1 files changed, 172 insertions, 60 deletions

diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 2cb116099b90..66e4c66cc63b 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -165,54 +165,89 @@ enum btrfs_trans_type {
         TRANS_USERSPACE,
 };
 
+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))
+                return 1;
+        return 0;
+}
+
 static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
-                                             int num_blocks, int type)
+                                                    u64 num_items, int type)
 {
-        struct btrfs_trans_handle *h =
-                kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS);
+        struct btrfs_trans_handle *h;
+        struct btrfs_transaction *cur_trans;
+        int retries = 0;
         int ret;
+again:
+        h = kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS);
+        if (!h)
+                return ERR_PTR(-ENOMEM);
 
         mutex_lock(&root->fs_info->trans_mutex);
-        if (!root->fs_info->log_root_recovering &&
-            ((type == TRANS_START && !root->fs_info->open_ioctl_trans) ||
-             type == TRANS_USERSPACE))
+        if (may_wait_transaction(root, type))
                 wait_current_trans(root);
+
         ret = join_transaction(root);
         BUG_ON(ret);
 
-        h->transid = root->fs_info->running_transaction->transid;
-        h->transaction = root->fs_info->running_transaction;
-        h->blocks_reserved = num_blocks;
+        cur_trans = root->fs_info->running_transaction;
+        cur_trans->use_count++;
+        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->alloc_exclude_nr = 0;
-        h->alloc_exclude_start = 0;
+        h->bytes_reserved = 0;
         h->delayed_ref_updates = 0;
+        h->block_rsv = NULL;
 
-        if (!current->journal_info && type != TRANS_USERSPACE)
-                current->journal_info = h;
+        smp_mb();
+        if (cur_trans->blocked && may_wait_transaction(root, type)) {
+                btrfs_commit_transaction(h, root);
+                goto again;
+        }
+
+        if (num_items > 0) {
+                ret = btrfs_trans_reserve_metadata(h, root, num_items,
+                                                   &retries);
+                if (ret == -EAGAIN) {
+                        btrfs_commit_transaction(h, root);
+                        goto again;
+                }
+                if (ret < 0) {
+                        btrfs_end_transaction(h, root);
+                        return ERR_PTR(ret);
+                }
+        }
 
-        root->fs_info->running_transaction->use_count++;
+        mutex_lock(&root->fs_info->trans_mutex);
         record_root_in_trans(h, root);
         mutex_unlock(&root->fs_info->trans_mutex);
+
+        if (!current->journal_info && type != TRANS_USERSPACE)
+                current->journal_info = h;
         return h;
 }
 
 struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
-                                                   int num_blocks)
+                                                   int num_items)
 {
-        return start_transaction(root, num_blocks, TRANS_START);
+        return start_transaction(root, num_items, TRANS_START);
 }
 struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root,
                                                    int num_blocks)
 {
-        return start_transaction(root, num_blocks, TRANS_JOIN);
+        return start_transaction(root, 0, TRANS_JOIN);
 }
 
 struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r,
                                                          int num_blocks)
 {
-        return start_transaction(r, num_blocks, TRANS_USERSPACE);
+        return start_transaction(r, 0, TRANS_USERSPACE);
 }
 
 /* wait for a transaction commit to be fully complete */
@@ -286,10 +321,36 @@ void btrfs_throttle(struct btrfs_root *root)
         mutex_unlock(&root->fs_info->trans_mutex);
 }
 
+static int should_end_transaction(struct btrfs_trans_handle *trans,
+                                  struct btrfs_root *root)
+{
+        int ret;
+        ret = btrfs_block_rsv_check(trans, root,
+                                    &root->fs_info->global_block_rsv, 0, 5);
+        return ret ? 1 : 0;
+}
+
+int btrfs_should_end_transaction(struct btrfs_trans_handle *trans,
+                                 struct btrfs_root *root)
+{
+        struct btrfs_transaction *cur_trans = trans->transaction;
+        int updates;
+
+        if (cur_trans->blocked || cur_trans->delayed_refs.flushing)
+                return 1;
+
+        updates = trans->delayed_ref_updates;
+        trans->delayed_ref_updates = 0;
+        if (updates)
+                btrfs_run_delayed_refs(trans, root, updates);
+
+        return should_end_transaction(trans, root);
+}
+
 static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
                           struct btrfs_root *root, int throttle)
 {
-        struct btrfs_transaction *cur_trans;
+        struct btrfs_transaction *cur_trans = trans->transaction;
         struct btrfs_fs_info *info = root->fs_info;
         int count = 0;
 
