sys/string_utils: add strscpy()

[benpicco]

Contribution description

strncpy is known to be dangerous as it will truncate the string without inserting a terminating zero.
The replacement strlcpy had the problem that the return value had to be compared to the destination buffer to detect truncation.

In Linux the consensus now seems to have settled on strscpy() which returns -E2BIG if the string was truncated.

https://lwn.net/Articles/905777/

Testing procedure

Issues/PRs references

[maribu]

nice one!

[maribu]

Anyway, whether or not the assert(0) is there is bike-shedding. As long as the API says "do not call it with a zero-length dest buffer, as you won't get a zero terminated string in dest" the API is as safe to use as it gets with C.

Thanks for this addition! :)

[maribu]

The guard in

#if !defined(BOARD_NATIVE) \
    && !(IS_USED(MODULE_PICOLIBC) && __BSD_VISIBLE) \
&& !(IS_USED(MODULE_NEWLIB) && __BSD_VISIBLE && !defined(MCU_ESP8266))
[...]
static inline void explicit_bzero(void *dest, size_t n_bytes)
[...]
#endif

doesn't trigger tests_external_board_dirs for the custom board native2. I wonder if just adding an CFLAGS += -DBOARD_NATIVE to the test is a sensible workaround. Adding an && !defined(BOARD_NATIVE2) just for a test app seems to be the wrong approach to me.

[benpicco]

But CPU_NATIVE should work?