On Tue, 24 Sep 2024, Tamar Christina wrote:
> > Can you explain how you get to see constant/external defs with
> > astmt_vec_info? That's somehow a violation of some inherentinvariant in
> > the vectorizer.
>
> I'm not sure I actually get any. It could be the condition is never hit
> with a stmt_vec_info. I had assumed however since the condition is part
> of a gimple_cond and if one of the arguments of the gimple_cond is loop
> bound, that the condition would be analyzed too.
>
> So if you're saying you never get a stmt_vec_info for invariants at this
> point (I assume you could see you see them in the corresponding slp
> tree) then maybe checking for the stmt_vec_info is enough.
>
> However, when I was looking around for how to check for externals I
> noticed other patterns also check for externals and constants. So I
> assumed that you could indeed get them.
You usually check that after doing vect_is_simple_use on the SSA name
or constant which internally makes all stmts with a stmt_vec_info
one of the internal def kinds.
So I guess you could do vect_is_simple_use on 'var' as well and check
the 'dt' it will populate
Richard.
> Kind regards,
> Tamar
>
>
> ________________________________
> From: Richard Biener <rguent...@suse.de>
> Sent: Tuesday, September 24, 2024 7:45 AM
> To: Tamar Christina <tamar.christ...@arm.com>
> Cc: gcc-patches@gcc.gnu.org <gcc-patches@gcc.gnu.org>; nd <n...@arm.com>;
> j...@ventanamicro.com <j...@ventanamicro.com>
> Subject: RE: [PATCH]middle-end: check explicitly for external or constants
> when checking for loop invariant [PR116817]
>
> On Mon, 23 Sep 2024, Tamar Christina wrote:
>
> > I had made the condition to strict before, here's an updated patch:
> >
> > Hi All,
> >
> > The previous check if a value was external was checking
> > !vect_get_internal_def (vinfo, var) but this of course isn't completely
> > right
> > as they could reductions etc.
> >
> > This changes the check to just explicitly look at externals and constants.
> > Note that reductions remain unhandled here, but we don't support codegen of
> > boolean reductions today anyway.
>
> Can you explain how you get to see constant/external defs with a
> stmt_vec_info? That's somehow a violation of some inherent
> invariant in the vectorizer.
>
> Richard.
>
> > So at the time we do then this would have the be handled as well in
> > lowering.
> >
> > Bootstrapped Regtested on aarch64-none-linux-gnu, arm-none-linux-gnueabihf,
> > x86_64-pc-linux-gnu -m32, -m64 and no issues.
> >
> > Ok for master?
> >
> > Thanks,
> > Tamar
> >
> > gcc/ChangeLog:
> >
> > PR tree-optimization/116817
> > * tree-vect-patterns.cc (vect_recog_bool_pattern): Check for const or
> > externals.
> >
> > gcc/testsuite/ChangeLog:
> >
> > PR tree-optimization/116817
> > * g++.dg/vect/pr116817.cc: New test.
> >
> > -- inline copy of patch --
> >
> > diff --git a/gcc/testsuite/g++.dg/vect/pr116817.cc
> > b/gcc/testsuite/g++.dg/vect/pr116817.cc
> > new file mode 100644
> > index
> > 0000000000000000000000000000000000000000..7e28982fb138c2cccc4f956aedb03fa454d9d858
> > --- /dev/null
> > +++ b/gcc/testsuite/g++.dg/vect/pr116817.cc
> > @@ -0,0 +1,16 @@
> > +/* { dg-do compile } */
> > +/* { dg-additional-options "-O3" } */
> > +
> > +int main_ulData0;
> > +unsigned *main_pSrcBuffer;
> > +int main(void) {
> > + int iSrc = 0;
> > + bool bData0;
> > + for (; iSrc < 4; iSrc++) {
> > + if (bData0)
> > + main_pSrcBuffer[iSrc] = main_ulData0;
> > + else
> > + main_pSrcBuffer[iSrc] = 0;
> > + bData0 = !bData0;
> > + }
> > +}
> > diff --git a/gcc/tree-vect-patterns.cc b/gcc/tree-vect-patterns.cc
> > index
> > e7e877dd2adb55262822f1660f8d92b42d44e6d0..f0298b2ab97a1e7dd0d943340e1389c3c0fa796e
> > 100644
> > --- a/gcc/tree-vect-patterns.cc
> > +++ b/gcc/tree-vect-patterns.cc
> > @@ -6062,12 +6062,15 @@ vect_recog_bool_pattern (vec_info *vinfo,
> > if (get_vectype_for_scalar_type (vinfo, type) == NULL_TREE)
> > return NULL;
> >
> > + stmt_vec_info var_def_info = vinfo->lookup_def (var);
> > if (check_bool_pattern (var, vinfo, bool_stmts))
> > var = adjust_bool_stmts (vinfo, bool_stmts, type, stmt_vinfo);
> > else if (integer_type_for_mask (var, vinfo))
> > return NULL;
> > else if (TREE_CODE (TREE_TYPE (var)) == BOOLEAN_TYPE
> > - && !vect_get_internal_def (vinfo, var))
> > + && (!var_def_info
> > + || STMT_VINFO_DEF_TYPE (var_def_info) == vect_external_def
> > + || STMT_VINFO_DEF_TYPE (var_def_info) ==
> > vect_constant_def))
> > {
> > /* If the condition is already a boolean then manually convert it
> > to a
> > mask of the given integer type but don't set a vectype. */
> >
>
> --
> Richard Biener <rguent...@suse.de>
> SUSE Software Solutions Germany GmbH,
> Frankenstrasse 146, 90461 Nuernberg, Germany;
> GF: Ivo Totev, Andrew McDonald, Werner Knoblich; (HRB 36809, AG Nuernberg)
>
--
Richard Biener <rguent...@suse.de>
SUSE Software Solutions Germany GmbH,
Frankenstrasse 146, 90461 Nuernberg, Germany;
GF: Ivo Totev, Andrew McDonald, Werner Knoblich; (HRB 36809, AG Nuernberg)
