pinctrl: initial support for GD32

[gmarull]

This PR adds a new pinctrl driver for GD32 SoCs for both AF and AFIO pinmux models.

Tested on:

• GD32F450I-EVAL
• GD32F403Z-EVAL (@nandojve)

It depends on:

• boards: arm: GD32F450I-EVAL: add initial support #39958
• pinctrl: introduce pin configuration autogenerator hal_gigadevice#2

[github-actions]

The following west manifest projects have been modified in this Pull Request:

Name	Old Revision	New Revision	Diff
hal_gigadevice	zephyrproject-rtos/hal_gigadevice@85f10a6	zephyrproject-rtos/hal_gigadevice@cb4cc10 (main)	zephyrproject-rtos/[email protected]

Note: This message is automatically posted and updated by the Manifest GitHub Action.

[gmarull]

@nandojve Support for AFIO (like GD32F403) should now be ready. However, I haven't been able to test it as I don't have that board (will try on VF103 using #34970)

[nandojve]

@nandojve Support for AFIO (like GD32F403) should now be ready. However, I haven't been able to test it as I don't have that board (will try on VF103 using #34970)

Great ! I'll try soon and report to you.

[nandojve]

Hi @gmarull ,

I tested using gd32f403 without sucess, no characters at serial port.
I copy usart0_default resulting node here to help you check if something is missing.

usart0_default: usart0_default {
	phandle = < 0x2 >;
	pins1 {
		pinmux = < 0x11290 >, < 0x111a0 >;
	};
};

[nandojve]

Hi @gmarull ,

Excellent work!

I start to look from manifest update and have only small comments.

[nandojve]

Hi @gmarull ,

After some fixes AFIO is working:

*** Booting Zephyr OS build v2.7.99-1382-g31b26bab517c  ***
Running test suite pinctrl_gd32
===================================================================
START - test_dt_extract
 PASS - test_dt_extract in 0.1 seconds
===================================================================
Test suite pinctrl_gd32 succeeded
===================================================================
PROJECT EXECUTION SUCCESSFUL

[nandojve]

bias-high-impedance is missing

[gmarull]

This option is not required, same as STM32 (see https://github.com/zephyrproject-rtos/zephyr/blob/main/dts/bindings/pinctrl/st%2Cstm32-pinctrl.yaml#L18). Not setting any bias and operating as input would be sufficient as far as I understand.

[nandojve]

Not setting any bias and operating as input would be sufficient as far as I understand.

I understand not setting any bias, which means bias-disable (default state), equals Analog. We should not work with a hidden state because people can made confusion. The configuration table have 4 options for Input and there are 3 options defined. If you look at

zephyr/dts/bindings/pinctrl/pincfg-node.yaml

Lines 25 to 28 in a5c2a61

	bias-high-impedance:
	required: false
	type: boolean
	description: high impedance mode ("third-state", "floating")

this is the option that Linux define for the Input Floating.

[gmarull]

The problem is that analog is treated via the pinmux value as a separate mode, not as an input mode. In case of DAC (an output signal), ANALOG still has to be used. For low-power mode, what is usually done is to set the analog mode (default config) on a pin, this is why I added the analog option for all pins in the generator:

/* ANALOG */
#define ANALOG_PA0 \
	GD32_PINMUX_AFIO('A', 0, ANALOG, NORMP)
#define ANALOG_PA1 \
	GD32_PINMUX_AFIO('A', 1, ANALOG, NORMP)
...

I know it can be a bit confusing, it's really STM32 inheritance (it's done this way, see for example https://elixir.bootlin.com/linux/latest/source/arch/arm/boot/dts/stm32mp15-pinctrl.dtsi#L62). I think that treating analog in a special way makes things more understandable when it comes to the analog signals, but not that clear for low-power modes. In such case, it would probably make more sense to configure the pin as an input and set bias-high-impedance. But familiarity with STM32 has value for users experienced in that platform.

[nandojve]

In this case, it is important give some background for users. The information you presented at last comment is relevant to users understand how this should be configured. Other important thing is that no one can determine from where user will start to read the documentation. I think it could be very valuable improve documentation and try connect those things. For instance, correlate the GPIO configuration table with the options available can be very valuable to users. The same way, explain to then the why Analog is configured differently will help. Maybe provide pointer that exists in Zephyr can help to. Take in consideration that a lot of users don't know Linux and devicetree.

[gmarull]

I have updated the bindings description explaining more details on low power modes, please check.

[nandojve]

bias-high-impedance is missing

[gmarull]

ditto

[nandojve]

bias-disable: Disable pull-up/down (default, not required).

What is this? Analog or High-Impedance? Again, it is not clear the meaning and it is possible improve here to a very clear pin state.

[nandojve]

bias-high-impedance is missing

[gmarull]

ditto

[gmarull]

@cameled reviews welcome

[nandojve]

Hi @gmarull ,

Thank you for this PR!

[mbolivar-nordic]

I didn't review this very closely, but it looks like it's following all the latest practices from a quick skim.