[Bug tree-optimization/46886] New: [4.5/4.6 Regression] gfortran.dg/array_constructor_9.f90 FAILs with -ftree-parallelize-loops -fstrict-overflow -fno-tree-ch

[Bug tree-optimization/46886] New: [4.5/4.6 Regression] gfortran.dg/array_constructor_9.f90 FAILs with -ftree-parallelize-loops -fstrict-overflow -fno-tree-ch

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

           Summary: [4.5/4.6 Regression]
                    gfortran.dg/array_constructor_9.f90 FAILs with
                    -ftree-parallelize-loops -fstrict-overflow
                    -fno-tree-ch
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: zsojka@seznam.cz
              Host: x86_64-pc-linux-gnu
            Target: x86_64-pc-linux-gnu


Created attachment 22707
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22707
reduced testcase (hopefully correctly)

Output:
$ gfortran -O -ftree-parallelize-loops=2 -fstrict-overflow -fno-tree-ch
testcase.f90                                   
$ valgrind -q ./a.out 
==7049== Invalid read of size 4
==7049==    at 0x400D33: main (testcase.f90:12)
==7049==  Address 0x6174788 is 0 bytes after a block of size 1,320 alloc'd
==7049==    at 0x4C265BF: realloc (in
/usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
==7049==    by 0x400C1A: main (testcase.f90:12)
==7049== 
==7049== Invalid write of size 4
==7049==    at 0x400D37: main (testcase.f90:12)
==7049==  Address 0x6174cf8 is 0 bytes after a block of size 1,320 alloc'd
==7049==    at 0x4C2648E: malloc (in
/usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
==7049==    by 0x400CDD: main (testcase.f90:12)
==7049== 

Tested revisions: (for the reduced testcase)
r167643 - fail
4.5 r166509 - fail (original testcase passes)
4.4 r166509 - OK

[Bug tree-optimization/46886] [4.5/4.6 Regression] gfortran.dg/array_constructor_9.f90 FAILs with -ftree-parallelize-loops -fstrict-overflow -fno-tree-ch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org
   Target Milestone|---                         |4.5.3

[Bug tree-optimization/46886] [4.5/4.6 Regression] gfortran.dg/array_constructor_9.f90 FAILs with -ftree-parallelize-loops -fstrict-overflow -fno-tree-ch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rakdver at gcc dot gnu.org,
                   |                            |spop at gcc dot gnu.org

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-01-18 17:00:29 UTC ---
Seems one extra incorrect iteration is added after GOMP_parallel_end:
<bb 16>:
  D.1592_69 = D.1591_183 * 4;

<bb 17>:
  # D.1592_15 = PHI <0(15), D.1592_69(16)>
  D.1749_71 = (<unnamed-unsigned:64>) D.1592_15;
  D.1750_72 = MAX_EXPR <D.1749_71, 1>;
  D.1751_73 = __builtin_malloc (D.1750_72);
  D.1594_74 = (void * restrict) D.1751_73;
  D.1898_232 = (<unnamed-unsigned:64>) D.1591_183;
  D.1899_233 = D.1591_183 >= 0;
  D.1900_234 = D.1898_232 > 199;
  D.1901_235 = D.1899_233 && D.1900_234;
  if (D.1901_235 != 0)
    goto <bb 18>;
  else
    goto <bb 19>;

<bb 18>:
  .paral_data_store.64.D.1910 = D.1578_1;
  .paral_data_store.64.D.1911 = D.1594_74;
  .paral_data_store.64.D.1912 = D.1898_232;
  __builtin_GOMP_parallel_start (build._loopfn.1, &.paral_data_store.64, 2);
  build._loopfn.1 (&.paral_data_store.64);
  __builtin_GOMP_parallel_end ();
  ivtmp.60_247 = D.1898_232;
  S.7_248 = (integer(kind=8)) ivtmp.60_247;
  D.1755_249 = *D.1578_1[S.7_248];
  MEM[(real(kind=4)[0:] *)D.1594_74][S.7_248] = D.1755_249;
  S.7_251 = S.7_248 + 1;
  ivtmp.60_252 = ivtmp.60_247 + 1;
S.7_248 is equal to the number of ints allocated, so stores after the end of
malloced area.  The loop is initially:
loop_4 (header = 18, latch = 19, niter = , upper_bound = 9223372036854775808,
estimate = 9223372036854775808)
{
  bb_18 (preds = {bb_17 bb_19 }, succs = {bb_20 bb_19 })
  {
  <bb 18>:
    # S.7_7 = PHI <0(17), S.7_80(19)>
    # .MEM_152 = PHI <.MEM_182(17), .MEM_185(19)>
    if (S.7_7 > D.1586_64)
      goto <bb 20>;
    else
      goto <bb 19>;

  }
  bb_19 (preds = {bb_18 }, succs = {bb_18 })
  {
  <bb 19>:
    # VUSE <.MEM_152>
    D.1755_79 = *D.1578_1[S.7_7];
    # .MEM_185 = VDEF <.MEM_152>
    MEM[(real(kind=4)[0:] *)D.1594_74][S.7_7] = D.1755_79;
    S.7_80 = S.7_7 + 1;
    goto <bb 18>;

