struct inode *inode)
 {
         struct btrfs_root *root = BTRFS_I(inode)->root;
+        struct btrfs_block_rsv *rsv;
         int ret;
         int err = 0;
         struct btrfs_trans_handle *trans;
@@ -6513,28 +6499,80 @@ static int 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, 5);
-        if (IS_ERR(trans))
-                return PTR_ERR(trans);
+        /*
+         * Yes ladies and gentelment, this is indeed ugly.  The fact is we have
+         * 3 things going on here
+         *
+         * 1) We need to reserve space for our orphan item and the space to
+         * delete our orphan item.  Lord knows we don't want to have a dangling
+         * orphan item because we didn't reserve space to remove it.
+         *
+         * 2) We need to reserve space to update our inode.
+         *
+         * 3) We need to have something to cache all the space that is going to
+         * be free'd up by the truncate operation, but also have some slack
+         * space reserved in case it uses space during the truncate (thank you
+         * very much snapshotting).
+         *
+         * And we need these to all be seperate.  The fact is we can use alot of
+         * space doing the truncate, and we have no earthly idea how much space
+         * we will use, so we need the truncate reservation to be seperate so it
+         * doesn't end up using space reserved for updating the inode or
+         * removing the orphan item.  We also need to be able to stop the
+         * transaction and start a new one, which means we need to be able to
+         * update the inode several times, and we have no idea of knowing how
+         * many times that will be, so we can't just reserve 1 item for the
+         * entirety of the opration, so that has to be done seperately as well.
+         * Then there is the orphan item, which does indeed need to be held on
+         * to for the whole operation, and we need nobody to touch this reserved
+         * space except the orphan code.
+         *
+         * So that leaves us with
+         *
+         * 1) root->orphan_block_rsv - for the orphan deletion.
+         * 2) rsv - for the truncate reservation, which we will steal from the
+         * transaction reservation.
+         * 3) fs_info->trans_block_rsv - this will have 1 items worth left for
+         * updating the inode.
+         */
+        rsv = btrfs_alloc_block_rsv(root);
+        if (!rsv)
+                return -ENOMEM;
+        btrfs_add_durable_block_rsv(root->fs_info, rsv);
 
-        btrfs_set_trans_block_group(trans, inode);
+        trans = btrfs_start_transaction(root, 4);
+        if (IS_ERR(trans)) {
+                err = PTR_ERR(trans);
+                goto out;
+        }
+
+        /*
+         * Reserve space for the truncate process.  Truncate should be adding
+         * space, but if there are snapshots it may end up using space.
+         */
+        ret = btrfs_truncate_reserve_metadata(trans, root, rsv);
+        BUG_ON(ret);
 
         ret = btrfs_orphan_add(trans, inode);
         if (ret) {
                 btrfs_end_transaction(trans, root);
-                return ret;
+                goto out;
         }
 
         nr = trans->blocks_used;
         btrfs_end_transaction(trans, root);
         btrfs_btree_balance_dirty(root, nr);
 
-        /* Now start a transaction for the truncate */
-        trans = btrfs_start_transaction(root, 0);
-        if (IS_ERR(trans))
-                return PTR_ERR(trans);
-        btrfs_set_trans_block_group(trans, inode);
-        trans->block_rsv = root->orphan_block_rsv;
+        /*
+         * Ok so we've already migrated our bytes over for the truncate, so here
+         * just reserve the one slot we need for updating the inode.
+         */
+        trans = btrfs_start_transaction(root, 1);
+        if (IS_ERR(trans)) {
+                err = PTR_ERR(trans);
+                goto out;
+        }
+        trans->block_rsv = rsv;
 
         /*
          * setattr is responsible for setting the ordered_data_close flag,
@@ -6558,24 +6596,17 @@ static int btrfs_truncate(struct inode *inode)
 
         while (1) {
                 if (!trans) {
-                        trans = btrfs_start_transaction(root, 0);
-                        if (IS_ERR(trans))
-                                return PTR_ERR(trans);
-                        btrfs_set_trans_block_group(trans, inode);
-                        trans->block_rsv = root->orphan_block_rsv;
-                }
+                        trans = btrfs_start_transaction(root, 3);
+                        if (IS_ERR(trans)) {
+                                err = PTR_ERR(trans);
+                                goto out;
+                        }
 
-                ret = btrfs_block_rsv_check(trans, root,
-                                            root->orphan_block_rsv, 0, 5);
-                if (ret == -EAGAIN) {
-                        ret = btrfs_commit_transaction(trans, root);
-                        if (ret)
-                                return ret;
-                        trans = NULL;
-                        continue;
-                } else if (ret) {
-                        err = ret;
-                        break;
+                        ret = btrfs_truncate_reserve_metadata(trans, root,
+                                                              rsv);
+                        BUG_ON(ret);
+
+                        trans->block_rsv = rsv;
                 }
 
                 ret = btrfs_truncate_inode_items(trans, root, inode,
@@ -6586,6 +6617,7 @@ static int btrfs_truncate(struct inode *inode)
                         break;
                 }
 
+                trans->block_rsv = &root->fs_info->trans_block_rsv;
                 ret = btrfs_update_inode(trans, root, inode);
                 if (ret) {
                         err = ret;
@@ -6599,6 +6631,7 @@ static int btrfs_truncate(struct inode *inode)
         }
 
         if (ret == 0 && inode->i_nlink > 0) {
+                trans->block_rsv = root->orphan_block_rsv;
                 ret = btrfs_orphan_del(trans, inode);
                 if (ret)
                         err = ret;
@@ -6610,15 +6643,20 @@ static int btrfs_truncate(struct inode *inode)
                 ret = btrfs_orphan_del(NULL, inode);
         }
 
