Hi Geert,

Am 24.11.2021 um 17:21 schrieb Geert Uytterhoeven [off-list ref]:

Hi Nikolaus,

On Wed, Nov 24, 2021 at 5:19 PM H. Nikolaus Schaller [off-list ref] wrote:

quoted

quoted

Am 23.11.2021 um 21:10 schrieb Paul Cercueil [off-list ref]:
Le mar., nov. 23 2021 at 19:13:59 +0100, H. Nikolaus Schaller [off-list ref] a écrit :

quoted

+    assigned-clock-rates = <48000000>, <0>, <54000000>, <0>, <27000000>;
};
&tcu {

@@ -509,6 +534,19 @@ pins_i2c4: i2c4 {
            bias-disable;
    };
+    pins_hdmi_ddc: hdmi_ddc {
+            function = "hdmi-ddc";
+            groups = "hdmi-ddc";
+            bias-disable;
+    };
+
+    /* switch to PF25 as gpio driving DDC_SDA low */
+    pins_hdmi_ddc_unwedge: hdmi_ddc {
+            function = "hdmi-ddc";
+            groups = "hdmi-ddc";
+            bias-disable;
+    };

Your pins_hdmi_ddc and pins_hdmi_ddc_unwedge are the exact same? You could just use the former and pass it to both pinctrl-0 and pinctrl-1.

This was forgotten to remove. We do not make use of the unwedge feature because I could not find out how to use pinctrl to switch this to gpio25 and drive it low.

Using gpio-hog?

well, AFAIR it activates the gpio permanently and is a propery of the gpio controller and not of pinmux.
The driver assumes it can use pinmux state switching to drive the DDC_SDA line low on demand.

Since it is unlikely that we need it at all (and we have no test case that it works) I think we simply can leave
this driver feature unused unless we get a test case.

BR and thanks,
Nikolaus