On Tuesday, June 29, 2021 20:04 EEST, Alex Bee [off-list ref] wrote: 
 

Hi Adrian, Hi Robin
Am 29.06.21 um 17:42 schrieb Robin Murphy:

quoted

On 2021-06-29 16:07, Adrian Ratiu wrote:

quoted

Hi Alex,

On Tue, 29 Jun 2021, Alex Bee [off-list ref] wrote:

quoted

Hi Adrian,
I've submitted similar patch already. Its part of [1]

Thank you for pointing out your patch, I was not aware of it as it was 
not merged yet (I tested latest linux-next).

It is a good sign that our code is almost identical except for the 
headphone detect pin which you mention.

quoted

There is no headphone detection pin in the schematics, btw.

There are two reasons I added it:

1. The es8316 codec failed to probe() in my testing unless the irq was 
defined. That might have been due to me testing the codec directly 
with the simple card driver instead of the graph driver.

Point 1 doesn't appear to be an issue anymore, the codec + graph 
driver probe correctly and audio supposedly (see below) works even 
without the irq.

I guess that just a "warning" that doesn't mean the driver failed to 
probe. (It the same for RockPro64, for example - where the interrupt 
line isn't connected as well)

quoted

quoted

2. I got the gpio bank 1 pin 0 location from various headphone 
detection commits in the vendor v4.4 kernel, like for example [1].

Are you 100% sure there is no hp detection pin wired on the rock-pi-4?
Even if it might be missing from the schematics, the vendor driver 
code apparently defines and uses it?

FWIW according to the schematics it *is* wired up on the Model C board, 
but not on the Model A or B.

Thanks for the pointer, I did look at v1.3 only, tbh.

I'm not sure I understand why we should pull up this pin.

Shouldn't we better use hp-det-gpio property of audio-graph-card to make 
real use of this pin?

That sounds like the best way forward, only on the model / revision which actually has the pin connected.

quoted

Robin.

quoted

 From booting with the irq defined I get the following in 
/proc/interrupts however I'm having an unrelated difficulty testing 
because the audio connector literally broke off the board like in [2].
90: 1  0  0  0  0  rockchip_gpio_irq   0 Level     es8316

According to the Model C schematics the codec interrupt (HP_INT) is 
connected to GPIO1_PA1 (in contrast to GPIO1_PA0 which is used for hp 
detection).

If you're OK I'll update my series and add those pins for Model C only 
(and will request the schematics of A+ and B+ which seem not to be 
published yet to check out what the manufacturer decided for those variants)

Yes, thank you very much! Please CC me on the new series, hopefully I can provide a tested-by if I receive the fixed HW by then.

Adrian

Alex

quoted

quoted

Team Radxa was very nice in offering me a replacement but until that 
arrives I can't test this anymore.

[1] 
https://github.com/radxa/kernel/commit/e945cad5c3ec82d171760465d3c7a84bb10ed1b7 


[2] https://forum.radxa.com/t/audio-jack-broke-off/935

quoted

[1] 
https://patchwork.kernel.org/project/linux-rockchip/cover/20210618181256.27992-1-knaerzche@gmail.com/ 


Best,

Alex

Am 28.06.21 um 13:30 schrieb Adrian Ratiu:

quoted

This adds the necessary pinctrl and nodes to enable the
analog audio on rk3399 rock-pi-4 SBCs using the es8316
codec and the audio-graph-card driver.

Signed-off-by: Adrian Ratiu <redacted>
---
  .../boot/dts/rockchip/rk3399-rock-pi-4.dtsi   | 38 
+++++++++++++++++++
  1 file changed, 38 insertions(+)

diff --git a/arch/arm64/boot/dts/rockchip/rk3399-rock-pi-4.dtsi 

b/arch/arm64/boot/dts/rockchip/rk3399-rock-pi-4.dtsi
index b28888ea9262..77781d9150ac 100644

--- a/arch/arm64/boot/dts/rockchip/rk3399-rock-pi-4.dtsi
+++ b/arch/arm64/boot/dts/rockchip/rk3399-rock-pi-4.dtsi

@@ -128,6 +128,12 @@ vdd_log: vdd-log {

          regulator-max-microvolt = <1400000>;
          vin-supply = <&vcc5v0_sys>;
      };
+
+    sound {
+        compatible = "audio-graph-card";
+        label = "rockchip,rk3399";
+        dais = <&i2s0_p0>;
+    };
  };
  &cpu_l0 {

@@ -422,6 +428,24 @@ &i2c1 {

      i2c-scl-rising-time-ns = <300>;
      i2c-scl-falling-time-ns = <15>;
      status = "okay";
+
+    es8316: codec@11 {
+        compatible = "everest,es8316";
+        reg = <0x11>;
+        clocks = <&cru SCLK_I2S_8CH_OUT>;
+        clock-names = "mclk";
+        pinctrl-names = "default";
+        pinctrl-0 = <&hp_det_pin>;
+        interrupt-parent = <&gpio1>;
+        interrupts = <RK_PA0 IRQ_TYPE_LEVEL_HIGH>;
+        #sound-dai-cells = <0>;
+
+        port {
+            es8316_p0_0: endpoint {
+                remote-endpoint = <&i2s0_p0_0>;
+            };
+        };
+    };
  };
  &i2c3 {

@@ -441,6 +465,14 @@ &i2s0 {

      rockchip,capture-channels = <2>;
      rockchip,playback-channels = <2>;
      status = "okay";
+
+    i2s0_p0: port {
+        i2s0_p0_0: endpoint {
+            dai-format = "i2s";
+            mclk-fs = <256>;
+            remote-endpoint = <&es8316_p0_0>;
+        };
+    };
  };
  &i2s1 {

@@ -556,6 +588,12 @@ wifi_host_wake_l: wifi-host-wake-l {

              rockchip,pins = <0 RK_PA3 RK_FUNC_GPIO &pcfg_pull_none>;
          };
      };
+
+    es8316 {
+        hp_det_pin: hp-det-pin {
+            rockchip,pins = <1 RK_PA0 RK_FUNC_GPIO &pcfg_pull_up>;
+        };
+    };
  };
  &pwm2 {

_______________________________________________
Linux-rockchip mailing list
Linux-rockchip@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-rockchip

_______________________________________________
Linux-rockchip mailing list
Linux-rockchip@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-rockchip