Re: [PATCH v4 13/21] target/loongarch: Add floating point move instruction translation

[Richard Henderson]

On 9/2/21 2:40 PM, Song Gao wrote:
> +static bool gen_mov(DisasContext *ctx, arg_fmt_fdfj *a,
> +                    void (*func)(TCGv, TCGv))
> +{
> +    TCGv dest = cpu_fpr[a->fd];
> +    TCGv src = cpu_fpr[a->fj];

Maybe clearer as gen_f2f, to match the others?

> +static void gen_movfrh2gr_s(TCGv dest, TCGv src)
> +{
> +    TCGv t0 = tcg_temp_new();
> +
> +    tcg_gen_extract_tl(dest, src, 32, 32);
> +
> +    tcg_temp_free(t0);
> +}

t0 is unused.
Use sextract_tl, or sari_tl...

> +TRANS(movfr2gr_s, gen_f2r, EXT_NONE, tcg_gen_ext32s_tl)
> +TRANS(movfr2gr_d, gen_f2r, EXT_NONE, tcg_gen_mov_tl)
> +TRANS(movfrh2gr_s, gen_f2r, EXT_SIGN,  gen_movfrh2gr_s)

... which then means you don't need the EXT_SIGN here, and can drop the 
parameter entirely.


r~