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: pci: return DevresLt from Bar::into_devres()
Date: Thu, 04 Jun 2026 12:08:40 +1000
Message-ID: <review-patch6-20260603011020.2073650-7-dakr@kernel.org>
In-Reply-To: <20260603011020.2073650-7-dakr@kernel.org>
References: <20260603011020.2073650-1-dakr@kernel.org> <20260603011020.2073650-7-dakr@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

The manual `ForLt` and `CovariantForLt` impls on `Bar<'static, SIZE>`:

```rust
impl<const SIZE: usize> ForLt for Bar<'static, SIZE> {
    type Of<'a> =3D Bar<'a, SIZE>;
}

// SAFETY: `Bar<'a, SIZE>` is covariant over `'a`; it holds `&'a Device<Bou=
nd>`,
// which is covariant.
unsafe impl<const SIZE: usize> CovariantForLt for Bar<'static, SIZE> {}
```

This bypasses the macro's compile-time covariance proof, relying instead on=
 a manual safety argument. For `Bar`, which holds only `&'a Device<Bound>` =
(covariant) and `PhantomData`/raw pointers/scalars, the covariance claim is=
 clearly correct.

The `into_devres()` simplification is a nice improvement =E2=80=94 the expl=
icit `transmute` is replaced by `DevresLt::new()` which handles it internal=
ly:

```rust
pub fn into_devres(self) -> Result<DevresBar<SIZE>> {
    let pdev =3D self.pdev;
    unsafe { DevresLt::new(pdev.as_ref(), self) }
}
```

**Minor note on the diff context:**

```rust
    types::{
        CovariantForLt,
        ForLt, //
    }, //
```

The double `//` trailing comments look like formatting artifacts from the m=
acro expansion or tooling. Cosmetic only.

---
Generated by Claude Code Patch Reviewer