On Wed, Feb 07, 2024 at 02:13:05PM +0100, Krzysztof Hałasa wrote:
> Laurent,
>
> Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx> writes:
>
> >> +++ b/drivers/media/i2c/adv7604.c
> >> @@ -3205,7 +3205,7 @@ static int adv76xx_parse_dt(struct adv76xx_state *state)
> >> np = state->i2c_clients[ADV76XX_PAGE_IO]->dev.of_node;
> >>
> >> /* Parse the endpoint. */
> >> - endpoint = of_graph_get_next_endpoint(np, NULL);
> >> + endpoint = of_graph_get_endpoint_by_regs(np, 0, -1);
> >
> > I think this should be port 1 for the adv7611 and port2 for the adv7612.
> > The adv7610 may need to use port 1 too, but the bindings likely need to
> > be updated.
>
> To be honest I have no idea about ADV7611 and 7612.
> The 7610 I have on Tinyrex "mobo" seems to be single port.
One issue is that the commit that added the adv7610 compatible string to
the DT bindings (commit 901a52c43359, "media: Add ADV7610 support for
adv7604 driver - DT docs.") didn't extend the conditional rules further
down in the file that defines what ports are needed. That's why I don't
know what was intended. I believe the adv7610 should be treated like the
adv7611, given that the driver has
{ .compatible = "adi,adv7610", .data = &adv76xx_chip_info[ADV7611] },
{ .compatible = "adi,adv7611", .data = &adv76xx_chip_info[ADV7611] },
Could you send a patch to address that in the DT bindings ?
> ADV7611 seems to be mostly a 7610 in a different package (LQFP 64
> instead of some BGA 76). The driver simply treats ADV7610 as a 7611.
>
> ADV7612 is apparently dual port (only one port can be used at a time)
> though:
>
> [ADV7612] = {
> .type = ADV7612,
> .has_afe = false,
> .max_port = ADV76XX_PAD_HDMI_PORT_A, /* B not supported */
> .num_dv_ports = 1, /* normally 2 */
>
>
> All related in-tree DTS entries (as of v6.8.0-rc1) seem to be ADV7612.
>
> To me it seems all known devices use the first port only.
--
Regards,
Laurent Pinchart
