Re: [PATCH RESEND v5 14/57] target/loongarch: Implement xvadd/xvsub

[gaosong]

在 2023/9/10 上午9:44, Richard Henderson 写道:
> On 9/7/23 01:31, Song Gao wrote:
> > --- a/target/loongarch/insn_trans/trans_vec.c.inc
> > +++ b/target/loongarch/insn_trans/trans_vec.c.inc
> > @@ -208,6 +208,16 @@ static bool gvec_vvv(DisasContext *ctx, arg_vvv 
> > *a, MemOp mop,
> >       return gvec_vvv_vl(ctx, a, 16, mop, func);
> >   }
> > +static bool gvec_xxx(DisasContext *ctx, arg_vvv *a, MemOp mop,
> > +                     void (*func)(unsigned, uint32_t, uint32_t,
> > +                                  uint32_t, uint32_t, uint32_t))
> > +{
> > +    if (!check_vec(ctx, 32)) {
> > +        return true;
> > +    }
> > +
> > +    return gvec_vvv_vl(ctx, a, 32, mop, func);
> > +}
>
> You can move check_vec into gvec_vvv_vl, removing it from gvec_vvv.
>
> > +static bool gen_vaddsub_q_vl(DisasContext *ctx, arg_vvv *a, uint32_t 
> > oprsz,
> > +                             void (*func)(TCGv_i64, TCGv_i64, TCGv_i64,
> > +                                          TCGv_i64, TCGv_i64, TCGv_i64))
> > +{
> > +    int i;
> > +    TCGv_i64 rh, rl, ah, al, bh, bl;
>
> Have check_vec here ...
>
> > +static bool gen_vaddsub_q(DisasContext *ctx, arg_vvv *a,
> > +                          void (*func)(TCGv_i64, TCGv_i64, TCGv_i64,
> > +                                       TCGv_i64, TCGv_i64, TCGv_i64))
> > +{
> > +    if (!check_vec(ctx, 16)) {
> > +        return true;
> > +    }
> > +
> > +    return gen_vaddsub_q_vl(ctx, a, 16, func);
> > +}
> > +
> > +static bool gen_xvaddsub_q(DisasContext *ctx, arg_vvv *a,
> > +                           void (*func)(TCGv_i64, TCGv_i64, TCGv_i64,
> > +                                        TCGv_i64, TCGv_i64, TCGv_i64))
> > +{
> > +    if (!check_vec(ctx, 32)) {
> > +        return true;
> > +    }
> > +    return gen_vaddsub_q_vl(ctx, a, 16, func);
> > +}
>
> ... instead of these two places.
Ok, I will correct all similar patches.

Thanks.
Song Gao