Remove ESP_CONSOLE_UART from sdkconfig for 666 S3 Display panels

[RetiredWizard]

This fixes #9569 boot loop for the Hacktablet and Makerfabs 7" TFT ESP32-S3 666 Dotclock display boards.

It looks to me like in addition to the Hacktablet and Makerfabs TFT 7" boards, the Espressif ESP32S3_LCD_EV and possibly (I don't know if non-S3 boards are affected) the Lilygo twatch_2020_v3 may be affected by this issue. I don't have either of those devices to test with but could add the same sdkconfig changes for those boards if that makes sense?

[jepler]

Thanks for submitting this! I have some questions though...

• Didn't this configuration work once upon a time?
• Does removing the sdk uart configuration have any downsides, like preventing visibility of early boot messages?
• I found 4 boards with CONFIG_ESP_CONSOLE_UART=y in sdkconfig but this PR changes 2. Any reason to leave the other two unchanged?

[RetiredWizard]

• Didn't this configuration work once upon a time?

Yes it did, the boot loop didn't occur in 9.1.2/9.1.3.

• Does removing the sdk uart configuration have any downsides, like preventing visibility of early boot messages?

On the two boards I have/tested, I was still able to connect to the REPL over the UART but you are right, the early boot messages are no longer visible 😢

• I found 4 boards with CONFIG_ESP_CONSOLE_UART=y in sdkconfig but this PR changes 2. Any reason to leave the other two unchanged?

Since I didn't know or track down the reason for the change, I went back and forth about adding the other boards since I couldn't test them as well, but I certainly could add them to the PR.

[dhalbert]

The early boot messages are no longer visible

I think this is fine, because this is just noise to most users anyway. I would prefer these be suppressed for ordinary use.

[RetiredWizard]

I would prefer these be suppressed for ordinary use.

Okay, but remember that at least on the boards I'm testing, their are two USB connectors, for ordinary use you would connect to the connector labeled USB which brings up the CIRCUITPY drive and shows up as /dev/ttyACM0, that connection doesn't show the early boot messages anyway. If you're connecting to the UART connector /dev/ttyUSB0 you don't get the CIRCUITPY drive and you're probably doing some debugging. That's the connection that I assume is affected by this PR.

EDIT Actually on further testing, the USB connector (ACM0, not affected by this PR) does show some of the early boot messages but (using tio) the connection disconnects almost immediately after starting the early boot log (cutting off most of the output) and then reconnects at which point only the code.py output is displayed.

[dhalbert]

@RetiredWizard Ah, sorry, I misunderstood. Maybe I am seeing the boot messages on the non-full-USB boards.

[RetiredWizard]

@RetiredWizard Ah, sorry, I misunderstood. Maybe I am seeing the boot messages on the non-full-USB boards.

Now I'm not sure, I did some retesting and there is a smaller set of the boot messages on the normal port.... But I don't think that output is impacted by this PR.

[dhalbert]

I would merge this but I think it's worth seeing which other similar boards still have ESP_CONSOLE_UART.

[RetiredWizard]

It looks to me like in addition to the Hacktablet and Makerfabs TFT 7" boards, the Espressif ESP32S3_LCD_EV and possibly (I don't know if non-S3 boards are affected) the Lilygo twatch_2020_v3 may be affected by this issue. I don't have either of those devices to test with but I'm happy to update the PR to include the same changes for those two boards as well.

I would feel better about changing the boards I can't test if I had an idea why this change is necessary for 9.2.x and not 9.1.2.

[RetiredWizard]

I've been testing #9634 using main (9.2.0-alpha.2351-39-g1e36e05380-dirty on 2024-09-16; MakerFabs-ESP32-S3-Parallel-TFT-With-Touch-7inch with ESP32S3) as a base and haven't been seeing the boot looping that this was supposed to address. I'm not sure why but I'll do some more testing to see if this has been fixed by something since I was seeing the issue.

[RetiredWizard]

Okay, I'm not seeing the issue anymore using Absolute Newest:

Adafruit CircuitPython 9.2.0-alpha.2351-37-g015a5cf922 on 2024-09-16; MakerFabs-ESP32-S3-Parallel-TFT-With-Touch-7inch with ESP32S3
Adafruit CircuitPython 9.2.0-alpha.2351-37-g015a5cf922 on 2024-09-16; ESP32-S3-DevKitC-1-N8R8-with-HACKTABLET with ESP32S3

[dhalbert]

@RetiredWizard Could you make the same change on espressif_esp32s3_lcd_ev? That board also specifies GPIO43 and GPIO44. However, the LILYGO watch board uses special UART pins, so I'd be inclined to leave that alone.

I would have made the same change myself but you have disallowed upstream maintainers from pushing to your repo.

[RetiredWizard]

While I'm at it, is circuitpy a good LWIP hostname for the espressif_esp32s3_lcd_ev board?

# LWIP
#
CONFIG_LWIP_LOCAL_HOSTNAME="circuitpy"
# end of LWIP

[RetiredWizard]

I just noticed that #9608 adds a new version 1.5 of the s3-lcd-ev board and in that PR Scott suggested along with all the lines this PR is removing from sdkconfig, the CONFIG_LWIP_LOCAL_HOSTNAME can be dropped as well.

Lots of boards have that configured, but I could go through and cut it from all the S3 boards as long as we're updating these three?

[dhalbert]

the CONFIG_LWIP_LOCAL_HOSTNAME can be dropped as well.

That is OK too, there are plenty of boards without that. It doesn't serve a lot of purpose.

[RetiredWizard]

Okay, I'll put the removal of the LWIP hostnames from S3 boards in a new PR so it doesn't hold this one up.

[dhalbert]

Thanks!