Claude review: rust: sync: Introduce LazyInit

[Claude Code Review Bot <claude-review@example.com>]

Overall Series Review

Subject: rust: sync: Introduce LazyInit
Author: Lyude Paul <lyude@redhat.com>
Patches: 6
Reviewed: 2026-06-04T16:11:03.339061

---

This is a well-structured 2-patch series from Lyude Paul introducing `LazyInit<T>`, a thread-safe lazy-initialization container for the kernel's Rust abstractions. The motivation is clear: the existing `SetOnce` cannot handle fallible initializers and doesn't allow racing callers to block until initialization completes. `LazyInit` solves both by using a `Mutex` for synchronization.

**Patch 1** adds a small utility method `Lock::get_mut_pinned()` that Patch 2 depends on. **Patch 2** is the bulk of the work — a new `LazyInit<T>` type with Mutex-based synchronization, fallible `PinInit` support, `reset()` capability, and thorough kunit tests.

The design is sound: the Mutex protects only initialization, while post-init access is lock-free (relying on `T: Send + Sync`). The lifetime extension pattern via `mem::transmute` in `data()` is the riskiest construct but is correctly justified. The `reset()` implementation using `ptr::drop_in_place` on `Inner` is subtle but correct given that `PinnedDrop` leaves `Inner` in a well-defined uninitialized state and the fields have no drop glue of their own.

There are a few documentation bugs and minor issues detailed below, but no correctness issues found.

---

---
Generated by Claude Code Patch Reviewer