[Bug fortran/50892] New: Internal compiler error: in gimplify_expr, at gimplify.c:7477

[Bug fortran/50892] New: Internal compiler error: in gimplify_expr, at gimplify.c:7477

[fkrogh#gcc at mathalacarte dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50892

             Bug #: 50892
           Summary: Internal compiler error: in gimplify_expr, at
                    gimplify.c:7477
    Classification: Unclassified
           Product: gcc
           Version: 4.6.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: fkrogh#gcc@mathalacarte.com


Created attachment 25641
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25641
File test.f90 gives the internal compile error

I'm using: GNU Fortran (Gentoo 4.6.1-r1 p1.0, pie-0.4.5) 4.6.1
Compilation flags are: -march=native -funroll-loops -pipe \
 -ftree-vectorize -floop-strip-mine \
 -floop-block -fimplicit-none -finit-real=NAN -ggdb

In addition to the internal compiler error, there is a problem mentioned in a
comment.  According to a Fortran language lawyer, Fortran 2008 allows two local
definitions of procedures to have the same binding label.  This problem is
illustrated in a comment.  I'm mainly filing this bug because of the internal
compiler error, which reads as follows:

 test.f90: In function 'test':
test.f90:11:0: internal compiler error: in gimplify_expr, at gimplify.c:7477
Please submit a full bug report,
with preprocessed source if appropriate.

My CPU is: Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz

Thanks,
   Fred

[Bug fortran/50892] Internal compiler error: in gimplify_expr, at gimplify.c:7477

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50892

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011-10-28
     Ever Confirmed|0                           |1
      Known to fail|                            |4.7.0

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-10-28 09:42:57 UTC ---
Confirmed.

#1  0x00000000009ab110 in gimplify_expr (expr_p=0x7ffff5b3d078, 
    pre_p=0x7fffffffbdc8, post_p=0x7fffffff84f8, gimple_test_f=
    0x93fdf9 <is_gimple_val(tree_node*)>, fallback=1)
    at /space/rguenther/src/svn/trunk/gcc/gimplify.c:7599
7599          gcc_assert (!VOID_TYPE_P (TREE_TYPE (*expr_p)));
(gdb) call debug_generic_expr (*expr_p)
*D.1735

#3  0x00000000009840a8 in gimplify_call_expr (expr_p=0x7ffff5a2d7a8, 
    pre_p=0x7fffffffbdc8, want_value=true)
    at /space/rguenther/src/svn/trunk/gcc/gimplify.c:2438
2438                                    EXPR_LOCATION (*expr_p));
(gdb) call debug_generic_expr (*expr_p)
strlen_void (*D.1735)

 <mem_ref 0x7ffff5a2d930
    type <void_type 0x7ffff5a31bd0 void VOID
        align 8 symtab 0 alias set -1 canonical type 0x7ffff5a31bd0
        pointer_to_this <pointer_type 0x7ffff5a31c78>>

    arg 0 <var_decl 0x7ffff5a2c640 D.1735
        type <pointer_type 0x7ffff5a31c78 type <void_type 0x7ffff5a31bd0 void>
            public unsigned DI
            size <integer_cst 0x7ffff5a20ec0 constant 64>
            unit size <integer_cst 0x7ffff5a20ee0 constant 8>
            align 64 symtab 0 alias set -1 canonical type 0x7ffff5a31c78
            pointer_to_this <pointer_type 0x7ffff5a44f18>>
        used unsigned ignored DI file t.f90 line 18 col 0 size <integer_cst
0x7ffff5a20ec0 64> unit size <integer_cst 0x7ffff5a20ee0 8>
        align 64 context <function_decl 0x7ffff5b38600 test>>
    arg 1 <integer_cst 0x7ffff5a34580 type <pointer_type 0x7ffff5a31c78>
constant 0>
    t.f90:18:0>

[Bug fortran/50892] Internal compiler error: in gimplify_expr, at gimplify.c:7477

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50892

Tobias Burnus <burnus at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-valid-code
                 CC|                            |burnus at gcc dot gnu.org
      Known to fail|                            |4.3.4

--- Comment #2 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-10-29 15:04:34 UTC ---
(In reply to comment #0)
> In addition to the internal compiler error, there is a problem mentioned in a
> comment.  According to a Fortran language lawyer, Fortran 2008 allows two local
> definitions of procedures to have the same binding label.  This problem is
> illustrated in a comment.

I think that's PR 48858.

> I'm mainly filing this bug because of the internal compiler error

Simplified test case below. Dump shows:


test ()
{
  character(kind=1)[1:1] * str;
  void * text;
  static void fortranchar (character(kind=1)[1:] * &, integer(kind=4), void *);

  {
    character(kind=1)[1:MAX_EXPR <D.1732, 0>] * pstr.0;
    integer(kind=4) D.1732;
    void * D.1730;

    D.1730 = text;
    D.1732 = strlen (*D.1730);
    fortranchar (&pstr.0, MAX_EXPR <D.1732, 0>, D.1730);
    str = (character(kind=1)[1:1] *) pstr.0;
  }
}



program test
  use, intrinsic :: ISO_C_Binding, only: c_ptr, c_int, c_char
  type(c_ptr) :: text
  character, pointer :: str
  interface
    pure integer(c_int) function strlen (c) bind(c,name='strlen')
      import
      type(c_ptr), intent(in), value :: c
    end function strlen
  end interface
  str => FortranChar(text)
contains
  function FortranChar ( C )
    type(c_ptr), intent(in), value :: C
    character(len=strlen(c),kind=c_char), pointer :: FortranChar
  end function FortranChar
end program test

[Bug fortran/50892] Internal compiler error: in gimplify_expr, at gimplify.c:7477

[fxcoudert at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=50892

Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> changed:

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

--- Comment #3 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
Further reduced testcase still fails. Shows that it's all about the character
length, not C binding or the C_PTR derived type:

program test
  character, pointer :: str
  str => g(42)
contains
  function g(i)
    integer, value :: i
    character(len=i), pointer :: g
  end function
end