Re: [PATCH 07/55] target/arm: Implement MVE WLSTP insn

[Richard Henderson]

On 6/7/21 9:57 AM, Peter Maydell wrote:
> +    WLS          1111 0 0000 100     rn:4 1100 . .......... 1 imm=%lob_imm 
> size=4
> +    {
> +      # This is WLSTP
> +      WLS        1111 0 0000 0 size:2 rn:4 1100 . .......... 1 imm=%lob_imm
> +      LE         1111 0 0000 0 f:1 0 1111 1100 . .......... 1 imm=%lob_imm
> +    }

I guess it doesn't matter, but I'd swap these two, as LE is the more specific 
encoding.

> @@ -8148,10 +8152,40 @@ static bool trans_WLS(DisasContext *s, arg_WLS *a)
>            */
>           return false;
>       }
> +    if (a->size != 4) {
> +        /* WLSTP */
> +        if (!dc_isar_feature(aa32_mve, s)) {
> +            return false;
> +        }
> +        /*
> +         * We need to check that the FPU is enabled here, but mustn't
> +         * call vfp_access_check() to do that because we don't want to
> +         * do the lazy state preservation in the "loop count is zero" case.
> +         * Do the check-and-raise-exception by hand.
> +         */
> +        if (s->fp_excp_el) {
> +            gen_exception_insn(s, s->pc_curr, EXCP_NOCP,
> +                               syn_uncategorized(), s->fp_excp_el);
> +        }

Surely return true here...

> +    if (a->size != 4) {
> +        /*
> +         * WLSTP: set FPSCR.LTPSIZE. This requires that we do the
> +         * lazy state preservation, new FP context creation, etc,
> +         * that vfp_access_check() does. We know that the actual
> +         * access check will succeed (ie it won't generate code that
> +         * throws an exception) because we did that check by hand earlier.
> +         */
> +        bool ok = vfp_access_check(s);
> +        assert(ok);

... otherwise this assert will trigger.


r~
r~