+        trans->block_rsv = &root->fs_info->trans_block_rsv;
         ret = btrfs_update_inode(trans, root, inode);
         if (ret && !err)
                 err = ret;
 
         nr = trans->blocks_used;
         ret = btrfs_end_transaction_throttle(trans, root);
+        btrfs_btree_balance_dirty(root, nr);
+
+out:
+        btrfs_free_block_rsv(root, rsv);
+
         if (ret && !err)
                 err = ret;
-        btrfs_btree_balance_dirty(root, nr);
 
         return err;
 }
@@ -6627,15 +6665,14 @@ static int btrfs_truncate(struct inode *inode)
  * create a new subvolume directory/inode (helper for the ioctl).
  */
 int btrfs_create_subvol_root(struct btrfs_trans_handle *trans,
-                             struct btrfs_root *new_root,
-                             u64 new_dirid, u64 alloc_hint)
+                             struct btrfs_root *new_root, u64 new_dirid)
 {
         struct inode *inode;
         int err;
         u64 index = 0;
 
         inode = btrfs_new_inode(trans, new_root, NULL, "..", 2, new_dirid,
-                                new_dirid, alloc_hint, S_IFDIR | 0700, &index);
+                                new_dirid, S_IFDIR | 0700, &index);
         if (IS_ERR(inode))
                 return PTR_ERR(inode);
         inode->i_op = &btrfs_dir_inode_operations;
@@ -6748,21 +6785,6 @@ void btrfs_destroy_inode(struct inode *inode)
                 spin_unlock(&root->fs_info->ordered_extent_lock);
         }
 
-        if (root == root->fs_info->tree_root) {
-                struct btrfs_block_group_cache *block_group;
-
-                block_group = btrfs_lookup_block_group(root->fs_info,
-                                                BTRFS_I(inode)->block_group);
-                if (block_group && block_group->inode == inode) {
-                        spin_lock(&block_group->lock);
-                        block_group->inode = NULL;
-                        spin_unlock(&block_group->lock);
-                        btrfs_put_block_group(block_group);
-                } else if (block_group) {
-                        btrfs_put_block_group(block_group);
-                }
-        }
-
         spin_lock(&root->orphan_lock);
         if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
                 printk(KERN_INFO "BTRFS: inode %llu still on the orphan list\n",
@@ -6948,8 +6970,6 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
                 goto out_notrans;
         }
 
-        btrfs_set_trans_block_group(trans, new_dir);
-
         if (dest != root)
                 btrfs_record_root_in_trans(trans, dest);
 
@@ -7131,16 +7151,13 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
         if (IS_ERR(trans))
                 return PTR_ERR(trans);
 
-        btrfs_set_trans_block_group(trans, dir);
-
         err = btrfs_find_free_ino(root, &objectid);
         if (err)
                 goto out_unlock;
 
         inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
                                 dentry->d_name.len, btrfs_ino(dir), objectid,
-                                BTRFS_I(dir)->block_group, S_IFLNK|S_IRWXUGO,
-                                &index);
+                                S_IFLNK|S_IRWXUGO, &index);
         if (IS_ERR(inode)) {
                 err = PTR_ERR(inode);
                 goto out_unlock;
@@ -7152,7 +7169,6 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
                 goto out_unlock;
         }
 
-        btrfs_set_trans_block_group(trans, inode);
         err = btrfs_add_nondir(trans, dir, dentry, inode, 0, index);
         if (err)
                 drop_inode = 1;
@@ -7163,8 +7179,6 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
                 inode->i_op = &btrfs_file_inode_operations;
                 BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
         }
-        btrfs_update_inode_block_group(trans, inode);
-        btrfs_update_inode_block_group(trans, dir);
         if (drop_inode)
                 goto out_unlock;
 
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 85e818ce00c5..74c80595d707 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -243,7 +243,7 @@ static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
                 ip->flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS);
         }
 
-        trans = btrfs_join_transaction(root, 1);
+        trans = btrfs_join_transaction(root);
         BUG_ON(IS_ERR(trans));
 
         ret = btrfs_update_inode(trans, root, inode);
@@ -414,8 +414,7 @@ static noinline int create_subvol(struct btrfs_root *root,
 
         btrfs_record_root_in_trans(trans, new_root);
 
-        ret = btrfs_create_subvol_root(trans, new_root, new_dirid,
-                                       BTRFS_I(dir)->block_group);
+        ret = btrfs_create_subvol_root(trans, new_root, new_dirid);
         /*
          * insert the directory item
          */
@@ -2489,12 +2488,10 @@ static long btrfs_ioctl_trans_start(struct file *file)
         if (ret)
                 goto out;
 
-        mutex_lock(&root->fs_info->trans_mutex);
-        root->fs_info->open_ioctl_trans++;
-        mutex_unlock(&root->fs_info->trans_mutex);
+        atomic_inc(&root->fs_info->open_ioctl_trans);
 
         ret = -ENOMEM;
-        trans = btrfs_start_ioctl_transaction(root, 0);
+        trans = btrfs_start_ioctl_transaction(root);
         if (IS_ERR(trans))
                 goto out_drop;
 
@@ -2502,9 +2499,7 @@ static long btrfs_ioctl_trans_start(struct file *file)
         return 0;
 
 out_drop:
-        mutex_lock(&root->fs_info->trans_mutex);
-        root->fs_info->open_ioctl_trans--;
-        mutex_unlock(&root->fs_info->trans_mutex);
+        atomic_dec(&root->fs_info->open_ioctl_trans);
         mnt_drop_write(file->f_path.mnt);
 out:
         return ret;
@@ -2738,9 +2733,7 @@ long btrfs_ioctl_trans_end(struct file *file)
 
         btrfs_end_transaction(trans, root);
 
