Claude review: drm/loongson: use managed KMS polling

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

Patch Review

**Analysis of the problem:**

Looking at the current `lsdc_pci_probe()` (line 296), `drm_kms_helper_poll_init(ddev)` is called, but there are three subsequent failure paths (lines 300, 309, 316) that all `return ret` without finalizing polling. The `lsdc_pci_remove()` function (line 324-330) also lacks a `drm_kms_helper_poll_fini()` call:

```c
static void lsdc_pci_remove(struct pci_dev *pdev)
{
	struct drm_device *ddev = pci_get_drvdata(pdev);

	drm_dev_unregister(ddev);
	drm_atomic_helper_shutdown(ddev);
}
```

**The fix:**

```diff
-	drm_kms_helper_poll_init(ddev);
+	drmm_kms_helper_poll_init(ddev);
```

This is the right approach. The managed function internally calls `drm_kms_helper_poll_init()` and then registers a cleanup action via `drmm_add_action_or_reset()`, so no manual `drm_kms_helper_poll_fini()` calls are needed in error paths or remove/shutdown.

**No issues found.** The function signature is `void drmm_kms_helper_poll_init(struct drm_device *dev)` — same as the non-managed version — so the one-line substitution is a complete fix.

The Fixes tag and Cc: stable are appropriate since this is a real resource leak present since the driver was introduced.

**Reviewed-by recommendation:** This patch is ready to merge.

---
Generated by Claude Code Patch Reviewer