Re: [PATCH 07/15] Hexagon (target/hexagon) use softfloat default NaN and tininess

[Richard Henderson]

On 3/24/21 8:50 PM, Taylor Simpson wrote:
> @@ -180,6 +183,8 @@ static FloatParts parts_silence_nan(FloatParts a, 
> float_status *status)
>   #if defined(TARGET_HPPA)
>       a.frac &= ~(1ULL << (DECOMPOSED_BINARY_POINT - 1));
>       a.frac |= 1ULL << (DECOMPOSED_BINARY_POINT - 2);
> +#elif defined(TARGET_HEXAGON)
> +    return parts_default_nan(status);

This part is incorrect.  You should be changing no_signaling_nans above, 
because:

4.3.11.2 Floating-Point Representation

... The ISA does not include IEEE754 signaling NaNs ...


r~