ovl: don't check sticky
authorMiklos Szeredi <mszeredi@redhat.com>
Fri, 16 Dec 2016 10:02:55 +0000 (11:02 +0100)
committerMiklos Szeredi <mszeredi@redhat.com>
Fri, 16 Dec 2016 10:02:55 +0000 (11:02 +0100)
Since commit 07a2daab49c5 ("ovl: Copy up underlying inode's ->i_mode to
overlay inode") sticky checking on overlay inode is performed by the vfs,
so checking against sticky on underlying inode is not needed.

Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
fs/overlayfs/dir.c

index a806fb1..fdc7b14 100644 (file)
@@ -717,28 +717,12 @@ out_unlock:
        return err;
 }
 
-static inline int ovl_check_sticky(struct dentry *dentry)
-{
-       struct inode *dir = ovl_dentry_real(dentry->d_parent)->d_inode;
-       struct inode *inode = ovl_dentry_real(dentry)->d_inode;
-
-       if (check_sticky(dir, inode))
-               return -EPERM;
-
-       return 0;
-}
-
 static int ovl_do_remove(struct dentry *dentry, bool is_dir)
 {
        enum ovl_path_type type;
        int err;
        const struct cred *old_cred;
 
-
-       err = ovl_check_sticky(dentry);
-       if (err)
-               goto out;
-
        err = ovl_want_write(dentry);
        if (err)
                goto out;
@@ -804,10 +788,6 @@ static int ovl_rename(struct inode *olddir, struct dentry *old,
 
        flags &= ~RENAME_NOREPLACE;
 
-       err = ovl_check_sticky(old);
-       if (err)
-               goto out;
-
        /* Don't copy up directory trees */
        old_type = ovl_path_type(old);
        err = -EXDEV;
@@ -815,10 +795,6 @@ static int ovl_rename(struct inode *olddir, struct dentry *old,
                goto out;
 
        if (new->d_inode) {
-               err = ovl_check_sticky(new);
-               if (err)
-                       goto out;
-
                if (d_is_dir(new))
                        new_is_dir = true;