Hi Sakari,
just a few questions about the below helper.
On 20-01-08 16:36, Marco Felsch wrote:
> Hi Sakari,
>
> On 19-11-16 01:34, Sakari Ailus wrote:
> > Hi Marco,
> >
> > On Mon, Sep 30, 2019 at 11:38:49AM +0200, Marco Felsch wrote:
>
> ...
>
> > > +int v4l2_fwnode_connector_alloc_parse(struct fwnode_handle *fwnode,
> > > + struct v4l2_fwnode_connector *connector)
> > > +{
> > > + struct fwnode_handle *remote_pp, *remote_ep;
> > > + const char *type_name;
> > > + unsigned int i = 0, ep_num = 0;
> > > + int err;
> > > +
> > > + memset(connector, 0, sizeof(*connector));
> > > +
> > > + remote_pp = fwnode_graph_get_remote_port_parent(fwnode);
> > > + if (!remote_pp)
> > > + return -ENOLINK;
>
> I will align the API with the v4l2_fwnode_endpoint_alloc_parse()
> function so the caller need to pass the connector fwnode handle.
Should we really align this with v4l2_fwnode_endpoint_alloc_parse()? I
have some concerns because this moves the step to the user. Using the
current approach makes it easier for the user.
Regards,
Marco
> > > +
> > > + /* Parse all common properties first. */
> > > + fwnode_graph_for_each_endpoint(remote_pp, remote_ep)
> > > + ep_num++;
> >
> > If there are no endpoints, ep_num will be zero and kmalloc_array() will
> > return NULL? There should be a way there are no endpoints, rather than
> > returning -ENOMEM.
> >
> > > +
> > > + connector->nr_of_links = ep_num;
> > > + connector->links = kmalloc_array(ep_num, sizeof(*connector->links),
> > > + GFP_KERNEL);
> > > + if (!connector->links) {
> > > + err = -ENOMEM;
> > > + goto err_put_fwnode;
> > > + }
> > > +
> > > + fwnode_graph_for_each_endpoint(remote_pp, remote_ep) {
> > > + err = v4l2_fwnode_parse_link(remote_ep, &connector->links[i]);
> >
> > If you start parsing a connector starting from another device connected to
> > it, nothing seems to prevent parsing the same links twice, in case the
> > connector is connected to more than one sub-device.
> >
> > Or do I miss something crucial here?
>
> Yes thats right but it seems that sharing connectors isn't supported at
> all. All bridge drivers using connectors implementing the connector
> handling by their self. Anyway, I will add a function parameter to check
> that we parse only the endpoints connected to the calling sub-dev.
>
> Regards,
> Marco
>
> > > + if (err) {
> > > + fwnode_handle_put(remote_ep);
> > > + goto err_free_links;
> > > + }
> > > + i++;
> > > + }
> > > +
> > > + /*
> > > + * Links reference counting guarantees access -> no duplication needed
> > > + */
> > > + fwnode_property_read_string(remote_pp, "label", &connector->label);
> > > +
> > > + /* The connector-type is stored within the compatible string. */
> > > + err = fwnode_property_read_string(remote_pp, "compatible", &type_name);
> > > + if (err) {
> > > + err = -EINVAL;
> > > + goto err_free_links;
> > > + }
> > > + connector->type = v4l2_fwnode_string_to_connector_type(type_name);
> > > +
> > > + /* Now parse the connector specific properties. */
> > > + switch (connector->type) {
> > > + case V4L2_CONN_COMPOSITE:
> > > + case V4L2_CONN_SVIDEO:
> > > + err = v4l2_fwnode_connector_parse_analog(remote_pp, connector);
> > > + break;
> > > + case V4L2_CONN_UNKNOWN:
> > > + default:
> > > + pr_err("Unknown connector type\n");
> > > + err = -EINVAL;
> > > + goto err_free_links;
> > > + }
> > > +
> > > + fwnode_handle_put(remote_pp);
> > > +
> > > + return err;
> > > +
> > > +err_free_links:
> > > + for (i = 0; i < ep_num; i++)
> > > + v4l2_fwnode_put_link(&connector->links[i]);
> > > + kfree(connector->links);
> > > +err_put_fwnode:
> > > + fwnode_handle_put(remote_pp);
> > > +
> > > + return err;
> > > +}
> > > +EXPORT_SYMBOL_GPL(v4l2_fwnode_connector_alloc_parse);
> > > +
>
>
--
Pengutronix e.K. | |
Steuerwalder Str. 21 | http://www.pengutronix.de/ |
31137 Hildesheim, Germany | Phone: +49-5121-206917-0 |
Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |
