[Bug fortran/26821] New: ice in varasm.c with certain flags

[Bug fortran/26821] New: ice in varasm.c with certain flags

[jvdelisle at gcc dot gnu dot org]

The following code ices with the given options:

$ cat icey.f
      SUBROUTINE IRCGS2(NC,QA,QB,QC,RTRMS)
      IMPLICIT DOUBLE PRECISION(A-H,O-Z)
      DIMENSION QA(NC), QB(NC), QC(NC), QCM(6000), QBM(6000), QAM(6000),
     * RTRMS(NC)
      IF (NC .GT. 6000) NC = 6000
      DO 30 I = 1,NC
         RTRMSI = RTRMS(I)
         QAM(I) = QA(I) / RTRMSI
         QBM(I) = QB(I) / RTRMSI
         QCM(I) = QC(I) / RTRMSI
   30 CONTINUE
      RETURN
      END

$ gfc -c -march=pentium4 -O -ffast-math -ftree-vectorize  icey.f
icey.f:12: internal compiler error: in output_constant_pool_2, at varasm.c:3339
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.

Adding -fno-automatic and there is no ice.

$ gfc -c -march=pentium4 -O -ffast-math -ftree-vectorize -fno-automatic  icey.f
$

Using -march=k8 instead of -march=pentium4 also ices.

Getting rid of the -march= and the ice goes away.

$ gfc -v Using built-in specs.
Target: i686-pc-linux-gnu
Configured with: ../main/configure --prefix=/home/jerry/gcc/usr
--enable-languages=c,fortran --disable-libmudflap
Thread model: posix
gcc version 4.2.0 20060323 (experimental)


-- 
           Summary: ice in varasm.c with certain flags
           Product: gcc
           Version: 4.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: jvdelisle at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26821

[Bug tree-optimization/26821] ice in varasm.c with certain flags

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2006-03-23 06:41 -------
There is definitely something wrong:
  reciptmp.126 = 1.0e+0 / M*((vector real8 *) rtrms + (vector real8 *)
ivtmp.111){misalignment: 0};


1.0/vector is wrong.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
          Component|fortran                     |tree-optimization


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26821

[Bug tree-optimization/26821] [4.1/4.2 Regression] ice in varasm.c with certain flags

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2006-03-23 06:45 -------
This is a regression from 4.0.3.

The problem is that we are converting vector1/vector2, vector3/vector2 into
temp = 1.0/vector2, vector1*temp, vector3*temp which is wrong as 1.0 is a float
but vector2 is a vector.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bonzini at gnu dot org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
           Keywords|                            |ice-on-valid-code
      Known to fail|                            |4.1.0 4.2.0
      Known to work|                            |4.0.3
   Last reconfirmed|0000-00-00 00:00:00         |2006-03-23 06:45:05
               date|                            |
            Summary|ice in varasm.c with certain|[4.1/4.2 Regression] ice in
                   |flags                       |varasm.c with certain flags
   Target Milestone|---                         |4.1.1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26821

[Bug tree-optimization/26821] [4.1/4.2 Regression] ice in varasm.c with certain flags

[pinskia at gcc dot gnu dot org]

------- Comment #3 from pinskia at gcc dot gnu dot org  2006-03-23 06:51 -------
And here is the C example:
void f(double *__restrict__ qa, double *__restrict__ qb, double *__restrict__
qc,
        double *__restrict__ rtrms)
{
  int i;
  static double qam[6000];
  static double qbm[6000];
  static double qcm[6000];
  for(i=0;i<6000;i++)
  {
    double a = rtrms[i];
    qam[i] = qa[i]/a;
    qbm[i] = qb[i]/a;
    qcm[i] = qc[i]/a;
  }
}


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26821

[Bug tree-optimization/26821] [4.1/4.2 Regression] ice in varasm.c with certain flags

[bonzini at gnu dot org]

-- 

bonzini at gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |bonzini at gnu dot org
                   |dot org                     |
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2006-03-23 06:45:05         |2006-03-23 08:12:34
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26821

