fs: Fix S_NOSEC handling
authorJan Kara <jack@suse.cz>
Thu, 21 May 2015 14:05:52 +0000 (16:05 +0200)
committerBen Hutchings <ben@decadent.org.uk>
Wed, 12 Aug 2015 14:33:17 +0000 (16:33 +0200)
commit 2426f3910069ed47c0cc58559a6d088af7920201 upstream.

file_remove_suid() could mistakenly set S_NOSEC inode bit when root was
modifying the file. As a result following writes to the file by ordinary
user would avoid clearing suid or sgid bits.

Fix the bug by checking actual mode bits before setting S_NOSEC.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
[bwh: Backported to 3.2: adjust filename]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
mm/filemap.c

index 556858c..6c009c2 100644 (file)
@@ -2007,8 +2007,8 @@ int file_remove_suid(struct file *file)
                error = security_inode_killpriv(dentry);
        if (!error && killsuid)
                error = __remove_suid(dentry, killsuid);
-       if (!error && (inode->i_sb->s_flags & MS_NOSEC))
-               inode->i_flags |= S_NOSEC;
+       if (!error)
+               inode_has_no_xattr(inode);
 
        return error;
 }