Attendees

• Alexandru Radovici
• Alyssa Haroldsen
• Amit Levy
• Brad Campbell
• Branden Ghena
• Jett Rink
• Johnathan Van Why
• Leon Schuermann
• Pat Pannuto
• Vadim Sukhomlinov

Updates

• Pat: Hudson is presenting the threat model paper at EuroSys, correct?
• Amit: He already presented, is flying back.
• Pat: I'd like an update on how that was received when possible.
• Amit: I can relay what he said to me. It went pretty well, I think. EuroSys is not a super embedded-familiar audience, so had to answer a few questions about why we don't use ASLR etc.. We also got some good questions from people who thought the challenge of having to trust liveness is interesting. Also got some questions after the talk about how we compare to Redleaf and whether Redleaf's fault-recovery mechanisms could work for our purposes. Some questions about whether there is interest in developing some public collection of safe-Rust-only crates with other systems like Redleaf and Theseus and working with them on requiring safe-Rust-only app code, which is an interesting idea.
• Amit: Are people familiar with Redleaf and Theseus? These are two research operating systems for not-microcontrollers also requiring Rust.
• Alyssa: I have not heard of them.
• Branden: [Linked https://mars-research.github.io/projects/redleaf/ in chat.]
• Amit: Expect to hear from Hudson next week. I think generally the topic of an ecosystem of safe-only core crates is a pretty interesting one. There are questions on if systems are similar enough to rely on that. My very limited experience on looking into that is that the differences between Tock and other things are not just in relying on unsafe, but also in not doing general dynamic allocation in the kernel. Probably a lot of things that would be reasonable dependencies would look pretty different if you can or cannot use the heap. Even if that is the case, maybe there is a set of related stuff that is relevant to the Rust userspace that has similar safety requirements, generally, but is allowed to do heap allocation and looks similar to what the other Rust kernels are doing.

[Editor's note: The meeting agenda was empty, and there was no topical discussion]