-        mutex_lock(&root->fs_info->trans_mutex);
-        root->fs_info->open_ioctl_trans--;
-        mutex_unlock(&root->fs_info->trans_mutex);
+        atomic_dec(&root->fs_info->open_ioctl_trans);
 
         mnt_drop_write(file->f_path.mnt);
         return 0;
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index fa2c5d87f219..f25b10a22a0a 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -677,6 +677,8 @@ struct backref_node *build_backref_tree(struct reloc_control *rc,
                 err = -ENOMEM;
                 goto out;
         }
+        path1->reada = 1;
+        path2->reada = 2;
 
         node = alloc_backref_node(cache);
         if (!node) {
@@ -1999,6 +2001,7 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc,
         path = btrfs_alloc_path();
         if (!path)
                 return -ENOMEM;
+        path->reada = 1;
 
         reloc_root = root->reloc_root;
         root_item = &reloc_root->root_item;
@@ -2139,10 +2142,10 @@ int prepare_to_merge(struct reloc_control *rc, int err)
         u64 num_bytes = 0;
         int ret;
 
-        mutex_lock(&root->fs_info->trans_mutex);
+        spin_lock(&root->fs_info->trans_lock);
         rc->merging_rsv_size += root->nodesize * (BTRFS_MAX_LEVEL - 1) * 2;
         rc->merging_rsv_size += rc->nodes_relocated * 2;
-        mutex_unlock(&root->fs_info->trans_mutex);
+        spin_unlock(&root->fs_info->trans_lock);
 again:
         if (!err) {
                 num_bytes = rc->merging_rsv_size;
@@ -2152,7 +2155,7 @@ again:
                         err = ret;
         }
 
-        trans = btrfs_join_transaction(rc->extent_root, 1);
+        trans = btrfs_join_transaction(rc->extent_root);
         if (IS_ERR(trans)) {
                 if (!err)
                         btrfs_block_rsv_release(rc->extent_root,
@@ -2211,9 +2214,9 @@ int merge_reloc_roots(struct reloc_control *rc)
         int ret;
 again:
         root = rc->extent_root;
-        mutex_lock(&root->fs_info->trans_mutex);
+        spin_lock(&root->fs_info->trans_lock);
         list_splice_init(&rc->reloc_roots, &reloc_roots);
-        mutex_unlock(&root->fs_info->trans_mutex);
+        spin_unlock(&root->fs_info->trans_lock);
 
         while (!list_empty(&reloc_roots)) {
                 found = 1;
@@ -3236,7 +3239,7 @@ truncate:
                 goto out;
         }
 
-        trans = btrfs_join_transaction(root, 0);
+        trans = btrfs_join_transaction(root);
         if (IS_ERR(trans)) {
                 btrfs_free_path(path);
                 ret = PTR_ERR(trans);
@@ -3300,6 +3303,7 @@ static int find_data_references(struct reloc_control *rc,
         path = btrfs_alloc_path();
         if (!path)
                 return -ENOMEM;
+        path->reada = 1;
 
         root = read_fs_root(rc->extent_root->fs_info, ref_root);
         if (IS_ERR(root)) {
@@ -3586,17 +3590,17 @@ next:
 static void set_reloc_control(struct reloc_control *rc)
 {
         struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
-        mutex_lock(&fs_info->trans_mutex);
+        spin_lock(&fs_info->trans_lock);
         fs_info->reloc_ctl = rc;
-        mutex_unlock(&fs_info->trans_mutex);
+        spin_unlock(&fs_info->trans_lock);
 }
 
 static void unset_reloc_control(struct reloc_control *rc)
 {
         struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
-        mutex_lock(&fs_info->trans_mutex);
+        spin_lock(&fs_info->trans_lock);
         fs_info->reloc_ctl = NULL;
-        mutex_unlock(&fs_info->trans_mutex);
+        spin_unlock(&fs_info->trans_lock);
 }
 
 static int check_extent_flags(u64 flags)
@@ -3645,7 +3649,7 @@ int prepare_to_relocate(struct reloc_control *rc)
         rc->create_reloc_tree = 1;
         set_reloc_control(rc);
 
-        trans = btrfs_join_transaction(rc->extent_root, 1);
+        trans = btrfs_join_transaction(rc->extent_root);
         BUG_ON(IS_ERR(trans));
         btrfs_commit_transaction(trans, rc->extent_root);
         return 0;
@@ -3668,6 +3672,7 @@ static noinline_for_stack int relocate_block_group(struct reloc_control *rc)
         path = btrfs_alloc_path();
         if (!path)
                 return -ENOMEM;
+        path->reada = 1;
 
         ret = prepare_to_relocate(rc);
         if (ret) {
@@ -3834,7 +3839,7 @@ restart:
         btrfs_block_rsv_release(rc->extent_root, rc->block_rsv, (u64)-1);
 
         /* get rid of pinned extents */
-        trans = btrfs_join_transaction(rc->extent_root, 1);
+        trans = btrfs_join_transaction(rc->extent_root);
         if (IS_ERR(trans))
                 err = PTR_ERR(trans);
         else
@@ -4093,6 +4098,7 @@ int btrfs_recover_relocation(struct btrfs_root *root)
         path = btrfs_alloc_path();
         if (!path)
                 return -ENOMEM;
+        path->reada = -1;
 
         key.objectid = BTRFS_TREE_RELOC_OBJECTID;
         key.type = BTRFS_ROOT_ITEM_KEY;
@@ -4159,7 +4165,7 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 
         set_reloc_control(rc);
 
-        trans = btrfs_join_transaction(rc->extent_root, 1);
+        trans = btrfs_join_transaction(rc->extent_root);
         if (IS_ERR(trans)) {
                 unset_reloc_control(rc);
                 err = PTR_ERR(trans);
@@ -4193,7 +4199,7 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 
         unset_reloc_control(rc);
 
