Nice!
I have no RP2040 yet to test it, so it will have to wait for that.
I have a few objections, though:

• the addition of __linux__ to the #ifdef checks in the async TCP code is not sensible in my opinion, as there will be no such support without completely rewriting the code.
• your code formatting preferences are different to ours, so all the replaced braces are irritating at least. I would ask you to maintain our formatting habits instead.
• there is a new #if HAS_FREERTOS section introduced in the middle of the RTU cpp code that I would prefer to protect the complete code.

Thanks for the quick reply.

• On code formatting, I did change it so that the braces match your formatting habits. This is what commit [2824eff] intends to do. My current diff does not show a difference. Could you tell me in which file you see a difference ?
• On __linux__ I've added all the platforms previously listed in the options.h file without checking if this would have worked on the platform. Happy to change it to defined(ESP32) || defined(ESP8266) if that that's the only platforms it works for.
• On #if HAS_FREERTOS, on second thought I believe this should be changed to #if ESP8266 || ESP32 as the code in this function is specific to these platforms.

My bad, sorry! I had a look only at the first commit.

Regarding those RTOS task calls, that could be valid for the RP2040 as well - IIRC there are two cores, too? On the ESP32 running the worker task on a dedicated core helps to maintain the bus speed at higher baud rates, the ESP32 can go up to 5Mbaud with it. The RTU server or client do not run on the ESP8266 for exactly the reason of having only one core.

The RP2040 has two cores indeed, however the function used xTaskCreatePinnedToCore is proprietary to Expressif. According to this, the "proper" RTOS way to do it is to set affinity. To mimic xTaskCreatePinnedToCore functionality I had to split it into two calls.

We are getting there. 😀
To swap out the Espressif task calls you are using the ARDUINO_ARCH... define. I would think the HAS_FREERTOS2040 would be more consistent instead. At one spot of these there is a isolated #elsif - shouldn't that be an #else?
Finally the addition of __linux__ still is in the async TCP files...

I will order a RP2040 for me to test. I would like it better to have the TCP parts also working on that MCU. Did you try those as well? I suppose your definition in options.h is protecting you from all the HAS_FREERTOS encapsulated code.

Excellent! Thank you so much. I expect to get the RP2040 on Saturday, so I may approve the PR beginning of next week.

• __linux__ removed.
• Agree on the platform vs RTOS flavour in terms of choosing the task creation method. I believe the flag HAS_FREERTOS should be named HAS_ESP_FREERTOS ;-) but that's a decision for you to make and probably too much refactoring.
• My RP2040 does not have wifi so I can't test the TCP part. I initially introduced a HAS_IP flag to distinguish between plaform and and capabilities but in the end thought this was too complicated.
• I'll craft a simple RP2040 example before the weekend so that you can test.

What are your recent changes about to achieve? Just renaming the config variable? If so, why is there no matching commit for options.h?

Just revisiting this project and cleaning up code on my end. Did not think about the side effect on this PR. I can remove it for the time being if you prefer.

@ecarjat Thank you for your contribution.