[PATCH v2 0/3] arm64: dts: mediatek: mt8390-genio-common: USB fixes

[PATCH v2 0/3] arm64: dts: mediatek: mt8390-genio-common: USB fixes

[Chen-Yu Tsai]

Hi,

This is v2 of the Genio 510 / 700 EVK USB fixes that came out of the
MTK XHCI binding fix [1].

Changes since v1:
- Patch 1 now also switches the type C port to a source-only port, and
  limits the source PDO to 5V 1A, matching the hardware (Sashiko)
- Patch 1 now connects the SuperSpeed and HighSpeed lines to the
  correct USB hubs
- New patch 3 fixing swapped USB hub labels
- Link to v1: https://lore.kernel.org/all/20261103083335.2944847-1-wenst@chromium.org/

Patch 1 corrects the USB type C port: the controller operates in host
mode only, the data lines go through the onboard USB hub, and the port
can only source power (5V 1A). The block diagrams for both the Genio
700 EVK and Genio 510 EVK were used as reference.

Patch 1 depends on "dt-bindings: usb: ti,usb8041: Fold in TI USB8020B" [2]
to allow graph ports under the TI USB8020B hub nodes. Without it,
dtbs_check will complain about the new "ports" nodes.

Patch 2 moves the pin mux for the USB hub supply enable pin to the
regulator.

Patch 3 fixes the swapped labels for the USB 2.0 and USB 3.0 hub nodes.

I don't have the actual board to test this on. The changes were only
compile tested.

This v2 patch series was partially put together using LLM. v1 was done
by hand. Sashiko review comments were fed to the LLM, as well as the
datasheet for TI USB8020B. I manually checked the PIDs in the datasheet
before giving the datasheet to the LLM. Same goes for the board block
diagrams. The changelogs in each patch and in the cover letter were
also put together using LLM.

Also, the LLM seems to do a decent job at mimicking my writing style,
even though it's a bit more verbose than what I'd go for.


ChenYu

[1] https://lore.kernel.org/all/20260916075745.3549953-11-wenst@chromium.org/
[2] https://lore.kernel.org/all/20261103095425.3097720-1-wenst@chromium.org/

Chen-Yu Tsai (3):
  arm64: dts: mediatek: mt8390-genio-common: Fix type C port mode and
    connection
  arm64: dts: mediatek: mt8390-genio-common: Split out USB hub enable
    pin mux
  arm64: dts: mediatek: mt8390-genio-common: Fix swapped USB hub labels

 .../dts/mediatek/mt8390-genio-common.dtsi     | 76 +++++++++++--------
 1 file changed, 43 insertions(+), 33 deletions(-)

-- 
2.56.0.455.g6c5a503b38-goog

[PATCH v2 1/3] arm64: dts: mediatek: mt8390-genio-common: Fix type C port mode and connection

[Chen-Yu Tsai]

The USB controller complex ssusb1 and xhci1 operate in host mode only,
as it is immediately connected to a USB hub. The user manual has a block
diagram showing this, and a USB usage table showing port 1 as "host
only".

The block diagrams for both the Genio 700 EVK and the Genio 510 EVK
show that the type C port's USB 2.0 data lines are connected to the
USB hub's second downstream port directly, while its USB 3.x data lines
go through the USB 3.1 / DP mux to the same hub port. The type C port's
VBUS is supplied by a 5V 1A boost converter and current limiter. The
port cannot sink power; the charger input is on a separate type C
connector.

Change the controller's operating mode to "host", and drop the USB role
switch property. Fix the USB type C port data line connections to go
through the USB hub. The TI TUSB8020B's USB 3.0 hub uses PID 0x8025,
and the USB 2.0 hub uses PID 0x8027, so the SuperSpeed lines connect to
the former and the HighSpeed lines to the latter.

Change the type C port to a source-only, host-only port. Drop the sink
related properties, and limit the source PDO to 5V 1A, matching the
hardware. Also drop the dual role and data swap flags from the source
PDO, since the port supports neither.

Fixes: 1afaeca17238 ("arm64: dts: mediatek: mt8390-genio-700: Add USB, TypeC Controller, MUX")
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
---
Changes since v1:
- Switched type C port to source-only, and dropped sink related
  properties (Sashiko)
- Limited source PDO to 5V 1A, matching the hardware, and dropped dual
  role and data swap flags
- Fixed swapped SuperSpeed / HighSpeed connections to the USB hubs
- Expanded commit message with details from the Genio 700 / 510 EVK
  block diagrams

Compile tested only.

