From mboxrd@z Thu Jan  1 00:00:00 1970
From: Claude Code Review Bot <claude-review@example.com>
To: dri-devel-reviews@example.com
Subject: Claude review: rust: rcu: add RcuBox type
Date: Thu, 04 Jun 2026 15:39:36 +1000
Message-ID: <review-patch2-20260530143541.229628-4-phasta@kernel.org>
In-Reply-To: <20260530143541.229628-4-phasta@kernel.org>
References: <20260530143541.229628-2-phasta@kernel.org> <20260530143541.229628-4-phasta@kernel.org>
X-Mailer: Claude Code Patch Reviewer
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0

Patch Review

**Author: Alice Ryhl**

Well-designed container. The `needs_drop::<T>()` branching in `Drop` is a n=
ice optimization =E2=80=94 using `kvfree_call_rcu` for trivially-droppable =
types avoids an unnecessary callback indirection. The `ForeignOwnableRcu` t=
rait extension is cleanly separated.

A few observations:

1. **`Send + Sync` bound on `Send`/`Sync` impls**: The bounds require `T: S=
end + Sync` for `RcuBox<T>: Send`. The `Send` requirement comes from the st=
ruct definition (`RcuBox<T: Send>`), and `Sync` is needed because the previ=
ous owner may still read `&T` during an RCU grace period after transferring=
 ownership. The safety comment explains this well.

2. **Trailing `//` comments** again appear in the `use` block:
```rust
use crate::{
    bindings,
    types::{
        ForeignOwnable,
        NotThreadSafe, //
    }, //
};
```

3. **The `rcu_borrow` safety contract** could be more precise about what "p=
lus one rcu grace period" means in practice:
```rust
/// * If [`from_foreign`] is called, then `'a` must not end after the call =
to `from_foreign`
///   plus one rcu grace period.
```
This is correct but subtle =E2=80=94 it means callers must hold the RCU rea=
d lock to ensure the grace period hasn't elapsed.

No blocking issues.

---

---
Generated by Claude Code Patch Reviewer