RFC: Should the MicroPython RP2 port implement a PICO_PANIC_FUNCTION?

[Gadgetoid]

For longer than I'd care to admit I let Pico SDK functions raise a panic and basically lock up MicroPython with no useful output. This happens when attempting to claim DMA channels when they're exhausted, and for all sorts of other reasons that are easy to hit when re-running C/C++ device driver code upon soft reset.

Defining PICO_PANIC_FUNCTION to print out the error using mp_plat_print and sit in a loop reliably (as far as I can tell) gets the panic message out over USB serial and has been a time saver for chasing down hardware resource related bugs/lockups.

I put this in the board config:

#define PICO_PANIC_FUNCTION mp_pico_panic

And this... somewhere:

void mp_pico_panic(const char *fmt, ...) {
    va_list args;

    if (fmt) {
        va_start(args, fmt);
        int ret = mp_vprintf(&mp_plat_print, fmt, args);
        (void)ret;
        va_end(args);
    }

    while (1) {
        mp_event_handle_nowait();
        sleep_ms(1);
    }
}

Should this be included in the RP2 port by default? I don't believe it has any drawbacks. It would probably be ideal if it supplemented the format string with "PANIC: " to make it clear why the board has stopped responding.

Is there any precedent for this sort of thing in other ports I could follow?

The standard panic implementation is here: https://github.com/raspberrypi/pico-sdk/blob/a1438dff1d38bd9c65dbd693f0e5db4b9ae91779/src/rp2_common/pico_platform_panic/panic.c#L64-L83