From 78cee9a8e4b42b3f585ea3bd1c076f5a76fee722 Mon Sep 17 00:00:00 2001 From: "J. Bruce Fields" Date: Tue, 22 Oct 2013 20:59:19 -0400 Subject: exportfs: more detailed comment for path_reconnect Reviewed-by: Christoph Hellwig Signed-off-by: J. Bruce Fields Signed-off-by: Al Viro --- fs/exportfs/expfs.c | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) (limited to 'fs/exportfs') diff --git a/fs/exportfs/expfs.c b/fs/exportfs/expfs.c index 6d0a7fa9abb3..87e6dca69e43 100644 --- a/fs/exportfs/expfs.c +++ b/fs/exportfs/expfs.c @@ -93,7 +93,19 @@ find_disconnected_root(struct dentry *dentry) /* * Make sure target_dir is fully connected to the dentry tree. * - * It may already be, as the flag isn't always updated when connection happens. + * On successful return, DCACHE_DISCONNECTED will be cleared on + * target_dir, and target_dir->d_parent->...->d_parent will reach the + * root of the filesystem. + * + * Whenever DCACHE_DISCONNECTED is unset, target_dir is fully connected. + * But the converse is not true: target_dir may have DCACHE_DISCONNECTED + * set but already be connected. In that case we'll verify the + * connection to root and then clear the flag. + * + * Note that target_dir could be removed by a concurrent operation. In + * that case reconnect_path may still succeed with target_dir fully + * connected, but further operations using the filehandle will fail when + * necessary (due to S_DEAD being set on the directory). */ static int reconnect_path(struct vfsmount *mnt, struct dentry *target_dir, char *nbuf) -- cgit