Devicetree: Macro to initialize driver for every device instance

[io-pa]

This pull request introduces the macro: DT_INST_FOREACH(inst_expr).

This makes it possible to do this:
DT_INST_FOREACH(I2C_NRFX_TWIM_DEVICE);

Instead of:

#ifdef DT_INST_N_0_nordic_nrf_twim
I2C_NRFX_TWIM_DEVICE(0);
#endif

#ifdef DT_INST_N_1_nordic_nrf_twim
I2C_NRFX_TWIM_DEVICE(1);
#endif

#ifdef DT_INST_N_2_nordic_nrf_twim
I2C_NRFX_TWIM_DEVICE(2);
#endif

#ifdef DT_INST_N_3_nordic_nrf_twim
I2C_NRFX_TWIM_DEVICE(3);
#endif

The macro calls the passed inst_expr for every compatible (and enabled) device in the device tree.
By using this approach the drivers don't have to make assumption about the maximum count of device instances anymore.
The expr argument can be either a macro or a function that take the instance number as argument.

[zephyrbot]

All checks are passing now.

checkpatch (informational only, not a failure)

-:62: WARNING:TRAILING_SEMICOLON: macros should not use a trailing semicolon
#62: FILE: include/devicetree.h:1451:
+#define DT_CALL_WITH_ARG(arg, expr) expr(arg);

- total: 0 errors, 1 warnings, 74 lines checked

NOTE: For some of the reported defects, checkpatch may be able to
      mechanically convert to the typical style using --fix or --fix-inplace.

Your patch has style problems, please review.

NOTE: Ignored message types: AVOID_EXTERNS BRACES CONFIG_EXPERIMENTAL CONST_STRUCT DATE_TIME FILE_PATH_CHANGES MINMAX NETWORKING_BLOCK_COMMENT_STYLE PRINTK_WITHOUT_KERN_LEVEL SPLIT_STRING VOLATILE

NOTE: If any of the errors are false positives, please report
      them to the maintainers.

Tip: The bot edits this comment instead of posting a new one, so you can check the comment's history to see earlier messages.

[stephanosio]

Thanks for submitting this. This will be very useful for initialising multiple device instances.

[mbolivar-nordic]

I think ideally this would be rebased on #23245 and added to devicetree.h instead. @galak?

[galak]

Generally looks go to me. Will let @mbolivar-nordic and @nordic-krch comment on the UTIL_PASS_ARG_TO_MACRO bits.

[gmarull]

This macro will be really helpful, thanks for this constribution. @io-pa it would be helpful to add some documentation/example.

[mbolivar-nordic]

Looks good to me

I think you could define UTIL_PASS_ARG_TO_MACRO as DT_PASS_ARG_TO_MACRO_INTERNAL at the end of devicetree.h to avoid having to extend the util.h API for this.

Can you please also:

• update doc/guides/dts/howtos.rst to use this in the code example in the "Create struct devices in a driver" section? IMO this should be the standard way to define devices now, whenever possible.
• add a test case for DT_INST_FOREACH to tests/lib/devicetree/src/main.c? Every macro in devicetree.h has at least one test case in that file. One idea would be to pass TEST_GPIO_INIT to DT_INST_FOREACH instead of the way it's done now.

[mbolivar-nordic]

Might want to change the commit log's "device.h" to "devicetree.h" if you get a chance also. Not a blocker.

[io-pa]

I changed the helper macro to add ; at the end, since it's now specific to DT_INST_FOREACH.
That way it supports also functions next to macros and macros that need to be ;-terminated.

[mbolivar-nordic]

Thank you!

[mbolivar-nordic]

Just checking -- this does the right thing if DT_NUM_INST(DT_DRV_COMPAT) is 0, right?

[io-pa]

Uses LISTIFY internally, and that uses REPEAT and REPEAT checks wether the variable is 0 in the beginning so should be fine. I just tested it though to be sure and it looks good.

[mbolivar-nordic]

Thanks for confirming.

[mbolivar-nordic]

Shippable failure looks ignorable.

[galak]

ignoring the shippable failure as its not related.