@@ -313,9 +374,21 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
                 count++;
         }
 
+        btrfs_trans_release_metadata(trans, root);
+
+        if (!root->fs_info->open_ioctl_trans &&
+            should_end_transaction(trans, root))
+                trans->transaction->blocked = 1;
+
+        if (cur_trans->blocked && !cur_trans->in_commit) {
+                if (throttle)
+                        return btrfs_commit_transaction(trans, root);
+                else
+                        wake_up_process(info->transaction_kthread);
+        }
+
         mutex_lock(&info->trans_mutex);
-        cur_trans = info->running_transaction;
-        WARN_ON(cur_trans != trans->transaction);
+        WARN_ON(cur_trans != info->running_transaction);
         WARN_ON(cur_trans->num_writers < 1);
         cur_trans->num_writers--;
 
@@ -603,6 +676,7 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans,
 
                         btrfs_free_log(trans, root);
                         btrfs_update_reloc_root(trans, root);
+                        btrfs_orphan_commit_root(trans, root);
 
                         if (root->commit_root != root->node) {
                                 switch_commit_root(root);
@@ -627,30 +701,30 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans,
 int btrfs_defrag_root(struct btrfs_root *root, int cacheonly)
 {
         struct btrfs_fs_info *info = root->fs_info;
-        int ret;
         struct btrfs_trans_handle *trans;
+        int ret;
         unsigned long nr;
 
-        smp_mb();
-        if (root->defrag_running)
+        if (xchg(&root->defrag_running, 1))
                 return 0;
-        trans = btrfs_start_transaction(root, 1);
+
         while (1) {
-                root->defrag_running = 1;
+                trans = btrfs_start_transaction(root, 0);
+                if (IS_ERR(trans))
+                        return PTR_ERR(trans);
+
                 ret = btrfs_defrag_leaves(trans, root, cacheonly);
+
                 nr = trans->blocks_used;
                 btrfs_end_transaction(trans, root);
                 btrfs_btree_balance_dirty(info->tree_root, nr);
                 cond_resched();
 
-                trans = btrfs_start_transaction(root, 1);
                 if (root->fs_info->closing || ret != -EAGAIN)
                         break;
         }
         root->defrag_running = 0;
-        smp_mb();
-        btrfs_end_transaction(trans, root);
-        return 0;
+        return ret;
 }
 
 #if 0
@@ -758,47 +832,63 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
         struct btrfs_root *root = pending->root;
         struct btrfs_root *parent_root;
         struct inode *parent_inode;
+        struct dentry *dentry;
         struct extent_buffer *tmp;
         struct extent_buffer *old;
         int ret;
-        u64 objectid;
-        int namelen;
+        int retries = 0;
+        u64 to_reserve = 0;
         u64 index = 0;
-
-        parent_inode = pending->dentry->d_parent->d_inode;
-        parent_root = BTRFS_I(parent_inode)->root;
+        u64 objectid;
 
         new_root_item = kmalloc(sizeof(*new_root_item), GFP_NOFS);
         if (!new_root_item) {
-                ret = -ENOMEM;
+                pending->error = -ENOMEM;
                 goto fail;
         }
+
         ret = btrfs_find_free_objectid(trans, tree_root, 0, &objectid);
-        if (ret)
+        if (ret) {
+                pending->error = ret;
                 goto fail;
+        }
+
+        btrfs_reloc_pre_snapshot(trans, pending, &to_reserve);
+        btrfs_orphan_pre_snapshot(trans, pending, &to_reserve);
+
+        if (to_reserve > 0) {
+                ret = btrfs_block_rsv_add(trans, root, &pending->block_rsv,
+                                          to_reserve, &retries);
+                if (ret) {
+                        pending->error = ret;
+                        goto fail;
+                }
+        }
 
         key.objectid = objectid;
-        /* record when the snapshot was created in key.offset */
-        key.offset = trans->transid;
-        btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
+        key.offset = (u64)-1;
+        key.type = BTRFS_ROOT_ITEM_KEY;
 
-        memcpy(&pending->root_key, &key, sizeof(key));
-        pending->root_key.offset = (u64)-1;
+        trans->block_rsv = &pending->block_rsv;
 
