Re: [musl] [PATCH 02/14] time64: Don't make aliases to nonexistent syscalls

[Rich Felker <dalias@libc.org>]

On Thu, Sep 03, 2020 at 07:22:57AM -0400, Stefan O'Rear wrote:
> riscv32 and future architectures lack the _time32 variants entirely, so
> don't try to use their numbers.
> ---
>  src/internal/syscall.h | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/src/internal/syscall.h b/src/internal/syscall.h
> index d5f294d4..66fc4e5c 100644
> --- a/src/internal/syscall.h
> +++ b/src/internal/syscall.h
> @@ -201,6 +201,7 @@ static inline long __alt_socketcall(int sys, int sock, int cp, long a, long b, l
>  #define SYS_sendfile SYS_sendfile64
>  #endif
>  
> +#ifdef SYS_timer_settime32
>  #ifndef SYS_timer_settime
>  #define SYS_timer_settime SYS_timer_settime32
>  #endif
> @@ -240,6 +241,7 @@ static inline long __alt_socketcall(int sys, int sock, int cp, long a, long b, l
>  #ifndef SYS_settimeofday
>  #define SYS_settimeofday SYS_settimeofday_time32
>  #endif
> +#endif

The existing expectation internally in musl is that archs that lack
legacy time32 syscalls have both the unadorned and _time64 macros
defined to the same value. The public headers don't have to be like
that but the internal ones do. See arch/x32/syscall_arch.h which does
it but in the opposite direction. This logic is all tested for x32 and
a big part of the point of that was preparing for rv32 and future
32-bit archs.

I'm actually missing how this patch worked as-written, since for
example timer_settime.c unconditionally assumes SYS_timer_settime is
defined, and if it's defined to anything other than the same value as
the time64 version, it will use the value as a fallback.

Rich