Building Hardened IoT Implementations With LangSec
summary
Language-theoretic approach separates recognition from processing and yields concrete results: a C/C++ hammer parser-combinator library (bindings for Ruby, Python, .NET) plus a generator/fuzzer pipeline (H_agee) that emits valid seeds from grammars. Implementations for MQTT, XMPP, AMQP (AMQP in Python 2.7) and SCADA protocols (DNP3, IEC C37) parse like the spec, run on a Raspberry Pi in microseconds, keep parser code under ~50 LOC, and survive state-of-the-art fuzzers (AFL/libFuzzer) with no crashes over 24 hours.
AI-written summary · about
chapters (8)
- 0:00Start
- 7:12Parsing & protocol anti-patterns
- 14:14A recap of the Chomsky hierarchy
- 16:53Principles & Desiderata
- 23:20AMQP Protocol
- 28:37Code that looks like grammar - DNP3
- 33:32Wiring hammer with
- 34:55Concluding Remarks
related talks
52mVisualizing Security via LANGSEC
29mBuilding a Hackbot: How Ethiack's Hackian Finds Real Vulnerabilities
44mTL;DR: Applying AI to Security
01h 05mTheory and Security
21mSecuring IoT's Trust Boundaries: Navigating Complexity with NCSC Patterns
52mStructure aware fuzzing
34mLow-level RASP: Protecting Applications Implemented in High-level Programming Languages
47mA New Era of SSRF - Exploiting URL Parser in Trending Programming Languages!
