This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug tree-optimization/63341] [4.8/4.9/5 Regression] Vectorization miscompilation with -mcpu=power7
- From: "jakub at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Tue, 23 Sep 2014 11:22:08 +0000
- Subject: [Bug tree-optimization/63341] [4.8/4.9/5 Regression] Vectorization miscompilation with -mcpu=power7
- Auto-submitted: auto-generated
- References: <bug-63341-4 at http dot gcc dot gnu dot org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63341
--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I see non-NULL offset passed to vect_create_data_ref_ptr by vectorizable_store
(for negative case), vectorizable_load (for this
dr_explicit_realign_optimized),
and to vect_create_addr_base_for_vector_ref in vect_gen_niters_for_prolog_loop
(for negative case) and vect_create_cond_for_align_checks (also for negative
case). So guess we'd need to check all these cases what we really want in
those cases.