[Bug tree-optimization/66704] New: [6 Regression] ICE in tree-vectorizer at tree-ssanames.c:457

["marxin at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66704

            Bug ID: 66704
           Summary: [6 Regression] ICE in tree-vectorizer at
                    tree-ssanames.c:457
           Product: gcc
           Version: 6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: marxin at gcc dot gnu.org
  Target Milestone: ---

Hello.

Starting from r224514, there's a new issue spotted on ppc64le:

./xgcc -B. ../../gcc/testsuite/gcc.target/powerpc/pr60137.c -mcpu=power8 -O3 -c
-mno-vsx

../../gcc/testsuite/gcc.target/powerpc/pr60137.c:13:6: internal compiler error:
tree check: expected ssa_name, have addr_expr in copy_ssa_name_fn, at
tree-ssanames.c:457
 void init_reg_sets_1(void)
      ^
0x11114f8b tree_check_failed(tree_node const*, char const*, int, char const*,
...)
        ../../gcc/tree.c:9379
0x102748ab tree_check(tree_node*, char const*, int, char const*, tree_code)
        ../../gcc/tree.h:2858
0x11029a4f copy_ssa_name_fn(function*, tree_node*, gimple_statement_base*)
        ../../gcc/tree-ssanames.c:457
0x116685cf copy_ssa_name
        ../../gcc/tree-ssanames.h:116
0x11677963 vect_setup_realignment(gimple_statement_base*,
gimple_stmt_iterator*, tree_node**, dr_alignment_support, tree_node*, loop**)
        ../../gcc/tree-vect-data-refs.c:4854
0x1106dd77 vectorizable_load
        ../../gcc/tree-vect-stmts.c:6548
0x11071717 vect_transform_stmt(gimple_statement_base*, gimple_stmt_iterator*,
bool*, _slp_tree*, _slp_instance*)
        ../../gcc/tree-vect-stmts.c:7506
0x1108d0ab vect_transform_loop(_loop_vec_info*)
        ../../gcc/tree-vect-loop.c:6238
0x110ad3a7 vectorize_loops()
        ../../gcc/tree-vectorizer.c:495
0x10f2046b execute
        ../../gcc/tree-ssa-loop.c:289

Problem can reproduced on gcc112.fsffrance.org machine.

Thanks,
Martin