Make #[naked] an unsafe attribute

[folkertdev]

tracking issue: #138997
blocked on #139797

Per #134213 (comment), the #[naked] attribute is now an unsafe attribute (in any edition).

This can only be merged when the above PRs are merged, I'd just like to see if there are any CI surprises here, and maybe there is early review feedback too.

r? @traviscross

[folkertdev]

Hmm, how should this this error be resolved?

error: unsafe attribute used without unsafe
##[error]   --> /cargo/registry/src/index.crates.io-1949cf8c6b5b557f/compiler_builtins-0.1.153/src/macros.rs:436:15
    |
436 |               #[naked]
    |                 ^^^^^ usage of unsafe attribute
    |
   ::: /cargo/registry/src/index.crates.io-1949cf8c6b5b557f/compiler_builtins-0.1.153/src/x86_64.rs:10:1
    |
10  | / intrinsics! {
11  | |     #[naked]
12  | |     #[cfg(all(
13  | |         any(
...   |
42  | | }
    | |_- in this macro invocation
    |
    = note: this error originates in the macro `intrinsics` (in Nightly builds, run with -Z macro-backtrace for more info)
help: wrap the attribute in `unsafe(...)`
    |
436 |             #[unsafe(naked)]
    |               +++++++     +

It's kind of a cyclic dependency: I can't update the compiler without updating compiler-builtins, but I can't update compiler builtins either without an updated compiler. What is the right approach here?

[traviscross]

Does #[cfg_attr(bootstrap, ..)] (docs) work here?

[bors]

☔ The latest upstream changes (presumably #139766) made this pull request unmergeable. Please resolve the merge conflicts.

[folkertdev]

I thought that would not work because of the error that was emitted earlier (something with building miri), but it actually does appear to work (or at least, we get a different error now).

@bjorn3 do you see a way to work around the latest error? apparently a compiler where cfg(bootstrap) is true is used to run the mini_core_hello_world cranelift test?

[bjorn3]

That shouldn't happen. cg_clif should be tested using a stage1 rustc.

[rust-log-analyzer]

The job x86_64-gnu-tools failed! Check out the build log: (web) (plain)

Click to see the possible cause of the failure (guessed by this bot)

    |
    = note: this error originates in the macro `intrinsics` (in Nightly builds, run with -Z macro-backtrace for more info)
help: wrap the attribute in `unsafe(...)`
    |
436 |             #[unsafe(naked)]
    |               +++++++     +

[RUSTC-TIMING] compiler_builtins test:false 1.759
error: could not compile `compiler_builtins` (lib) due to 1 previous error
fatal error: failed to build sysroot: sysroot build failed
Command has failed. Rerun with -v to see more details.
Build completed unsuccessfully in 0:00:28
  local time: Mon Apr 14 11:54:07 UTC 2025
  network time: Mon, 14 Apr 2025 11:54:07 GMT
##[error]Process completed with exit code 1.
Post job cleanup.

[folkertdev]

well, the #[cfg(bootstrap)] thing does not work after all. So I think the only viable solution is to allow but not require unsafe(naked) for a brief period, so that compiler-builtins can be updated.

[bors]

☔ The latest upstream changes (presumably #139826) made this pull request unmergeable. Please resolve the merge conflicts.