-        trans = btrfs_join_transaction(rc->extent_root, 1);
+        trans = btrfs_join_transaction(rc->extent_root);
         if (IS_ERR(trans))
                 err = PTR_ERR(trans);
         else
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index dc80f7156923..2d5c6d2aa4e4 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -35,6 +35,7 @@ static noinline void put_transaction(struct btrfs_transaction *transaction)
 {
         WARN_ON(atomic_read(&transaction->use_count) == 0);
         if (atomic_dec_and_test(&transaction->use_count)) {
+                BUG_ON(!list_empty(&transaction->list));
                 memset(transaction, 0, sizeof(*transaction));
                 kmem_cache_free(btrfs_transaction_cachep, transaction);
         }
@@ -49,46 +50,72 @@ static noinline void switch_commit_root(struct btrfs_root *root)
 /*
  * either allocate a new transaction or hop into the existing one
  */
-static noinline int join_transaction(struct btrfs_root *root)
+static noinline int join_transaction(struct btrfs_root *root, int nofail)
 {
         struct btrfs_transaction *cur_trans;
+
+        spin_lock(&root->fs_info->trans_lock);
+        if (root->fs_info->trans_no_join) {
+                if (!nofail) {
+                        spin_unlock(&root->fs_info->trans_lock);
+                        return -EBUSY;
+                }
+        }
+
         cur_trans = root->fs_info->running_transaction;
-        if (!cur_trans) {
-                cur_trans = kmem_cache_alloc(btrfs_transaction_cachep,
-                                             GFP_NOFS);
-                if (!cur_trans)
-                        return -ENOMEM;
-                root->fs_info->generation++;
-                atomic_set(&cur_trans->num_writers, 1);
-                cur_trans->num_joined = 0;
-                cur_trans->transid = root->fs_info->generation;
-                init_waitqueue_head(&cur_trans->writer_wait);
-                init_waitqueue_head(&cur_trans->commit_wait);
-                cur_trans->in_commit = 0;
-                cur_trans->blocked = 0;
-                atomic_set(&cur_trans->use_count, 1);
-                cur_trans->commit_done = 0;
-                cur_trans->start_time = get_seconds();
-
-                cur_trans->delayed_refs.root = RB_ROOT;
-                cur_trans->delayed_refs.num_entries = 0;
-                cur_trans->delayed_refs.num_heads_ready = 0;
-                cur_trans->delayed_refs.num_heads = 0;
-                cur_trans->delayed_refs.flushing = 0;
-                cur_trans->delayed_refs.run_delayed_start = 0;
-                spin_lock_init(&cur_trans->delayed_refs.lock);
-
-                INIT_LIST_HEAD(&cur_trans->pending_snapshots);
-                list_add_tail(&cur_trans->list, &root->fs_info->trans_list);
-                extent_io_tree_init(&cur_trans->dirty_pages,
-                                     root->fs_info->btree_inode->i_mapping);
-                spin_lock(&root->fs_info->new_trans_lock);
-                root->fs_info->running_transaction = cur_trans;
-                spin_unlock(&root->fs_info->new_trans_lock);
-        } else {
+        if (cur_trans) {
+                atomic_inc(&cur_trans->use_count);
                 atomic_inc(&cur_trans->num_writers);
                 cur_trans->num_joined++;
+                spin_unlock(&root->fs_info->trans_lock);
+                return 0;
         }
+        spin_unlock(&root->fs_info->trans_lock);
+
+        cur_trans = kmem_cache_alloc(btrfs_transaction_cachep, GFP_NOFS);
+        if (!cur_trans)
+                return -ENOMEM;
+        spin_lock(&root->fs_info->trans_lock);
+        if (root->fs_info->running_transaction) {
+                kmem_cache_free(btrfs_transaction_cachep, cur_trans);
+                cur_trans = root->fs_info->running_transaction;
+                atomic_inc(&cur_trans->use_count);
+                atomic_inc(&cur_trans->num_writers);
+                cur_trans->num_joined++;
+                spin_unlock(&root->fs_info->trans_lock);
+                return 0;
+        }
+        atomic_set(&cur_trans->num_writers, 1);
+        cur_trans->num_joined = 0;
+        init_waitqueue_head(&cur_trans->writer_wait);
+        init_waitqueue_head(&cur_trans->commit_wait);
+        cur_trans->in_commit = 0;
+        cur_trans->blocked = 0;
+        /*
+         * One for this trans handle, one so it will live on until we
+         * commit the transaction.
+         */
+        atomic_set(&cur_trans->use_count, 2);
+        cur_trans->commit_done = 0;
+        cur_trans->start_time = get_seconds();
+
+        cur_trans->delayed_refs.root = RB_ROOT;
+        cur_trans->delayed_refs.num_entries = 0;
+        cur_trans->delayed_refs.num_heads_ready = 0;
+        cur_trans->delayed_refs.num_heads = 0;
+        cur_trans->delayed_refs.flushing = 0;
+        cur_trans->delayed_refs.run_delayed_start = 0;
+        spin_lock_init(&cur_trans->commit_lock);
+        spin_lock_init(&cur_trans->delayed_refs.lock);
+
+        INIT_LIST_HEAD(&cur_trans->pending_snapshots);
+        list_add_tail(&cur_trans->list, &root->fs_info->trans_list);
+        extent_io_tree_init(&cur_trans->dirty_pages,
+                             root->fs_info->btree_inode->i_mapping);
+        root->fs_info->generation++;
+        cur_trans->transid = root->fs_info->generation;
+        root->fs_info->running_transaction = cur_trans;
+        spin_unlock(&root->fs_info->trans_lock);
 
         return 0;
 }
@@ -99,39 +126,28 @@ static noinline int join_transaction(struct btrfs_root *root)
  * to make sure the old root from before we joined the transaction is deleted
  * when the transaction commits
  */
-static noinline int record_root_in_trans(struct btrfs_trans_handle *trans,
-                                         struct btrfs_root *root)
+int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
+                               struct btrfs_root *root)
 {
         if (root->ref_cows && root->last_trans < trans->transid) {
                 WARN_ON(root == root->fs_info->extent_root);
                 WARN_ON(root->commit_root != root->node);
 
+                spin_lock(&root->fs_info->fs_roots_radix_lock);
+                if (root->last_trans == trans->transid) {
+                        spin_unlock(&root->fs_info->fs_roots_radix_lock);
+                        return 0;
+                }
+                root->last_trans = trans->transid;
                 radix_tree_tag_set(&root->fs_info->fs_roots_radix,
                            (unsigned long)root->root_key.objectid,
                            BTRFS_ROOT_TRANS_TAG);
-                root->last_trans = trans->transid;
+                spin_unlock(&root->fs_info->fs_roots_radix_lock);
                 btrfs_init_reloc_root(trans, root);
         }
         return 0;
 }
 
