Attendees

Agenda

Meeting overview

• Recording: We’ll record the meeting, if you object, please let us now. Link and password will be added to this page (see sections further down).

• Goal with today’s meeting: Better understand the gaps in developing for Arm using Rust and whether Linaro, its members and community can and should invest in this.

Linux kernel's use of Rust

• The kernel makes use of and depends on unstable features. Do we know what those features are? Can we name something?

  • Rust for Linux - Miguel Ojeda, will share his view.

    • The PR for tracking unstable features.

• Also, creating connections for certain subsystems like clk, regulator, pinmux, gpio, etc., which are important for supporting System-on-Chip (SoC) devices. This might be more important in the future.

  • Perhaps that partly would address the ONCD report?
    Press Release: Future Software Should Be Memory Safe | ONCD | The White House

Tier support

Tier support is generally pretty good, but it wouldn't hurt to review and also to see whether potential Arm Tier-3 should be moved up to Tier-2. Any direct suggestions on what to propose? The matrices shown here https://rust-lang.github.io/rustup-components-history seems to be helpful, but Tier-3 is mostly lacking information.

https://rust-lang.github.io/rustup-components-history/

Bare-metal drivers and HAL

It's difficult to re-use drivers written for bare-metal systems and adding support for HAL interfaces is going too slow (SoC vendors typically not involved).

Files, presentations and recording

• Recording

  • https://linaro-org.zoom.us/rec/share/R6cRyV95rcxzZk5aKg2Hhn4GdUuNnLBoE-cfqeh2G2Gw16_pVWyIm_ek5nbGA-e-.BHdEt06Ea-KpI7jO

    Passcode: DfW0#zEg

Notes

• In general, please see recording.

• We were running out of time to discuss the Tier support and the bare-metal drivers/HAL etc. Will need to follow up.

• Summary from Rust-for-Linux / Miguel:

  • Rust toolchain people need to know about specific compiler flag that we use and need.

  • There is a bit of a gap of Arm presence when it comes to the toolchain side of Rust.

  • Rust-for-Linux people, within limits and availability offers support to help Linux kernel maintainer interested in rewriting their subsystems using Rust.

  • We need justifications and users to be able to justify rewriting some parts.

  • CI is important for the Rust support, currently there are testing done via TuxSuite (Linaro) and KernelCI.

Chat

• Please see recording.