Serial.end() not reattaching pin to GPIO matrix

[luketanti]

After trying to switch back two UART pins to GPIO by using the Serial.end() function, the pins do not respond to digitalWrite commands.

I have found out that by using gpio_matrix_out(gpio_number, SIG_GPIO_OUT_IDX, false, false); I am able to use the pin again after ending the Serial.

[stickbreaker]

@luketanti after using Serial.end() did you do a pinMode(pin,OUTPUT) before doing a digitalWrite(pin,value)?

Chuck.

[luketanti]

@luketanti after using Serial.end() did you do a pinMode(pin,OUTPUT) before doing a digitalWrite(pin,value)?

Chuck.

Yes, offcourse.

Here are two snippets. The first one is when the issue occurs and the second is how I managed to get the pin to respond to digitalWrite commands.

  Serial2.end();
  pinMode(K_TX, OUTPUT);
  pinMode(K_RX, INPUT);
  digitalWrite(K_TX, HIGH);
  delay(298);
  digitalWrite(K_TX, LOW);
  delay(23);
  digitalWrite(K_TX, HIGH);
  delay(23);

  Serial2.begin(10417, SERIAL_8N1, K_RX, K_TX);  // K-LINE

  Serial2.end();
  gpio_matrix_out(K_TX, SIG_GPIO_OUT_IDX, false, false);
  pinMode(K_TX, OUTPUT);
  pinMode(K_RX, INPUT);
  digitalWrite(K_TX, HIGH);
  delay(298);
  digitalWrite(K_TX, LOW);
  delay(23);
  digitalWrite(K_TX, HIGH);
  delay(23);

  Serial2.begin(10417, SERIAL_8N1, K_RX, K_TX);  // K-LINE

[atanisoft]

Did you try gpio_reset(K_TX); instead of gpio_matrix_out?

[luketanti]

I have tried to compile with gpio_reset(K_TX); but it failed.

exit status 1
'gpio_reset' was not declared in this scope

Shouldn't Serial.end() automatically release the pins allocated to other use?

[atanisoft]

Sorry that should have been gpio_reset_pin. And yes I'd think that Serial.end() should release the pin to normal/default state. I believe there was an older issue for the same.

[luketanti]

Yes, I can confirm that it also works.

gpio_reset_pin(GPIO_NUM_17); in my case.

Can this line be added to the UART HAL?
And in future Arduino core updates this would be included?

[atanisoft]

@luketanti send in a PR that does this for both RX and TX in the esp32-hal-uart.c file in the uartEnd() method and I'm sure @me-no-dev will consider it.

[me-no-dev]

huh? this is rather really strange. this is already called in uartEnd() when the pins are detached from the matrix (gpio_matrix_out(gpio_number, SIG_GPIO_OUT_IDX, false, false);)

[me-no-dev]

• SIG_GPIO_OUT_IDX equals 0x100 (256). see here: https://github.com/espressif/esp-idf/blob/aaf12390eb14b95589acd98db5c268a2e56bb67e/components/soc/esp32/include/soc/gpio_sig_map.h#L421
• Arduino's function to detach from the matrix is here: https://github.com/espressif/arduino-esp32/blob/master/cores/esp32/esp32-hal-matrix.c#L28
• MATRIX_DETACH_OUT_SIG in Arduino also equals 0x100. see here: https://github.com/espressif/arduino-esp32/blob/master/cores/esp32/esp32-hal-matrix.c#L19
• The call to detach is coming from here: https://github.com/espressif/arduino-esp32/blob/master/cores/esp32/esp32-hal-uart.c#L127-L142
• And those are called in uartEnd() here: https://github.com/espressif/arduino-esp32/blob/master/cores/esp32/esp32-hal-uart.c#L242-L243
• uartEnd() is called by Arduino's HardwareSeral::end() here: https://github.com/espressif/arduino-esp32/blob/master/cores/esp32/HardwareSerial.cpp#L87

[stickbreaker]

@me-no-dev my 2 cents: The only difference I see between I2C(working) and UART de-init is a call to pinMode(pin,INPUT | PULLUP);

Has anyone other than the OP validated this behavior?

That is why I asked my silly question above.

Chuck.

[stickbreaker]

@luketanti do you call Serial.flush(); before you call Serial.end();?
Could it be a case where the UART is still in use (ISR active)?

Chuck.

[bertmelis]

flush blocks untill uart's FSM is idle.

[stickbreaker]

@bertmelis yep, that is its purpose. I believe reconfiguring UART should not happen until it is idle.

[stale]

[STALE_SET] This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 14 days if no further activity occurs. Thank you for your contributions.