-int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
-                               struct btrfs_root *root)
-{
-        if (!root->ref_cows)
-                return 0;
-
-        mutex_lock(&root->fs_info->trans_mutex);
-        if (root->last_trans == trans->transid) {
-                mutex_unlock(&root->fs_info->trans_mutex);
-                return 0;
-        }
-
-        record_root_in_trans(trans, root);
-        mutex_unlock(&root->fs_info->trans_mutex);
-        return 0;
-}
-
 /* wait for commit against the current transaction to become unblocked
  * when this is done, it is safe to start a new transaction, but the current
  * transaction might not be fully on disk.
@@ -140,21 +156,23 @@ static void wait_current_trans(struct btrfs_root *root)
 {
         struct btrfs_transaction *cur_trans;
 
+        spin_lock(&root->fs_info->trans_lock);
         cur_trans = root->fs_info->running_transaction;
         if (cur_trans && cur_trans->blocked) {
                 DEFINE_WAIT(wait);
                 atomic_inc(&cur_trans->use_count);
+                spin_unlock(&root->fs_info->trans_lock);
                 while (1) {
                         prepare_to_wait(&root->fs_info->transaction_wait, &wait,
                                         TASK_UNINTERRUPTIBLE);
                         if (!cur_trans->blocked)
                                 break;
-                        mutex_unlock(&root->fs_info->trans_mutex);
                         schedule();
-                        mutex_lock(&root->fs_info->trans_mutex);
                 }
                 finish_wait(&root->fs_info->transaction_wait, &wait);
                 put_transaction(cur_trans);
+        } else {
+                spin_unlock(&root->fs_info->trans_lock);
         }
 }
 
@@ -167,10 +185,16 @@ enum btrfs_trans_type {
 
 static int may_wait_transaction(struct btrfs_root *root, int type)
 {
-        if (!root->fs_info->log_root_recovering &&
-            ((type == TRANS_START && !root->fs_info->open_ioctl_trans) ||
-             type == TRANS_USERSPACE))
+        if (root->fs_info->log_root_recovering)
+                return 0;
+
+        if (type == TRANS_USERSPACE)
+                return 1;
+
+        if (type == TRANS_START &&
+            !atomic_read(&root->fs_info->open_ioctl_trans))
                 return 1;
+
         return 0;
 }
 
@@ -184,36 +208,44 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
 
         if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR)
                 return ERR_PTR(-EROFS);
+
+        if (current->journal_info) {
+                WARN_ON(type != TRANS_JOIN && type != TRANS_JOIN_NOLOCK);
+                h = current->journal_info;
+                h->use_count++;
+                h->orig_rsv = h->block_rsv;
+                h->block_rsv = NULL;
+                goto got_it;
+        }
 again:
         h = kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS);
         if (!h)
                 return ERR_PTR(-ENOMEM);
 
-        if (type != TRANS_JOIN_NOLOCK)
-                mutex_lock(&root->fs_info->trans_mutex);
         if (may_wait_transaction(root, type))
                 wait_current_trans(root);
 
-        ret = join_transaction(root);
+        do {
+                ret = join_transaction(root, type == TRANS_JOIN_NOLOCK);
+                if (ret == -EBUSY)
+                        wait_current_trans(root);
+        } while (ret == -EBUSY);
+
         if (ret < 0) {
                 kmem_cache_free(btrfs_trans_handle_cachep, h);
-                if (type != TRANS_JOIN_NOLOCK)
-                        mutex_unlock(&root->fs_info->trans_mutex);
                 return ERR_PTR(ret);
         }
 
         cur_trans = root->fs_info->running_transaction;
-        atomic_inc(&cur_trans->use_count);
-        if (type != TRANS_JOIN_NOLOCK)
-                mutex_unlock(&root->fs_info->trans_mutex);
 
         h->transid = cur_trans->transid;
         h->transaction = cur_trans;
         h->blocks_used = 0;
-        h->block_group = 0;
         h->bytes_reserved = 0;
         h->delayed_ref_updates = 0;
+        h->use_count = 1;
         h->block_rsv = NULL;
+        h->orig_rsv = NULL;
 
         smp_mb();
         if (cur_trans->blocked && may_wait_transaction(root, type)) {
@@ -241,11 +273,8 @@ again:
                 }
         }
 
-        if (type != TRANS_JOIN_NOLOCK)
-                mutex_lock(&root->fs_info->trans_mutex);
-        record_root_in_trans(h, root);
-        if (type != TRANS_JOIN_NOLOCK)
-                mutex_unlock(&root->fs_info->trans_mutex);
+got_it:
+        btrfs_record_root_in_trans(h, root);
 
         if (!current->journal_info && type != TRANS_USERSPACE)
                 current->journal_info = h;
@@ -257,22 +286,19 @@ struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
 {
         return start_transaction(root, num_items, TRANS_START);
 }
-struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root,
-                                                   int num_blocks)
+struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root)
 {
         return start_transaction(root, 0, TRANS_JOIN);
 }
 
-struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root,
-                                                          int num_blocks)
+struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root)
 {
         return start_transaction(root, 0, TRANS_JOIN_NOLOCK);
 }
 
-struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r,
-                                                         int num_blocks)
+struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *root)
 {
-        return start_transaction(r, 0, TRANS_USERSPACE);
+        return start_transaction(root, 0, TRANS_USERSPACE);
 }
 
