Precursor
summary
Precursor is a compact, inspectable mobile platform that compiles open hardware into an FPGA-based SoC using Migen/LiteX, pushing the point-of-check toward users. Hardware highlights: a single-core RISC‑V CPU, 16 MB RAM, 120 MB flash, 200 dpi 336×536 B/W display, physical keyboard and a translucent PCB for visual inspection. Crypto: a Curve25519 accelerator that drops ECDH from ~100 ms to ~2.3 ms, dual TRNGs xored, AES blocks and key ROMs. Design separates trusted domain from untrusted I/O, supports self‑destruct fuses, open toolchain and a Rust OS (Zeus) in development; crowdfunding underway.
AI-written summary · about
chapters (49)
- 0:00Intro
- 1:31Software Has Evidence-Based Tools to Transfer Trust
- 1:52Hashing
- 2:10Signing
- 2:21Key Point: In Software, We can Transfer Trust, So that Place of Check is Near the User
- 5:11Removing Tamper Stickers with hot air
- 6:01Cut to the Chase: It's a Big Attack Surface
- 6:58Can Open Source Save Us?
- 7:14Problem: Place of Check Too far from Place of Use
- 7:41Open Factory Test (Trustable Factory) Is Only a Marginal Improvement
- 8:08The Big Problem: You Can't "Hash" Hardware
- 9:04Problem #1: A Building-Sized Microscope
- 9:43Problem #2: Verifying One Chip Verifies Only One Chip
- 10:33Can We Build an Evidence-Based Case To Trust Our Computers?
- 11:26Three Principles For Evidence-Based Trust in Hardware
- 12:01Problem: Complexity is Complicated
- 13:01Simplicity is Limiting
- 14:22Point of Use Verification Tradeoff: Ease of Verification vs. Features & Usability
- 15:57Why a Device, and Not a Chip?
- 16:16The Secure I/0 Problem
- 17:21Empower Verification At Multiple Levels
- 17:51Self-Provisioning and Sealing
- 18:49Sealing Can (Should?) be Physical Too
- 19:24Precursor: A Case Study in Verifiable Hardware
- 20:09Precursor: What Functions?
- 21:33Precursor: Simple in Construction
- 22:05Physical Keyboard
- 22:37Verification Difficulty: Trivial
- 23:03Touch Keyboard Verification: Very Hard
- 23:34Verifiable LCD
- 23:56Verifiable Screen
- 26:03The PCB: Designed Along Attack Surfaces
- 27:50T-Domain Attack Surfaces Illustrated
- 28:40The Hardest Problem: Evidence-Based Trust and the CPU (or SoC)
- 30:16FPGA: Narrowing the TOCTOU Gap by Compiling Your Own SoC
- 31:10FPGA Features "ASLR for Hardware": Pseudo-Random Mapping of Design to Device
- 32:03How Pseudo-Random Mapping Helps
- 33:10Closing the Loop: Prjxray
- 34:32A Look Inside the SoC
- 35:35The Core Complex
- 37:47LiteX is a Powerful SoC Integration Framework
- 39:13Debug Complex
- 40:10Memory-Mapped I/O
- 40:53Cryptography Complex
- 41:54Engine 25519
- 42:49Dual TRNG Architecture
- 43:42Precursor: Embodying the Three Princples for Evidence-Based Trust in Hardware
- 44:13Precursor: Close the Hardware TOCTOU Gap by Moving Verification Closer to the Point of Use
- 45:20Precursor Current Development Status
related talks
49mCryptech - Open Hardware Security Modules for a safer, open world
06mDay 1 Keynote
49mBuilding Open Hardware with Open Software
24mEuropean Processor Initiative
32mIntel Pathfinder for RISC-V & OpenHW CORE-V CVA6 Open-Source Core
12mIntroducing Sonata — the new open source platform for CHERIoT development
21mBuilding an Open-Source Eink Laptop
12mRISC-V Compatible Processor IP by Syntacore
