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: drm/connector: check if destroy function exist
Date: Mon, 25 May 2026 20:21:15 +1000
Message-ID: <review-patch1-20260521103329.140064-1-andreas@kemnade.info>
In-Reply-To: <20260521103329.140064-1-andreas@kemnade.info>
References: <20260521103329.140064-1-andreas@kemnade.info> <20260521103329.140064-1-andreas@kemnade.info>
X-Mailer: Claude Code Patch Reviewer
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0

Patch Review

**Correctness: Good.**

The `drmm_connector_init()` function at line 525 of `drm_connector.c` expli=
citly documents and enforces that `.destroy` must be NULL:

```c
/* line 520 */ * The @drm_connector_funcs.destroy hook must be NULL.
...
/* line 533 */ if (drm_WARN_ON(dev, funcs && funcs->destroy))
                   return -EINVAL;
```

This means any connector using `drmm_connector_init()` (like `drm_bridge_co=
nnector` at `drm_bridge_connector.c:993`) will have `funcs->destroy =3D=3D =
NULL`. Cleanup for these connectors is handled via `drmm_add_action_or_rese=
t()` calling `drm_connector_cleanup_action()` (line 540), not through `.des=
troy`.

The two call sites fixed are:

1. **`drm_connector_free()` (line 194)** =E2=80=94 the `kref` put callback:
```c
if (connector->funcs->destroy)
    connector->funcs->destroy(connector);
```

2. **`drm_connector_free_work_fn()` (line 206)** =E2=80=94 the deferred-fre=
e worker:
```c
if (connector->funcs->destroy)
    connector->funcs->destroy(connector);
```

Both are correct. When `.destroy` is NULL, there is nothing to call =E2=80=
=94 `drmm` handles the cleanup separately through the managed action.

**Minor observations:**

- The commit message subject has a minor grammatical issue: "check if destr=
oy function exist" should be "check if destroy function exists". Trivial, n=
ot blocking.
- It might be worth also checking `connector->funcs` itself for NULL, but i=
n practice `funcs` is always set during `drm_connector_init_only()` and sho=
uld never be NULL at this point. The current check is sufficient.
- No concern about memory leaks: when `.destroy` is NULL (the `drmm` path),=
 cleanup is already registered via `drmm_add_action_or_reset` calling `drm_=
connector_cleanup()`, and the memory was allocated with `drmm_kzalloc()` wh=
ich is also automatically freed. Skipping `.destroy` here is the correct be=
havior =E2=80=94 calling it would have been a double-free.

**Verdict: Correct, minimal, appropriate fix. The `Fixes:` tag is accurate.=
**

---
Generated by Claude Code Patch Reviewer