Re: [PATCH v5 46/49] target/ppc: Implement xvcvbf16spn and xvcvspbf16 instructions

[Richard Henderson <richard.henderson@linaro.org>]

On 2/25/22 11:09, matheus.ferst@eldorado.org.br wrote:
> +void helper_XVCVSPBF16(CPUPPCState *env, ppc_vsr_t *xt, ppc_vsr_t *xb)
> +{
> +    ppc_vsr_t t = { };
> +    int i, status;
> +
> +    for (i = 0; i < 4; i++) {
> +        t.VsrH(2 * i + 1) = float32_to_bfloat16(xb->VsrW(i), &env->fp_status);
> +    }
> +
> +    status = get_float_exception_flags(&env->fp_status);
> +    if (unlikely(status & float_flag_invalid_snan)) {
> +        float_invalid_op_vxsnan(env, GETPC());
> +    }
> +
> +    *xt = t;
> +    do_float_check_status(env, GETPC());
> +}

Missing reset_fpstatus.  Otherwise.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~


PS: Which is reminding me that cleaning that up has been on the to-do list for a long 
time. We should be able to rely on env->fp_status.float_exception_flags being 0 between 
and at the start of each fp operation. In do_float_check_status, we would reset 
float_exception_flags in the expected unlikely case that it is ever non-zero.