... not harmless in this case - we have a string in the end of buffer
already.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>