On Fri, Dec 10, 2021 at 04:47:07PM +0530, Jagan Teki wrote:
> Having component_add for running all drm bind callbacks returns
> error or unbound due to chain of DSI devices connected across
> bridge topology on a display pipeline.
I'm not sure what that means?
> In a typical bridge oriented display pipeline where the host is
> connected to the bridge converter and that indeed connected to
> a panel.
>
> DRM => SUN6I DSI Host => Chipone ICN6211 => BananaPi Panel
>
> The bridge converter is looking for a panel to probe first and
> then attach the host. The host attach is looking for a bridge
> converter to probe and preserve bridge pointer, at this movement
^ moment ?
> the host is trying to bind the all callbacks and one of the bind
> callback in the DSI host is trying to find the bridge using the
> bridge pointer in sun6i_dsi_attach call.
>
> chipone_probe().start
> drm_of_find_panel_or_bridge
> mipi_dsi_attach
> sun6i_dsi_attach
> drm_of_find_panel_or_bridge
> chipone_probe().done
>
> sun6i_dsi_probe().start
> mipi_dsi_host_register
> component_add
> sun6i_dsi_probe().done
>
> However, the movement when panel defers the probe, will make the
> bridge converter defer the host attach call which eventually found
> a NULL bridge pointer during DSI component bind callback.
>
> So, in order to prevent this scenario of binding invalid bridge,
> wait for DSI devices on the pipeline to probe first and start the
> binding process by moving component_add in host probe to attach call.
>
> chipone_probe().start
> drm_of_find_panel_or_bridge
> mipi_dsi_attach
> sun6i_dsi_attach
> drm_of_find_panel_or_bridge
> component_add
> chipone_probe().done
>
> sun6i_dsi_probe().start
> mipi_dsi_host_register
> sun6i_dsi_probe().done
>
> Signed-off-by: Jagan Teki <jagan@xxxxxxxxxxxxxxxxxxxx>
> ---
> Changes for v6:
> - none
> Changes for v5:
> - new patch
>
> drivers/gpu/drm/sun4i/sun6i_mipi_dsi.c | 120 +++++++++++++------------
> 1 file changed, 61 insertions(+), 59 deletions(-)
>
> diff --git a/drivers/gpu/drm/sun4i/sun6i_mipi_dsi.c b/drivers/gpu/drm/sun4i/sun6i_mipi_dsi.c
> index 4bdcce8f1d84..9cf91dcac3f2 100644
> --- a/drivers/gpu/drm/sun4i/sun6i_mipi_dsi.c
> +++ b/drivers/gpu/drm/sun4i/sun6i_mipi_dsi.c
> @@ -959,11 +959,63 @@ static int sun6i_dsi_dcs_read(struct sun6i_dsi *dsi,
> return 1;
> }
>
> +static int sun6i_dsi_bind(struct device *dev, struct device *master,
> + void *data)
> +{
> + struct drm_device *drm = data;
> + struct sun6i_dsi *dsi = dev_get_drvdata(dev);
> + int ret;
> +
> + drm_encoder_helper_add(&dsi->encoder,
> + &sun6i_dsi_enc_helper_funcs);
> + ret = drm_simple_encoder_init(drm, &dsi->encoder,
> + DRM_MODE_ENCODER_DSI);
> + if (ret) {
> + dev_err(dsi->dev, "Couldn't initialise the DSI encoder\n");
> + return ret;
> + }
> + dsi->encoder.possible_crtcs = BIT(0);
> +
> + drm_connector_helper_add(&dsi->connector,
> + &sun6i_dsi_connector_helper_funcs);
> + ret = drm_connector_init(drm, &dsi->connector,
> + &sun6i_dsi_connector_funcs,
> + DRM_MODE_CONNECTOR_DSI);
> + if (ret) {
> + dev_err(dsi->dev,
> + "Couldn't initialise the DSI connector\n");
> + goto err_cleanup_connector;
> + }
> +
> + drm_connector_attach_encoder(&dsi->connector, &dsi->encoder);
> +
> + return 0;
> +
> +err_cleanup_connector:
> + drm_encoder_cleanup(&dsi->encoder);
> + return ret;
> +}
> +
> +static void sun6i_dsi_unbind(struct device *dev, struct device *master,
> + void *data)
> +{
> + struct sun6i_dsi *dsi = dev_get_drvdata(dev);
> +
> + drm_encoder_cleanup(&dsi->encoder);
> +}
> +
> +static const struct component_ops sun6i_dsi_ops = {
> + .bind = sun6i_dsi_bind,
> + .unbind = sun6i_dsi_unbind,
> +};
Just use a forward declaration there, it will make the patch more
straightforward.
Maxime
Attachment:
signature.asc
Description: PGP signature
