[Bug tree-optimization/127805] [17 Regression] ICE in in do_rpo_vn_1, at tree-ssa-sccvn.cc:8962 (since r17-5018-g50b1269d94bac5)

[burnus at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127805

Tobias Burnus <burnus at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[17 Regression] ICE in in   |[17 Regression] ICE in in
                   |do_rpo_vn_1, at             |do_rpo_vn_1, at
                   |tree-ssa-sccvn.cc:8962      |tree-ssa-sccvn.cc:8962
                   |                            |(since
                   |                            |r17-5018-g50b1269d94bac5)
                 CC|                            |alfierichards at gcc dot 
gnu.org,
                   |                            |ams at gcc dot gnu.org,
                   |                            |tnfchris at gcc dot gnu.org

--- Comment #3 from Tobias Burnus <burnus at gcc dot gnu.org> ---
Bisecting points at:

commit r17-5018-g50b1269d94bac5be3bbdfb88b07f370a4497e561
commit 50b1269d94bac5be3bbdfb88b07f370a4497e561
Author:     Alfie Richards
AuthorDate: Thu Oct 8 15:49:01 2026 +0000
Commit:     Alfie Richards
CommitDate: Fri Oct 9 11:48:32 2026 +0000

    fix: vect: Fix for avx512 and non-partial vectorization [PR 126301]

    I was incorrect previously thinking PR 126301 only affected partial
    vectors.

    This duplicates Tamars fix for the other two control flow branches.

            PR tree-optimization/126301

    gcc/ChangeLog:

            * tree-vect-loop-manip.cc (vect_use_loop_latch_condition_p):
            Remove restriction for avx512 and non-partial vectors.
            (vect_reuse_loop_latch): New function.
            (vect_set_loop_condition_partial_vectors_avx512):
            Add logic for do loop transformation.
            (vect_set_loop_condition_normal): Likewise.

    Co-Authored-By: Tamar Christina