gpio_hold_en not working as expected since 2.0.0

[dynek]

Board

ESP-WROOM-32

Device Description

I tested both with a 30 pin and 38 pin dev modules:

Hardware Configuration

There is absolutely nothing attached to the ESP, only an oscilloscope on GPIO 33 and GND.

Version

v2.0.3

IDE Name

PlateformIO

Operating System

macOS 10.15.7

Flash frequency

40Mhz

PSRAM enabled

no

Upload speed

460800

Description

Below's code sets GPIO 33 HIGH, instructs the ESP to hold it in its current state, and sleep for 2 seconds.
Upon wake-up I expect GPIO 33 to be held high, except if / until I call gpio_hold_dis.

When using Arduino-ESP32 version 1.0.6, it works as mentioned.
Since 2.0.0, as soon as the ESP wakes up, GPIOs are reset to their default state.

Sketch

pinMode(GPIO_NUM_33, OUTPUT);
digitalWrite(GPIO_NUM_33, HIGH);
gpio_hold_en(GPIO_NUM_33);
esp_sleep_enable_timer_wakeup(2000000);
esp_deep_sleep_start();

Debug Message

M/A

Other Steps to Reproduce

N/A

I have checked existing issues, online documentation and the Troubleshooting Guide

• I confirm I have checked existing issues, online documentation and Troubleshooting guide.

[lbernstone]

I think you now need to call gpio_deep_sleep_hold_en?

[dynek]

Isn't this used to hold all rtc gpios in their current state? I want to to specify which ones, reason why I use gpio_hold_en.

[SuGlider]

@dynek - This is not an Arduino Issue... but I'm surprised it works with Arduino 1.0.6, because as @lbernstone said you must call gpio_deep_sleep_hold_en() as well, for any both Arduino Cores (1.0.6 and 2.0.0+).

Check the documentation:

Arduino 1.0.6 uses IDF 3.3:
https://espressif-docs.readthedocs-hosted.com/projects/esp-idf/en/v3.3.4/api-reference/peripherals/gpio.html#_CPPv412gpio_hold_en10gpio_num_t

Arduino 2.0.0+ uses IDF 4.4:
https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-reference/peripherals/gpio.html#_CPPv412gpio_hold_en10gpio_num_t

[SuGlider]

@dynek - check out this link #2712

[VojtechBartoska]

@dynek Do you still need help?

[dynek]

@VojtechBartoska I still need to give a try, I've been super busy lately.
I was under the impression (when I tried it some weeks ago) that gpio_deep_sleep_hold_en() would hold all RTC GPIO in their current state, not just the ones specified with gpio_hold_en().
Also what's weird in my case is that the GPIO is actually held in the right state, it's just switching back to its default state when the ESP comes back to life, after sleeping.
But I'll give it a try asap, thanks for the hints!

[VojtechBartoska]

@dynek thanks! Looking forward for your results.

[dynek]

Just tried it and it doesn't work better.
I tried with:

gpio_deep_sleep_hold_en();
gpio_hold_en(GPIO_NUM_33);
esp_sleep_enable_timer_wakeup(2000000);
esp_deep_sleep_start();

And:

gpio_hold_en(GPIO_NUM_33);
gpio_deep_sleep_hold_en();
esp_sleep_enable_timer_wakeup(2000000);
esp_deep_sleep_start();

Using gpio_deep_sleep_hold_en or not has the same effect.
GPIO is held in its current state, but as soon as the ESP restarts it goes LOW again.

[TD-er]

The documentation is a bit confusing here, but when looking at the gpio_hold_dis function, it looks like what you're seeing is intended behavior.

Not sure when the gpio_hold_dis is called, but maybe you can set the pin to the desired state in the setup() function and call gpio_hold_dis yourself right after?

What does seem contradicting in the documentation is that the gpio_hold_en is apparently not active during deep sleep, but should be activated right after deep sleep.
However the documentation of gpio_deep_sleep_hold_en does contradict this.
So it seems as if gpio_hold_dis is somehow called anyway when waking from deep sleep.