[Bug fortran/27524] New: -fbounds-check interracts *strangely* with an array of size 1

[Bug fortran/27524] New: -fbounds-check interracts *strangely* with an array of size 1

[P dot Schaffnit at access dot rwth-aachen dot de]

Hi!

I've spotted something quite strange with -fbounds-check and an array which
happens to have a size of 1 (a quite unlikely configuration admitedly, but
could be the tip of s(omething worse, worth investigating I guess).

I was able to reproduce this behaviour with a small variation on Paul Thomas's
example for PR 27124 (is it related to that in any way?)): if the size of the
array is 1 *and* -fbounds-check is specified it returns some junk (1075191960
for me), and otherwise (I mean, another size of the array and/or without
-fbounds-check) it returns the expected 10.

Philippe

PS: a small example exhibiting this behaviour:
  PROGRAM Test
    IMPLICIT NONE
    INTEGER, PARAMETER :: Array_Size = 1
    INTEGER :: Array(Array_Size)
    Array = 100
    Array = Function_Test ( Array, 10 )
    WRITE ( 6, * ) Array
    CONTAINS
      FUNCTION Function_Test ( Array, Scalar )
        INTEGER, INTENT(IN) :: Array(1:Array_Size), Scalar
        INTEGER :: Function_Test(1:Array_Size)
        Function_Test = Array / Scalar
      END FUNCTION Function_Test
  END PROGRAM Test

PPS: how I start it: gfortran -g -fbounds-check Test.f90 -o Test ; ./Test

PPPS: I'm using reasonably up-to-date sources:
gfortran --v
Using built-in specs.
Target: i686-pc-linux-gnu
Configured with: /USER/philippe/Irix/Gcc_Sources/configure
--prefix=/usr1/MICRESS/Philippe/Tools/Gcc --enable-languages=c,fortran
--disable-maintainer-mode --disable-shared
--with-mpfr=usr1/MICRESS/Philippe/Tools/Mpfr
--with-gmp=/usr1/MICRESS/Philippe/Tools/Gmp
Thread model: posix
gcc version 4.2.0 20060508 (experimental)


-- 
           Summary: -fbounds-check interracts *strangely* with an array of
                    size 1
           Product: gcc
           Version: 4.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: P dot Schaffnit at access dot rwth-aachen dot de
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27524

[Bug fortran/27524] -fbounds-check interracts *strangely* with an array of size 1

[P dot Schaffnit at access dot rwth-aachen dot de]

------- Comment #1 from P dot Schaffnit at access dot rwth-aachen dot de  2006-05-10 11:34 -------

Sorry, I missed it before, but this could definitly be a dupplicate of 19777.

Philippe


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27524

[Bug libfortran/27524] -fbounds-check interracts *strangely* with an array of size 1

[fxcoudert at gcc dot gnu dot org]

------- Comment #2 from fxcoudert at gcc dot gnu dot org  2006-05-19 16:59 -------
Not sure if it's a duplicate. Here's a reduced testcase, anyhow:

    print *, F()
    contains
      function F()
        integer :: F(1)
        f = 1
      end function F
    end

$ gfortran -v
Using built-in specs.
Target: i386-linux
Configured with: ../gcc/configure --prefix=/tmp/gfortran-20060519/irun
--enable-languages=c,fortran --host=i386-linux
--with-gmp=/tmp/gfortran-20060519/gfortran_libs
Thread model: posix
gcc version 4.2.0 20060519 (experimental)
$ gfortran bounds.f90 && ./a.out               
           1
$ gfortran bounds.f90 -fbounds-check && ./a.out
     5090100
zsh: bus error  ./a.out


-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |fxcoudert at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2006-05-19 16:59:16
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27524

[Bug libfortran/27524] -fbounds-check interracts *strangely* with an array of size 1

[fxcoudert at gcc dot gnu dot org]

------- Comment #3 from fxcoudert at gcc dot gnu dot org  2006-05-24 09:20 -------
>       function F()
>         integer :: F(1)
>         f = 1
>       end function F

Compiling this simple function without and with -fbounds-check and dumping the
GIMPLE tree generated gives the following diff:

--- 003t.original.noboundscheck 2006-05-24 11:09:37.000000000 +0200
+++ 003t.original.boundscheck   2006-05-24 11:09:32.000000000 +0200
@@ -14,8 +14,16 @@ f (__result)
     D.917 = __result->dim[0].stride;
     stride.0 = D.917 == 0 ? 1 : D.917;
     __result.0 = (int4[0:D.918] *) __result->data;
+    if ((logical4) __builtin_expect (__result->dim[0].ubound !=
__result->dim[0].lbound, 0))
+      {
+        _gfortran_runtime_error ("Array bound mismatch", "test.f90", 1);
+      }
+    else
+      {
+        (void) 0;
+      }
     size.2 = NON_LVALUE_EXPR <stride.0>;
-    offset.1 = -stride.0;
+    offset.1 = __result->dim[0].lbound - __result->dim[0].ubound;
     D.918 = size.2 - 1;
     D.919 = (bit_size_type) (<unnamed type>) size.2 * 


I think the cuplrit here is gfc_trans_dummy_array_bias(), around line 3728
(where the offset is built), although I'm not skilled enough in all this to
understand how exactly to fix it.


