Week 0025

gpt-disk-rs

A developer contacted me to suggest changing the license from just Apache-2.0 to Apache-2.0 or MIT. This improves compatibility with GPLv2 and is in alignment with much of the Rust ecosystem. That change has now been made and new versions of the crates have been released. (Took me a few tries to get the license README links correct, so ended up with a few extra dot releases of uguid, but should be all good now.)

Spent some time testing and reviewing a PR that adds support for a couple more disk protocols. The nonblocking stuff was a bit tricky.
Finally put up a PR for open_protocol_exclusive, which I think will provide about the best interface possible for safely opening UEFI protocols given the limitations of the UEFI design. (I think ideally UEFI would have a reference count on open protocols, rather than just exclusive vs free-for-all, but that's outside the scope of what I think we can accomplish in uefi-rs.)
Put up a PR to remove use of the unstable try_trait_v2 feature. This has a small impact on the API, but I think removing it is the right thing to do to avoid potential future breakage on nightly due to the Try API still being under discussion.

Opened another compiler_builtins PR to fix some more UEFI intrinsics. For some reason the ABI intrinsics that convert from i128/u128 to f32/f64 are different depending on whether softfloat is enabled.
Opened a Rust PR to update the cc crate to latest. That fixes some issues with compiling compiler_builtins for the UEFI targets.
Opened a Rust PR to add documentation for the no-std target build option. Since https://github.com/rust-lang/rust/pull/99765 was merged this option isn't needed for UEFI anymore, but can hopefully help future people trying to build for new targets that don't support std (until such time as https://github.com/rust-lang/rust/pull/99765 is fixed).
Opened a Rust PR to update the compiler_builtins crate to latest. This will fix some (but not all) UEFI intrinsic ABI issues.