Btrfs: fix orphan transaction on the freezed filesystem

With the following debug patch:

 static int btrfs_freeze(struct super_block *sb)
 {
+ 	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
+	struct btrfs_transaction *trans;
+
+	spin_lock(&fs_info->trans_lock);
+	trans = fs_info->running_transaction;
+	if (trans) {
+		printk("Transid %llu, use_count %d, num_writer %d\n",
+			trans->transid, atomic_read(&trans->use_count),
+			atomic_read(&trans->num_writers));
+	}
+	spin_unlock(&fs_info->trans_lock);
 	return 0;
 }

I found there was a orphan transaction after the freeze operation was done.

It is because the transaction may not be committed when the transaction handle
end even though it is the last handle of the current transaction. This design
avoid committing the transaction frequently, but also introduce the above
problem.

So I add btrfs_attach_transaction() which can catch the current transaction
and commit it. If there is no transaction, it will return ENOENT, and do not
anything.

This function also can be used to instead of btrfs_join_transaction_freeze()
because it don't increase the writer counter and don't start a new transaction,
so it also can fix the deadlock between sync and freeze.

Besides that, it is used to instead of btrfs_join_transaction() in
transaction_kthread(), because if there is no transaction, the transaction
kthread needn't anything.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>

1 files changed, 30 insertions, 15 deletions

diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index d0a2b7e4938..69139a356f7 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -53,7 +53,7 @@ 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, int nofail)
+static noinline int join_transaction(struct btrfs_root *root, int type)
 {
         struct btrfs_transaction *cur_trans;
         struct btrfs_fs_info *fs_info = root->fs_info;
@@ -67,7 +67,13 @@ loop:
         }
 
         if (fs_info->trans_no_join) {
-                if (!nofail) {
+                /* 
+                 * If we are JOIN_NOLOCK we're already committing a current
+                 * transaction, we just need a handle to deal with something
+                 * when committing the transaction, such as inode cache and
+                 * space cache. It is a special case.
+                 */
+                if (type != TRANS_JOIN_NOLOCK) {
                         spin_unlock(&fs_info->trans_lock);
                         return -EBUSY;
                 }
@@ -87,6 +93,13 @@ loop:
         }
         spin_unlock(&fs_info->trans_lock);
 
+        /*
+         * If we are ATTACH, we just want to catch the current transaction,
+         * and commit it. If there is no transaction, just return ENOENT.
+         */
+        if (type == TRANS_ATTACH)
+                return -ENOENT;
+
         cur_trans = kmem_cache_alloc(btrfs_transaction_cachep, GFP_NOFS);
         if (!cur_trans)
                 return -ENOMEM;
@@ -340,27 +353,28 @@ again:
          * because we're already holding a ref.  We need this because we could
          * have raced in and did an fsync() on a file which can kick a commit
          * and then we deadlock with somebody doing a freeze.
+         *
+         * If we are ATTACH, it means we just want to catch the current
+         * transaction and commit it, so we needn't do sb_start_intwrite(). 
          */
-        if (type != TRANS_JOIN_NOLOCK &&
-            !__sb_start_write(root->fs_info->sb, SB_FREEZE_FS, false)) {
-                if (type == TRANS_JOIN_FREEZE) {
-                        kmem_cache_free(btrfs_trans_handle_cachep, h);
-                        return ERR_PTR(-EPERM);
-                }
+        if (type < TRANS_JOIN_NOLOCK)
                 sb_start_intwrite(root->fs_info->sb);
-        }
 
         if (may_wait_transaction(root, type))
                 wait_current_trans(root);
 
         do {
-                ret = join_transaction(root, type == TRANS_JOIN_NOLOCK);
+                ret = join_transaction(root, type);
                 if (ret == -EBUSY)
                         wait_current_trans(root);
         } while (ret == -EBUSY);
 
         if (ret < 0) {
-                sb_end_intwrite(root->fs_info->sb);
+                /* We must get the transaction if we are JOIN_NOLOCK. */
+                BUG_ON(type == TRANS_JOIN_NOLOCK);
+
+                if (type < TRANS_JOIN_NOLOCK)
+                        sb_end_intwrite(root->fs_info->sb);
                 kmem_cache_free(btrfs_trans_handle_cachep, h);
                 return ERR_PTR(ret);
         }
@@ -432,9 +446,9 @@ struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *root
         return start_transaction(root, 0, TRANS_USERSPACE, 0);
 }
 
-struct btrfs_trans_handle *btrfs_join_transaction_freeze(struct btrfs_root *root)
+struct btrfs_trans_handle *btrfs_attach_transaction(struct btrfs_root *root)
 {
-        return start_transaction(root, 0, TRANS_JOIN_FREEZE, 0);
+        return start_transaction(root, 0, TRANS_ATTACH, 0);
 }
 
 /* wait for a transaction commit to be fully complete */
@@ -605,7 +619,7 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
                 }
         }
 
-        if (lock)
+        if (trans->type < TRANS_JOIN_NOLOCK)
                 sb_end_intwrite(root->fs_info->sb);
 
         WARN_ON(cur_trans != info->running_transaction);
@@ -1678,7 +1692,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
         put_transaction(cur_trans);
         put_transaction(cur_trans);
 
-        sb_end_intwrite(root->fs_info->sb);
+        if (trans->type < TRANS_JOIN_NOLOCK)
+                sb_end_intwrite(root->fs_info->sb);
 
         trace_btrfs_transaction_commit(root);