https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126926
--- Comment #2 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:a9c22e6d049bf10e44b6a76384a1eb00991c3eb8 commit r17-3364-ga9c22e6d049bf10e44b6a76384a1eb00991c3eb8 Author: Richard Biener <[email protected]> Date: Tue Aug 18 11:02:12 2026 +0200 tree-optimization/126926 - back-to-back SLP instance scheduling When we have two SLP instances interacting at the c_13 definition point of a vector which is both a vector CTOR root for one instance and the leaf (existing vector) of a reduction instance, we transform the CTOR instance first replacing the definition in vectorize_slp_instance_root_stmt. That disrupts the SLP scheduling process of the 2nd as the replacement now appears to be outside of the SLP region. Avoid this by not replacing the vector constructing stmt but only its RHS. PR tree-optimization/126926 * tree-vect-slp.cc (vectorize_slp_instance_root_stmt): Only replace the RHS of the vector constructing scalar stmt. * gcc.dg/vect/bb-slp-pr126926.c: New testcase.
