Explore a novel approach to protecting sensitive data in software systems using language-level information flow control (IFC) in this conference talk from Strange Loop. Dive into LIO, an IFC-security language embedded in Haskell, and learn how it leverages monads and strong typing to enable developers to specify and enforce security policies on data. Discover Hails, a server-side web framework built on LIO, which allows for declarative policy specification alongside the data model, separate from application functionality. Understand how these tools can help construct web applications that inherently protect user data privacy and integrity. Examine real-world examples, including a blogging web app, to see how this approach can prevent data breaches caused by overlooked security checks. Gain insights into the benefits of using Haskell for secure programming, the concept of secure sub-languages, and the practical implementation details of this security-focused development paradigm.
Overview
Syllabus
Intro
Facebook missed a single security check...
Putting user privacy at risk: a recipe
A blogging web app example
Change how we build software
Yes! At least for web apps...
Hails: secure Haskell web framework
Extend MVC with security policy
How do we specify policy? • Policy specified as function from row to label Label concisely encodes read/write restriction on data
Where to enforce label restrictions?
How do we enforce labels?
Two kinds of code: MPs and VCs
extensibility
Is the TCB actually smaller?
Implementation details
Why Haskell?
Thinking about secure sub-language
Review: the 10 monad
How can we do anything useful?
Tracking reads & restricting writes
The LIO secure sub-language
What do we get with this?
What's the catch?
Summary
Taught by
Strange Loop Conference
