Re: [Qemu-devel] [PATCH v3 14/16] target-m68k: add explicit single and double precision operations

[Richard Henderson]

On 02/07/2017 11:59 AM, Laurent Vivier wrote:
> +    case 0: /* fmove */
> +        break;
> +    case 0x40: /* fsmove */
> +        gen_helper_redf32_FP0(cpu_env);
> +        gen_helper_extf32_FP0(cpu_env);
> +        break;
> +    case 0x44: /* fdmove */
> +        gen_helper_redf64_FP0(cpu_env);
> +        gen_helper_extf64_FP0(cpu_env);
>          break;

This is going to produce double-rounding errors.  Better to properly set the 
rounding precision first and convert once.


r~