[Bug tree-optimization/59817] New: ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[Bug tree-optimization/59817] New: ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[anton at samba dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

            Bug ID: 59817
           Summary: ICE in extract_affine_chrec with -O2
                    -ftree-loop-linear
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: anton at samba dot org

The following testcase:


c -O2 -ftree-loop-linear 

      SUBROUTINE PREPD(ICAST,ICAS,ICASX,ICAS1,ICAS2,NDET,NM,III,IMP,
     *                 CASMIN)
      LOGICAL CASMIN
      DIMENSION ICAST(NDET,NM),IMP(NM)
      IF(CASMIN) THEN
         DO K=1,NDET
            DO L=1,NM
               IF(L.EQ.K-1) ICAST(K,L) = 1
            END DO
         END DO
      END IF


fails on a powerpc64-linux build from today with:


# gfortran -c -O2 -ftree-loop-linear testcase.f
testcase.f: In function ‘prepd’:
testcase.f:3:0: internal compiler error: in extract_affine_chrec, at
graphite-sese-to-poly.c:620
       SUBROUTINE PREPD(ICAST,ICAS,ICASX,ICAS1,ICAS2,NDET,NM,III,IMP,
 ^
0x10cf684f extract_affine_chrec
        ../../gcc/gcc/graphite-sese-to-poly.c:619
0x10cf684f extract_affine
        ../../gcc/gcc/graphite-sese-to-poly.c:803
0x10cf62fb extract_affine
        ../../gcc/gcc/graphite-sese-to-poly.c:842
0x10cf843f pdr_add_memory_accesses
        ../../gcc/gcc/graphite-sese-to-poly.c:1486
0x10cf843f build_poly_dr
        ../../gcc/gcc/graphite-sese-to-poly.c:1583
0x10cf843f build_pbb_drs
        ../../gcc/gcc/graphite-sese-to-poly.c:1846
0x10cf843f build_scop_drs
        ../../gcc/gcc/graphite-sese-to-poly.c:1929
0x10cfa8d7 build_poly_scop(scop*)
        ../../gcc/gcc/graphite-sese-to-poly.c:3171
0x10cdcabb graphite_transform_loops()
        ../../gcc/gcc/graphite.c:300
0x10cdd227 graphite_transforms
        ../../gcc/gcc/graphite.c:332
0x10cdd227 execute
        ../../gcc/gcc/graphite.c:416
>From gcc-bugs-return-440375-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Wed Jan 15 00:47:36 2014
Return-Path: <gcc-bugs-return-440375-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 31692 invoked by alias); 15 Jan 2014 00:47:36 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 31646 invoked by uid 48); 15 Jan 2014 00:47:33 -0000
From: "pinskia at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug target/59780] ICE in aarch64_split_128bit_move
Date: Wed, 15 Jan 2014 00:47:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: target
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords: ice-checking, ice-on-valid-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: pinskia at gcc dot gnu.org
X-Bugzilla-Status: NEW
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields: keywords bug_status cf_reconfirmed_on everconfirmed
Message-ID: <bug-59780-4-o91zWAHMyr@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-59780-4@http.gcc.gnu.org/bugzilla/>
References: <bug-59780-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2014-01/txt/msg01517.txt.bz2
Content-length: 610

http://gcc.gnu.org/bugzilla/show_bug.cgi?idY780

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-checking,
                   |                            |ice-on-valid-code
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-01-15
     Ever confirmed|0                           |1

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Confirmed.

[Bug tree-optimization/59817] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|powerpc64-linux             |
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-01-15
     Ever confirmed|0                           |1
      Known to fail|                            |4.8.2, 4.9.0

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Also on x86_64.  The CHREC is
  {-(unsigned long) stride.2_13, +, (unsigned long) stride.2_13 + 1}_1

also fails on the 4.8 branch for me at least.

