This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch, vectorizer] Add checks that vectype exists - resubmitting
- From: "Richard Guenther" <richard dot guenther at gmail dot com>
- To: "Ira Rosen" <IRAR at il dot ibm dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Tue, 23 Oct 2007 13:02:20 +0200
- Subject: Re: [patch, vectorizer] Add checks that vectype exists - resubmitting
- References: <OFE099A637.15BD5951-ONC225737C.002A0B78-C225737C.002BDFD0@LocalDomain> <OF54811D23.3F9B5A0B-ONC225737D.003121B1-C225737D.00328924@il.ibm.com>
On 10/23/07, Ira Rosen <IRAR@il.ibm.com> wrote:
> Hi,
>
> Ira Rosen/Haifa/IBM wrote on 22/10/2007 09:59:13:
>
> > Hi,
> >
> > Following Uros's fix to PR tree-optimization/33742 (http://gcc.gnu.
> > org/ml/gcc-patches/2007-10/msg00707.html), checks that vectype
> > exists should be also added in the analysis of SLP. In
> > transformation, vectype must not be NULL, so I added gcc_asserts to
> > verify vectype existence.
>
> Some of the cases in tree-vect-transform.c occur during the analysis, so I
> removed gcc_asserts in such cases and added checks that vectype exists (and
> we now fail to vectorize, otherwise).
>
> Bootstrapped and tested on ppc-linux.
> OK for mainline?
This is ok.
Thanks,
Richard.
> Thanks,
> Ira
>
> ChangeLog:
>
> * tree-vect-analyze.c (vect_build_slp_tree): Return false if
> vectype cannot be determined for the scalar type.
> (vect_analyze_slp_instance): Likewise.
> * tree-vect-transform.c (vect_model_reduction_cost): Change the
> return type to bool. Return false if vectype cannot be determined
> for the scalar type.
> (vect_get_constant_vectors): Return false if vectype cannot be
> determined for the scalar type.
> (get_initial_def_for_induction, vect_get_vec_def_for_operand,
> get_initial_def_for_reduction, vect_create_epilog_for_reduction,
> vectorizable_reduction): Likewise.
>
>
> (See attached file: vectype.txt)
>