The v2 changes were done with the help of an LLM coding assistant:
the source-only connector and PDO changes, the swapped hub endpoint
fix, most of the expanded commit message, and these notes. The input
was Sashiko's review, the TI USB8020B datasheet and the Genio 510 / 700
EVK block diagrams. The PIDs in the datasheet and the block diagrams
were checked by hand.

This patch depends on "dt-bindings: usb: ti,usb8041: Fold in TI
USB8020B" [1] to allow graph ports under the TI USB8020B hub nodes.

The downstream port number of the USB hub is based on the order shown
in the block diagram. I don't have a board or the schematics to confirm
this.

The hub_2_0 and hub_3_0 labels are swapped relative to the actual hubs
(usb451,8025 is the USB 3.0 hub). This is fixed in a separate patch.

[1] https://lore.kernel.org/all/20261103095425.3097720-1-wenst@chromium.org/
---
 .../dts/mediatek/mt8390-genio-common.dtsi     | 54 ++++++++++---------
 1 file changed, 30 insertions(+), 24 deletions(-)

diff --git a/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi b/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
index 669eaec662f3..d10ee1603103 100644
--- a/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
+++ b/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
@@ -512,18 +512,11 @@ rt1715@4e {
 		connector {
 			compatible = "usb-c-connector";
 			label = "USB-C";
-			data-role = "dual";
-			op-sink-microwatt = <10000000>;
-			power-role = "dual";
-			try-power-role  = "sink";
+			data-role = "host";
+			power-role = "source";
 			pd-revision = /bits/ 8 <0x03 0x00 0x01 0x08>;
 
-			sink-pdos = <PDO_FIXED(5000, 2000,
-					       PDO_FIXED_DUAL_ROLE |
-					       PDO_FIXED_DATA_SWAP)>;
-			source-pdos = <PDO_FIXED(5000, 2000,
-						 PDO_FIXED_DUAL_ROLE |
-						 PDO_FIXED_DATA_SWAP)>;
+			source-pdos = <PDO_FIXED(5000, 1000, 0)>;
 
 			altmodes {
 				displayport {
@@ -539,14 +532,14 @@ ports {
 				port@0 {
 					reg = <0>;
 					typec_con_hs: endpoint {
-						remote-endpoint = <&mtu3_hs1_role_sw>;
+						remote-endpoint = <&usb_2_0_p2>;
 					};
 				};
 
 				port@1 {
 					reg = <1>;
 					typec_con_ss: endpoint {
-						remote-endpoint = <&xhci_ss_ep>;
+						remote-endpoint = <&usb_3_0_p2>;
 					};
 				};
 
@@ -1528,19 +1521,12 @@ &xhci0 {
 };
 
 &ssusb1 {
-	dr_mode = "otg";
-	usb-role-switch;
+	dr_mode = "host";
 	wakeup-source;
 	vusb33-supply = <&mt6359_vusb_ldo_reg>;
 	pinctrl-0 = <&usb1_default_pins>;
 	pinctrl-names = "default";
 	status = "okay";
-
-	port {
-		mtu3_hs1_role_sw: endpoint {
-			remote-endpoint = <&typec_con_hs>;
-		};
-	};
 };
 
 &xhci1 {
@@ -1555,6 +1541,19 @@ hub_2_0: hub@1 {
 		peer-hub = <&hub_3_0>;
 		reset-gpios = <&pio 7 GPIO_ACTIVE_HIGH>;
 		vdd-supply = <&usb_hub_fixed_3v3>;
+
+		ports {
+			#address-cells = <1>;
+			#size-cells = <0>;
+
+			port@2 {
+				reg = <2>;
+
+				usb_3_0_p2: endpoint {
+					remote-endpoint = <&typec_con_ss>;
+				};
+			};
+		};
 	};
 
 	hub_3_0: hub@2 {
@@ -1563,11 +1562,18 @@ hub_3_0: hub@2 {
 		peer-hub = <&hub_2_0>;
 		reset-gpios = <&pio 7 GPIO_ACTIVE_HIGH>;
 		vdd-supply = <&usb_hub_fixed_3v3>;
-	};
 
-	port {
-		xhci_ss_ep: endpoint {
-			remote-endpoint = <&typec_con_ss>;
+		ports {
+			#address-cells = <1>;
+			#size-cells = <0>;
+
+			port@2 {
+				reg = <2>;
+
+				usb_2_0_p2: endpoint {
+					remote-endpoint = <&typec_con_hs>;
+				};
+			};
 		};
 	};
 };
-- 
2.56.0.455.g6c5a503b38-goog

[PATCH v2 2/3] arm64: dts: mediatek: mt8390-genio-common: Split out USB hub enable pin mux

[Chen-Yu Tsai]

"pins-usb-hub-3v3-en" listed in "usb1-default-pins" is actually the
enable pin for the USB hub's supply regulator.

