Build-time TMC address conflict detection

[sjasonsmith]

Description

This checks for conflicts between TMC UART addresses at build time. This works for all supported axis, for both HW and SW serial.

Benefits

Provides conflict messages at build-time.

Sample hardware serial error:

Marlin\src\module\stepper\trinamic.cpp:843:31: error: static assertion failed: X_SLAVE_ADDRESS conflicts with another driver using the same X_HARDWARE_SERIAL
     #define SA_NO_TMC_HW_C(A) static_assert(no_tmc_hw_serial_conflict(TMC_HW_DETAIL_ARGS(A)), TMC_HWSERIAL_CONFLICT_MSG(A));
                               ^
Marlin\src\module\stepper\trinamic.cpp:844:5: note: in expansion of macro 'SA_NO_TMC_HW_C'
     SA_NO_TMC_HW_C(X);SA_NO_TMC_HW_C(X2);
     ^~~~~~~~~~~~~~

Sample SW serial error:

Marlin\src\module\stepper\trinamic.cpp:873:31: error: static assertion failed: X_SLAVE_ADDRESS conflicts with another driver using the same X_SERIAL_RX_PIN or X_SERIAL_TX_PIN
     #define SA_NO_TMC_SW_C(A) static_assert(no_tmc_sw_serial_conflict(TMC_SW_DETAIL_ARGS(A)), TMC_SWSERIAL_CONFLICT_MSG(A));
                               ^
Marlin\src\module\stepper\trinamic.cpp:874:5: note: in expansion of macro 'SA_NO_TMC_SW_C'
     SA_NO_TMC_SW_C(X);SA_NO_TMC_SW_C(X2);
     ^~~~~~~~~~~~~~

Configurations

N/A

Related Issues

N/A

[sjasonsmith]

This was inspired by @ellensp's attempt to solve this problem yesterday: #19445
That was a great start, but using simple #if and #error statements gets really huge when checking both HW/SW on all 16 steppers a user can define in Marlin.

The technique I used here allows a lot of flexibility, and doesn't require hundreds/thousands of #error statements.
It is really unfortunate that it isn't compatible with the compiler used for AVR, though. I marked this as a draft because I'd like to replace the constexpr for loop with a recursive constexpr algorithm, to see if it can be made to work on AVR, since we still have a lot of TMC users there.

[sjasonsmith]

I did this in trinamic.cpp rather than SanityChecks for a couple reasons:

1. This is a rather large build-time algorithm. It would be nice not to repeat it for every file.
2. This uses some #if/#define statements already in trinamic.cpp to make using TERN easier.

As we add more complex build-time checks relying on static_assert, perhaps it would be useful to make a SanityCheck.cpp. To get maximum benefit from that we would need to be able to force it to build first, to avoid sanity checking after almost all of Marlin has built.

[sjasonsmith]

I have reworked the constexpr functions so they will work with the GCC 5.x used by esp32 and AVR.

[sjasonsmith]

I used a GTR configuration with 11 drivers to play with various conflicts involving mixtures of HW/SW, serial port number, RX/TX pins, etc. In all cases the error messages reports seemed to accurately reflect the conflicts I had present in the config.

[sjasonsmith]

@thinkyhead I had escaped the + becaused it didn't work without it.

[thinkyhead]

Yep, turns out the macOS version of sed is throwing me off, but with gsed it works perfectly.

[mrschmiklz]

I am getting this error on compile. How should I approach the fix?

[sjasonsmith]

I am getting this error on compile. How should I approach the fix?

This means you have TMC2208 drivers sharing a serial port, but without unique SLAVE_ADDRESS definitions in your Configuration.h file. If you are using a board with built-in drivers such as an SKR Mini E3 V2.0, I suggest starting with one of the example configurations which already has these configured properly.

If you have a different setup and are not sure how to proceed, I suggest you seek help at one of the following support venues. I myself frequent the Discord channel.

• MarlinFW.org Marlin Documentation
• RepRap.org Marlin Forum
• Tom's 3D Forums
• Facebook Group "Marlin Firmware"
• Facebook Group "Marlin Firmware for 3D Printers"
• Marlin Configuration on YouTube
• Marlin Discord server. Join link: https://discord.gg/n5NJ59y