Re: [Qemu-devel] [Qemu-arm] [PATCH v2 46/67] target/arm: Implement SVE load and broadcast quadword

[Peter Maydell]

On 17 February 2018 at 18:23, Richard Henderson
<address@hidden> wrote:
> Signed-off-by: Richard Henderson <address@hidden>
> ---
>  target/arm/translate-sve.c | 51 
> ++++++++++++++++++++++++++++++++++++++++++++++
>  target/arm/sve.decode      |  9 ++++++++
>  2 files changed, 60 insertions(+)

> +static void trans_LD1RQ_zprr(DisasContext *s, arg_rprr_load *a, uint32_t 
> insn)
> +{
> +    TCGv_i64 addr;
> +    int msz = dtype_msz(a->dtype);
> +
> +    if (a->rm == 31) {
> +        unallocated_encoding(s);
> +        return;
> +    }
> +
> +    addr = new_tmp_a64(s);
> +    tcg_gen_shli_i64(addr, cpu_reg(s, a->rm), msz);
> +    tcg_gen_add_i64(addr, addr, cpu_reg_sp(s, a->rn));
> +    do_ldrq(s, a->rd, a->pg, addr, msz);
> +}
> +
> +static void trans_LD1RQ_zpri(DisasContext *s, arg_rpri_load *a, uint32_t 
> insn)
> +{
> +    TCGv_i64 addr = new_tmp_a64(s);
> +
> +    tcg_gen_addi_i64(addr, cpu_reg_sp(s, a->rn), a->imm * 16);

It confused me initially here that the calculation of the offset
for the +immediate and the +scalar cases isn't the same, but that
is indeed what the architecture does. Maybe
       /* Unlike LD1RQ_zprr, offset scaling is constant rather
        * than based on msz.
        */
?

> +    do_ldrq(s, a->rd, a->pg, addr, dtype_msz(a->dtype));
> +}
>

Otherwise
Reviewed-by: Peter Maydell <address@hidden>

thanks
-- PMM