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: driver core: remove drvdata() and driver_type
Date: Tue, 28 Apr 2026 13:51:40 +1000
Message-ID: <review-patch2-20260427221002.2143861-3-dakr@kernel.org>
In-Reply-To: <20260427221002.2143861-3-dakr@kernel.org>
References: <20260427221002.2143861-1-dakr@kernel.org> <20260427221002.2143861-3-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

This is a clean removal patch. All the pieces fit together:

**C-side removal** (`drivers/base/base.h`):

The `struct driver_type` and the `driver_type` field in `struct device_priv=
ate` are cleanly removed behind `#ifdef CONFIG_RUST`. This shrinks `struct =
device_private` by 16 bytes for Rust-enabled kernels.

**Rust-side removal** (`rust/kernel/device.rs`):

Removed items:
- `set_type_id()` =E2=80=94 was called from `set_drvdata()`
- `match_type_id()` =E2=80=94 was called from `drvdata()`
- `drvdata()` =E2=80=94 the public type-checked accessor

Retained items (correctly):
- `set_drvdata()` =E2=80=94 still used internally by bus `probe_callback`s
- `drvdata_borrow()` / `drvdata_unchecked()` =E2=80=94 still used internall=
y by bus `remove_callback`s
- `drvdata_obtain()` =E2=80=94 still used for taking ownership during unbind

The `static_assert!` comparing `size_of::<bindings::driver_type>()` to `siz=
e_of::<TypeId>()` is correctly removed since `struct driver_type` no longer=
 exists.

`set_drvdata` no longer calls `set_type_id`:

```rust
pub fn set_drvdata<T: 'static>(&self, data: impl PinInit<T, Error>) -> Resu=
lt {
    let data =3D KBox::pin_init(data, GFP_KERNEL)?;
    unsafe { bindings::dev_set_drvdata(self.as_raw(), data.into_foreign().c=
ast()) };
    Ok(())
}
```

This is safe because the remaining callers of `drvdata_borrow`/`drvdata_unc=
hecked` are all `unsafe` internal methods where the type `T` is known stati=
cally by the bus abstraction code =E2=80=94 no runtime type check is needed.

**Summary**: Both patches are correct. The series achieves a genuine simpli=
fication: lifetime constraints that were previously upheld by convention (d=
rvdata must outlive all callers) are now enforced structurally (registratio=
n data is owned by `Registration`, and `Registration::drop` unbinds the chi=
ld before freeing the data).

---
Generated by Claude Code Patch Reviewer