Rollup merge of #62514 - stephaneyfx:box-ffi, r=nikomatsakis

Clarify `Box<T>` representation and its use in FFI

This officializes what was only shown as a code example in [the unsafe code guidelines](https://rust-lang.github.io/unsafe-code-guidelines/layout/function-pointers.html?highlight=box#use) and follows [the discussion](https://github.com/rust-lang/unsafe-code-guidelines/issues/157) in the corresponding repository.

It is also related to [the issue](https://github.com/rust-lang/rust/issues/52976) regarding marking `Box<T>` `#[repr(transparent)]`.

If the statement this PR adds is incorrect or a more in-depth discussion is warranted, I apologize. Should it be the case, the example in the unsafe code guidelines should be amended and some document should make it clear that it is not sound/supported.

src/liballoc/boxed.rs

@@ -61,7 +61,60 @@
 //! T` obtained from [`Box::<T>::into_raw`] may be deallocated using the
 //! [`Global`] allocator with [`Layout::for_value(&*value)`].
 //!
+//! So long as `T: Sized`, a `Box<T>` is guaranteed to be represented
+//! as a single pointer and is also ABI-compatible with C pointers
+//! (i.e. the C type `T*`). This means that if you have extern "C"
+//! Rust functions that will be called from C, you can define those
+//! Rust functions using `Box<T>` types, and use `T*` as corresponding
+//! type on the C side. As an example, consider this C header which
+//! declares functions that create and destroy some kind of `Foo`
+//! value:
 //!
+//! ```c
+//! /* C header */
+//!
+//! /* Returns ownership to the caller */
+//! struct Foo* foo_new(void);
+//!
+//! /* Takes ownership from the caller; no-op when invoked with NULL */
+//! void foo_delete(struct Foo*);
+//! ```
+//!
+//! These two functions might be implemented in Rust as follows. Here, the
+//! `struct Foo*` type from C is translated to `Box<Foo>`, which captures
+//! the ownership constraints. Note also that the nullable argument to
+//! `foo_delete` is represented in Rust as `Option<Box<Foo>>`, since `Box<Foo>`
+//! cannot be null.
+//!
+//! ```
+//! #[repr(C)]
+//! pub struct Foo;
+//!
+//! #[no_mangle]
+//! pub extern "C" fn foo_new() -> Box<Foo> {
+//!     Box::new(Foo)
+//! }
+//!
+//! #[no_mangle]
+//! pub extern "C" fn foo_delete(_: Option<Box<Foo>>) {}
+//! ```
+//!
+//! Even though `Box<T>` has the same representation and C ABI as a C pointer,
+//! this does not mean that you can convert an arbitrary `T*` into a `Box<T>`
+//! and expect things to work. `Box<T>` values will always be fully aligned,
+//! non-null pointers. Moreover, the destructor for `Box<T>` will attempt to
+//! free the value with the global allocator. In general, the best practice
+//! is to only use `Box<T>` for pointers that originated from the global
+//! allocator.
+//!
+//! **Important.** At least at present, you should avoid using
+//! `Box<T>` types for functions that are defined in C but invoked
+//! from Rust. In those cases, you should directly mirror the C types
+//! as closely as possible. Using types like `Box<T>` where the C
+//! definition is just using `T*` can lead to undefined behavior, as
+//! described in [rust-lang/unsafe-code-guidelines#198][ucg#198].
+//!
+//! [ucg#198]: https://github.com/rust-lang/unsafe-code-guidelines/issues/198
 //! [dereferencing]: ../../std/ops/trait.Deref.html
 //! [`Box`]: struct.Box.html
 //! [`Box<T>`]: struct.Box.html