Claude review: drm/bridge: stm_lvds: Do not fail atomic_check on disabled connector

[Claude Code Review Bot <claude-review@example.com>]

Patch Review

**Correctness: Good.**

The patch does two things:
1. Changes the `!conn_state->crtc` check from returning `-EINVAL` to returning `0`.
2. Moves the check *above* the `list_empty(&connector->modes)` check.

Both changes are correct:

- **Returning 0 instead of -EINVAL**: A NULL `crtc` in the new connector state simply means the connector is being disabled (or is already disabled). This is not an error. Other drivers in the tree follow the same pattern — e.g., `analogix_dp_core.c:906`, `vmwgfx_stdu.c:895`, `vkms_writeback.c:47` all do `if (!conn_state->crtc) return 0;`.

- **Moving the check earlier**: This is the right ordering. When the connector is disabled, there is no point validating modes or checking crtc state — none of that applies. Early-exit avoids a spurious "empty modes list" debug message or other irrelevant validation on a disabled connector.

Looking at the post-patch function (lines 896–923), the resulting flow is clean:

```c
conn_state = drm_atomic_get_new_connector_state(state, connector);
if (!conn_state)
    return -EINVAL;

if (!conn_state->crtc)        /* disabled connector — nothing to validate */
    return 0;

if (list_empty(&connector->modes)) {  /* only relevant when enabled */
    ...
}
```

The subsequent code at lines 912–921 dereferences `conn_state->crtc` (via `drm_atomic_get_crtc_state`), so the NULL check *must* precede it — which it did before and still does after the patch. No null-deref risk.

**Commit message**: Clear and well-written. The Fixes tag correctly references the original commit that introduced the driver. The explanation of the suspend/resume failure scenario is helpful.

**No issues found.**

---
Generated by Claude Code Patch Reviewer