[Bug c++/106794] New: ice in vect_transform_slp_perm_load_1

[Bug c++/106794] New: ice in vect_transform_slp_perm_load_1

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106794

            Bug ID: 106794
           Summary: ice in vect_transform_slp_perm_load_1
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dcb314 at hotmail dot com
  Target Milestone: ---

Created attachment 53525
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=53525&action=edit
gzipped C++ source code

For the attached C++ code, recent gcc does this:

$ /home/dcb/gcc/results/bin/gcc -c -w -O3 -ansi -march=bdver2  -ffast-math
bug839.cc
during GIMPLE pass: vect
spheremesh.cpp: In member function ‘void SphereMesh::generateNormals()’:
spheremesh.cpp:123:6: internal compiler error: in operator[], at vec.h:889
0x137b3fd vect_transform_slp_perm_load_1(vec_info*, _slp_tree*, vec<unsigned
int, va_heap, vl_ptr>&, vec<tree_node*
, va_heap, vl_ptr> const&, gimple_stmt_iterator*, poly_int<1u, unsigned long>,
bool, bool, unsigned int*, unsigned 
int*, bool)
        ../../trunk.git/gcc/tree-vect-slp.cc:0
0x137a5c0 vect_optimize_slp_pass::internal_node_cost(_slp_tree*, int, unsigned
int)
        ../../trunk.git/gcc/tree-vect-slp.cc:4490
0x137cf1b vect_optimize_slp_pass::forward_pass()
        ../../trunk.git/gcc/tree-vect-slp.cc:4951
0x137fe93 vect_optimize_slp_pass::run()
        ../../trunk.git/gcc/tree-vect-slp.cc:5498

A reduction is currently running. I will try to identify a git hash range.

[Bug c++/106794] ice in vect_transform_slp_perm_load_1

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106794

--- Comment #1 from David Binderman <dcb314 at hotmail dot com> ---
Git hash range seems to be 4fbe3e6aa74dae5c..3de9fb3235998a05, a distance of 28
commits.

[Bug tree-optimization/106794] [13 Regression] ice in vect_transform_slp_perm_load_1

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106794

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |13.0

[Bug tree-optimization/106794] [13 Regression] ice in vect_transform_slp_perm_load_1

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106794

--- Comment #2 from David Binderman <dcb314 at hotmail dot com> ---
Reduced code seems to be:

template <class T> struct Vector3 {
  Vector3();
  Vector3(T, T, T);
  T length() const;
  T x, y, z;
};
template <class T>
Vector3<T>::Vector3(T _x, T _y, T _z) : x(_x), y(_y), z(_z) {}
Vector3<float> cross(Vector3<float> a, Vector3<float> b) {
  return Vector3<float>(a.y * b.z - a.z * b.y, a.z * b.x - a.x * b.z,
                        a.x * b.y - a.y * b.x);
}
template <class T> T Vector3<T>::length() const { return z; }
int generateNormals_i;
float generateNormals_p2_0, generateNormals_p0_0;
struct SphereMesh {
  void generateNormals();
  float vertices;
};
void SphereMesh::generateNormals() {
  Vector3<float> *faceNormals = new Vector3<float>;
  for (int j; j; j++) {
    float *p0 = &vertices + 3, *p1 = &vertices + j * 3, *p2 = &vertices + 3,
          *p3 = &vertices + generateNormals_i + j * 3;
    Vector3<float> v0(p1[0] - generateNormals_p0_0, p1[1] - 1, p1[2] - 2),
        v1(0, 1, 2);
    if (v0.length())
      v1 = Vector3<float>(p3[0] - generateNormals_p2_0, p3[1] - p2[1],
                          p3[2] - p2[2]);
    else
      v1 = Vector3<float>(generateNormals_p0_0 - p3[0], p0[1] - p3[1],
                          p0[2] - p3[2]);
    Vector3<float> faceNormal = cross(v0, v1);
    faceNormals[j] = faceNormal;
  }
}

Flags -c -O3 -ffast-math -march=bdver2 required.

[Bug tree-optimization/106794] [13 Regression] ICE in vect_transform_slp_perm_load_1 since r13-2288-g61c4c989034548f4

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106794

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2022-08-31
                 CC|                            |marxin at gcc dot gnu.org,
                   |                            |rsandifo at gcc dot gnu.org
             Status|UNCONFIRMED                 |NEW
            Summary|[13 Regression] ice in      |[13 Regression] ICE in
                   |vect_transform_slp_perm_loa |vect_transform_slp_perm_loa
                   |d_1                         |d_1 since
                   |                            |r13-2288-g61c4c989034548f4

--- Comment #3 from Martin Liška <marxin at gcc dot gnu.org> ---
Started with r13-2288-g61c4c989034548f4, the same as PR106787.

[Bug tree-optimization/106794] [13 Regression] ICE in vect_transform_slp_perm_load_1 since r13-2288-g61c4c989034548f4

[rsandifo at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106794

rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |rsandifo at gcc dot gnu.org
             Status|NEW                         |ASSIGNED

--- Comment #4 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> ---
Testing a patch.

[Bug tree-optimization/106794] [13 Regression] ICE in vect_transform_slp_perm_load_1 since r13-2288-g61c4c989034548f4

[rsandifo at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106794

rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |gscfq@t-online.de

--- Comment #5 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> ---
*** Bug 106914 has been marked as a duplicate of this bug. ***

[Bug tree-optimization/106794] [13 Regression] ICE in vect_transform_slp_perm_load_1 since r13-2288-g61c4c989034548f4

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106794

--- Comment #6 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The trunk branch has been updated by Richard Sandiford <rsandifo@gcc.gnu.org>:

https://gcc.gnu.org/g:1005c89431cc78a68762643e1bc7249c3959927e

commit r13-2738-g1005c89431cc78a68762643e1bc7249c3959927e
Author: Richard Sandiford <richard.sandiford@arm.com>
Date:   Tue Sep 20 15:27:47 2022 +0100

    vect: Fix SLP layout handling of masked loads [PR106794]

    PR106794 shows that I'd forgotten about masked loads when
    doing the SLP layout changes.  These loads can't currently
    be permuted independently of their mask input, so during
    construction they never get a load permutation.

    (If we did support permuting masked loads in future, the mask
    would need to be in the right order for the load, rather than in
    the order implied by the result of the permutation.  Since masked
    loads can't be partly or fully scalarised in the way that normal
    permuted loads can be, there's probably no benefit to fusing the
    permutation and the load.  Permutation after the fact is probably
    good enough.)

    gcc/
            PR tree-optimization/106794
            PR tree-optimization/106914
            * tree-vect-slp.cc (vect_optimize_slp_pass::internal_node_cost):
            Only consider loads that already have a permutation.
            (vect_optimize_slp_pass::start_choosing_layouts): Assert that
            loads with permutations are leaf nodes.  Prevent any kind of
grouped
            access from changing layout if it doesn't have a load permutation.

    gcc/testsuite/
            * gcc.dg/vect/pr106914.c: New test.
            * g++.dg/vect/pr106794.cc: Likewise.

[Bug tree-optimization/106794] [13 Regression] ICE in vect_transform_slp_perm_load_1 since r13-2288-g61c4c989034548f4

[rsandifo at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106794

rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> changed:

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

--- Comment #7 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> ---
Fixed.