Tock OT Notes 2/14/2020

Present:

• Laura Abbott, Oxide Computer
• Brad Campbell (Chair), University of Virginia
• Jon Flatley, Google
• Alistair Francis, Western Digital
• Garret Kelly, Google
• Philip Levis, Stanford University/Google
• Patrick Mooney, Oxide Computer

TODO

• @alistair23: Document PMP testing strategy on QEMU
• @bradjc: integrate elf2tab header offset into libtock-c/riscv
• All: review OpenTitan peripheral design docs
  • Provide comments to OT HW team

Summary of Goal Brainstorming

• Setting up software validation and CI
  • Extending the TravisCI setup we have for Tock already?
• Continuing to push bug reports to OT HW developers
• How to leverage Device Interface Functions (DIFs) to develop Tock peripheral drivers
  • Want to use DIFs initially and switch to full Rust versions later
  • Should there be an API inside a peripheral driver to allow easy switching between DIF (in C) and a Rust implementation?
  • Need a structure for using existing Tock crates with DIF-based drivers
• Managing relocations for apps
  • NxK approach workable, but far from desirable
  • LLVM PIC a better option
  • Still unclear what the developer effort would be
  • Need to follow up with @jrvanwhy

Discussion

• Should we keep arty-e21 platform?
  • Useful as the only M/U mode core
  • Once we have a suitable replacement platform likely can remove arty-e21
  • Might be useful to keep CLIC driver
• Work coordination
  • Weekly call to keep track of generally what people are working on
  • OK to have multiple implementations if conflicts happen