Re: [PATCH RESEND v2] arm64: dts: mediatek: add mmc2 support for mt8365-evk
From: Alexandre Mergnat
Date: Wed Mar 26 2025 - 06:14:49 EST
Hello Angelo,
Thanks for the review :D
I've some comment below:
On 24/03/2025 15:09, AngeloGioacchino Del Regno wrote:
Il 24/03/25 14:54, Alexandre Mergnat ha scritto:
Adds support for the MMC2 interface on the MT8365 EVK board.
It introduces a fixed regulator for the MMC2 VDD33 supply and configures
the MMC2 node with a 4-bit bus width, high-speed capabilities, UHS
modes, and appropriate power supplies. Enabled SDIO IRQ, wakeup source,
and kept power during suspend (to save firmware module) for wireless
chip functionality.
Signed-off-by: Alexandre Mergnat <amergnat@xxxxxxxxxxxx>
---
Changes in v2:
- Apply alphabetical order to pinctrl property items.
- Improve commit message
- Link to v1: https://lore.kernel.org/r/20250109-mmc2-support-v1-1-9b9d1b1ae35d@xxxxxxxxxxxx
---
arch/arm64/boot/dts/mediatek/mt8365-evk.dts | 103 +++++++++++++++++++++++++---
1 file changed, 94 insertions(+), 9 deletions(-)
diff --git a/arch/arm64/boot/dts/mediatek/mt8365-evk.dts
b/arch/arm64/boot/dts/mediatek/mt8365-evk.dts
index 7d90112a7e274..a87f1b3ed6500 100644
--- a/arch/arm64/boot/dts/mediatek/mt8365-evk.dts
+++ b/arch/arm64/boot/dts/mediatek/mt8365-evk.dts
@@ -53,6 +53,15 @@ memory@40000000 {
reg = <0 0x40000000 0 0xc0000000>;
};
+ mmc2_vdd33: mmc2_vdd33-regulator {
+ compatible = "regulator-fixed";
+ regulator-name = "mmc2_vdd33";
mmc2-vdd33 please
Ok
+ regulator-min-microvolt = <3300000>;
+ regulator-max-microvolt = <3300000>;
+ gpio = <&pio 121 0>;
+ enable-active-high;
+ };
+
usb_otg_vbus: regulator-0 {
compatible = "regulator-fixed";
regulator-name = "otg_vbus";
Then, I will change this too.
@@ -197,6 +206,28 @@ &mmc1 {
status = "okay";
};
+&mmc2 {
+ assigned-clock-parents = <&topckgen CLK_TOP_MSDCPLL>;
+ assigned-clocks = <&topckgen CLK_TOP_MSDC50_2_SEL>;
+ bus-width = <4>;
+ cap-sd-highspeed;
+ cap-sdio-irq;
+ hs400-ds-delay = <0x12012>;
+ keep-power-in-suspend;
+ max-frequency = <200000000>;
+ non-removable;
+ pinctrl-0 = <&mmc2_default_pins>;
+ pinctrl-1 = <&mmc2_uhs_pins>;
+ pinctrl-names = "default", "state_uhs";
+ sd-uhs-sdr104;
+ sd-uhs-sdr25;
+ sd-uhs-sdr50;
+ vmmc-supply = <&mmc2_vdd33>;
+ vqmmc-supply = <&mt6357_vcn18_reg>;
+ wakeup-source;
+ status = "okay";
+};
+
&mt6357_pmic {
interrupts-extended = <&pio 145 IRQ_TYPE_LEVEL_HIGH>;
interrupt-controller;
@@ -324,8 +355,8 @@ cmd-dat-pins {
<MT8365_PIN_94_MSDC0_DAT6__FUNC_MSDC0_DAT6>,
<MT8365_PIN_93_MSDC0_DAT7__FUNC_MSDC0_DAT7>,
<MT8365_PIN_98_MSDC0_CMD__FUNC_MSDC0_CMD>;
- input-enable;
bias-pull-up;
+ input-enable;
This is a cleanup and goes to a different commit
Agree
};
rst-pins {
@@ -337,8 +368,8 @@ rst-pins {
mmc0_uhs_pins: mmc0-uhs-pins {
clk-pins {
pinmux = <MT8365_PIN_99_MSDC0_CLK__FUNC_MSDC0_CLK>;
- drive-strength = <MTK_DRIVE_10mA>;
bias-pull-down = <MTK_PUPD_SET_R1R0_10>;
+ drive-strength = <MTK_DRIVE_10mA>;
While at it, in a cleanup commit, if you could also remove those MTK_DRIVE_xxx and
use just the number that'd be great.