-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2006-05-19 16:59:16         |2006-05-24 09:20:48
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27524

[Bug libfortran/27524] -fbounds-check interacts with array function

[fxcoudert at gcc dot gnu dot org]

------- Comment #4 from fxcoudert at gcc dot gnu dot org  2006-05-24 20:52 -------
(In reply to comment #3)
> I think the cuplrit here is gfc_trans_dummy_array_bias(), around line 3728
> (where the offset is built), although I'm not skilled enough in all this to
> understand how exactly to fix it.

OK, I think I made it. It's a typo, where the stride variable is used as a
temporary variable for creating a bounds-checking tree, but its value is needed
(and reused) later, leading to the bug.


Index: trans-array.c
===================================================================
--- trans-array.c       (revision 113849)
+++ trans-array.c       (working copy)
@@ -3564,7 +3564,7 @@
   tree dumdesc;
   tree tmp;
   tree stmt;
-  tree stride;
+  tree stride, stride2;
   tree stmt_packed;
   tree stmt_unpacked;
   tree partial;
@@ -3711,9 +3711,9 @@

              tmp = fold_build2 (MINUS_EXPR, gfc_array_index_type,
                                 ubound, lbound);
-              stride = build2 (MINUS_EXPR, gfc_array_index_type,
+              stride2 = build2 (MINUS_EXPR, gfc_array_index_type,
                               dubound, dlbound);
-              tmp = fold_build2 (NE_EXPR, gfc_array_index_type, tmp, stride);
+              tmp = fold_build2 (NE_EXPR, gfc_array_index_type, tmp, stride2);
              gfc_trans_runtime_check (tmp, gfc_strconst_bounds, &block);
            }
        }


-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |fxcoudert at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2006-05-24 09:20:48         |2006-05-24 20:52:53
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27524

[Bug libfortran/27524] -fbounds-check interacts with array function

[fxcoudert at gcc dot gnu dot org]

------- Comment #5 from fxcoudert at gcc dot gnu dot org  2006-05-26 21:18 -------
Subject: Bug 27524

Author: fxcoudert
Date: Fri May 26 21:18:45 2006
New Revision: 114142

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=114142
Log:
        PR fortran/27524

        * trans-array.c (gfc_trans_dummy_array_bias): Don't use stride as
        a temporary variable when -fbounds-check is enabled, since its
        value will be needed later.

        * gfortran.dg/bounds_check_1.f90: New test.

Added:
    trunk/gcc/testsuite/gfortran.dg/bounds_check_1.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/trans-array.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27524

[Bug libfortran/27524] [4.1 only] -fbounds-check interacts with array function

[fxcoudert at gcc dot gnu dot org]

-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|                            |4.1.2
      Known to work|                            |4.2.0
            Summary|-fbounds-check interacts    |[4.1 only] -fbounds-check
                   |with array function         |interacts with array
                   |                            |function
   Target Milestone|---                         |4.1.2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27524

[Bug libfortran/27524] [4.1 only] -fbounds-check interacts with array function

[fxcoudert at gcc dot gnu dot org]

------- Comment #6 from fxcoudert at gcc dot gnu dot org  2006-05-27 08:29 -------
*** Bug 27760 has been marked as a duplicate of this bug. ***


-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |vivekrao4 at yahoo dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27524

[Bug libfortran/27524] [4.1 only] -fbounds-check interacts with array function

[fxcoudert at gcc dot gnu dot org]

------- Comment #7 from fxcoudert at gcc dot gnu dot org  2006-06-03 16:07 -------
Subject: Bug 27524

Author: fxcoudert
Date: Sat Jun  3 16:06:51 2006
New Revision: 114342

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=114342
Log:
        PR fortran/19777
        * trans-array.c (gfc_conv_array_ref): Don't perform out-of-bounds
        checking for assumed-size arrrays.
        * gfortran.dg/bounds_check_2.f: New test.

        PR fortran/27524
        * trans-array.c (gfc_trans_dummy_array_bias): Don't use stride as
        a temporary variable when -fbounds-check is enabled, since its
        value will be needed later.
        * gfortran.dg/bounds_check_1.f90: New test.

Added:
    branches/gcc-4_1-branch/gcc/testsuite/gfortran.dg/bounds_check_1.f90
      - copied unchanged from r114142,
trunk/gcc/testsuite/gfortran.dg/bounds_check_1.f90
    branches/gcc-4_1-branch/gcc/testsuite/gfortran.dg/bounds_check_2.f
      - copied unchanged from r114153,
trunk/gcc/testsuite/gfortran.dg/bounds_check_2.f
Modified:
    branches/gcc-4_1-branch/gcc/fortran/ChangeLog
    branches/gcc-4_1-branch/gcc/fortran/trans-array.c
    branches/gcc-4_1-branch/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27524

[Bug libfortran/27524] -fbounds-check interacts with array function

[fxcoudert at gcc dot gnu dot org]

------- Comment #8 from fxcoudert at gcc dot gnu dot org  2006-06-03 16:07 -------
Fixed.


-- 

fxcoudert at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
      Known to fail|4.1.2                       |
      Known to work|4.2.0                       |4.2.0 4.1.2
         Resolution|                            |FIXED
            Summary|[4.1 only] -fbounds-check   |-fbounds-check interacts
                   |interacts with array        |with array function
                   |function                    |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27524