Teeny Tiny patch

[Nipfswd]

No description provided.

[ReaQwQ]

Thanks for the patch! This is a helpful developer-experience cleanup, especially adding rust-toolchain.toml and making run.sh handle target setup more explicitly.

I found one lint issue that needs to be fixed before merging, plus two small cleanup/consistency notes. Nothing here looks conceptually wrong; the main issue is that Rust recently changed the CPUID intrinsic safety contract for x86_64/i686-style targets, so the added unsafe block now trips this repo's warning-deny lint.

Validation I ran locally:

• cargo fmt --check: passed
• cargo check --target x86_64-unknown-uefi: passed, with an unused_unsafe warning
• cargo clippy --all-targets --all-features -- -D warnings: failed on the unused unsafe block
• just lint: failed at the same unused unsafe block
• just architecture-boundaries: passed
• userland clippy for x86_64-unknown-none: passed
• bash -n run.sh: passed

One small consistency request: could we also mirror the new target setup in run.bat so Windows users get the same setup path as run.sh?

The top-level kernel build runs build.rs, which builds userland demos for x86_64-unknown-none, so Windows users can still hit the manual “install this target” failure path even though run.sh now handles it. Keeping x86_64-unknown-none in rust-toolchain.toml looks right; this is mainly about keeping run.sh and run.bat consistent.

[ReaQwQ]

Could you remove this unsafe block?

I see why it was added, since __cpuid used to be an unsafe intrinsic. Recent stdarch changed that for x86_64/i686-style targets, though, so with the toolchain selected by this PR, core::arch::x86_64::__cpuid(1) is safe and this currently fails clippy as unused_unsafe.

Removing the unsafe { ... } wrapper and the safety comment should fix the lint failure.

Reference: rust-lang/stdarch#1935

[ReaQwQ]

Tiny cleanup: this line seems to include a hidden private-use character after nul (EF 80 8D in UTF-8). Could you replace it with plain nul, or remove the line if it was accidental?

While touching this file, adding a final newline would also keep the diff cleaner.

[ReaQwQ]

Question/nit: since channel = "nightly" isn't date-pinned, this toolchain could
silently shift over time — today's unused_unsafe situation with __cpuid is
actually a good example of how a nightly-only safety-contract change can flip a
build from passing to failing between two people's local runs (or between CI runs
on different days).

Would it make sense to pin this to a dated nightly (e.g. nightly-2026-06-15) for
reproducibility? Not blocking this PR, just flagging for discussion.