[Bug tree-optimization/106963] New: [13 Regression] ICE in vect_gen_perm_mask_checked, at tree-vect-stmts.cc:8606

[Bug tree-optimization/106963] New: [13 Regression] ICE in vect_gen_perm_mask_checked, at tree-vect-stmts.cc:8606

[asolokha at gmx dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106963

            Bug ID: 106963
           Summary: [13 Regression] ICE in vect_gen_perm_mask_checked, at
                    tree-vect-stmts.cc:8606
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: asolokha at gmx dot com
                CC: liuhongt at gcc dot gnu.org
  Target Milestone: ---
            Target: x86_64-unknown-linux-gnu

gcc 13.0.0 20220911 snapshot (g:0ea5e3f4542832b8da016b152695e64a2a386309) ICEs
when compiling the following testcase, reduced from
gcc/testsuite/gcc.target/i386/pr103144-neg-1.c, w/ -mavx -O2:

void
foo_neg_const (int *a)
{
  int i, b = 1;

  for (i = 0; i < 1000; i++)
    {
      a[i] = b;
      b = -b;
    }
}

% x86_64-unknown-linux-gnu-gcc-13.0.0 -mavx -O2 -c l9k774zu.c
during GIMPLE pass: vect
l9k774zu.c: In function 'foo_neg_const':
l9k774zu.c:2:1: internal compiler error: in vect_gen_perm_mask_checked, at
tree-vect-stmts.cc:8606
    2 | foo_neg_const (int *a)
      | ^~~~~~~~~~~~~
0x86e97b vect_gen_perm_mask_checked(tree_node*, vec_perm_indices const&)
       
/var/tmp/portage/sys-devel/gcc-13.0.0_p20220911/work/gcc-13-20220911/gcc/tree-vect-stmts.cc:8606
0x1177c4b vect_create_nonlinear_iv_init
       
/var/tmp/portage/sys-devel/gcc-13.0.0_p20220911/work/gcc-13-20220911/gcc/tree-vect-loop.cc:8360
0x1177c4b vectorizable_nonlinear_induction
       
/var/tmp/portage/sys-devel/gcc-13.0.0_p20220911/work/gcc-13-20220911/gcc/tree-vect-loop.cc:8803
0x1179a67 vectorizable_induction(_loop_vec_info*, _stmt_vec_info*, gimple**,
_slp_tree*, vec<stmt_info_for_cost, va_heap, vl_ptr>*)
       
/var/tmp/portage/sys-devel/gcc-13.0.0_p20220911/work/gcc-13-20220911/gcc/tree-vect-loop.cc:8943
0x1ef3690 vect_transform_stmt(vec_info*, _stmt_vec_info*,
gimple_stmt_iterator*, _slp_tree*, _slp_instance*)
       
/var/tmp/portage/sys-devel/gcc-13.0.0_p20220911/work/gcc-13-20220911/gcc/tree-vect-stmts.cc:11319
0x1186e81 vect_transform_loop(_loop_vec_info*, gimple*)
       
/var/tmp/portage/sys-devel/gcc-13.0.0_p20220911/work/gcc-13-20220911/gcc/tree-vect-loop.cc:10643
0x11c99c1 vect_transform_loops
       
/var/tmp/portage/sys-devel/gcc-13.0.0_p20220911/work/gcc-13-20220911/gcc/tree-vectorizer.cc:1007
0x11c9fff try_vectorize_loop_1
       
/var/tmp/portage/sys-devel/gcc-13.0.0_p20220911/work/gcc-13-20220911/gcc/tree-vectorizer.cc:1153
0x11c9fff try_vectorize_loop
       
/var/tmp/portage/sys-devel/gcc-13.0.0_p20220911/work/gcc-13-20220911/gcc/tree-vectorizer.cc:1183
0x11ca3b4 execute
       
/var/tmp/portage/sys-devel/gcc-13.0.0_p20220911/work/gcc-13-20220911/gcc/tree-vectorizer.cc:1299

[Bug tree-optimization/106963] [13 Regression] ICE in vect_gen_perm_mask_checked, at tree-vect-stmts.cc:8606

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106963

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |13.0
           Keywords|                            |needs-bisection

[Bug tree-optimization/106963] [13 Regression] ICE in vect_gen_perm_mask_checked, at tree-vect-stmts.cc:8606

[crazylht at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106963

Hongtao.liu <crazylht at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |crazylht at gmail dot com

--- Comment #1 from Hongtao.liu <crazylht at gmail dot com> ---
When init_expr is INTEGER_CST or REAL_CST, we don't need can_vec_perm_const_p
since there's no real vec_perm needed, but vec_gen_perm_mask_checked will
gcc_assert (can_vec_perm_const_p). So it's better to use vec_gen_perm_mask_any
here.

[Bug tree-optimization/106963] [13 Regression] ICE in vect_gen_perm_mask_checked, at tree-vect-stmts.cc:8606 since r13-2503-gc13223b790bbc5e4

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106963

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|needs-bisection             |
                 CC|                            |marxin at gcc dot gnu.org
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2022-09-20
             Status|UNCONFIRMED                 |NEW
            Summary|[13 Regression] ICE in      |[13 Regression] ICE in
                   |vect_gen_perm_mask_checked, |vect_gen_perm_mask_checked,
                   |at tree-vect-stmts.cc:8606  |at tree-vect-stmts.cc:8606
                   |                            |since
                   |                            |r13-2503-gc13223b790bbc5e4

--- Comment #2 from Martin Liška <marxin at gcc dot gnu.org> ---
Btw. started with r13-2503-gc13223b790bbc5e4.

[Bug tree-optimization/106963] [13 Regression] ICE in vect_gen_perm_mask_checked, at tree-vect-stmts.cc:8606 since r13-2503-gc13223b790bbc5e4

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106963

--- Comment #3 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by hongtao Liu <liuhongt@gcc.gnu.org>:

https://gcc.gnu.org/g:eddbb81f1254f025d6f1b6b8e7554aa9a3830302

commit r13-2754-geddbb81f1254f025d6f1b6b8e7554aa9a3830302
Author: liuhongt <hongtao.liu@intel.com>
Date:   Tue Sep 20 13:22:28 2022 +0800

    Don't check can_vec_perm_const_p for nonlinear iv_init when it's constant.

    When init_expr is INTEGER_CST or REAL_CST, can_vec_perm_const_p is not
    necessary since there's no real vec_perm needed, but
    vec_gen_perm_mask_checked will gcc_assert (can_vec_perm_const_p). So
    it's better to use vec_gen_perm_mask_any in
    vect_create_nonlinear_iv_init.

    gcc/ChangeLog:

            PR tree-optimization/106963
            * tree-vect-loop.cc (vect_create_nonlinear_iv_init): Use
            vec_gen_perm_mask_any instead of vec_gen_perm_mask_check.

    gcc/testsuite/ChangeLog:

            * gcc.target/i386/pr106963.c: New test.

[Bug tree-optimization/106963] [13 Regression] ICE in vect_gen_perm_mask_checked, at tree-vect-stmts.cc:8606 since r13-2503-gc13223b790bbc5e4

[crazylht at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106963

--- Comment #4 from Hongtao.liu <crazylht at gmail dot com> ---
Fixed in GCC13.

[Bug tree-optimization/106963] [13 Regression] ICE in vect_gen_perm_mask_checked, at tree-vect-stmts.cc:8606 since r13-2503-gc13223b790bbc5e4

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106963

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.