Fix c_char on AIX

[taiki-e]

Closes #122985

[rustbot]

r? @scottmcm

rustbot has assigned @scottmcm.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[bjorn3]

cc @ecnelises and @bzEq as target maintainers

[scottmcm]

Re-rolling to hopefully get someone with target experience:
r? libs

[Mark-Simulacrum]

r? @Mark-Simulacrum

@rustbot author (while libc is merged)

Per rust-lang/libc#3631 (comment), it looks like target maintainers are happy with this change, and our policy is that no further review for tier-3 targets is required.

[taiki-e]

@rustbot ready

#124560 which includes libc side change has been merged.

[riking]

You still need to remove the Github draft status 🙂

[ecnelises]

Thanks for taking care of this! I did not notice libc has been bumped.

[quadruple-output]

It might be obvious, but I would like to note that there is an inconsistency, resulting in syntax errors on AIX while the libc side (#124560) has already been merged, but this PR is still open.
Type conflicts can arise where core::ffi meets libc. This is the case at least for crates nix and rustix. The following sequence of commands results in compile errors when executed in an empty directory:

cargo init
cargo add rustix
cargo check

(tested with rustix-0.38.34 on AIX)

  --> ~/.cargo/registry/src/index.crates.io-d11c229612889eed/rustix-0.38.34/src/backend/libc/conv.rs:16:5
   |
15 | pub(super) fn c_str(c: &CStr) -> *const c::c_char {
   |                                  ---------------- expected `*const u8` because of return type
16 |     c.as_ptr()
   |     ^^^^^^^^^^ expected `*const u8`, found `*const i8`
   |
   = note: expected raw pointer `*const u8`
              found raw pointer `*const i8`

[Mark-Simulacrum]

@bors r+ rollup

[bors]

📌 Commit c31ec4f has been approved by Mark-Simulacrum

It is now in the queue for this repository.