Tracking Issue for `cfg_target_has_atomic_equal_alignment`

[Amanieu]

This is a tracking issue for the target_has_atomic_equal_alignment CFG flag which was added in #76965 for Atomic*::from_mut.

It indicates whether an AtomicX has the same alignment as the equivalent X integer type. On some 32-bit platforms (e.g. x86) AtomicI64 has an alignment of 8 while i64 is only aligned to 4 bytes.

---

Potential design concerns:

• Open question of whether a dedicated cfg is necessary over cfg(accessible) on the from_mut method specifically, which is what is currently gated on this cfg

[WaffleLapkin]

T-lang nomination

I would like to suggest stabilizing this feature.

It is currently blocking stabilization of atomic_from_mut feature (Atomic*::{from_mut, from_mut_slice, get_mut_slice}), as from_mut and from_mut_slice use this cfg and users would need to gate their uses on the same cfg.

Open question of whether a dedicated cfg is necessary over cfg(accessible) on the from_mut method specifically, which is what is currently gated on this cfg

It appears cfg(accessible) is still nowhere near to being stable and I don't see the reason to block this on it.

[joshtriplett]

@WaffleLapkin We discussed this in today's @rust-lang/lang meeting.

Please do post a stabilization PR for this, and we'll FCP it.