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: make Driver trait lifetime-parameterized
Date: Thu, 07 May 2026 13:02:04 +1000
Message-ID: <review-patch10-20260506215113.851360-11-dakr@kernel.org>
In-Reply-To: <20260506215113.851360-11-dakr@kernel.org>
References: <20260506215113.851360-1-dakr@kernel.org> <20260506215113.851360-11-dakr@kernel.org>
X-Mailer: Claude Code Patch Reviewer
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit
MIME-Version: 1.0

Patch Review

This is where the ergonomic payoff starts. The `pci::Driver` trait becomes:
```rust
pub trait Driver<'bound>: Send {
    fn probe(
        pdev: &'bound Device<Core>,
        info: &'bound Self::IdInfo,
    ) -> impl PinInit<Self, Error> + 'bound;
    fn unbind(pdev: &'bound Device<Core>, this: Pin<&'bound Self>) { ... }
}
```

The `Adapter` is parameterized over `F: ForLt` with the HRTB bound `for<'bound> F::Of<'bound>: Driver<'bound>`. The `module_pci_driver!` macro wraps the type: `ForLt!($type)`.

Well structured. The HRTB on `Adapter` is the key mechanism: it says "for any lifetime `'bound`, the instantiated type implements `Driver<'bound>`", which is what lets the framework pick the right lifetime at probe time.

No issues found.

### PATCHES 11-16: make Driver lifetime-parameterized (platform, auxiliary, auxiliary Registration, usb, i2c)

All follow the same pattern as patch 10 for their respective bus types. The auxiliary `Registration<F: ForLt>` generalization (patch 13) is notable -- it extends the lifetime-bound pattern beyond driver private data to registration data, which is a good demonstration of the generality.

The sample in patch 14 showcases this well, storing a `&'bound pci::Device<Bound>` in auxiliary registration data.

No issues found.

---
Generated by Claude Code Patch Reviewer