Re: [Qemu-devel] [PATCH 05/10] target-arm: A64: Add "Floating-point data-processing (3 source)" insns

[Richard Henderson]

On 12/28/2013 01:49 PM, Peter Maydell wrote:
> +    /* These are fused multiply-add, and must be done as one
> +     * floating point operation with no rounding between the
> +     * multiplication and addition steps.
> +     * NB that doing the negations here as separate steps is
> +     * correct : an input NaN should come out with its sign bit
> +     * flipped if it is a negated-input.
> +     */
> +    if (o1 == true) {
> +        gen_helper_vfp_negs(tcg_op3, tcg_op3);
> +    }
> +
> +    if (o0 != o1) {
> +        gen_helper_vfp_negs(tcg_op1, tcg_op1);
> +    }
> +
> +    gen_helper_vfp_muladds(tcg_res, tcg_op1, tcg_op2, tcg_op3, fpst);

Do you really get the wrong result passing the appropriate
float_muladd_negate_* constant?

I realize that the A32 target doesn't do this either, but shouldn't that be
cleaned up?  It's certainly confusing to the reader, wondering why they aren't
used...


r~