[Patch, Fortran, F03] PR 80983: memory leak when calling procedure-pointer component with allocatable result

[Patch, Fortran, F03] PR 80983: memory leak when calling procedure-pointer component with allocatable result

[Janus Weil]

[-- Attachment #1: Type: text/plain, Size: 772 bytes --]

Hi all,

the attached patch fixes a runtime memory leak with procedure-pointer
components (PPCs). gfortran correctly deallocates scalar allocatable
function results (also for procedure pointers), but fails to do so for
PPCs, which is cured by my patch.

[Note: Since gfortran translates any type-bound procedure call into a
PPC call internally, the patch also cures memory leaks with TBPs.]

It regtests cleanly on x86_64-linux-gnu. Ok for trunk?

Cheers,
Janus


2017-06-15  Janus Weil  <janus@gcc.gnu.org>

    PR fortran/70983
    * trans-expr.c (gfc_conv_procedure_call): Deallocate the result of
    scalar allocatable procedure-pointer components.


2017-06-15  Janus Weil  <janus@gcc.gnu.org>

    PR fortran/70983
    * gfortran.dg/proc_ptr_comp_51.f90: New test.

[-- Attachment #2: pr80983.diff --]
[-- Type: text/plain, Size: 745 bytes --]

Index: gcc/fortran/trans-expr.c
===================================================================
--- gcc/fortran/trans-expr.c	(revision 249130)
+++ gcc/fortran/trans-expr.c	(working copy)
@@ -6132,7 +6132,8 @@ gfc_conv_procedure_call (gfc_se * se, gfc_symbol *
      after use. This necessitates the creation of a temporary to
      hold the result to prevent duplicate calls.  */
   if (!byref && sym->ts.type != BT_CHARACTER
-      && sym->attr.allocatable && !sym->attr.dimension && !comp)
+      && ((sym->attr.allocatable && !sym->attr.dimension && !comp)
+	  || (comp && comp->attr.allocatable && !comp->attr.dimension)))
     {
       tmp = gfc_create_var (TREE_TYPE (se->expr), NULL);
       gfc_add_modify (&se->pre, tmp, se->expr);

[-- Attachment #3: proc_ptr_comp_51.f90 --]
[-- Type: text/x-fortran, Size: 824 bytes --]

! { dg-do compile }
!
! PR 80983: [F03] memory leak when calling procedure-pointer component with allocatable result
!
! Contributed by Janus Weil <janus@gcc.gnu.org>

program test
  implicit none

  type :: concrete_type
    procedure (alloc_integer), pointer, nopass :: alloc
  end type

  procedure (alloc_integer), pointer :: pp

  type(concrete_type) :: concrete

  print *, alloc_integer()     ! case #1: plain function

  pp => alloc_integer
  print *, pp()                ! case #2: procedure pointer

  concrete % alloc => alloc_integer
  print *, concrete % alloc()  ! case #3: procedure-pointer component

contains

   function alloc_integer() result(res)
      integer, allocatable :: res
      allocate(res, source=13)
   end function

end

! { dg-final { scan-tree-dump-times "__builtin_free" 3 "original" } }

Re: [Patch, Fortran, F03] PR 80983: memory leak when calling procedure-pointer component with allocatable result

[Thomas Koenig]

Hi Janus,

> It regtests cleanly on x86_64-linux-gnu. Ok for trunk?

OK.

Thanks for the patch!

Regards

	Thomas

Re: [Patch, Fortran, F03] PR 80983: memory leak when calling procedure-pointer component with allocatable result

[Janus Weil]

2017-06-15 15:33 GMT+02:00 Thomas Koenig <tkoenig@netcologne.de>:
> Hi Janus,
>
>> It regtests cleanly on x86_64-linux-gnu. Ok for trunk?
>
>
> OK.
>
> Thanks for the patch!

Thanks, Thomas! Committed as r249227.

Cheers,
Janus

Re: [Patch, Fortran, F03] PR 80983: memory leak when calling procedure-pointer component with allocatable result

[Christophe Lyon]

Hi,

On 15 June 2017 at 23:18, Janus Weil <janus@gcc.gnu.org> wrote:
> 2017-06-15 15:33 GMT+02:00 Thomas Koenig <tkoenig@netcologne.de>:
>> Hi Janus,
>>
>>> It regtests cleanly on x86_64-linux-gnu. Ok for trunk?
>>
>>
>> OK.
>>
>> Thanks for the patch!
>
> Thanks, Thomas! Committed as r249227.
>

There's a problem with the new test. It says:
gfortran.dg/proc_ptr_comp_51.f90   -O  : dump file does not exist
UNRESOLVED: gfortran.dg/proc_ptr_comp_51.f90   -O
scan-tree-dump-times original "__builtin_free" 3

You probably either want to add:
! { dg-options "-fdump-tree-original" }
or to remove:
! { dg-final { scan-tree-dump-times "__builtin_free" 3 "original" } }

Thanks,

Christophe

> Cheers,
> Janus

Re: [Patch, Fortran, F03] PR 80983: memory leak when calling procedure-pointer component with allocatable result

[Janus Weil]

2017-06-16 11:11 GMT+02:00 Christophe Lyon <christophe.lyon@linaro.org>:
> Hi,
>
> On 15 June 2017 at 23:18, Janus Weil <janus@gcc.gnu.org> wrote:
>> 2017-06-15 15:33 GMT+02:00 Thomas Koenig <tkoenig@netcologne.de>:
>>> Hi Janus,
>>>
>>>> It regtests cleanly on x86_64-linux-gnu. Ok for trunk?
>>>
>>>
>>> OK.
>>>
>>> Thanks for the patch!
>>
>> Thanks, Thomas! Committed as r249227.
>>
>
> There's a problem with the new test. It says:
> gfortran.dg/proc_ptr_comp_51.f90   -O  : dump file does not exist
> UNRESOLVED: gfortran.dg/proc_ptr_comp_51.f90   -O
> scan-tree-dump-times original "__builtin_free" 3
>
> You probably either want to add:
> ! { dg-options "-fdump-tree-original" }
> or to remove:
> ! { dg-final { scan-tree-dump-times "__builtin_free" 3 "original" } }

Thanks for your attention, Christophe. I wanted the first of these two
options, but apparently forgot that line ... :(

Fixed with r249243.

Cheers,
Janus