Re: [PATCH, rs6000] Fix tests that are failing in gcc.target/powerpc/bfp with -m32

[Segher Boessenkool <segher at kernel dot crashing dot org>]

Hi!

On Fri, Apr 13, 2018 at 03:15:09PM -0500, Kelvin Nilsen wrote:
> Twelve failures have been occuring in the bfp test directory during -m32
> regression testing.

> This patch:
> 
> 1. Changes the expected error messages in certain test programs.
> 
> 2. Disables certain test programs from being exercised on 32-bit targets.
> 
> 3. Adds a "note" error message to explain the mapping from overloaded
> built-in functions
> to non-overloaded built-in functions.

Ah, very good plan!

> @@ -6932,14 +6933,20 @@
>          && rs6000_builtin_type_compatible (types[1], desc->op2))
>        {
>          if (rs6000_builtin_decls[desc->overloaded_code] != NULL_TREE)
> -          return altivec_build_resolved_builtin (args, n, desc);
> +          {
> +        result = altivec_build_resolved_builtin (args, n, desc);
> +        /* overloaded_code is set above */
> +        if (!rs6000_builtin_is_supported_p (overloaded_code))
> +          unsupported_builtin = true;
> +        else
> +          return result;
> +          }
>          else
>            unsupported_builtin = true;
>        }
>        }

The indentation (here and elsewhere) is off; probably because of the way
you sent this?  Please check.

> +    else
> +      /* on 64-bit, i seem to end up here */
> +      error ("builtin function %qs not supported in this compiler "
> +         "configuration", name);

I think you don't want that comment?

> +    /* If an error-representing  result tree was returned from
> +       altivec_build_resolved_builtin above, use it.  */
> +    return (result != NULL)? result: error_mark_node;

Spaces before ? and :.  x != 0 as a condition is pretty silly btw, but
that's just style ;-)

Okay for trunk with those nits fixed.  Thanks!


Segher