Claude review: drm/probe-helper: honour connector->force in drm_helper_probe_detect()

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

Patch Review

**Correctness — epoch counter not updated on first call (minor issue)**

When `drm_helper_probe_detect()` is called via the `ctx != NULL` path, the existing code bumps `epoch_counter` when the returned status differs from `connector->status`:

```c
if (ret != connector->status)
    connector->epoch_counter += 1;
```

The new early return skips this entirely. On the very first call (when `connector->status` is `connector_status_unknown` and force is `DRM_FORCE_OFF`), the returned `connector_status_disconnected` differs from the stored status but the epoch counter is not bumped. This means callers like `check_connector_changed()` (line 988) won't see the status transition:

```c
connector->status = drm_helper_probe_detect(connector, NULL, false);
if (old_epoch_counter == connector->epoch_counter) {
    ...
    return false;
}
```

In practice this is likely benign — the initial status update for forced connectors typically happens through `drm_helper_probe_single_connector_modes()` which sets `connector->status` directly (line 592-594) before ever reaching this path. But it would be more robust to update the epoch counter in the early-return path too, matching the existing pattern. Consider:

```c
if (connector->force) {
    int status;

    drm_dbg_kms(dev, "[CONNECTOR:%d:%s] force=%d, skipping detect\n",
                connector->base.id, connector->name,
                connector->force);
    if (connector->force == DRM_FORCE_ON ||
        connector->force == DRM_FORCE_ON_DIGITAL)
        status = connector_status_connected;
    else
        status = connector_status_disconnected;
    if (status != connector->status)
        connector->epoch_counter += 1;
    return status;
}
```

**The `ctx == NULL` path — `drm_helper_probe_detect_ctx()` also has epoch counter logic**

When `ctx` is NULL, the code falls through to `drm_helper_probe_detect_ctx()` which has its own `epoch_counter` bump at line 377-378. The new early return is placed *before* the `ctx` dispatch, so it correctly covers both paths. Good.

**Style nit — `else` after `return`**

The `else` on line 108 is unnecessary since the `if` branch returns:

```c
if (connector->force == DRM_FORCE_ON ||
    connector->force == DRM_FORCE_ON_DIGITAL)
    return connector_status_connected;
else
    return connector_status_disconnected;
```

This does match the pattern used in `drm_helper_probe_single_connector_modes()` (line 590-594) which also uses if/else, so it's consistent with existing code style. Not a blocker, but it could be simplified to just:

```c
if (connector->force == DRM_FORCE_ON ||
    connector->force == DRM_FORCE_ON_DIGITAL)
    return connector_status_connected;
return connector_status_disconnected;
```

**Debug message format — `force=%d`**

Printing the raw enum value (`force=%d`) is functional but not immediately readable in logs. The log output `force=1` means `DRM_FORCE_OFF` (since `UNSPECIFIED=0, OFF=1, ON=2, ON_DIGITAL=3`). This is fine for developers who know the enum, but a human-readable string would be friendlier. Minor nit, not a blocker.

**Overall assessment**: The patch is correct and addresses a real issue. The epoch counter concern is the only substantive point — it should either be handled or explicitly called out as intentionally skipped. With that addressed, this is a clean, well-motivated fix.

---
Generated by Claude Code Patch Reviewer