|There's more than one way to do things|
Re: Web Security for CGI programsby andreychek (Parson)
|on Jul 03, 2001 at 05:56 UTC||Need Help??|
In addition to some sort of user authentication, nothing beats good 'ol taint mode. If you're dealing with CGI, just use
at the top of your Perl scripts. If you're in mod_perl, you need add the following line to your httpd.conf:
Taint mode forces you to clean up data that was received from an outside source, before you can do anything too damaging with it. To untaint data, you just have to send it through a regex.. if it matches, the resulting $1 will be untainted. Yes, it is possible to write cheesy regexes to get around this, but then that's the programmers fault for not trying.
You can get all sorts of juicy bits of information on taint mode, and other security measures, with "perldoc perlsec". Additionally, there are two modules, taint and untaint which are utility functions for working in taint mode. You don't need them, but they may prove helpful depending on what you are trying to accomplish.
When the browser connects to the server for the first time, a new session ID is generated for it. Additionally, a hash is generated (made of random characters, much like the session ID). The session ID and Hash are stored on the server, AND given to the browser on it's first request (which is probably the login page).
When the form finally submits the info to the server, your server does the same process. It digs the users password out of the database (or whatever you are using), runs it through MD5, appends that to the hash it has associated with your Session ID, and then runs the resulting string through MD5. Now, if that matches what the browser passed you, the password is correct!
This method isn't the most secure method in the world, by any means. SSL would be better then this. For even better protection SSL could be used in addition to this method. But as you said, SSL isn't always available. If somebody is sniffing your connection and you don't have SSL, they can sniff the hash that the server sent you. Now, they would never ever see the password in plaintext. But with the hash, they might have a better chance at being able to brute force your password. Would it be easy for them to do this? No. But if somebody really wanted to break in, that would be a possible route they could use. Just something to ponder...