Technical Articles
F5 SMEs share good practice.
cancel
Showing results for 
Search instead for 
Did you mean: 
Custom Alert Banner
jwham20
Nimbostratus
Nimbostratus

->Part of the F5/Owasp Top Ten Series 
 

At the top of the  Owasp list is Injections. Their definition is

“Injection flaws, such as SQL, OS, and LDAP injection, occur when untrusted data is sent to an interpreter as part of a command or query. The attacker’s hostile data can trick the interpreter into executing unintended commands or accessing unauthorized data. “

 

Long story short, it’s is allowing unsanitized input into a program field that has the potential for execution. (which is darn near everywhere these days)

Everyone knows the story of little bobby:

0151T000003d4BYQAY.jpg

In all honesty, I thought that Bobby’s Mom had a very valid point. 

 

Let’s Inject:

Basic injection attacks are fairly simple to perform. We find an input parameter and try to send it something nefarious. In my labs, I’ve got a nice little auction site up and running. Everyone loves an auction.

To bid on items, of course, you need to be authenticated.  Well, being the evilHacker,  I want to get in without using my credentials. This is where the injection comes in. Using either passive intelligence gathering or just guessing due to the common usages,  I decide to try a simple SQL-Injection attack:

The input we are injecting into is the USERNAME field:

Username: ‘ or 1=1 #

 

Pre-Injection:

0151T000003d4BZQAY.jpg
Post Injection:
0151T000003d4BaQAI.jpg

Huh…  a logged in user of ‘ or 1=1 #?  Rut Row Shaggy!  So what is going on here?  Lets look at the code at play:

<php yadda yadda yadda

$query = "select id from users where nick='$username' and password='".md5($MD5_PREFIX.$password)."' and suspended=0";

It says:

Find me the user who’s username and password matches the input (username, plus some MD5 fun on the password) AND whose account is not suspended. How nice.

 

So what evilHacker did was make that simple query say:

$query = "select id from users where nick=`‘ or 1=1 # and password='".md5($MD5_PREFIX.$password)."' and suspended=0";

Now it says:

Find me the user <no one> or 1=1 (1=1 is a truth statement). In essence you get a select all records that exist in the table users. 

Not a very strong front door eh?  



 

Let’s fix

Wouldn’t it be nice if they could just fix it at the code level and be done with it? Well, this one they can (fairly simple escaping of characters). But we all know is reality, most code changes require scrums, waterfalls, validations, testing, and a flood of tears. 

In our case, we already have the Virtual server for this website on the LTM/ASM (Virtual Edition 11.1).  It’s a few steps to get the ASM in place to defend:

1. Create the HTTP Class for the Virtual Server:

         A. Local Traffic –> Virtual Servers –> Profiles –>Protocol –> HTTP Class –>Create

         B. Give it a name and select Enabled for the “Application Security”

0151T000003d4BbQAI.jpg

2. Go to Application Security –>Security Policies.  Select “Configure Security Policy”

0151T000003d4BcQAI.jpg
3. For our case, we are going to “Create a Policy Manually”
4. Set the Policy Language. Here we are using UTF-8.
0151T000003d4BdQAI.jpg
5. Select the Signature Lists you want to use. For our site, we are going to run with the defaults and for the sake of the demonstration, I turned off staging.  This way, we get immediate blocks (yay satisfaction!)0151T000003d4BeQAI.jpg
6. Now we configure the Wildcard Tightening. It will put a wildcard in place so that we can have a chance to learn parameters as they are used. Hit next and finished.0151T000003d4BfQAI.jpg
7. Now, we apply that HTTP Class to the virtual server. This hooks it into the asm. 
      A. Local Traffic –> Virtual Servers –> Your VS – >Resources –>HTTPClass Profiles
      B. Add your profile and update
0151T000003d4BgQAI.jpg
8. Now we are passing traffic through the ASM, I hit the page and log in. In the ASM Learning section, I now see the parameters for username and password. 
    A. Application Security –>Policy Building –>Manual –> Traffic Learning.  Click on    Parameters and we see them.
    B. For now, I hit accept all.
0151T000003d4BhQAI.jpg
9. Now the parameters are in the policy, the policy is listening transparently. I want to take the out of staging, so I get immediate blocks. I go to
    Policy –> Parameters –> Parameter List

and select each parameter I want to remove from staging. Here, I do password and username.
0151T000003d4BiQAI.jpg
10. Now, I want to put the policy in blocking mode, to see this bad boy in action.
    A. Click Policy
    B Set Enforcement Mode to Blocking
    C. Profit?  Or hit Save,  then apply policy in the top right
0151T000003d4BjQAI.jpg

 

Now:

Pre-Injection:

0151T000003d4BkQAI.jpg
Post Injection:

0151T000003d4BlQAI.jpg

Why the Block?

Now the coolest part,  we, as the admins, can see why the block happened.

0151T000003d4BmQAI.jpg

We go to Application Security –> Reporting –>Requests.  Put the Support ID into the filter.

It returns the full request,  why it was blocked, and the options to learn it as a false positive.

 

Pretty cool huh?  This is only the tip of the iceberg for what fun we can have with the ASM.

 

Part of the F5/Owasp Top Ten Series

Comments
swo0sh_gt_13163
Altostratus
Altostratus
Hey Josh, Cool explanation. What exactly ASM doesm when ASM admin clicks on "Learn" if the incident reported is false positive? Will it keep it under staging one more time and once the ERP is over, it will ask you to enforce it? Or it will create an exception for this parameter for a particular signature? Cheers! Darshan
Ndlovumm
Cirrus
Cirrus

Great explanation

Version history
Last update:
‎29-Feb-2012 13:00
Updated by:
Contributors