Improving stability for TCPM using boards that are not self-powered
summary
Boards that use cheap FUSB302-like controllers and have no backup power hit reboot loops because USB‑PD sources time out after ~5s and trigger a hard reset that cuts Vbus. Work includes an upstream U‑Boot port of TCPM (lands last year) to initialize PD early, kernel changes to TCPM (~8,000 LOC) that add an explicit soft reset request when source capabilities don’t arrive and print warnings before hard resets, and discussion of alternatives: TF‑A firmware handling, faster boot, or better self-powered detection.
AI-written summary · about
chapters (12)
- 0:00Introduction to TCPM
- 1:06Autonomous vs non-autonomous controllers
- 3:11Problem with FUSB302 chips
- 7:38Solutions to the 5-second issue
- 9:24U-Boot state machine implementation
- 11:34Power supply spec compliance issues
- 16:30Power delivery for alternate modes
- 17:34Hard reset risks and detection
- 20:04Spec compliance vs real-world devices
- 21:52Why chips lack firmware
- 24:47Moving state to Trusted Firmware A
- 25:50Q&A and closing
related talks
45mPowering up "discoverable bus-attached" devices on DT-based platforms
43mDebugging USB Type-C Power Delivery: Beauty and the Beast
52mMaking Pi ServerReady: time for cheap Arm infrastructure be standards-compliant
01h 01mBoosting Software Efficiency : A Case Study of 100% Performance Improvement in ...
27mPMFault: Voltage Fault Injection on Server Platforms Through the PMBus
42mBooting Automotive ECUs Really Fast with Modern Security Features
33mTizenRT: A Reliable IoT Platform
46mDevice Power Management: Journey to 5uA