[Bug tree-optimization/26821] [4.1/4.2 Regression] ice in varasm.c with certain flags

[bonzini at gnu dot org]

------- Comment #4 from bonzini at gnu dot org  2006-03-23 10:56 -------
Reproducible with  -O -ffast-math -ftree-vectorize -msse -msse2 which means the
testcase can easily go in the vect testsuite.  Testing a patch.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26821

[Bug tree-optimization/26821] [4.1/4.2 Regression] ice in varasm.c with certain flags

[mmitchel at gcc dot gnu dot org]

-- 

mmitchel at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26821

[Bug tree-optimization/26821] [4.1/4.2 Regression] ice in varasm.c with certain flags

[bonzini at gcc dot gnu dot org]

------- Comment #5 from bonzini at gnu dot org  2006-04-18 08:08 -------
Subject: Bug 26821

Author: bonzini
Date: Tue Apr 18 08:08:47 2006
New Revision: 113025

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=113025
Log:
2006-04-18  Paolo Bonzini  <bonzini@gnu.org>

        PR tree-optimization/26821
        * tree-ssa-math-opts.c (get_constant_one): New.
        (insert_reciprocals): Use it.

Added:
    trunk/gcc/testsuite/gcc.dg/vect/vect-recip.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/tree-ssa-math-opts.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26821

[Bug tree-optimization/26821] [4.1 Regression] ice in varasm.c with certain flags

[bonzini at gnu dot org]

------- Comment #7 from bonzini at gnu dot org  2006-04-18 13:25 -------
patch committed.


-- 

bonzini at gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
      Known to work|4.0.3 4.2.0                 |4.0.3 4.1.1 4.2.0
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26821

[Bug tree-optimization/26821] [4.1 Regression] ice in varasm.c with certain flags

[bonzini at gcc dot gnu dot org]

------- Comment #6 from bonzini at gnu dot org  2006-04-18 13:24 -------
Subject: Bug 26821

Author: bonzini
Date: Tue Apr 18 13:24:45 2006
New Revision: 113036

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=113036
Log:
2006-04-18  Paolo Bonzini  <bonzini@gnu.org>

        PR tree-optimization/26821

        Backport from mainline:

        2006-04-18  Paolo Bonzini  <bonzini@gnu.org>

        PR tree-optimization/26821
        * tree-ssa-math-opts.c (get_constant_one): New.
        (insert_reciprocals): Use it.

        2006-02-02  Paolo Bonzini  <bonzini@gnu.org>

        * tree-flow-inline.h (bsi_after_labels): Rewrite, return
        what its name says.
        * lambda-code.c (perfect_nestify): Use bsi_insert_before on
        bsi_after_labels iterator.
        * tree-if-conv.c (find_phi_replacement_condition,
        replace_phi_with_cond_modify_expr): Likewise.
        * tree-scalar-evolution.c (scev_const_prop): Likewise.
        * tree-ssa-loop-ivopts.c (compute_phi_arg_on_exit): Likewise.   

2006-04-18  Paolo Bonzini  <bonzini@gnu.org>

        PR tree-optimization/26821
        * gcc.dg/vect/vect-recip.c: New test.


Added:
    branches/gcc-4_1-branch/gcc/testsuite/gcc.dg/vect/vect-recip.c
Modified:
    branches/gcc-4_1-branch/gcc/ChangeLog
    branches/gcc-4_1-branch/gcc/lambda-code.c
    branches/gcc-4_1-branch/gcc/testsuite/ChangeLog
    branches/gcc-4_1-branch/gcc/tree-flow-inline.h
    branches/gcc-4_1-branch/gcc/tree-if-conv.c
    branches/gcc-4_1-branch/gcc/tree-scalar-evolution.c
    branches/gcc-4_1-branch/gcc/tree-ssa-loop-ivopts.c
    branches/gcc-4_1-branch/gcc/tree-ssa-math-opts.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26821