[Qemu-devel] [PATCH 02/14] target-mips: use the softfloat floatXX_muladd functions

[Tom de Vries]

Aurelien,

> @@ -8307,7 +8307,7 @@ static void gen_flt3_arith (DisasContext *ctx, uint32_t 
> opc,
>              gen_load_fpr64(ctx, fp0, fs);
>              gen_load_fpr64(ctx, fp1, ft);
>              gen_load_fpr64(ctx, fp2, fr);
> -            gen_helper_float_muladd_d(fp2, cpu_env, fp0, fp1, fp2);
> +            gen_helper_float_madd_d(fp2, cpu_env, fp0, fp1, fp2);
>              tcg_temp_free_i64(fp0);
>              tcg_temp_free_i64(fp1);
>              gen_store_fpr64(ctx, fp2, fd);

AFAIU:
- you're replacing here a non-fused mAC with a fused MAC and
- for all mips cores (except the r8000) the madd.d is non-fused.
So shouldn't we use a non-fused MAC here?

Thanks,
- Tom