Optionally implement defmt::Format

[bjoernQ]

Not a huge thing but since Debug is already implemented on these two types, having defmt::Format would be useful (at least for us and at least for TrapFrame ).

Given defmt is 1.0 this shouldn't be a problem, I guess. Let me know if you want something changed here or if you in general dislike the optional dependency.

[romancardenas]

LGTM! defmt is a very typical dependency and feature, so I guess it is OK to merge this PR. Thank you!

[romancardenas]

By the way, @bjoernQ , are you planning to use riscv-rt's trap frame? I was planning to work on a mechanism to skip it and let crates such as esp-riscv-rt define their own trap handling workflow, but if you are planning to migrate to our trap frame, then it is not necessary.

[bjoernQ]

We reduced our TrapFrame to the one you define - so we can use it.

I guess for now we would need our own trap-handler still - I am wondering how we could avoid that.

This is the draft PR on esp-hal: esp-rs/esp-hal#3857

[romancardenas]

Looks amazing!

You can use the #[riscv::core_interrupt] macro over your enumeration of interrupt sources. The macro is supposedly able to create the vector table and the default _start_INTERRUPT_trap symbols. I will take a look and find if we need to adapt something.