Why is using thread locals in Django bad?
I am using thread locals to store the current user and request objects. This way I can have easy access to the request from anywhere in the programme (e.g. dynamic forms) without having to pass them around.
To implement the thread locals storage in a middleware, I followed a tutorial on the Django site: http://code.djangoproject.com/wiki/CookBookThreadlocalsAndUser?version=18
This document has since been modified to suggest avoiding this technique: http://code.djangoproject.com/wiki/CookBookThreadlocalsAndUser?version=20
From the article:
From a design point of view, threadlocals are essentially global variables, and are subject to all the usual problems of portability and predictability that global variables usually entail.
More importantly, from a security point of view, threadlocals pose a huge risk. By providing an data store that exposes the state of other threads, you provide a way for one thread in your web server to potentially modify the state of another thread in the system. If the threadlocal data contains descriptions of users or other authentication-related data, that data could be used as the basis for an attack that grants access to an unauthorized user, or exposes private details of a user. While it is possible to build a threadlocal system that is safe from this sort of attack, it's a lot easier to be defensive and build a system that isn't subject to any such vulnerability in the first place.
I understand why global variables can be bad, but in this case I'm running my own code on my own server so I can't see what danger two global variables pose.
Can someone explain the security issue involved? I have asked many people how they would hack my application if they read this article and know I'm using thread locals, yet no one has been able to tell me. I am starting to suspect that this is an opinion held by hair-splitting purists who love to pass objects explicitly.
Answer
I disagree entirely. TLS is extremely useful. It should be used with care, just as globals should be used with care; but saying it shouldn't be used at all is just as ridiculous as saying globals should never be used.
For example, I store the currently active request in TLS. This makes it accessible from my logging class, without having to pass the request around through every single interface--including many that don't care about Django at all. It lets me make log entries from anywhere in the code; the logger outputs to a database table, and if a request happens to be active when a log is made, it logs things like the active user and what was being requested.
If you don't want one thread to have the capability of modifying another thread's TLS data, then set your TLS up to prohibit this, which probably requires using a native TLS class. I don't find that argument convincing, though; if an attacker can execute arbitrary Python code as your backend, your system is already fatally compromised--he could monkey patch anything to be run later as a different user, for example.
Obviously, you'll want to clear any TLS at the end of a request; in Django, that means clearing it in process_response and process_exception in a middleware class.