[Bug tree-optimization/59817] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[asolokha at gmx dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

Arseny Solokha <asolokha at gmx dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |asolokha at gmx dot com

--- Comment #2 from Arseny Solokha <asolokha at gmx dot com> ---
The following testcase is another way to trigger this segfault:

int kd;

void
n2(void)
{
  static int so;
  static short int i5;
  int wj;
  int *il;
  int *nk = &so;
  for (wj = 0; wj < 2; ++wj)
    *nk = ((i5 += *il) || kd );
}

I can reproduce it on x86_64 w/ 4.8.2 and 4.9.0-alpha20140406.

[Bug tree-optimization/59817] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[asolokha at gmx dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

--- Comment #3 from Arseny Solokha <asolokha at gmx dot com> ---
void
xl(void)
{
  static int j3;
  for (j3 = 0; j3 < 1; ++j3) {
    static int f2;
    static int w7;
    short int b5;
    int ok;
    f2 = (b5 += ok) ? (w7 = 0): (w7 ? 0 : (f2 = ok));
  }
}

4.7.3 doesn't fail, so this is surely a regression.

[Bug tree-optimization/59817] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
The ICE tells us that the parameter isn't recorded as parameter in the SESE
region:

(gdb) p debug_scop_params (s, 2)
# parameters (
# Parameter names are provided
1
# Parameter names
_10 pretmp_30 offset.3_15 
#)
$28 = void
(gdb) p debug_scop_params (s, 5)
# parameters (
# Parameter names are provided
1
# Parameter names
_10 pretmp_30 offset.3_15 
#)
$29 = void

(gdb) p debug_scop (s, 1)
SCoP 1
#(
# Language
Gimple
# Context (
#eq     p0     p1     p2    cst
XXX isl
# )
# Parameter names are provided
1
# Parameter names
_10 pretmp_30 offset.3_15 
# Number of statements
2

# Iteration domain of bb_8 (
#eq     i0     p0     p1     p2    cst
XXX isl
#)
# Access informations are provided
1
# Read access informations
0
# Write access informations
1
# data accesses (
#  eq   alias   sub0     i0     p0     p1     p2    cst
#)
# Statement body is not provided
0

# Iteration domain of bb_10 (
#eq     i0     i1     p0     p1     p2    cst
XXX isl
#)
# Access informations are not provided
0
# Statement body is not provided
0
#)
$32 = void

not sure if I interpret the above correctly but bb8 and bb10 seem to be in
that scop.  The parameter is refered from bb 8.

Ah, scan_tree_for_params does

scan_tree_for_params (sese s, tree e)
{
  if (e == chrec_dont_know)
    return;

  switch (TREE_CODE (e))
    {
    case POLYNOMIAL_CHREC:
      scan_tree_for_params (s, CHREC_LEFT (e));
      break;

thus it misses to scan CHREC_RIGHT.  The MULT_EXPR handling is also suspicious.

Fixing the above shifts the ICE to

#1  0x000000000132e1ae in extract_affine_chrec (s=0x1f89e10, e=
    <polynomial_chrec 0x7ffff6d91ae0>, space=0x1f8b910)
    at /space/rguenther/src/svn/trunk/gcc/graphite-sese-to-poly.c:619
619       gcc_assert (isl_pw_aff_is_cst (rhs)
(gdb) l
614       isl_aff *loop = isl_aff_set_coefficient_si
615         (isl_aff_zero_on_domain (ls), isl_dim_in, pos, 1);
616       isl_pw_aff *l = isl_pw_aff_from_aff (loop);
617
618       /* Before multiplying, make sure that the result is affine.  */
619       gcc_assert (isl_pw_aff_is_cst (rhs)
620                   || isl_pw_aff_is_cst (l));
621
622       return isl_pw_aff_add (lhs, isl_pw_aff_mul (rhs, l));
623     }

so maybe it simply doesn't handle CHRECs with variable stride but fails to
disregard those.

It indeed fails to as scev_is_linear_expression doesn't have the
same restriction on CHRECs as graphite_can_represent_scev asserts.
Thus completing that one as well is required.

