Padding Oracle - Yet another post

I thought I will take some time to write up something about Padding Oracle attacks and how to exploit the vulnerability based on some scenarios that I encountered in the field.

Before we start, this issue has nothing to do with Oracle database (just in case you mistook) or is specific to a platform (e.g. .NET).

Padding Oracle – The vulnerability

Padding oracle has been covered in detail by many leading experts. And so, I am not going to explain the technical side of the issue. If you want to refresh your memories, check out these blog posts

Padding Oracle – Identification and Exploitation

Its not often that you would come across Padding oracle issue these days and the ones I cane across have always been .NET based. And so, I will cover how to quickly identify and exploit the vulnerability on a .NET application – MS10-070.

Identification, in a hurry

(Update: 10 Aug 2013) A word of caution before we proceed. Dan (@commonexploits) pointed out that the tools described below for testing for vulnerability do report .NET 4.5 to be vulnerable, even though they are not. So, I advice you to take your head while testing and not to rely solely on the tools. Also, I think the test without tools (described below) may help solving this issue. However, I am yet to test this out. If you are in a position to test it and find promising results let me know and I will update it on my site.

Testing without tools

  • First, get the link for WebResource.axd or ScriptResource.axd
  • Delete the value of 'd' parameter and send the request. Keep the response with you.
  • Next, put a random value to 'd' parameter and send the request again. Get the response and compare it with the previous results.

If the vulnerability exists, then you will see some difference in the pages.

Testing with tools

To identify a potentially vulnerable site, first scour the .NET website for a link referencing to WebResource.axd and ScriptResource.axd. Obtain the value of the 'd' parameter within this link.

There are two scripts you can use to quickly check if the site is vulnerable or not.

  1. Check Patch by Giorgio Fedon at Minded Security
  2. MS10-070_check by Bernado Damele
  3. dotnetpaddingoracle by Gabriel Caudrelier at NCC Group (Update: 10 Aug 2013). Note: This runs on Python v3.

Run either of the script and provide the value of 'd' parameter within WebResource.axd and/or ScriptResource.axd. The script will check and will let you know if the site is potentially vulnerable or not.

Exploitation – Scenario based approach

Scenario # 1 – Normal cases

Normal cases involve the application responding with a 500 status codes when there is an error in decryption.

Tools to use

First task is to find the encrypted cipher block for the arbitrary file to download. As a proof of concept, we will try to extract the web.config file.

We use PadBuster from Brian Holyfield at Gotham Digital Security to aid in obtaining the encrypted cipher block.

Command: 12sfion23c0293jrm_asdmals232d2lms23f2; 16 –encoding 3 –plaintext "|||~/web.config" –prefix 12sfion23c0293jrm_asdmals232d2lms23f2


After few minutes, PadBuster will ask you to select one from a list of response signatures – it will have ** beside it.

Wait for few minutes until PadBuster does its business and in the end will give you the encrypted cipher block for "|||~/web.config".

Next task is to brute force and gain access to web.config. We now have three different choices of tools – PadBuster by Brian Holyfield. Web.Config bruter by Giorgio Fedon and ASPX_PO_CHOTEXT_ATTACK script by Agustin Azubel.


PadBuster 16 –encoding 3 –noiv –bruteforce –log –prefix 12sfion23c0293jrm_asdmals232d2lms23f2

Look out for a relatively larger (or smaller) content length (which would be web.config)

Web.config_bruter 16

Once web.config is obtained, the script will print it on the screen and stop.


ruby aspx_po_chotext_attack.rb

As you can see here, we don't have to give the direct link to WebResource.axd or ScriptResource.axd. Just give the link to the page which will have ScriptResource.axd link. The script will find it and then do the decryption and then brute forcing for you.


Scenario # 2 – Custom error messages enabled

When this issue was first discovered and disclosed, Microsoft released a quick workaround (link) – Use custom error messages. When custom error messages are enabled, things get very tricky as you might not get the usual 500 status code when decryption fails.

But, if you don't configure custom error message properly like reference two different types of error files for the type of error (e.g. filenotavailable.html for 404, error.aspx for 500) then this difference can be used to exploit. Also, look for any difference in the pages being returned (e.g. words).

Brian Holyfield has written some steps that can be taken to exploit even when custom error messages are enabled. Check this link for more info. Read the section "Bypassing the Workarounds".

Hope all the information I collected and provided helps you.

Happy hacking..