 /* wait for a transaction commit to be fully complete */
@@ -280,17 +306,13 @@ static noinline int wait_for_commit(struct btrfs_root *root,
                                     struct btrfs_transaction *commit)
 {
         DEFINE_WAIT(wait);
-        mutex_lock(&root->fs_info->trans_mutex);
         while (!commit->commit_done) {
                 prepare_to_wait(&commit->commit_wait, &wait,
                                 TASK_UNINTERRUPTIBLE);
                 if (commit->commit_done)
                         break;
-                mutex_unlock(&root->fs_info->trans_mutex);
                 schedule();
-                mutex_lock(&root->fs_info->trans_mutex);
         }
-        mutex_unlock(&root->fs_info->trans_mutex);
         finish_wait(&commit->commit_wait, &wait);
         return 0;
 }
@@ -300,59 +322,56 @@ int btrfs_wait_for_commit(struct btrfs_root *root, u64 transid)
         struct btrfs_transaction *cur_trans = NULL, *t;
         int ret;
 
-        mutex_lock(&root->fs_info->trans_mutex);
-
         ret = 0;
         if (transid) {
                 if (transid <= root->fs_info->last_trans_committed)
-                        goto out_unlock;
+                        goto out;
 
                 /* find specified transaction */
+                spin_lock(&root->fs_info->trans_lock);
                 list_for_each_entry(t, &root->fs_info->trans_list, list) {
                         if (t->transid == transid) {
                                 cur_trans = t;
+                                atomic_inc(&cur_trans->use_count);
                                 break;
                         }
                         if (t->transid > transid)
                                 break;
                 }
+                spin_unlock(&root->fs_info->trans_lock);
                 ret = -EINVAL;
                 if (!cur_trans)
-                        goto out_unlock;  /* bad transid */
+                        goto out;  /* bad transid */
         } else {
                 /* find newest transaction that is committing | committed */
+                spin_lock(&root->fs_info->trans_lock);
                 list_for_each_entry_reverse(t, &root->fs_info->trans_list,
                                             list) {
                         if (t->in_commit) {
                                 if (t->commit_done)
-                                        goto out_unlock;
+                                        goto out;
                                 cur_trans = t;
+                                atomic_inc(&cur_trans->use_count);
                                 break;
                         }
                 }
+                spin_unlock(&root->fs_info->trans_lock);
                 if (!cur_trans)
-                        goto out_unlock;  /* nothing committing|committed */
+                        goto out;  /* nothing committing|committed */
         }
 
-        atomic_inc(&cur_trans->use_count);
-        mutex_unlock(&root->fs_info->trans_mutex);
-
         wait_for_commit(root, cur_trans);
 
-        mutex_lock(&root->fs_info->trans_mutex);
         put_transaction(cur_trans);
         ret = 0;
-out_unlock:
-        mutex_unlock(&root->fs_info->trans_mutex);
+out:
         return ret;
 }
 
 void btrfs_throttle(struct btrfs_root *root)
 {
-        mutex_lock(&root->fs_info->trans_mutex);
-        if (!root->fs_info->open_ioctl_trans)
+        if (!atomic_read(&root->fs_info->open_ioctl_trans))
                 wait_current_trans(root);
-        mutex_unlock(&root->fs_info->trans_mutex);
 }
 
 static int should_end_transaction(struct btrfs_trans_handle *trans,
@@ -370,6 +389,7 @@ int btrfs_should_end_transaction(struct btrfs_trans_handle *trans,
         struct btrfs_transaction *cur_trans = trans->transaction;
         int updates;
 
+        smp_mb();
         if (cur_trans->blocked || cur_trans->delayed_refs.flushing)
                 return 1;
 
@@ -388,6 +408,11 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
         struct btrfs_fs_info *info = root->fs_info;
         int count = 0;
 
+        if (--trans->use_count) {
+                trans->block_rsv = trans->orig_rsv;
+                return 0;
+        }
+
         while (count < 4) {
                 unsigned long cur = trans->delayed_ref_updates;
                 trans->delayed_ref_updates = 0;
@@ -410,9 +435,11 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 
         btrfs_trans_release_metadata(trans, root);
 
-        if (lock && !root->fs_info->open_ioctl_trans &&
-            should_end_transaction(trans, root))
+        if (lock && !atomic_read(&root->fs_info->open_ioctl_trans) &&
+            should_end_transaction(trans, root)) {
                 trans->transaction->blocked = 1;
+                smp_wmb();
+        }
 
         if (lock && cur_trans->blocked && !cur_trans->in_commit) {
                 if (throttle)
@@ -703,9 +730,9 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans,
  */
 int btrfs_add_dead_root(struct btrfs_root *root)
 {
-        mutex_lock(&root->fs_info->trans_mutex);
+        spin_lock(&root->fs_info->trans_lock);
         list_add(&root->root_list, &root->fs_info->dead_roots);
-        mutex_unlock(&root->fs_info->trans_mutex);
+        spin_unlock(&root->fs_info->trans_lock);
         return 0;
 }
 
@@ -721,6 +748,7 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans,
         int ret;
         int err = 0;
 
+        spin_lock(&fs_info->fs_roots_radix_lock);
         while (1) {
                 ret = radix_tree_gang_lookup_tag(&fs_info->fs_roots_radix,
                                                  (void **)gang, 0,
@@ -733,6 +761,7 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans,
                         radix_tree_tag_clear(&fs_info->fs_roots_radix,
                                         (unsigned long)root->root_key.objectid,
                                         BTRFS_ROOT_TRANS_TAG);
+                        spin_unlock(&fs_info->fs_roots_radix_lock);
 
                         btrfs_free_log(trans, root);
                         btrfs_update_reloc_root(trans, root);
@@ -753,10 +782,12 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans,
                         err = btrfs_update_root(trans, fs_info->tree_root,
                                                 &root->root_key,
                                                 &root->root_item);
+                        spin_lock(&fs_info->fs_roots_radix_lock);
                         if (err)
                                 break;
                 }
         }
+        spin_unlock(&fs_info->fs_roots_radix_lock);
         return err;
 }
 
@@ -851,7 +882,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
         parent = dget_parent(dentry);
         parent_inode = parent->d_inode;
         parent_root = BTRFS_I(parent_inode)->root;
-        record_root_in_trans(trans, parent_root);
+        btrfs_record_root_in_trans(trans, parent_root);
 
