Optimize Rc<T>::default

[slanterns]

The missing piece of #131460.

Also refactored Arc<T>::default by using a safe NonNull::from(Box::leak(_)) to replace the unnecessarily unsafe call to NonNull::new_unchecked(Box::into_raw(_)). The remaining unsafety is coming from [Rc|Arc]::from_inner, which is safe from the construction of [Rc|Arc]Inner.

[rustbot]

r? @Mark-Simulacrum

rustbot has assigned @Mark-Simulacrum.
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

[slanterns]

@rustbot author

[slanterns]

@rc_default_inplace = unnamed_addr alias ptr (), ptr @arc_default_inplace

Whoa

[slanterns]

CI green. @rustbot ready
r? @ibraheemdev since you reviewed the previous PR :)

[ibraheemdev]

cc @rust-lang/opsem regarding NonNull::from(Box::leak(..)), I'm not sure we want to introduce this pattern into std (rust-lang/unsafe-code-guidelines#316)

I'm also happy to approve this without the Box::leak change if you wish.

[slanterns]

The current implementation of Rc::new is already using such pattern:

rust/library/alloc/src/rc.rs

Lines 393 to 404 in 86d69c7

	pub fn new(value: T) -> Rc<T> {
	// There is an implicit weak pointer owned by all the strong
	// pointers, which ensures that the weak destructor never frees
	// the allocation while the strong destructor is running, even
	// if the weak pointer is stored inside the strong one.
	unsafe {
	Self::from_inner(
	Box::leak(Box::new(RcInner { strong: Cell::new(1), weak: Cell::new(1), value }))
	.into(),
	)
	}
	}

And it's also what was suggested in #47336 (comment).

[ibraheemdev]

Alright, this seems reasonable then. Thanks!

@bors r+

[bors]

📌 Commit 0a963ab has been approved by ibraheemdev

It is now in the queue for this repository.