[Bug middle-end/65533] New: [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[Bug middle-end/65533] New: [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533

            Bug ID: 65533
           Summary: [5 Regression] 252.eon in SPEC CPU 2000 failed to
                    build
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
          Assignee: unassigned at gcc dot gnu.org
          Reporter: hjl.tools at gmail dot com

On Linux/x86-64, r221604 gave

g++ -c -o ggHMatrix3.o   -DSPEC_CPU2000_LP64 -DHAS_ERRLIST    -I. -DNDEBUG    
-O3 -funroll-loops -ffast-math   ggHMatrix3.cc
ggHMatrix3.cc: In function 'ggHPoint3 operator*(const ggHMatrix3&, const
ggHPoint3&)':
ggHMatrix3.cc:303:11: internal compiler error: in quick_push, at vec.h:867
 ggHPoint3 operator*(const ggHMatrix3 &m, const ggHPoint3 &p) {
           ^
0xedabf3 vec<_slp_tree*, va_heap, vl_embed>::quick_push(_slp_tree* const&)
    ../../src-trunk/gcc/vec.h:867
0xedabf3 vec<_slp_tree*, va_heap, vl_ptr>::quick_push(_slp_tree* const&)
    ../../src-trunk/gcc/vec.h:1525
0xedabf3 vec<_slp_tree*, va_heap, vl_ptr>::safe_push(_slp_tree* const&)
    ../../src-trunk/gcc/vec.h:1538
0xedabf3 vect_build_slp_tree
    ../../src-trunk/gcc/tree-vect-slp.c:955
0xeda6a9 vect_build_slp_tree
    ../../src-trunk/gcc/tree-vect-slp.c:1054
0xeda445 vect_build_slp_tree
    ../../src-trunk/gcc/tree-vect-slp.c:1010
0xedc2bc vect_analyze_slp_instance
    ../../src-trunk/gcc/tree-vect-slp.c:1636
0xedda56 vect_analyze_slp(_loop_vec_info*, _bb_vec_info*, unsigned int)
    ../../src-trunk/gcc/tree-vect-slp.c:1768
0xeddf29 vect_slp_analyze_bb_1
    ../../src-trunk/gcc/tree-vect-slp.c:2300
0xeddf29 vect_slp_analyze_bb(basic_block_def*)
    ../../src-trunk/gcc/tree-vect-slp.c:2421
0xee0332 execute
    ../../src-trunk/gcc/tree-vectorizer.c:657
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <http://gcc.gnu.org/bugs.html> for instructions.

[Bug middle-end/65533] [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[hjl.tools at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2015-03-24
                 CC|                            |rguenther at suse dot de
   Target Milestone|---                         |5.0
     Ever confirmed|0                           |1

--- Comment #1 from H.J. Lu <hjl.tools at gmail dot com> ---
It was caused by r221592.

[Bug middle-end/65533] [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
C testcase:
struct A { int a[2]; };
struct B { double b[2]; };
struct C { double c[4][1]; };

static void inline
bar (struct B *x, double y, double z)
{
  x->b[0] = y;
  x->b[1] = z;
}

void baz (struct B *);

void
foo (struct C *x, struct A *y)
{
  struct B d;
  bar (&d, x->c[1][0] * y->a[0] + x->c[0][1] * y->a[1] + x->c[0][0] *
x->c[0][1],
       x->c[0][0] * y->a[0] + x->c[0][1] * y->a[1] + x->c[0][1] * y->a[0] +
x->c[0][0]);
  baz (&d);
}

[Bug middle-end/65533] [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Working on a patch...

[Bug tree-optimization/65533] [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1
          Component|middle-end                  |tree-optimization

[Bug tree-optimization/65533] [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

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

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 35120
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35120&action=edit
gcc5-pr65533.patch

Untested fix.

[Bug tree-optimization/65533] [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #4)
> Created attachment 35120 [details]
> gcc5-pr65533.patch
> 
> Untested fix.

Hmm, looks basically ok, though doing

  vect_free_slp_tree (child);

from the cleanup before

      /* If the SLP build for operand zero failed and operand zero
         and one can be commutated try that for the scalar stmts
         that failed the match.  */
      if (i == 0
...

and simply re-allocating child with

      child = vect_create_new_slp_node (oprnd_info->def_stmts);
      if (!child)
        {
          vect_free_oprnd_info (oprnds_info);
          return false;
        }

might be "simpler".

Your patch is ok if it is already tested.

[Bug tree-optimization/65533] [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533

--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> ---
Looks like a latent issue on the branch(es) btw.

[Bug tree-optimization/65533] [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[rguenther at suse dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533

--- Comment #8 from rguenther at suse dot de <rguenther at suse dot de> ---
On Tue, 24 Mar 2015, jakub at gcc dot gnu.org wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533
> 
> --- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
> (In reply to Richard Biener from comment #5)
> > Hmm, looks basically ok, though doing
> > 
> >   vect_free_slp_tree (child);
> > 
> > from the cleanup before
> > 
> >       /* If the SLP build for operand zero failed and operand zero
> >          and one can be commutated try that for the scalar stmts
> >          that failed the match.  */
> >       if (i == 0
> > ...
> > 
> > and simply re-allocating child with
> > 
> >       child = vect_create_new_slp_node (oprnd_info->def_stmts);
> >       if (!child)
> >         {
> >           vect_free_oprnd_info (oprnds_info);
> >           return false;
> >         }
> > 
> > might be "simpler".
> 
> Well, it would be more costly (having to deallocate and allocate everything
> again), and not really much shorter in the source.  Only if we in the future
> popluate early not just the SLP_TREE_CHILDREN vector, but various further
> things your version might be better from maintanance POV.
> 
> > Your patch is ok if it is already tested.
> 
> Not yet tested, but bootstraps/regtests already in progress.  If you feel
> strongly about this, I can surely kill those, write a new patch and redo that.

No, I don't feel strongly about it - for clarity the function should
get some refactoring but not at this stage.

[Bug tree-optimization/65533] [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Author: jakub
Date: Tue Mar 24 10:45:09 2015
New Revision: 221622

URL: https://gcc.gnu.org/viewcvs?rev=221622&root=gcc&view=rev
Log:
    PR tree-optimization/65533
    * tree-vect-slp.c (vect_build_slp_tree): Before re-trying
    with swapped operands, call vect_free_slp_tree on
    SLP_TREE_CHILDREN of child and truncate the SLP_TREE_CHILDREN
    vector.

    * gcc.dg/pr65533.c: New test.

Added:
    trunk/gcc/testsuite/gcc.dg/pr65533.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/tree-vect-slp.c

[Bug tree-optimization/65533] [5 Regression] 252.eon in SPEC CPU 2000 failed to build

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65533

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

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

--- Comment #10 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed.