https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128189
--- Comment #4 from GCC Commits <cvs-commit at gcc dot gnu.org> --- The master branch has been updated by Richard Biener <[email protected]>: https://gcc.gnu.org/g:0091900417570a0d70c9d382ec4491dbe12fb01f commit r17-5672-g0091900417570a0d70c9d382ec4491dbe12fb01f Author: Richard Biener <[email protected]> Date: Wed Oct 28 15:11:48 2026 +0100 tree-optimization/128189 - ICE with OMP SIMD vectorization of uniform arg When we cannot compute a vector type for an uniform arg of a SIMD function we'll ICE, latest when assigning the vector type to the SLP node if it is invariant. Give up for such uniform args. If passing the scalar argument would be OK we should refrain from tracking it in a SLP node (or generally enhance that to handle "scalar lanes"). PR tree-optimization/128189 * tree-vect-stmts.cc (vectorizable_simd_clone_call): When the vector type of an argument cannot be computed, fail. * gcc.dg/vect/vect-simd-pr128189.c: New testcase.