Index: gcc/graphite-scop-detection.c
===================================================================
--- gcc/graphite-scop-detection.c       (revision 209292)
+++ gcc/graphite-scop-detection.c       (working copy)
@@ -219,7 +219,14 @@ graphite_can_represent_scev (tree scev)

   switch (TREE_CODE (scev))
     {
+    case NEGATE_EXPR:
+    case BIT_NOT_EXPR:
+    CASE_CONVERT:
+    case NON_LVALUE_EXPR:
+      return graphite_can_represent_scev (TREE_OPERAND (scev, 0));
+
     case PLUS_EXPR:
+    case POINTER_PLUS_EXPR:
     case MINUS_EXPR:
       return graphite_can_represent_scev (TREE_OPERAND (scev, 0))
        && graphite_can_represent_scev (TREE_OPERAND (scev, 1));
@@ -247,7 +254,8 @@ graphite_can_represent_scev (tree scev)
     }

   /* Only affine functions can be represented.  */
-  if (!scev_is_linear_expression (scev))
+  if (tree_contains_chrecs (scev, NULL)
+      || !scev_is_linear_expression (scev))
     return false;

   return true;

[Bug tree-optimization/59817] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

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

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

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

[Bug tree-optimization/59817] [4.8/4.9 Regression] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2
      Known to work|                            |4.7.3
   Target Milestone|---                         |4.8.3
            Summary|ICE in extract_affine_chrec |[4.8/4.9 Regression] ICE in
                   |with -O2 -ftree-loop-linear |extract_affine_chrec with
                   |                            |-O2 -ftree-loop-linear

[Bug tree-optimization/59817] [4.8/4.9/4.10 Regression] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> ---
Author: rguenth
Date: Mon Apr 14 08:09:06 2014
New Revision: 209355

URL: http://gcc.gnu.org/viewcvs?rev=209355&root=gcc&view=rev
Log:
2014-04-14  Richard Biener  <rguenther@suse.de>

    PR tree-optimization/59817
    PR tree-optimization/60453
    * graphite-scop-detection.c (graphite_can_represent_scev): Complete
    recursion to catch all CHRECs in the scalar evolution and restrict
    the predicate for the remains appropriately.

    * gfortran.dg/graphite/pr59817.f: New testcase.
    * gcc.dg/graphite/pr59817-1.c: Likewise.
    * gcc.dg/graphite/pr59817-2.c: Likewise.

Added:
    trunk/gcc/testsuite/gcc.dg/graphite/pr59817-1.c
    trunk/gcc/testsuite/gcc.dg/graphite/pr59817-2.c
    trunk/gcc/testsuite/gfortran.dg/graphite/pr59817.f
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/graphite-scop-detection.c
    trunk/gcc/testsuite/ChangeLog

[Bug tree-optimization/59817] [4.8/4.9 Regression] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sergamena at inbox dot ru

--- Comment #8 from Richard Biener <rguenth at gcc dot gnu.org> ---
*** Bug 60453 has been marked as a duplicate of this bug. ***

[Bug tree-optimization/59817] [4.8/4.9 Regression] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.10.0
            Summary|[4.8/4.9/4.10 Regression]   |[4.8/4.9 Regression] ICE in
                   |ICE in extract_affine_chrec |extract_affine_chrec with
                   |with -O2 -ftree-loop-linear |-O2 -ftree-loop-linear

--- Comment #7 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed on trunk sofar.

[Bug tree-optimization/59817] [4.8 Regression] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

--- Comment #9 from Richard Biener <rguenth at gcc dot gnu.org> ---
Author: rguenth
Date: Tue Apr 22 13:22:32 2014
New Revision: 209628

URL: http://gcc.gnu.org/viewcvs?rev=209628&root=gcc&view=rev
Log:
2014-04-22  Richard Biener  <rguenther@suse.de>

    Backport from mainline
    2014-04-14  Richard Biener  <rguenther@suse.de>

    PR tree-optimization/59817
    PR tree-optimization/60453
    * graphite-scop-detection.c (graphite_can_represent_scev): Complete
    recursion to catch all CHRECs in the scalar evolution and restrict
    the predicate for the remains appropriately.

