Btrfs: optimize fsyncs on old files

The fsync log has code to make sure all of the parents of a file are in the
log along with the file.  It uses a minimal log of the parent directory
inodes, just enough to get the parent directory on disk.

If the transaction that originally created a file is fully on disk,
and the file hasn't been renamed or linked into other directories, we
can safely skip the parent directory walk.  We know the file is on disk
somewhere and we can go ahead and just log that single file.

This is more important now because unrelated unlinks in the parent directory
might make us force a commit if we try to log the parent.

Signed-off-by: Chris Mason <chris.mason@oracle.com>

diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index 1b7f04a8f168521d7c8001eb12a3ee7ed48f989c..fc9b87a7975bd38f76d7e22c84cfe010e3aafa42 100644 (file)
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -2835,6 +2835,17 @@ static noinline int check_parent_dirs_for_sync(struct btrfs_trans_handle *trans,
        int ret = 0;
        struct btrfs_root *root;
 
+       /*
+        * for regular files, if its inode is already on disk, we don't
+        * have to worry about the parents at all.  This is because
+        * we can use the last_unlink_trans field to record renames
+        * and other fun in this file.
+        */
+       if (S_ISREG(inode->i_mode) &&
+           BTRFS_I(inode)->generation <= last_committed &&
+           BTRFS_I(inode)->last_unlink_trans <= last_committed)
+                       goto out;
+
        if (!S_ISDIR(inode->i_mode)) {
                if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb)
                        goto out;
@@ -2904,8 +2915,19 @@ int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 
        ret = btrfs_log_inode(trans, root, inode, inode_only);
        BUG_ON(ret);
-       inode_only = LOG_INODE_EXISTS;
 
+       /*
+        * for regular files, if its inode is already on disk, we don't
+        * have to worry about the parents at all.  This is because
+        * we can use the last_unlink_trans field to record renames
+        * and other fun in this file.
+        */
+       if (S_ISREG(inode->i_mode) &&
+           BTRFS_I(inode)->generation <= last_committed &&
+           BTRFS_I(inode)->last_unlink_trans <= last_committed)
+                       goto no_parent;
+
+       inode_only = LOG_INODE_EXISTS;
        while (1) {
                if (!parent || !parent->d_inode || sb != parent->d_inode->i_sb)
                        break;
@@ -2921,6 +2943,7 @@ int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 
                parent = parent->d_parent;
        }
+no_parent:
        ret = 0;
        btrfs_end_log_trans(root);
 end_no_trans:
@@ -3069,6 +3092,19 @@ void btrfs_record_unlink_dir(struct btrfs_trans_handle *trans,
                             struct inode *dir, struct inode *inode,
                             int for_rename)
 {
+       /*
+        * when we're logging a file, if it hasn't been renamed
+        * or unlinked, and its inode is fully committed on disk,
+        * we don't have to worry about walking up the directory chain
+        * to log its parents.
+        *
+        * So, we use the last_unlink_trans field to put this transid
+        * into the file.  When the file is logged we check it and
+        * don't log the parents if the file is fully on disk.
+        */
+       if (S_ISREG(inode->i_mode))
+               BTRFS_I(inode)->last_unlink_trans = trans->transid;
+
        /*
         * if this directory was already logged any new
         * names for this file/dir will get recorded
@@ -3114,6 +3150,13 @@ int btrfs_log_new_name(struct btrfs_trans_handle *trans,
 {
        struct btrfs_root * root = BTRFS_I(inode)->root;
 
+       /*
+        * this will force the logging code to walk the dentry chain
+        * up for the file
+        */
+       if (S_ISREG(inode->i_mode))
+               BTRFS_I(inode)->last_unlink_trans = trans->transid;
+
        /*
         * if this inode hasn't been logged and directory we're renaming it
         * from hasn't been logged, we don't need to log it