  }
}
and try_get_loop_niter computes niter->niter as (<unnamed-unsigned:64>)
D.1591_183.  On:
__attribute__((noinline, noclone)) void
foo (int *__restrict__ p, int *__restrict__ q, int n)
{
  int i;
  for (i = 0; i < n; i++)
    p[i] = q[i];
}

int
main (void)
{
  int *p = __builtin_calloc (1024, sizeof (int));
  int *q = __builtin_calloc (1024, sizeof (int));
  foo (p, q, 1024);
  return 0;
}
try_get_loop_niter instead computes niter->niter as n - 1 (which is strange to
me), but then it works even with the extra iteration added after
GOMP_parallel_end (as that iteration is done with iv set to n - 1).

[Bug tree-optimization/46886] [4.5/4.6 Regression] gfortran.dg/array_constructor_9.f90 FAILs with -ftree-parallelize-loops -fstrict-overflow -fno-tree-ch

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-02-04 07:11:53 UTC ---
Sebastian, thanks for PR46194 fix, could you please also look at this one?

[Bug tree-optimization/46886] [4.5/4.6 Regression] gfortran.dg/array_constructor_9.f90 FAILs with -ftree-parallelize-loops -fstrict-overflow -fno-tree-ch

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

Sebastian Pop <spop at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2011.02.04 07:57:49
         AssignedTo|unassigned at gcc dot       |spop at gcc dot gnu.org
                   |gnu.org                     |
     Ever Confirmed|0                           |1

--- Comment #3 from Sebastian Pop <spop at gcc dot gnu.org> 2011-02-04 07:57:49 UTC ---
Sure, I'll have a look.

[Bug tree-optimization/46886] [4.5/4.6 Regression] gfortran.dg/array_constructor_9.f90 FAILs with -ftree-parallelize-loops -fstrict-overflow -fno-tree-ch

[sebpop at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

--- Comment #4 from sebpop at gmail dot com <sebpop at gmail dot com> 2011-02-04 20:30:46 UTC ---
On Tue, Jan 18, 2011 at 11:00, jakub at gcc dot gnu.org
<gcc-bugzilla@gcc.gnu.org> wrote:
> Seems one extra incorrect iteration is added after GOMP_parallel_end:

That extra iteration is added by this call:

  /* Ensure that the exit condition is the first statement in the loop.  */
  transform_to_exit_first_loop (loop, reduction_list, nit);


/* Moves the exit condition of LOOP to the beginning of its header, and
   duplicates the part of the last iteration that gets disabled to the
   exit of the loop.  NIT is the number of iterations of the loop
   (used to initialize the variables in the duplicated part).

   TODO: the common case is that latch of the loop is empty and immediately
   follows the loop exit.  In this case, it would be better not to copy the
   body of the loop, but only move the entry of the loop directly before the
   exit check and increase the number of iterations of the loop by one.
   This may need some additional preconditioning in case NIT = ~0.
   REDUCTION_LIST describes the reductions in LOOP.  */

static void
transform_to_exit_first_loop (struct loop *loop, htab_t
reduction_list, tree nit)

[Bug tree-optimization/46886] [4.5/4.6 Regression] gfortran.dg/array_constructor_9.f90 FAILs with -ftree-parallelize-loops -fstrict-overflow -fno-tree-ch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2

[Bug tree-optimization/46886] [4.5/4.6 Regression] gfortran.dg/array_constructor_9.f90 FAILs with -ftree-parallelize-loops -fstrict-overflow -fno-tree-ch

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

--- Comment #5 from Sebastian Pop <spop at gcc dot gnu.org> 2011-02-14 04:19:03 UTC ---
Patch:
http://gcc.gnu.org/ml/gcc-patches/2011-02/msg00858.html

[Bug tree-optimization/46886] [4.5/4.6/4.7 Regression] gfortran.dg/array_constructor_9.f90 FAILs with -ftree-parallelize-loops -fstrict-overflow -fno-tree-ch

[pault at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

Paul Thomas <pault at gcc dot gnu.org> changed:

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

--- Comment #6 from Paul Thomas <pault at gcc dot gnu.org> 2011-03-21 12:17:18 UTC ---
(In reply to comment #5)
> Patch:
> http://gcc.gnu.org/ml/gcc-patches/2011-02/msg00858.html

This looks like a low hanging fruit, Sebastian.  Are you going to commit it?

Cheers

Paul Thomas

[Bug tree-optimization/46886] [4.5/4.6/4.7 Regression] wrong code with -ftree-parallelize-loops -fno-tree-ch

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

--- Comment #7 from Zdenek Sojka <zsojka at seznam dot cz> 2011-04-14 19:03:46 UTC ---
Created attachment 23989
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23989
C testcase (reduced from libgomp.c/autopar-1.c)

$ gcc -O -ftree-parallelize-loops=4 -fno-tree-ch -fno-tree-dominator-opts
testcase.c
$ ./a.out 
Aborted

[Bug tree-optimization/46886] [4.5/4.6/4.7 Regression] wrong code with -ftree-parallelize-loops -fno-tree-ch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.5.3                       |4.5.4

--- Comment #8 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-04-28 14:51:28 UTC ---
GCC 4.5.3 is being released, adjusting target milestone.

