Tag: security

Automated Rsync – without compromising security

Rsync is one of the most popular tools to synchronize data between two computers, and used mostly in taking backups using this sync feature. It’s easy to use and only uploads the changed files when a sync is necessary, so it’s effective in saving bandwidth and time too. To run rsync, you need root or a properly privileged user which can access that specific path in the remote machine. And to setup this privilege, you can either use a should-never-be-used root account or an user chrooted using a jail shell. But if, by any chance, current machine is compromised then remote machine is compromised too. Because anyone can connect to your remote machine using those credentials from the current machine. To avoid plain text credentials (or the credentials written in a shell script), people usually use ssh keys to establish a connection between two machines. Still, your remote machine is unsecured if anything goes wrong in the current machine from where you’re taking the backup.

So a fullproof solution is to use ssh keys and properly chroot the remote user so that it can only access the backup files and nothing else. However, setting up a jailed shell is a not-for-everyone type task and takes time and experience to accomplish properly. So here is a quick work around that you can implement in your remote machine to prevent the connecting user from doing anything malicious but only tasks those are needed to perform the rsync backup. Let’s see how we can do that

Before continuing, let’s name our two machines. The one which should be backed up, lets name that Workstation. And the one which is storing the backups is BackupServer. Also for now, let’s assume that you are backing up complete “/var/www” folder in the WorkStation.

Step 1: Create SSH key in the BackupServer
Log into the BackupServer and run the following command in ssh terminal. But remember, if you already have a key in ~/.ssh/id_rsa.pub then IGNORE IGNORE IGNORE this step.

[sourcecode language=”shell”]
ssh-keygen -t rsa
[/sourcecode]

It will prompt for a passphrase, REMEMBER to just hit the enter without typing anything.

Step 2: Display and Copy the SSH key from BackupServer
Run the following command to display the ssh public key in the terminal, and then copy it.
(more…)

Create personalized phar files in PHP

Created a screencast to show how you can create phar files, most importantly personalized phar files to store some information inside it and protect it using user’s password. Those information is usable only when user providers a correct password.

For packaging, I used http://box-project.org which is an excellent phar packager. I’ve also used two functions from Josh Hartman’s blog to encrypt and decrypt data using Rijndael algorithm.

www.youtube.com/watch?v=V-1NgA5T4Bw

How to login a user programatically in Symfony2

Sometime, you may need to log in an user manually from code, instead of generic form based log in. To do it, you need to use Two Security component “UsernamePasswordToken” and “InteractiveLoginEvent”. We will also use another exception object “UsernameNotFoundException” if the user is not found.

[sourcecode language=”php”]
use Symfony\Component\Security\Core\Authentication\Token\UsernamePasswordToken;
use Symfony\Component\Security\Http\Event\InteractiveLoginEvent;
use Symfony\Component\Security\Core\Exception\UsernameNotFoundException;
[/sourcecode]

Now from your controller, you can login an user like this

[sourcecode language=”php”]
$em = $this->getDoctrine();
$repo = $em->getRepository("UserBundle:User"); //Entity Repository
$user = $repo->loadUserByUsername($username);
if (!$user) {
throw new UsernameNotFoundException("User not found");
} else {
$token = new UsernamePasswordToken($user, null, "your_firewall_name", $user->getRoles());
$this->get("security.context")->setToken($token); //now the user is logged in

//now dispatch the login event
$request = $this->get("request");
$event = new InteractiveLoginEvent($request, $token);
$this->get("event_dispatcher")->dispatch("security.interactive_login", $event);
}
[/sourcecode]

Dispatching the “security.interactive_login” is important, because then every listeners which are bound to this event will work appropriately. But the actual login happens when you call setToken(…). Also make sure that you pass the correct firewall name which you had defined in your security.yml.

You can also recall the logged in user any time using the following code

[sourcecode language=”php”]
//anywhere
$user = $this->get(‘security.context’)->getToken()->getUser();

//or from a controller
$user = $this->getUser();
[/sourcecode]

That’s it. Hope you will find it handy 🙂

Massive security flaw in Facebook and why should they fix it immediately before I take your girlfriend out to dinner tonight

Ok, Facebook Groups has a huge security flaw by which any group member  can pretend to be anyone else in that group, and post in the group on behalf of that user. It is FATAL. I’ve reported to Facebook and I hope they should take immediate action for it.

I had disclosed it in details hoping that they will notice it and fix it quickly, and taking it down again. So if any Facebook official wants to know in details, drop me a mail to hasin_at_leevio_dot_com or better check today’s submitted bug reports with a “MASSIVE SECURITY FLAW” text inside it.

Peace.
*update: submitted this again to facebook.com via their whitehat program and someone named Alex contacted me. He asked me a few questions on how to reproduce the flaw and he said that they are looking into it.

Story about Blue E, iFramed Web Application, Wastage of 6 hours, Missed Lunch and what not!

Long story short, I was developing a Facebook page application which runs from inside a page tab. Such applications runs inside an iframe on Facebook Fanpage. Everything was going perfectly, until, we started checking it in IE!! The symptoms were simple, PHP Session is not working when user logs in. The application works perfectly in Safari, Opera, Chrome and Firefox and only the Elite of the Elites Internet Explorer is not accepting any cookie generated from the application.

I was hungry, almost lunch time – client was kicking my ass, so I couldn’t even think to grab a bite. I was VERY HUNGRY, clueless, lost and I was feeling like yelling at everything that walks in front of me. Oh boy, I was absolutely clueless about what was done wrong to satisfy this King Blue E.

After discussing my problem with Uncle G for over 5 hours, trial and error, do this and that, I realized that the problem is actually related to p3p privacy policy and how Internet Explorer deals with it. It doesn’t accept any cookie from any web application which is running inside an iframe within another application. Let me clarify

1. Web application A with domain a.com has an iFrame in one of its page
2. Inside that iFrame, it loads another web application B with domain b.com
3. Now Internet Explorer doesnt accept any cookie which generates by Web Application B, resulting a catastrophe, a real disaster.

The Solution: send the following header in your web application which runs inside the iFrame.

[sourcecode language=”php”]
<?php
header(‘P3P: CP="CAO PSA OUR"’);
header(‘P3P: CP="HONK"’);
?>
[/sourcecode]

and it fixes this weird behavior of Internet Explorer. Freaking p3p! wastage of 6 hours, remaining hungry, missing lunch, get kicked on the ass and what not.

I hate you Internet Explorer. You are the worst thing ever made in the history of web development. I sincerely hate you. And Microsoft, with all the Blue E fan-boys out there, GO TO HELL and let us live our life.

Morons!