[Bug tree-optimization/87974] [9 Regression] ICE in vect_get_vec_def_for_stmt_copy

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87974

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
No, it's different.  We somehow fail to make a 2nd PHI in the vectorization of
the inner loop.