[Bug ipa/110334] [13/14 Regresssion] unused functions not eliminated before LTO streaming

[hubicka at ucw dot cz via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110334

--- Comment #9 from Jan Hubicka <hubicka at ucw dot cz> ---
Just so it is somewhere, here is a testcase that we can't inline leaf
functions to always_inlines unless we do some tracking of what calls
were formerly indirect calls.

We really overloaded always_inline from the original semantics "drop
inlining heuristics" into "be sure that result is inlined" while for
the second it does not make sense to take its address.
Clang apparently simply does not error on failes always inlines which
makes its life easier.

int n;
typedef void (*fnptr)();
fnptr get_me();
__attribute__ ((always_inline))
inline void test(void)
{
        if (n < 10)
          (get_me())();
        n++;
        return;
}
fnptr get_me()
{
        return test;
}
void
foo()
{
        test();
}