Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [High] drm/bridge_connector: unconditional detect() after hpd_notify()
  breaks msm_dp
- [Medium] drm/bridge_connector: suppressed IRQ_HPD uevent on downstream
  EDID change
--

commit 639a7eb6e945d13bc1ca0a8e003c6cf0a496e228
Author: Yongxing Mou <[email protected]>

drm/bridge_connector: suppress hotplug for IRQ_HPD without status changes

This commit calls the bridge .detect() callback after HPD notification to
update connector->status and skips sending a hotplug event on IRQ_HPD when
the connector status has not changed.

> diff --git a/drivers/gpu/drm/display/drm_bridge_connector.c 
> b/drivers/gpu/drm/display/drm_bridge_connector.c
> index 4ade63306653f..1d4502126bff3 100644
> --- a/drivers/gpu/drm/display/drm_bridge_connector.c
> +++ b/drivers/gpu/drm/display/drm_bridge_connector.c
> @@ -160,15 +160,25 @@ static void drm_bridge_connector_handle_hpd(struct 
> drm_bridge_connector *drm_bri
>                                           enum drm_connector_dp_event event)
>  {
>       struct drm_connector *connector = &drm_bridge_connector->base;
> +     struct drm_bridge *detect = drm_bridge_connector->bridge_detect;
>       struct drm_device *dev = connector->dev;
> +     enum drm_connector_status old_status, new_status;
> +     bool level_hpd;
> +     bool changed;
>  
> -     mutex_lock(&dev->mode_config.mutex);
> -     connector->status = status;
> -     mutex_unlock(&dev->mode_config.mutex);
> +     level_hpd = event == DRM_CONNECTOR_NO_EXTRA_STATUS;
>  
>       drm_bridge_connector_hpd_notify(connector, status, event);
>  
> -     drm_kms_helper_connector_hotplug_event(connector);
> +     mutex_lock(&dev->mode_config.mutex);
> +     old_status = connector->status;
> +     new_status = detect ? detect->funcs->detect(detect, connector) : status;

[Severity: High]
Does calling detect->funcs->detect() unconditionally after
drm_bridge_connector_hpd_notify() leak a runtime PM reference on
dp->pdev->dev in msm_dp on every connect and IRQ_HPD event?

On plug and IRQ_HPD notifications, msm_dp_bridge_hpd_notify() ->
msm_dp_hpd_plug_handle() acquires a runtime PM reference and sets
dp->plugged = true.  Calling msm_dp_bridge_detect() right after acquires a
second reference and skips releasing it at exit because priv->plugged is
true:

drivers/gpu/drm/msm/dp/dp_display.c:msm_dp_bridge_detect() {
        ...
        ret = pm_runtime_resume_and_get(&dp->pdev->dev);
        ...
        priv->plugged = true;
        ...
end:
        if (!priv->plugged) {
                if (phy_deinit) {
                        msm_dp_aux_enable_xfers(priv->aux, false);
                        msm_dp_display_host_phy_exit(priv);
                }

                pm_runtime_put_sync(&dp->pdev->dev);
        }
}

msm_dp_hpd_unplug_handle() only drops one reference when clearing
dp->plugged, so the extra reference is never released and the DP controller
cannot runtime suspend.  If a DPCD read in msm_dp_bridge_detect() fails
instead, it clears priv->plugged = false while releasing only its own
reference, causing a later msm_dp_hpd_unplug_handle() to return early and
skip unplug cleanup and the initial reference release.

On disconnect notifications, running msm_dp_bridge_detect() after
msm_dp_hpd_unplug_handle() also re-enables AUX transfers or re-initializes
the PHY after unplug, and bridges whose .detect() callback does not sense
HPD (such as lt9611uxc_bridge_detect() when hpd_supported is false) will
overwrite connector_status_disconnected with connector_status_connected.

> +     connector->status = new_status;
> +     changed = new_status != old_status;
> +     mutex_unlock(&dev->mode_config.mutex);
> +
> +     if (changed || level_hpd)
> +             drm_kms_helper_connector_hotplug_event(connector);
>  }

[Severity: Medium]
Can checking only new_status != old_status suppress hotplug events when a
downstream DP branch device changes EDID over IRQ_HPD while remaining
connected?

When a branch device signals DP_DOWNSTREAM_PORT_STATUS_CHANGED while
connector->status is already connector_status_connected,
msm_dp_irq_hpd_handle() -> msm_dp_display_process_hpd_high() reads the new
EDID and calls drm_edid_connector_update(), which increments
connector->epoch_counter.  Because new_status and old_status are both
connector_status_connected and level_hpd is false, changed is false and
userspace never receives the hotplug notification.

When new_status != old_status, updating connector->status directly here
without incrementing connector->epoch_counter also causes subsequent
drm_helper_probe_detect() calls to see no status change and skip bumping
connector->epoch_counter.

-- 
Sashiko AI review ยท 
https://sashiko.dev/#/patchset/[email protected]?part=19

Reply via email to