Sure
};
cmd-dat-pins {
@@ -351,21 +382,21 @@ cmd-dat-pins {
<MT8365_PIN_94_MSDC0_DAT6__FUNC_MSDC0_DAT6>,
<MT8365_PIN_93_MSDC0_DAT7__FUNC_MSDC0_DAT7>,
<MT8365_PIN_98_MSDC0_CMD__FUNC_MSDC0_CMD>;
- input-enable;
- drive-strength = <MTK_DRIVE_10mA>;
bias-pull-up = <MTK_PUPD_SET_R1R0_01>;
+ drive-strength = <MTK_DRIVE_10mA>;
+ input-enable;
};
ds-pins {
pinmux = <MT8365_PIN_104_MSDC0_DSL__FUNC_MSDC0_DSL>;
- drive-strength = <MTK_DRIVE_10mA>;
bias-pull-down = <MTK_PUPD_SET_R1R0_10>;
+ drive-strength = <MTK_DRIVE_10mA>;
};
rst-pins {
pinmux = <MT8365_PIN_97_MSDC0_RSTB__FUNC_MSDC0_RSTB>;
- drive-strength = <MTK_DRIVE_10mA>;
bias-pull-up;
+ drive-strength = <MTK_DRIVE_10mA>;
};
};
@@ -386,16 +417,16 @@ cmd-dat-pins {
<MT8365_PIN_91_MSDC1_DAT2__FUNC_MSDC1_DAT2>,
<MT8365_PIN_92_MSDC1_DAT3__FUNC_MSDC1_DAT3>,
<MT8365_PIN_87_MSDC1_CMD__FUNC_MSDC1_CMD>;
- input-enable;
bias-pull-up = <MTK_PUPD_SET_R1R0_01>;
+ input-enable;
};
};
mmc1_uhs_pins: mmc1-uhs-pins {
clk-pins {
pinmux = <MT8365_PIN_88_MSDC1_CLK__FUNC_MSDC1_CLK>;
- drive-strength = <8>;
bias-pull-down = <MTK_PUPD_SET_R1R0_10>;
+ drive-strength = <8>;
};
cmd-dat-pins {
@@ -404,9 +435,63 @@ cmd-dat-pins {
<MT8365_PIN_91_MSDC1_DAT2__FUNC_MSDC1_DAT2>,
<MT8365_PIN_92_MSDC1_DAT3__FUNC_MSDC1_DAT3>,
<MT8365_PIN_87_MSDC1_CMD__FUNC_MSDC1_CMD>;
- input-enable;
+ bias-pull-up = <MTK_PUPD_SET_R1R0_01>;
drive-strength = <6>;
+ input-enable;
+ };
+ };
+
+ mmc2_default_pins: mmc2-default-pins {
+ clk-pins {
+ pinmux = <MT8365_PIN_81_MSDC2_CLK__FUNC_MSDC2_CLK>;
+ bias-pull-down = <MTK_PUPD_SET_R1R0_10>;
+ drive-strength = <4>;
+ };
+
+ cmd-dat-pins {
+ pinmux = <MT8365_PIN_82_MSDC2_DAT0__FUNC_MSDC2_DAT0>,
+ <MT8365_PIN_83_MSDC2_DAT1__FUNC_MSDC2_DAT1>,
+ <MT8365_PIN_84_MSDC2_DAT2__FUNC_MSDC2_DAT2>,
+ <MT8365_PIN_85_MSDC2_DAT3__FUNC_MSDC2_DAT3>,
+ <MT8365_PIN_80_MSDC2_CMD__FUNC_MSDC2_CMD>;
bias-pull-up = <MTK_PUPD_SET_R1R0_01>;
+ drive-strength = <4>;
+ input-enable;
+ };
+
+ sys-en-pins {
+ pinmux = <MT8365_PIN_120_DMIC1_CLK__FUNC_GPIO120>;
My schematics say that the DMIC1_CLK pin is PERST_N, DMIC_DAT0 is PWR_EN: what's
the intention here?!
In any case, this is not a mmc2 pin, but something else :-)
First of all, name in schematic are wrong, I noticed that when I've port these feature from kernel
5.15 to 6.6. After digging and talking to MTK, they confirm me that PERST_N is actually an enable
pin (or power_on).
Secondly, MT7663_PERST_N (MT8365_PIN_120_DMIC1_CLK__FUNC_GPIO120) is part of mmc2 pin since is it
connected to the MMC2 connector pin number 52.
Cheers,
Angelo
--
Regards,
Alexandre