Rust bindings

[josephlr]

This uses 5 extern "C" entrypoint functions and a single Platform trait to bind the the reference C code.

TODO: Add instructions on publishing crate.

[crocomegan]

Overall, the changes LGTM. A few architectural thoughts:

• Since our ultimate goal is to phase out the C FFIs in favor of native Rust functions, we should avoid calling ExecuteCommand() with raw pointers directly in the long term. Instead, we should introduce a Rust dispatcher (shim wrapper). The wrapper will route commands to our Rust logic as it gets implemented while gracefully falling back to the legacy C FFIs for unimplemented paths.
• Ideally we want more Rust code in tpm-rs. I would recommend adding the Rust dispatcher and these high-level traits defined in platform.rs directly into the tpm-rs repo.
• In that case, I would also like to see more integration tests added to the tpm-rs repo instead of the tests/basic.rs.

[josephlr]

Overall, the changes LGTM. A few architectural thoughts:

• Since our ultimate goal is to phase out the C FFIs in favor of native Rust functions, we should avoid calling ExecuteCommand() with raw pointers directly in the long term. Instead, we should introduce a Rust dispatcher (shim wrapper). The wrapper will route commands to our Rust logic as it gets implemented while gracefully falling back to the legacy C FFIs for unimplemented paths.

Agreed. However, the wrapper should live in tpm2-ref so that we can keep this crate as low-level as possible. The tpm2-ref crate will not expose raw pointers.

• Ideally we want more Rust code in tpm-rs. I would recommend adding the Rust dispatcher and these high-level traits defined in platform.rs directly into the tpm-rs repo.

Agreed, moved everything into a single Platform trait, without any default impl

• In that case, I would also like to see more integration tests added to the tpm-rs repo instead of the tests/basic.rs.

Agreed, removed all tests except a linking-based test in src/tests.rs. That just verifies that everything builds and doesn't segfault.

[crocomegan]

Can you add a comment about how this is calculated. Or we can also extern this value from C (if any).