    * gfortran.dg/graphite/pr59817.f: New testcase.
    * gcc.dg/graphite/pr59817-1.c: Likewise.
    * gcc.dg/graphite/pr59817-2.c: Likewise.

Added:
    branches/gcc-4_9-branch/gcc/testsuite/gcc.dg/graphite/pr59817-1.c
    branches/gcc-4_9-branch/gcc/testsuite/gcc.dg/graphite/pr59817-2.c
    branches/gcc-4_9-branch/gcc/testsuite/gfortran.dg/graphite/pr59817.f
Modified:
    branches/gcc-4_9-branch/gcc/ChangeLog
    branches/gcc-4_9-branch/gcc/graphite-scop-detection.c
    branches/gcc-4_9-branch/gcc/testsuite/ChangeLog

[Bug tree-optimization/59817] [4.8 Regression] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

--- Comment #10 from Richard Biener <rguenth at gcc dot gnu.org> ---
Author: rguenth
Date: Mon May  5 13:48:42 2014
New Revision: 210070

URL: http://gcc.gnu.org/viewcvs?rev=210070&root=gcc&view=rev
Log:
2014-05-05  Richard Biener  <rguenther@suse.de>

    Backport from mainline
    2014-04-23  Richard Biener  <rguenther@suse.de>

    PR middle-end/60895
    * tree-inline.c (declare_return_variable): Use mark_addressable.

    * g++.dg/torture/pr60895.C: New testcase.

    2014-04-07  Richard Biener  <rguenther@suse.de>

    PR middle-end/60750
    * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
    for noreturn calls.
    * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.

    * g++.dg/torture/pr60750.C: New testcase.
    * gcc.dg/tree-ssa/20040517-1.c: Adjust.

    2014-04-14  Richard Biener  <rguenther@suse.de>

    PR tree-optimization/59817
    PR tree-optimization/60453
    * graphite-scop-detection.c (graphite_can_represent_scev): Complete
    recursion to catch all CHRECs in the scalar evolution and restrict
    the predicate for the remains appropriately.

    * gfortran.dg/graphite/pr59817.f: New testcase.
    * gcc.dg/graphite/pr59817-1.c: Likewise.
    * gcc.dg/graphite/pr59817-2.c: Likewise.

    2014-04-17  Richard Biener  <rguenther@suse.de>

    PR tree-optimization/60836
    * tree-vect-loop.c (vect_create_epilog_for_reduction): Force
    initial PHI args to be gimple values.

    * g++.dg/vect/pr60836.cc: New testcase.

Added:
    branches/gcc-4_8-branch/gcc/testsuite/g++.dg/torture/pr60750.C
    branches/gcc-4_8-branch/gcc/testsuite/g++.dg/torture/pr60895.C
    branches/gcc-4_8-branch/gcc/testsuite/g++.dg/vect/pr60836.cc
    branches/gcc-4_8-branch/gcc/testsuite/gcc.dg/graphite/pr59817-1.c
    branches/gcc-4_8-branch/gcc/testsuite/gcc.dg/graphite/pr59817-2.c
    branches/gcc-4_8-branch/gcc/testsuite/gfortran.dg/graphite/pr59817.f
Modified:
    branches/gcc-4_8-branch/gcc/ChangeLog
    branches/gcc-4_8-branch/gcc/graphite-scop-detection.c
    branches/gcc-4_8-branch/gcc/testsuite/ChangeLog
    branches/gcc-4_8-branch/gcc/testsuite/gcc.dg/tree-ssa/20040517-1.c
    branches/gcc-4_8-branch/gcc/tree-cfgcleanup.c
    branches/gcc-4_8-branch/gcc/tree-inline.c
    branches/gcc-4_8-branch/gcc/tree-ssa-operands.c
    branches/gcc-4_8-branch/gcc/tree-vect-loop.c

[Bug tree-optimization/59817] [4.8 Regression] ICE in extract_affine_chrec with -O2 -ftree-loop-linear

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59817

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

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

--- Comment #11 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.