         /*
          * insert the directory item
@@ -869,7 +900,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
         ret = btrfs_update_inode(trans, parent_root, parent_inode);
         BUG_ON(ret);
 
-        record_root_in_trans(trans, root);
+        btrfs_record_root_in_trans(trans, root);
         btrfs_set_root_last_snapshot(&root->root_item, trans->transid);
         memcpy(new_root_item, &root->root_item, sizeof(*new_root_item));
         btrfs_check_and_init_root_item(new_root_item);
@@ -967,20 +998,20 @@ static void update_super_roots(struct btrfs_root *root)
 int btrfs_transaction_in_commit(struct btrfs_fs_info *info)
 {
         int ret = 0;
-        spin_lock(&info->new_trans_lock);
+        spin_lock(&info->trans_lock);
         if (info->running_transaction)
                 ret = info->running_transaction->in_commit;
-        spin_unlock(&info->new_trans_lock);
+        spin_unlock(&info->trans_lock);
         return ret;
 }
 
 int btrfs_transaction_blocked(struct btrfs_fs_info *info)
 {
         int ret = 0;
-        spin_lock(&info->new_trans_lock);
+        spin_lock(&info->trans_lock);
         if (info->running_transaction)
                 ret = info->running_transaction->blocked;
-        spin_unlock(&info->new_trans_lock);
+        spin_unlock(&info->trans_lock);
         return ret;
 }
 
@@ -1004,9 +1035,7 @@ static void wait_current_trans_commit_start(struct btrfs_root *root,
                                     &wait);
                         break;
                 }
-                mutex_unlock(&root->fs_info->trans_mutex);
                 schedule();
-                mutex_lock(&root->fs_info->trans_mutex);
                 finish_wait(&root->fs_info->transaction_blocked_wait, &wait);
         }
 }
@@ -1032,9 +1061,7 @@ static void wait_current_trans_commit_start_and_unblock(struct btrfs_root *root,
                                     &wait);
                         break;
                 }
-                mutex_unlock(&root->fs_info->trans_mutex);
                 schedule();
-                mutex_lock(&root->fs_info->trans_mutex);
                 finish_wait(&root->fs_info->transaction_wait,
                             &wait);
         }
@@ -1072,7 +1099,7 @@ int btrfs_commit_transaction_async(struct btrfs_trans_handle *trans,
 
         INIT_DELAYED_WORK(&ac->work, do_async_commit);
         ac->root = root;
-        ac->newtrans = btrfs_join_transaction(root, 0);
+        ac->newtrans = btrfs_join_transaction(root);
         if (IS_ERR(ac->newtrans)) {
                 int err = PTR_ERR(ac->newtrans);
                 kfree(ac);
@@ -1080,22 +1107,18 @@ int btrfs_commit_transaction_async(struct btrfs_trans_handle *trans,
         }
 
         /* take transaction reference */
-        mutex_lock(&root->fs_info->trans_mutex);
         cur_trans = trans->transaction;
         atomic_inc(&cur_trans->use_count);
-        mutex_unlock(&root->fs_info->trans_mutex);
 
         btrfs_end_transaction(trans, root);
         schedule_delayed_work(&ac->work, 0);
 
         /* wait for transaction to start and unblock */
-        mutex_lock(&root->fs_info->trans_mutex);
         if (wait_for_unblock)
                 wait_current_trans_commit_start_and_unblock(root, cur_trans);
         else
                 wait_current_trans_commit_start(root, cur_trans);
         put_transaction(cur_trans);
-        mutex_unlock(&root->fs_info->trans_mutex);
 
         return 0;
 }
@@ -1139,38 +1162,41 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
         ret = btrfs_run_delayed_refs(trans, root, 0);
         BUG_ON(ret);
 
-        mutex_lock(&root->fs_info->trans_mutex);
+        spin_lock(&cur_trans->commit_lock);
         if (cur_trans->in_commit) {
+                spin_unlock(&cur_trans->commit_lock);
                 atomic_inc(&cur_trans->use_count);
-                mutex_unlock(&root->fs_info->trans_mutex);
                 btrfs_end_transaction(trans, root);
 
                 ret = wait_for_commit(root, cur_trans);
                 BUG_ON(ret);
 
-                mutex_lock(&root->fs_info->trans_mutex);
                 put_transaction(cur_trans);
-                mutex_unlock(&root->fs_info->trans_mutex);
 
                 return 0;
         }
 
         trans->transaction->in_commit = 1;
         trans->transaction->blocked = 1;
+        spin_unlock(&cur_trans->commit_lock);
         wake_up(&root->fs_info->transaction_blocked_wait);
 
+        spin_lock(&root->fs_info->trans_lock);
         if (cur_trans->list.prev != &root->fs_info->trans_list) {
                 prev_trans = list_entry(cur_trans->list.prev,
                                         struct btrfs_transaction, list);
                 if (!prev_trans->commit_done) {
                         atomic_inc(&prev_trans->use_count);
-                        mutex_unlock(&root->fs_info->trans_mutex);
+                        spin_unlock(&root->fs_info->trans_lock);
 
                         wait_for_commit(root, prev_trans);
 
-                        mutex_lock(&root->fs_info->trans_mutex);
                         put_transaction(prev_trans);
+                } else {
+                        spin_unlock(&root->fs_info->trans_lock);
                 }
+        } else {
+                spin_unlock(&root->fs_info->trans_lock);
         }
 
         if (now < cur_trans->start_time || now - cur_trans->start_time < 1)
@@ -1178,12 +1204,12 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 
         do {
                 int snap_pending = 0;
+
                 joined = cur_trans->num_joined;
                 if (!list_empty(&trans->transaction->pending_snapshots))
                         snap_pending = 1;
 
                 WARN_ON(cur_trans != trans->transaction);
-                mutex_unlock(&root->fs_info->trans_mutex);
 
                 if (flush_on_commit || snap_pending) {
                         btrfs_start_delalloc_inodes(root, 1);
@@ -1206,14 +1232,15 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
                 prepare_to_wait(&cur_trans->writer_wait, &wait,
                                 TASK_UNINTERRUPTIBLE);
 
