Re: some problems with gcc/testsuite/gfortran.dg/ tests in 4.3-20061111 with OSX 10.3

[Paul Thomas <paulthomas2 at wanadoo dot fr>]

Dominique,
> The file gcc/testsuite/gfortran.dg/actual_array_interface_1.f90 in the 
> latest 4.3 snashot (20061111) is empty, as well as:
Yup! I agree.  Sorry about that.  I have attached the full patch and the 
straight up version of the testcase.  I even checked that it still works 
:-)

I cannot do anything about this until the weekend, at earliest.  If 
anybody else is doing a commit and has the time or inclination, please 
include this missing testcase.
> without using any CPU time. Note that the problem disappears if I remove 
> the PRINT's in the functions SUB and BAR.
The code for sub is:

sub (__result, .__result, i)
{
 int4 ..__result;
 int4 stride.8;
 int4 offset.9;
 int4 size.10;
 char[0:D.1048][1:..__result] * __result.0;
 bit_size_type D.1046;
 <unnamed type> D.1047;
 int4 D.1048;
 bit_size_type D.1049;
 <unnamed type> D.1050;

 ..__result = *i;
 {
   int4 D.1045;

   ..__result = *i;
   D.1045 = __result->dim[0].stride;
   stride.8 = D.1045 == 0 ? 1 : D.1045;
   __result.0 = (char[0:D.1048][1:..__result] *) __result->data;
   size.10 = NON_LVALUE_EXPR <stride.8>;
   offset.9 = -stride.8;
   D.1046 = (bit_size_type) (<unnamed type>) ..__result * 8;
   D.1047 = (<unnamed type>) ..__result;
   D.1048 = size.10 - 1;
   D.1049 = SAVE_EXPR <D.1046> * (bit_size_type) (<unnamed type>) size.10;
   D.1050 = SAVE_EXPR <D.1047> * (<unnamed type>) size.10;
 }
 {
   struct __st_parameter_dt dt_parm.7;

   dt_parm.7.common.filename = "intrinsic_actual_2.f90";
   dt_parm.7.common.line = 24;
   dt_parm.7.common.unit = 6;
   dt_parm.7.common.flags = 128;
   _gfortran_st_write (&dt_parm.7);
   {
     int4 D.1044;

     D.1044 = ..__result;
     _gfortran_transfer_integer (&dt_parm.7, &D.1044, 4);
   }
   _gfortran_st_write_done (&dt_parm.7);
 }
}

which looks OK apart from the repetition of ..__result = *i;  However, I 
can see nothing that would make the program hang - can anybody else?

Thanks

Paul
! { dg-do run }
! Tests the fix for PR29490, in which the interface was getting
! messed up over the lower bound on 't' in 'john' and causing
! an ICE in consequence.
!
! Contributed by Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>
!
  program fred
  !---------------------------------
  real              :: dezz(1:10)
  real, allocatable :: jack(:)
  !
  allocate(jack(10)); jack = 9.
  dezz = john(jack,1)
  print*,'dezz = ',dezz

  contains
    !---------------------------------
    function john(t,il)
    !---------------------------------
    real  :: t(il:)
    real  :: john(1:10)
    john = 10.
    end function john
  end

Index: gcc/fortran/trans-expr.c
===================================================================
*** gcc/fortran/trans-expr.c	(révision 117879)
--- gcc/fortran/trans-expr.c	(copie de travail)
*************** gfc_set_interface_mapping_bounds (stmtbl
*** 1297,1303 ****
    for (n = 0; n < GFC_TYPE_ARRAY_RANK (type); n++)
      {
        GFC_TYPE_ARRAY_STRIDE (type, n) = gfc_conv_array_stride (desc, n);
!       if (GFC_TYPE_ARRAY_UBOUND (type, n) == NULL_TREE)
  	{
  	  dim = gfc_rank_cst[n];
  	  tmp = fold_build2 (MINUS_EXPR, gfc_array_index_type,
--- 1297,1310 ----
    for (n = 0; n < GFC_TYPE_ARRAY_RANK (type); n++)
      {
        GFC_TYPE_ARRAY_STRIDE (type, n) = gfc_conv_array_stride (desc, n);
!       if (GFC_TYPE_ARRAY_LBOUND (type, n) == NULL_TREE)
! 	{
! 	  GFC_TYPE_ARRAY_LBOUND (type, n)
! 		= gfc_conv_descriptor_lbound (desc, dim);
! 	  GFC_TYPE_ARRAY_UBOUND (type, n)
! 		= gfc_conv_descriptor_ubound (desc, dim);
! 	}
!       else if (GFC_TYPE_ARRAY_UBOUND (type, n) == NULL_TREE)
  	{
  	  dim = gfc_rank_cst[n];
  	  tmp = fold_build2 (MINUS_EXPR, gfc_array_index_type,
Index: gcc/testsuite/gfortran.dg/actual_array_interface_1.f90
===================================================================
*** gcc/testsuite/gfortran.dg/actual_array_interface_1.f90	(révision 0)
--- gcc/testsuite/gfortran.dg/actual_array_interface_1.f90	(révision 0)
***************
*** 0 ****
--- 1,27 ----
+ ! { dg-do compile }
+ ! Tests the fix for PR29490, in which the creation of the
+ ! interface expression for the first argument of the call to
+ ! 'john' would cause an ICE because GFC_TYPE_ARRAY_LBOUND
+ ! was NULL.
+ !
+ ! Contributed by Philip Mason <pmason@ricardo.com>
+ !
+   !---------------------------------
+   program fred
+   !---------------------------------
+   real              :: dezz(1:10)
+   real, allocatable :: jack(:)
+   !
+   allocate(jack(10)); jack = 9.
+   dezz = john(jack,1)
+   print*,'dezz = ',dezz
+ 
+   contains
+     !---------------------------------
+     function john(t,il)
+     !---------------------------------
+     real  :: t(il:)
+     real  :: john(1:10)
+     john = 10.
+     end function john
+   end