The comment about the `weak` field in `ArcArcInner` is imprecise

[xmh0511]

#[repr(C)]
struct ArcInner<T: ?Sized> {
    strong: Atomic<usize>,

    // the value usize::MAX acts as a sentinel for temporarily "locking" the
    // ability to upgrade weak pointers or downgrade strong ones; this is used
    // to avoid races in `make_mut` and `get_mut`.
    weak: Atomic<usize>,

    data: T,
}

By examining the implementation of Weak::upgrade, it becomes apparent that the operation does not verify whether the weak is locked or not. It just verifies whether the strong is zero or not. So, "locking the ability to upgrade weak pointers" sounds like the operation cannot succeed if weak is locked. IIUC, the actual intent should be that the success of the locking indicates that no other weak pointers exist, so it's impossible that there is a potential concurrent upgrade operation.

The suggested comment might be:

// the value usize::MAX acts as a sentinel for temporarily "locking" the
// weak count. This prevents `Arc::downgrade` from creating new Weak pointers
// while a check for uniqueness is in progress.
//
// This lock is essential for the correctness of `Arc::get_mut` and `Arc::is_unique`.
// By successfully acquiring the lock (which is only possible if the weak count is 1),
// we prove that no other `Weak` pointers exist at that moment. This guarantees
// that no concurrent `Weak::upgrade` can occur to violate the uniqueness of the
// mutable reference we are about to create.