[Bug tree-optimization/46886] [4.5/4.6/4.7 Regression] wrong code with -ftree-parallelize-loops -fno-tree-ch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

--- Comment #9 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-02-08 14:16:34 UTC ---
Created attachment 26612
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=26612
alternative patch

Alternative patch I am testing.

[Bug tree-optimization/46886] [4.5/4.6 Regression] wrong code with -ftree-parallelize-loops -fno-tree-ch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.7.0
         AssignedTo|spop at gcc dot gnu.org     |rguenth at gcc dot gnu.org
            Summary|[4.5/4.6/4.7 Regression]    |[4.5/4.6 Regression] wrong
                   |wrong code with             |code with
                   |-ftree-parallelize-loops    |-ftree-parallelize-loops
                   |-fno-tree-ch                |-fno-tree-ch
      Known to fail|4.7.0                       |

--- Comment #11 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-02-08 15:28:42 UTC ---
Fixed on trunk sofar.

[Bug tree-optimization/46886] [4.5/4.6/4.7 Regression] wrong code with -ftree-parallelize-loops -fno-tree-ch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

--- Comment #10 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-02-08 15:28:06 UTC ---
Author: rguenth
Date: Wed Feb  8 15:28:01 2012
New Revision: 184010

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=184010
Log:
2012-03-08  Richard Guenther  <rguenther@suse.de>

    PR tree-optimization/46886
    * tree-flow.h (do_while_loop_p): Declare.
    * tree-ssa-loop-ch.c (do_while_loop_p): Export.
    * tree-parloops.c (parallelize_loops): Only parallelize do-while
    loops.

    * testsuite/libgomp.c/pr46886.c: New testcase.

Added:
    trunk/libgomp/testsuite/libgomp.c/pr46886.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/tree-flow.h
    trunk/gcc/tree-parloops.c
    trunk/gcc/tree-ssa-loop-ch.c
    trunk/libgomp/ChangeLog

[Bug tree-optimization/46886] [4.5/4.6 Regression] wrong code with -ftree-parallelize-loops -fno-tree-ch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

--- Comment #12 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-02-15 11:05:31 UTC ---
Author: rguenth
Date: Wed Feb 15 11:05:26 2012
New Revision: 184263

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=184263
Log:
2012-02-15  Richard Guenther  <rguenther@suse.de>

    Backport from mainline
    2012-02-08  Richard Guenther  <rguenther@suse.de>

    PR tree-optimization/46886
    * tree-flow.h (do_while_loop_p): Declare.
    * tree-ssa-loop-ch.c (do_while_loop_p): Export.
    * tree-parloops.c (parallelize_loops): Only parallelize do-while
    loops.

    * testsuite/libgomp.c/pr46886.c: New testcase.

Added:
    branches/gcc-4_6-branch/libgomp/testsuite/libgomp.c/pr46886.c
Modified:
    branches/gcc-4_6-branch/gcc/ChangeLog
    branches/gcc-4_6-branch/gcc/tree-flow.h
    branches/gcc-4_6-branch/gcc/tree-parloops.c
    branches/gcc-4_6-branch/gcc/tree-ssa-loop-ch.c
    branches/gcc-4_6-branch/libgomp/ChangeLog

[Bug tree-optimization/46886] [4.5/4.6 Regression] wrong code with -ftree-parallelize-loops -fno-tree-ch

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
      Known to work|                            |4.6.3
         Resolution|                            |FIXED
   Target Milestone|4.5.4                       |4.6.3

--- Comment #13 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-02-15 11:06:01 UTC ---
Fixed for 4.6.3.

[Bug tree-optimization/46886] [4.5/4.6 Regression] wrong code with -ftree-parallelize-loops -fno-tree-ch

[razya at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46886

--- Comment #14 from razya at gcc dot gnu.org 2012-04-22 10:36:18 UTC ---
Author: razya
Date: Sun Apr 22 10:36:13 2012
New Revision: 186667

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=186667
Log:
2012-04-20  Razya Ladelsky  <razya@il.ibm.com>
                 Correcting transform_to_exit_first_loop + fix to
                 PR tree-optimization/46886
                 * tree-parloops.c (transform_to_exit_first_loop):
                 * Remove setting of number of iterations according to
                 * the loop pattern.
                 Duplicate from entry to exit->src instead of
loop->latch.
                 (pallelize_loops): Remove the condition preventing
do-while loops.
                 * tree-cfg.c (bool bb_in_region_p): New.
                 (gimple_duplicate_sese_tail): Adjust duplication of the
the subloops.
                 Adjust redirection of the duplicated iteration.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/tree-cfg.c
    trunk/gcc/tree-parloops.c