Split it out as a separate pinmux node attached to the regulator
instead.

Fixes: 7982bf7ec2e3 ("arm64: dts: mediatek: add device-tree for Genio 700 EVK board")
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
---
Compile tested only.
---
 .../boot/dts/mediatek/mt8390-genio-common.dtsi     | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)

diff --git a/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi b/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
index d10ee1603103..63b4af476959 100644
--- a/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
+++ b/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
@@ -216,6 +216,8 @@ touch0_fixed_3v3: regulator-5 {
 
 	usb_hub_fixed_3v3: regulator-6 {
 		compatible = "regulator-fixed";
+		pinctrl-names = "default";
+		pinctrl-0 = <&usb_hub_3v3_pins>;
 		regulator-name = "vhub_3v3";
 		regulator-min-microvolt = <3300000>;
 		regulator-max-microvolt = <3300000>;
@@ -1249,16 +1251,18 @@ pins-vbus {
 
 	};
 
+	usb_hub_3v3_pins: usb-hub-3v3-pins {
+		pins-usb-hub-3v3-en {
+			pinmux = <PINMUX_GPIO112__FUNC_B_GPIO112>;
+			output-high;
+		};
+	};
+
 	usb1_default_pins: usb1-default-pins {
 		pins-valid {
 			pinmux = <PINMUX_GPIO88__FUNC_I0_VBUSVALID_1P>;
 			input-enable;
 		};
-
-		pins-usb-hub-3v3-en {
-			pinmux = <PINMUX_GPIO112__FUNC_B_GPIO112>;
-			output-high;
-		};
 	};
 
 	usb2_default_pins: usb2-default-pins {
-- 
2.56.0.455.g6c5a503b38-goog

[PATCH v2 3/3] arm64: dts: mediatek: mt8390-genio-common: Fix swapped USB hub labels

[Chen-Yu Tsai]

The TI TUSB8020B USB 3.0 hub on the Genio 510 / 700 EVKs presents
itself as two USB hubs, a USB 3.0 hub with the default PID 0x8025 and
a USB 2.0 hub with PID 0x8027 (the USB 3.0 PID XOR 0x2), as described
in the datasheet.

The hub nodes have the correct compatible strings, and are attached to
the correct XHCI ports (port 1 being SuperSpeed and port 2 being
HighSpeed), but the labels are swapped: the USB 3.0 hub (usb451,8025)
is labeled "hub_2_0", and the USB 2.0 hub (usb451,8027) is labeled
"hub_3_0".

Swap the labels so they match the hubs they are attached to. This
change has no functional impact.

Fixes: 20d959fa6fd0 ("arm64: dts: mediatek: mt8390-genio-700-evk: add keys and USB HUB")
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
---
Changes since v1:
- New patch

Compile tested only.

This patch, including the commit message, was written by an LLM coding
assistant at my request, based on the PID scheme described in the TI
USB8020B datasheet. The PIDs in the datasheet were checked by hand.
---
 arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi b/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
index 63b4af476959..e5c7608dca6b 100644
--- a/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
+++ b/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
@@ -1539,10 +1539,10 @@ &xhci1 {
 	#address-cells = <1>;
 	#size-cells = <0>;
 
-	hub_2_0: hub@1 {
+	hub_3_0: hub@1 {
 		compatible = "usb451,8025";
 		reg = <1>;
-		peer-hub = <&hub_3_0>;
+		peer-hub = <&hub_2_0>;
 		reset-gpios = <&pio 7 GPIO_ACTIVE_HIGH>;
 		vdd-supply = <&usb_hub_fixed_3v3>;
 
@@ -1560,10 +1560,10 @@ usb_3_0_p2: endpoint {
 		};
 	};
 
-	hub_3_0: hub@2 {
+	hub_2_0: hub@2 {
 		compatible = "usb451,8027";
 		reg = <2>;
-		peer-hub = <&hub_2_0>;
+		peer-hub = <&hub_3_0>;
 		reset-gpios = <&pio 7 GPIO_ACTIVE_HIGH>;
 		vdd-supply = <&usb_hub_fixed_3v3>;
 
