[BUG] SKR2 serial partially working

[LawnMo]

Did you test the latest bugfix-2.0.x code?

Yes, and the problem still exists.

Bug Description

I have a BTT SKR2 board and an esp12s (btt RRF Wifi module for SKR2).

According to the motherboard documentation, there's a double jumper to select wether U-disk or Spi Wifi is enabled, and it mentions that "U-disk jumper makes the wifi module work only on UART, only on Marlin).
Which is why I decided to give the wifi module a try, I breadboarded the thing with 10k ohms and a serial-to-usb adapter and properly flashed it with ESP3D 2.1.1 (also flashed the "spiffs size detection" sketch, it has 4mb and RX/TX on the esp12s is working).
I then plugged it to the motherboard and adapted my config (serial_port{1/2} to 3, LCD screen doesn't show any M117 messages from the ESP (no IP address, nothing), but ESP3D terminal shows some serial messages from marlin :
if I use an usb flashdrive, plugged/mounted on the board, ESP3D won't get any messages from marlin, as soon as I eject/unplug my flashdrive, ESP3D gets spammed by "waiting for media...".
Which is when I tested different media settings, in case the flashdrive usb port has pins/timers/serial shared with the Wifi module header pins, when I use the onboard SD OR my 12864 SDcard reader OR multi_volume with #define DEFAULT_VOLUME SV_SD_ONBOARD I always get marlin's serial messages.

I then decided to look at what the ESP_xxx pins were attributed to in https://github.com/MarlinFirmware/Marlin/blob/bugfix-2.0.x/Marlin/src/pins/stm32f4/pins_BTT_SKR_V2_0_common.h#L516

Only :

#define ESP_WIFI_MODULE_RESET_PIN           PC14
#define ESP_WIFI_MODULE_GPIO0_PIN           PB10

seems to be used by marlin in https://github.com/MarlinFirmware/Marlin/blob/bugfix-2.0.x/Marlin/src/HAL/shared/esp_wifi.cpp
pin definition is missing ESP_WIFI_MODULE_GPIO2_PIN but that pin on the motherboard is physically hardwired to VCC so it's pulled up anyway (and the esp module boots/connect to wifi) so it doesn't look to be the issue.

(ESP-RX)  PD8
(ESP-TX)  PD9

should be working fine, but since ESP-TX (sending commands to marlin) doesn't work, I tried a few things found online :

• making ESP3D wait 10s (and 15s)
• powered by usb with nothing attached but the ESP, powered by 12v
• with/without usb cable and pronterface attached
• changing the SKR2 and ESP baudrate (125k, 250k, mismatched baudrate shows weird symbols)
• uncommenting ESP_WIFI_MODULE_RESET_PIN
• swapping SERIAL_PORT / PORT1 / PORT2 to -1, 1, 3 in all possible orders
• filling a bug report to ESP3D (luc-github seems to think it's a marlin issue)
• inverting jumpers from U-disk to Wifi, and retested most of the above points

pins_debugging / M43 shows PD8/PD9 (ESP rx/tx pins) uses "USART1..3" (I forgot to save M43's output, I might rebuild and post it if that's relevant).

SKR2 uses a script to create a generic platformio variant and I couldn't (didn't really try to) find its' pin map, which is where I got lost ^^.

Bug Timeline

new bug, SKR2 release

Expected behavior

working RX/TX serial between ESP and marlin

Actual behavior

ESP-RX / PD8 pin works (marlin serial messages shows in ESP3D's terminal), ESP-TX / PD9 pin doesn't.

Steps to Reproduce

1. any config that uses a SKR2 board
2. any SERIAL_PORTx set to 3
3. ESP12s with ESP3d 2.1.1 on it (default pio build, 160mhz)
4. plug the ESP into the SKR2's ESP header
5. no ESP-TX / esp3d can't send commands

Version of Marlin Firmware

bugfix-2.0.x

Printer model

diy

Electronics

BTT SKR2

Add-ons

BTT RRF Wifi module (an ESP12s)

Your Slicer

No response

Host Software

SD Card (headless)

Additional information & file uploads

cfg.zip

[alex2501]

I got the same issue with this combination and found the root cause. There is a conflict with pin PB11 which is defined for USART3 in variant too.

You can change line 217 in file Marlin/buildroot/share/PlatformIO/variants/MARLIN_F4x7Vx/PeripheralPins.c:
From
{PB_11, USART3, STM_PIN_DATA(STM_MODE_AF_PP, GPIO_PULLUP, GPIO_AF7_USART3)},
To
//{PB_11, USART3, STM_PIN_DATA(STM_MODE_AF_PP, GPIO_PULLUP, GPIO_AF7_USART3)},

For me it is working, but don't know if this would affect any other boards using this variant and/or if this is a correct fix.

[LawnMo]

Good catch! Thanks for sharing :)

I might wait for the confirmation that it's safe to apply, but it's nice to know my board is not DOA.
I guess it can be ifdef'd for SKR2 A&B so it doesn't affect other boards 👍

edit: might be better to open a pull request see where it goes, @alex2501

[rhapsodyv]

I don’t know why all the peripheral pins are commented out. There’s no harm letting everything uncommented.

[LawnMo]

I don’t know why all the peripheral pins are commented out. There’s no harm letting everything uncommented.

Did you mean the opposite or do you mean alex2501's workaround is coincidental ?

[rhapsodyv]

I mean that @alex2501 patch is right, but we should not needed it if the variant file was right. Someone commented those lines, but it should not be commented at first place.

[LawnMo]

still happens.

[thinkyhead]

I wonder if this could be related to #22181 …. In that case the MMU2 uses a serial port on an SKR board, and somehow this is messing with endstops.

[LawnMo]

Didn't test it more than homing and checking temperatures but : feb6d28 appears to fix it, thanks.

[github-actions]

This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.