Coupling between the time() C function and the underlying hardware

Question

This is more a general question. How is the coupling done between the time() c-function and the underlying hardware? Suppose that you compile for a microcontroller not containing any RTC, what value do you get back? Or if you use your own RTC that keeps running with battery on a separate chip? How do you update time() after repowering? Is there any method for redirecting?

I did a trial in Mbed with RTC and time() function with MCU_STM32U575xG.

See also https://forums.mbed.com/t/time-null-of-rtc-example-not-synchronizing-with-rtc-when-using-battery/22248/10

Answer 1

How is the coupling done between the time() c-function and the underlying hardware?

There is no coupling "done". You have to "do it".

what value do you get back?

I would expect a linking error, or maybe I would expect time() to return (time_t)-1 with errno set to ENOSYS.

Returning -1 and ENOSYS is the default nosys implementation of newlib C standard library implementation so commonly used with GNU arm-none-eabi-gcc compiler toolchain here https://github.com/bminor/newlib/blob/master/libgloss/libnosys/gettod.c#L18 .

I did a trial in Mbed with RTC and time() function with MCU_STM32U575xG.

In the case of mbed-os, time function is implemented here https://github.com/mbed-ce/mbed-os/blob/master/platform/source/mbed_rtc_time.cpp#L120 . The specifics of the function and what _rtc_read() function calls is configured with mbed-os configuration. You have to configure mbed-os and provide it with the configuration and compile with mbed-os.

Bottom line, there is no coupling. You have to make the coupling if you want it.