[Bug middle-end/53144] New: [4.8 Regression] gcc.c-torture/execute/vector-compare-1.c

[Bug middle-end/53144] New: [4.8 Regression] gcc.c-torture/execute/vector-compare-1.c

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

             Bug #: 53144
           Summary: [4.8 Regression]
                    gcc.c-torture/execute/vector-compare-1.c
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: hjl.tools@gmail.com
                CC: bonzini@gnu.org


On Linux/x86-64, revision 186905:

http://gcc.gnu.org/ml/gcc-cvs/2012-04/msg00858.html

caused:

WARNING: program timed out.
FAIL: gcc.c-torture/execute/vector-compare-1.c compilation,  -O3
-fomit-frame-pointer 
WARNING: program timed out.
FAIL: gcc.c-torture/execute/vector-compare-1.c compilation,  -O3
-fomit-frame-pointer -funroll
-loops 
WARNING: program timed out.
FAIL: gcc.c-torture/execute/vector-compare-1.c compilation,  -O3
-fomit-frame-pointer -funroll
-all-loops -finline-functions 
WARNING: program timed out.
FAIL: gcc.c-torture/execute/vector-compare-1.c compilation,  -O3 -g

[Bug middle-end/53144] [4.8 Regression] gcc.c-torture/execute/vector-compare-1.c

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

Hans-Peter Nilsson <hp at gcc dot gnu.org> changed:

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

--- Comment #1 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2012-04-28 00:27:12 UTC ---
cris-elf too (186897:186908); no SIMD, FWIW.

[Bug middle-end/53144] [4.8 Regression] gcc.c-torture/execute/vector-compare-1.c

[bonzini at gnu dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

--- Comment #2 from Paolo Bonzini <bonzini at gnu dot org> 2012-04-28 06:19:31 UTC ---
In the case of PR53138 this was a preexisting bug in handling -(a < b), made
more explicit by the new phi-opt conversion.  Looking at vector-compare-1.c and
pass ordering (phiopt runs before PRE, which hangs while doing PPRE) it is
possible that this is the case too.

I'll make a reduced testcase as soon as possible.

[Bug middle-end/53144] [4.7/4.8 Regression] PPRE infinite loop

[bonzini at gnu dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

Paolo Bonzini <bonzini at gnu dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-04-28
      Known to work|                            |4.4.6
            Summary|[4.8 Regression]            |[4.7/4.8 Regression] PPRE
                   |gcc.c-torture/execute/vecto |infinite loop
                   |r-compare-1.c               |
     Ever Confirmed|0                           |1
      Known to fail|                            |4.7.0, 4.8.0

--- Comment #3 from Paolo Bonzini <bonzini at gnu dot org> 2012-04-28 15:46:26 UTC ---
The minimal testcase fails on 4.7 too. PPRE is going through an infinite loop.

typedef unsigned char __attribute__((vector_size(4))) uvec;

int main (int argc, char *argv[]) {
    int i;
    int x = 0;
    uvec uc0 = (uvec) {argc, 1,  2,  10};
    unsigned char uc1[4] = {0, 3, 2, 200};
    signed char ucg[4] = {1, 0, 0, 0 };
    signed char ucl[4] = {0, 1, 0, 1 };

#define uc0_ ((unsigned char *)&uc0)

    for (i = 0; i < 4; i ++) {
      x |= ucg[i] != (uc0_[i] > uc1[i]);
      x |= ucl[i] != (uc0_[i] < uc1[i]);
    }
    return x;
}

H.J., can you try bisecting this?

[Bug middle-end/53144] [4.7/4.8 Regression] PPRE infinite loop

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

--- Comment #4 from H.J. Lu <hjl.tools at gmail dot com> 2012-04-28 15:53:10 UTC ---
(In reply to comment #3)
> 
> H.J., can you try bisecting this?

Sure.  BTW, I created a new git based bisect tool. It works
quite well.

[Bug middle-end/53144] [4.7/4.8 Regression] PPRE infinite loop

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.7.1

--- Comment #5 from H.J. Lu <hjl.tools at gmail dot com> 2012-04-28 18:58:58 UTC ---
It is caused by revision 179593:

http://gcc.gnu.org/ml/gcc-cvs/2011-10/msg00187.html

[Bug middle-end/53144] [4.7/4.8 Regression] PPRE infinite loop

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

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

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

--- Comment #6 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-05-02 10:42:51 UTC ---
Mine.

[Bug middle-end/53144] [4.7/4.8 Regression] PPRE infinite loop

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

--- Comment #7 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-05-02 12:17:48 UTC ---
We are repeatedly inserting

Inserted pretmp.34_83 = MEM[(unsigned char *)&uc0];
 in predecessor 2

causing insert iteration.  Reproduces with -fno-tree-partial-pre, too, so a
regular PRE issue.

[Bug middle-end/53144] [4.7/4.8 Regression] PPRE infinite loop

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

--- Comment #8 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-05-02 14:08:18 UTC ---
I have a patch.

[Bug middle-end/53144] [4.7/4.8 Regression] PPRE infinite loop

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

--- Comment #9 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-05-03 08:37:24 UTC ---
Author: rguenth
Date: Thu May  3 08:37:18 2012
New Revision: 187078

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=187078
Log:
2012-05-03  Richard Guenther  <rguenther@suse.de>

    PR tree-optimization/53144
    * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_constant_for_pieces):
    Rename to ...
    (vn_reference_lookup_or_insert_for_pieces): ... this.  Properly deal
    with SSA name values.
    (vn_reference_lookup_3): Adjust callers.

    * gcc.dg/torture/pr53144.c: New testcase.

Added:
    trunk/gcc/testsuite/gcc.dg/torture/pr53144.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/tree-ssa-sccvn.c

[Bug middle-end/53144] [4.7/4.8 Regression] PPRE infinite loop

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

--- Comment #10 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-05-03 08:49:09 UTC ---
Author: rguenth
Date: Thu May  3 08:49:03 2012
New Revision: 187080

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=187080
Log:
2012-05-03  Richard Guenther  <rguenther@suse.de>

    PR tree-optimization/53144
    * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_constant_for_pieces):
    Rename to ...
    (vn_reference_lookup_or_insert_for_pieces): ... this.  Properly deal
    with SSA name values.
    (vn_reference_lookup_3): Adjust callers.

    * gcc.dg/torture/pr53144.c: New testcase.

Added:
    branches/gcc-4_7-branch/gcc/testsuite/gcc.dg/torture/pr53144.c
Modified:
    branches/gcc-4_7-branch/gcc/ChangeLog
    branches/gcc-4_7-branch/gcc/testsuite/ChangeLog
    branches/gcc-4_7-branch/gcc/tree-ssa-sccvn.c

[Bug middle-end/53144] [4.7/4.8 Regression] PPRE infinite loop

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53144

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
      Known to work|                            |4.7.1, 4.8.0
         Resolution|                            |FIXED

--- Comment #11 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-05-03 08:50:16 UTC ---
Fixed.