Safety-critical systems from the inside
summary
Shows concrete engineering choices: design hardware so a valve defaults closed so gas flow stops (fail-safe), use two CPUs exchanging a timed square wave to detect clock faults and three-channel voting (aerospace-style) to mask single failures, move a TCP/IP stack onto a separate processor to implement mixed criticality, store a value plus its bitwise negation to detect memory corruption, and prefer Ada/SPARK or MISRA subsets when formal checks (no runtime errors) and long-term reproducible builds are required.
AI-written summary · about
chapters (29)
- 0:00Intro
- 0:26MACIEJ GAJDZICA
- 1:36SYSTEM WHOSE MALFUNCTI CAN LEAD TO
- 2:26SAFETY VS SECURITY
- 3:13SAFETY VS RELIABILITY
- 4:23Risk: cutting fingers
- 5:13Risk: burning everything
- 6:54SAFETY INTEGRITY LEVEL SIL
- 8:29FAIL SAFE
- 10:50CPU ERRORS
- 11:58HOW TO DETECT CLOCK FAIL
- 13:35REDUNDANCY
- 15:14SAFE COMMUNICATION
- 18:26PROBLEM
- 20:14SOLUTION
- 21:46MIXED CRITICALITY
- 22:41DATA CORRUPTION
- 25:55uint32_t const value; uint32_t const value_inv
- 30:40FORMAL PROOF
- 31:58ADA SPARK
- 33:12LANGUAGE SUBSETS
- 35:33EFFECTIVE DOCUMENTATION
- 36:57VERSION MANAGEMENT
- 40:19PEOPLE AND PROCESSES
- 41:29ROOT CAUSE ANALYSIS
- 43:16FEEDBACK LOOP
- 44:36Engineering a Safer World
- 49:29PROJECT ROLES - SIL4
- 52:59ADDITIONAL RESOURCES
related talks
50mWhen Human Life Depends on Software - Introduction to Safety-Critical Systems
42mDecoding Safe(ty) Linux Architectural Approaches for Critical Systems
46mA Systems Approach to Safety and Cybersecurity
49mWhen Software Determines Human Life — On Safety-Critical Systems
43mLearning to Live with Errors
27mLet's Bring System Dynamics Back to CS!
58mBuilding antifragile systems using Secure by Design
54mArchitecture & Responsible Technology