Re: [PATCH v2 2/5] target/ppc: change xs[n]madd[am]sp to use float64r32_muladd

[Matheus K. Ferst]

On 03/03/2022 15:49, Richard Henderson wrote:
> On 3/3/22 07:20, matheus.ferst@eldorado.org.br wrote:
> > From: Matheus Ferst <matheus.ferst@eldorado.org.br>
> >
> > Change VSX Scalar Multiply-Add/Subtract Type-A/M Single Precision
> > helpers to use float64r32_muladd. This method should correctly handle
> > all rounding modes, so the workaround for float_round_nearest_even can
> > be dropped.
> >
> > Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
> > Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
> > ---
> >   target/ppc/fpu_helper.c | 93 ++++++++++++++++-------------------------
> >   1 file changed, 35 insertions(+), 58 deletions(-)
> >
> > diff --git a/target/ppc/fpu_helper.c b/target/ppc/fpu_helper.c
> > index 8f970288f5..c973968ed6 100644
> > --- a/target/ppc/fpu_helper.c
> > +++ b/target/ppc/fpu_helper.c
> > @@ -1916,22 +1916,19 @@ void helper_xsdivqp(CPUPPCState *env, uint32_t 
> > opcode,
> >    *   fld   - vsr_t field (VsrD(*) or VsrW(*))
> >    *   sfprf - set FPRF
> >    */
> > -#define VSX_RE(op, nels, tp, fld, sfprf, 
> > r2sp)                                \
> > +#define VSX_RE(op, nels, tp, op_tp, fld, 
> > sfprf)                               \
> >   void helper_##op(CPUPPCState *env, ppc_vsr_t *xt, ppc_vsr_t 
> > *xb)              \
> >   
> > {                                                                             
> > \
> >       ppc_vsr_t t = { 
> > };                                                        \
> > -    int 
> > i;                                                                    \
> > +    int i, 
> > flags;                                                             \
> >                                                                                 
> > \
> >       
> > helper_reset_fpstatus(env);                                               
> > \
> >                                                                                 
> > \
> >       for (i = 0; i < nels; i++) 
> > {                                              \
> > -        if (unlikely(tp##_is_signaling_nan(xb->fld, 
> > &env->fp_status))) {      \
> > +        t.fld = op_tp##_div(tp##_one, xb->fld, 
> > &env->fp_status);              \
> > +        flags = 
> > get_float_exception_flags(&env->fp_status);                   \
> > +        if (unlikely(flags & float_flag_invalid_snan)) {
>
> You seem to have squashed the change to recip-estimate into the same 
> patch as muladd.
>
>
> > -#define VSX_RSQRTE(op, nels, tp, fld, sfprf, 
> > r2sp)                           \
> > +#define VSX_RSQRTE(op, nels, tp, op_tp, fld, 
> > sfprf)                          \
>
> And recip-sqrt-estimate.
>
> I guess it's ok to squash, since it's all related, but you should update 
> the patch
> description if you leave it this way.

Sorry, I cherry-picked the wrong branch. This patch should just be a 
rebase of v1. I'll send the changes to 
VSX_{ADD_SUB,MUL,DIV,RE,SQRT,RSQRTE} in a separate patch series since 
it's not test-related.

Thanks,
Matheus K. Ferst
Instituto de Pesquisas ELDORADO <http://www.eldorado.org.br/>
Analista de Software
Aviso Legal - Disclaimer <https://www.eldorado.org.br/disclaimer.html>