+        dentry = pending->dentry;
+        parent_inode = dentry->d_parent->d_inode;
+        parent_root = BTRFS_I(parent_inode)->root;
         record_root_in_trans(trans, parent_root);
+
         /*
          * insert the directory item
          */
-        namelen = strlen(pending->name);
         ret = btrfs_set_inode_index(parent_inode, &index);
         BUG_ON(ret);
         ret = btrfs_insert_dir_item(trans, parent_root,
-                            pending->name, namelen,
-                            parent_inode->i_ino,
-                            &pending->root_key, BTRFS_FT_DIR, index);
+                                dentry->d_name.name, dentry->d_name.len,
+                                parent_inode->i_ino, &key,
+                                BTRFS_FT_DIR, index);
         BUG_ON(ret);
 
-        btrfs_i_size_write(parent_inode, parent_inode->i_size + namelen * 2);
+        btrfs_i_size_write(parent_inode, parent_inode->i_size +
+                                         dentry->d_name.len * 2);
         ret = btrfs_update_inode(trans, parent_root, parent_inode);
         BUG_ON(ret);
 
@@ -815,22 +905,32 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
         free_extent_buffer(old);
 
         btrfs_set_root_node(new_root_item, tmp);
-        ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
-                                new_root_item);
-        BUG_ON(ret);
+        /* record when the snapshot was created in key.offset */
+        key.offset = trans->transid;
+        ret = btrfs_insert_root(trans, tree_root, &key, new_root_item);
         btrfs_tree_unlock(tmp);
         free_extent_buffer(tmp);
+        BUG_ON(ret);
 
-        ret = btrfs_add_root_ref(trans, parent_root->fs_info->tree_root,
-                                 pending->root_key.objectid,
+        /*
+         * insert root back/forward references
+         */
+        ret = btrfs_add_root_ref(trans, tree_root, objectid,
                                  parent_root->root_key.objectid,
-                                 parent_inode->i_ino, index, pending->name,
-                                 namelen);
+                                 parent_inode->i_ino, index,
+                                 dentry->d_name.name, dentry->d_name.len);
         BUG_ON(ret);
 
+        key.offset = (u64)-1;
+        pending->snap = btrfs_read_fs_root_no_name(root->fs_info, &key);
+        BUG_ON(IS_ERR(pending->snap));
+
+        btrfs_reloc_post_snapshot(trans, pending);
+        btrfs_orphan_post_snapshot(trans, pending);
 fail:
         kfree(new_root_item);
-        return ret;
+        btrfs_block_rsv_release(root, &pending->block_rsv, (u64)-1);
+        return 0;
 }
 
 /*
@@ -878,6 +978,16 @@ int btrfs_transaction_in_commit(struct btrfs_fs_info *info)
         return ret;
 }
 
+int btrfs_transaction_blocked(struct btrfs_fs_info *info)
+{
+        int ret = 0;
+        spin_lock(&info->new_trans_lock);
+        if (info->running_transaction)
+                ret = info->running_transaction->blocked;
+        spin_unlock(&info->new_trans_lock);
+        return ret;
+}
+
 int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
                              struct btrfs_root *root)
 {
@@ -899,6 +1009,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
         ret = btrfs_run_delayed_refs(trans, root, 0);
         BUG_ON(ret);
 
+        btrfs_trans_release_metadata(trans, root);
+
         cur_trans = trans->transaction;
         /*
          * set the flushing flag so procs in this transaction have to
@@ -951,9 +1063,6 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
                         snap_pending = 1;
 
                 WARN_ON(cur_trans != trans->transaction);
-                prepare_to_wait(&cur_trans->writer_wait, &wait,
-                                TASK_UNINTERRUPTIBLE);
-
                 if (cur_trans->num_writers > 1)
                         timeout = MAX_SCHEDULE_TIMEOUT;
                 else if (should_grow)
@@ -976,6 +1085,9 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
                  */
                 btrfs_run_ordered_operations(root, 1);
 
+                prepare_to_wait(&cur_trans->writer_wait, &wait,
+                                TASK_UNINTERRUPTIBLE);
+
                 smp_mb();
                 if (cur_trans->num_writers > 1 || should_grow)
                         schedule_timeout(timeout);
@@ -1103,9 +1215,9 @@ int btrfs_clean_old_snapshots(struct btrfs_root *root)
 
                 if (btrfs_header_backref_rev(root->node) <
                     BTRFS_MIXED_BACKREF_REV)
-                        btrfs_drop_snapshot(root, 0);
+                        btrfs_drop_snapshot(root, NULL, 0);
                 else
-                        btrfs_drop_snapshot(root, 1);
+                        btrfs_drop_snapshot(root, NULL, 1);
         }
         return 0;
 }