-- 
2.56.0.455.g6c5a503b38-goog

Re: [PATCH v2 1/3] arm64: dts: mediatek: mt8390-genio-common: Fix type C port mode and connection

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [Low] arm64: dts: mediatek: missing PDO_FIXED_USB_COMM in mt8390
  source-pdos
- [Low] arm64: dts: mediatek: mt8390-genio: hub ports nodes violate
  ti,usb8020b schema
--

commit b958daa221ebf8e15b905e203df3fc6aa4fbfaae
Author: Chen-Yu Tsai <wenst@chromium.org>

arm64: dts: mediatek: mt8390-genio-common: Fix type C port mode and connection

This commit updates the USB controller and Type-C connector on MT8390
Genio boards to host-only and source-only mode, routing the connector's
High-Speed and SuperSpeed endpoints through the onboard TI TUSB8020B hub.

> diff --git a/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi b/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
> index 669eaec662f33..d10ee1603103f 100644
> --- a/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
> +++ b/arch/arm64/boot/dts/mediatek/mt8390-genio-common.dtsi
> @@ -512,18 +512,11 @@ rt1715@4e {
>  		connector {
>  			compatible = "usb-c-connector";
>  			label = "USB-C";
> -			data-role = "dual";
> -			op-sink-microwatt = <10000000>;
> -			power-role = "dual";
> -			try-power-role  = "sink";
> +			data-role = "host";
> +			power-role = "source";
>  			pd-revision = /bits/ 8 <0x03 0x00 0x01 0x08>;
>  
> -			sink-pdos = <PDO_FIXED(5000, 2000,
> -					       PDO_FIXED_DUAL_ROLE |
> -					       PDO_FIXED_DATA_SWAP)>;
> -			source-pdos = <PDO_FIXED(5000, 2000,
> -						 PDO_FIXED_DUAL_ROLE |
> -						 PDO_FIXED_DATA_SWAP)>;
> +			source-pdos = <PDO_FIXED(5000, 1000, 0)>;

[Severity: Low]
Should source-pdos pass PDO_FIXED_USB_COMM instead of 0?  This connector
routes High-Speed and SuperSpeed data lines to the onboard USB hub, so
passing 0 causes tcpm_pd_send_source_caps() and
usb_communication_capable_show() to advertise the port as incapable of
USB data communication.

[ ... ]

> @@ -1555,6 +1541,19 @@ hub_2_0: hub@1 {
>  		peer-hub = <&hub_3_0>;
>  		reset-gpios = <&pio 7 GPIO_ACTIVE_HIGH>;
>  		vdd-supply = <&usb_hub_fixed_3v3>;
> +
> +		ports {
> +			#address-cells = <1>;
> +			#size-cells = <0>;
> +
> +			port@2 {
> +				reg = <2>;
> +
> +				usb_3_0_p2: endpoint {
> +					remote-endpoint = <&typec_con_ss>;
> +				};
> +			};
> +		};
>  	};
>  
>  	hub_3_0: hub@2 {
> @@ -1563,11 +1562,18 @@ hub_3_0: hub@2 {
>  		peer-hub = <&hub_2_0>;
>  		reset-gpios = <&pio 7 GPIO_ACTIVE_HIGH>;
>  		vdd-supply = <&usb_hub_fixed_3v3>;
> -	};
>  
> -	port {
> -		xhci_ss_ep: endpoint {
> -			remote-endpoint = <&typec_con_ss>;
> +		ports {
> +			#address-cells = <1>;
> +			#size-cells = <0>;
> +
> +			port@2 {
> +				reg = <2>;
> +
> +				usb_2_0_p2: endpoint {
> +					remote-endpoint = <&typec_con_hs>;
> +				};
> +			};
>  		};
>  	};
>  };

[Severity: Low]
Do the new ports child nodes under hub@1 (usb451,8025) and hub@2
(usb451,8027) cause dtbs_check to fail against
Documentation/devicetree/bindings/usb/ti,usb8020b.yaml?  That schema only
references usb-device.yaml# with additionalProperties: false, and the
prerequisite commit "dt-bindings: usb: ti,usb8041: Fold in TI USB8020B"
noted in the patch commentary is not present in the tree yet.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261106060621.2706658-1-wenst@chromium.org?part=1