Re: [PATCH v7 28/34] drm/bridge_connector: Implement hw readout for connector

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] drm/bridge_connector: unconditional SRO hooks with no-op readout when unsupported
--

commit b75d5f60d55bc26fbc8505149b2ac1fa7543a9a1
Author: Maxime Ripard <mripard@kernel.org>

drm/bridge_connector: Implement hw readout for connector

This commit adds hardware state readout support to drm_bridge_connector by
introducing DRM_BRIDGE_OP_CONNECTOR_HW_READOUT and the
atomic_sro_connector_readout callback on drm_bridge_funcs.

> diff --git a/drivers/gpu/drm/display/drm_bridge_connector.c b/drivers/gpu/drm/display/drm_bridge_connector.c
> index 81f3e26f6fdf5..45806ef077081 100644
> --- a/drivers/gpu/drm/display/drm_bridge_connector.c
> +++ b/drivers/gpu/drm/display/drm_bridge_connector.c

[ ... ]

> @@ -244,6 +253,26 @@ drm_bridge_connector_create_state(struct drm_connector *connector)
>  	return conn_state;
>  }
>  
> +static int
> +drm_bridge_connector_readout_state(struct drm_connector *connector,
> +				   struct drm_atomic_sro_state *state,
> +				   struct drm_connector_state *conn_state)
> +{
> +	struct drm_bridge_connector *bridge_connector =
> +		to_drm_bridge_connector(connector);
> +	struct drm_bridge *readout =
> +		bridge_connector->bridge_connector_hw_readout;
> +	int ret;
> +
> +	if (readout) {
> +		ret = readout->funcs->atomic_sro_connector_readout(readout, state, conn_state);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	return 0;
> +}
> +
>  static enum drm_connector_color_format
>  drm_bridge_connector_color_format(const struct drm_connector_state *conn_state)
>  {
> @@ -270,6 +299,8 @@ drm_bridge_connector_color_format(const struct drm_connector_state *conn_state)
>  static const struct drm_connector_funcs drm_bridge_connector_funcs = {
>  	.fill_modes = drm_helper_probe_single_connector_modes,
>  	.atomic_create_state = drm_bridge_connector_create_state,
> +	.atomic_sro_readout_state = drm_bridge_connector_readout_state,
> +	.atomic_sro_compare_state = drm_atomic_helper_connector_compare_state,

[Severity: High]
Does unconditionally setting .atomic_sro_readout_state and
.atomic_sro_compare_state on drm_bridge_connector_funcs while returning 0 in
drm_bridge_connector_readout_state() when bridge_connector_hw_readout is NULL
cause a regression for bridge chains without DRM_BRIDGE_OP_CONNECTOR_HW_READOUT?

In drivers/gpu/drm/drm_atomic_state_readout.c, drm_atomic_sro_can_readout()
and drm_atomic_sro_can_compare() determine whether a connector supports state
readout by checking whether its callbacks are populated:

drivers/gpu/drm/drm_atomic_state_readout.c:drm_atomic_sro_can_readout() {
	...
	drm_connector_list_iter_begin(dev, &conn_iter);
	drm_for_each_connector_iter(connector, &conn_iter) {
		if (!connector->funcs->atomic_sro_readout_state) {
			drm_dbg_atomic(dev, "Connector %s missing readout callback",
				       connector->name);
			drm_connector_list_iter_end(&conn_iter);
			return false;
		}
	}
	drm_connector_list_iter_end(&conn_iter);
	...
}

Because drm_bridge_connector_funcs always defines both callbacks, every
drm_bridge_connector is treated as supporting hardware state readout even when
no bridge in the chain sets DRM_BRIDGE_OP_CONNECTOR_HW_READOUT (leaving
bridge_connector->bridge_connector_hw_readout NULL).

When state readout runs on a pipeline where bridge_connector_hw_readout is
NULL (for example, when bridges implement private state readout without any
bridge setting DRM_BRIDGE_OP_CONNECTOR_HW_READOUT, on a multi-connector device
where only one bridge chain supports connector readout, or when
drm.atomic_readout is set to DRM_ATOMIC_READOUT_SKIP_MISSING_READOUT),
drm_bridge_connector_readout_state() silently returns 0 and leaves
conn_state->crtc and conn_state->best_encoder set to NULL.

During initial readout in drm_mode_config_create_initial_state() ->
drm_atomic_helper_sro_readout_state() -> drm_atomic_helper_sro_build_state(),
this causes drm_bridge_atomic_readout_priv_state() in
drivers/gpu/drm/drm_bridge.c to skip reading out all bridges on that encoder's
chain because find_connector_state_for_encoder() matches encoder against
conn_state->best_encoder:

drivers/gpu/drm/drm_bridge.c:drm_bridge_atomic_readout_priv_state() {
	...
	conn_state = find_connector_state_for_encoder(state, encoder);
	if (!conn_state)
		return 0;

	if (!conn_state->crtc)
		return 0;
	...
}

It also installs an inconsistent atomic state via
drm_atomic_sro_install_state() with an active CRTC (crtc_state->active == true
and crtc_state->connector_mask == 0) and a disconnected connector state
instead of failing readout and falling back to hw_reset(). Furthermore, during
post-commit verification in commit_tail() ->
drm_atomic_sro_readout_and_compare(), every blocking commit enabling that
connector compares the committed active state against the unpopulated readout
state via drm_atomic_helper_connector_compare_state(),
drm_atomic_helper_crtc_compare_state(), and
drm_atomic_helper_bridge_compare_state(), triggering state mismatch errors.

Should drm_bridge_connector_init() only expose a drm_connector_funcs table
with .atomic_sro_readout_state and .atomic_sro_compare_state when
bridge_connector->bridge_connector_hw_readout is non-NULL (similar to how
drm_bridge_attach() selects between drm_bridge_priv_state_funcs and
drm_bridge_priv_state_funcs_with_sro)?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261029-drm-state-readout-v7-0-25e5163d75c3@kernel.org?part=28