-                smp_mb();
                 if (atomic_read(&cur_trans->num_writers) > 1)
                         schedule_timeout(MAX_SCHEDULE_TIMEOUT);
                 else if (should_grow)
                         schedule_timeout(1);
 
-                mutex_lock(&root->fs_info->trans_mutex);
                 finish_wait(&cur_trans->writer_wait, &wait);
+                spin_lock(&root->fs_info->trans_lock);
+                root->fs_info->trans_no_join = 1;
+                spin_unlock(&root->fs_info->trans_lock);
         } while (atomic_read(&cur_trans->num_writers) > 1 ||
                  (should_grow && cur_trans->num_joined != joined));
 
@@ -1258,9 +1285,6 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
         btrfs_prepare_extent_commit(trans, root);
 
         cur_trans = root->fs_info->running_transaction;
-        spin_lock(&root->fs_info->new_trans_lock);
-        root->fs_info->running_transaction = NULL;
-        spin_unlock(&root->fs_info->new_trans_lock);
 
         btrfs_set_root_node(&root->fs_info->tree_root->root_item,
                             root->fs_info->tree_root->node);
@@ -1281,10 +1305,13 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
                sizeof(root->fs_info->super_copy));
 
         trans->transaction->blocked = 0;
+        spin_lock(&root->fs_info->trans_lock);
+        root->fs_info->running_transaction = NULL;
+        root->fs_info->trans_no_join = 0;
+        spin_unlock(&root->fs_info->trans_lock);
 
         wake_up(&root->fs_info->transaction_wait);
 
-        mutex_unlock(&root->fs_info->trans_mutex);
         ret = btrfs_write_and_wait_transaction(trans, root);
         BUG_ON(ret);
         write_ctree_super(trans, root, 0);
@@ -1297,22 +1324,21 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 
         btrfs_finish_extent_commit(trans, root);
 
-        mutex_lock(&root->fs_info->trans_mutex);
-
         cur_trans->commit_done = 1;
 
         root->fs_info->last_trans_committed = cur_trans->transid;
 
         wake_up(&cur_trans->commit_wait);
 
+        spin_lock(&root->fs_info->trans_lock);
         list_del_init(&cur_trans->list);
+        spin_unlock(&root->fs_info->trans_lock);
+
         put_transaction(cur_trans);
         put_transaction(cur_trans);
 
         trace_btrfs_transaction_commit(root);
 
-        mutex_unlock(&root->fs_info->trans_mutex);
-
         btrfs_scrub_continue(root);
 
         if (current->journal_info == trans)
@@ -1334,9 +1360,9 @@ int btrfs_clean_old_snapshots(struct btrfs_root *root)
         LIST_HEAD(list);
         struct btrfs_fs_info *fs_info = root->fs_info;
 
-        mutex_lock(&fs_info->trans_mutex);
+        spin_lock(&fs_info->trans_lock);
         list_splice_init(&fs_info->dead_roots, &list);
-        mutex_unlock(&fs_info->trans_mutex);
+        spin_unlock(&fs_info->trans_lock);
 
         while (!list_empty(&list)) {
                 root = list_entry(list.next, struct btrfs_root, root_list);
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index 804c88639e5d..02564e6230ac 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -28,10 +28,12 @@ struct btrfs_transaction {
          * transaction can end
          */
         atomic_t num_writers;
+        atomic_t use_count;
 
         unsigned long num_joined;
+
+        spinlock_t commit_lock;
         int in_commit;
-        atomic_t use_count;
         int commit_done;
         int blocked;
         struct list_head list;
@@ -45,13 +47,14 @@ struct btrfs_transaction {
 
 struct btrfs_trans_handle {
         u64 transid;
-        u64 block_group;
         u64 bytes_reserved;
+        unsigned long use_count;
         unsigned long blocks_reserved;
         unsigned long blocks_used;
         unsigned long delayed_ref_updates;
         struct btrfs_transaction *transaction;
         struct btrfs_block_rsv *block_rsv;
+        struct btrfs_block_rsv *orig_rsv;
 };
 
 struct btrfs_pending_snapshot {
@@ -66,19 +69,6 @@ struct btrfs_pending_snapshot {
         struct list_head list;
 };
 
-static inline void btrfs_set_trans_block_group(struct btrfs_trans_handle *trans,
-                                               struct inode *inode)
-{
-        trans->block_group = BTRFS_I(inode)->block_group;
-}
-
-static inline void btrfs_update_inode_block_group(
-                                          struct btrfs_trans_handle *trans,
-                                          struct inode *inode)
-{
-        BTRFS_I(inode)->block_group = trans->block_group;
-}
-
 static inline void btrfs_set_inode_last_trans(struct btrfs_trans_handle *trans,
                                               struct inode *inode)
 {
@@ -92,12 +82,9 @@ int btrfs_end_transaction_nolock(struct btrfs_trans_handle *trans,
                                  struct btrfs_root *root);
 struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
                                                    int num_items);
-struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root,
-                                                  int num_blocks);
-struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root,
-                                                          int num_blocks);
-struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r,
-                                                         int num_blocks);
+struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root);
+struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root);
+struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *root);
 int btrfs_wait_for_commit(struct btrfs_root *root, u64 transid);
 int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans,
                                      struct btrfs_root *root);
diff --git a/fs/btrfs/xattr.c b/fs/btrfs/xattr.c
index f3107e4b4d56..5366fe452ab0 100644
--- a/fs/btrfs/xattr.c
+++ b/fs/btrfs/xattr.c
@@ -158,8 +158,6 @@ int __btrfs_setxattr(struct btrfs_trans_handle *trans,
         if (IS_ERR(trans))
                 return PTR_ERR(trans);
 
-        btrfs_set_trans_block_group(trans, inode);
-
         ret = do_setxattr(trans, inode, name, value, size, flags);
         if (ret)
                 goto out;