[Bug fortran/66079] [6 Regression] memory leak with source allocation in internal subprogram

[Bug fortran/66079] [6 Regression] memory leak with source allocation in internal subprogram

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |6.0
            Summary|[6.0 Regression] memory     |[6 Regression] memory leak
                   |leak with source allocation |with source allocation in
                   |in internal subprogram      |internal subprogram

[Bug fortran/66079] [6 Regression] memory leak with source allocation in internal subprogram

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

Paul Thomas <pault at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2015-05-16
                 CC|                            |pault at gcc dot gnu.org
           Assignee|unassigned at gcc dot gnu.org      |pault at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #1 from Paul Thomas <pault at gcc dot gnu.org> ---
Hi Damian,

The reason for this memory leak is rather obvious:

newrealvec ()
{
  struct subdata * d;

  try
    {
      d = 0B;
      {
        struct subdata D.3386;
        struct subdata D.3385;
        struct subdata subdata.0;

        subdata.0.b = (integer(kind=4) *) __builtin_malloc (4);
        *subdata.0.b = 0;
        D.3385 = subdata.0;
        D.3386 = D.3385;
        if (d != 0B)
          {
            _gfortran_runtime_error_at (&"At line 10 of file
pr66079.f90"[1]{lb: 1 sz: 1}, &"Attempting to allocate already allocated
variable \'%s\'"[1]{lb: 1 sz: 1}, &"d"[1]{lb: 1 sz: 1});
          }
        else
          {
            d = (struct subdata *) __builtin_malloc (8);
            if (d == 0B)
              {
                _gfortran_os_error (&"Allocation would exceed memory
limit"[1]{lb: 1 sz: 1});
              }
          }
        d->b = 0B;
        {
          struct subdata subdata.1;

          subdata.1.b = (integer(kind=4) *) __builtin_malloc (4);
          *subdata.1.b = 0;
          *d = subdata.1;
        }
      }
    }
  finally
    {
      if (d != 0B)
        {
          if (d->b != 0B)
            {
              __builtin_free ((void *) d->b);
            }
          d->b = 0B;
          __builtin_free ((void *) d);
        }
    }
}

subdata.1 gets freed in the finally block but subdata.0, which is redundant in
any case, does not.

Confirmed - thanks for the report.... I just have to find out why it is we wind
up with two subdata's.

Cheers

Paul

[Bug fortran/66079] [6 Regression] memory leak with source allocation in internal subprogram

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

--- Comment #2 from Paul Thomas <pault at gcc dot gnu.org> ---
Created attachment 35555
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35555&action=edit
Initial fix

This is a partial fix, in that it deals with the initial problem.

  type subdata
    integer, allocatable :: b
  endtype
  block
    call newRealVec
  end block
contains
  subroutine newRealVec
    type(subdata), allocatable :: d, e, f
    allocate(d,source=subdata(1)) ! was lost, now OK
    allocate(e,source=d) ! OK
    allocate(f,source=create (99)) ! 8 bytes go AWOL
    if (d%b .ne. 1) call abort
    if (e%b .ne. 1) call abort
    if (f%b .ne. 99) call abort
    f = create (42) ! another 8 bytes go AWOL
    print *, f%b
  end subroutine
  function create (arg) result(res)
    integer :: arg
    type(subdata), allocatable :: res
    allocate(res, source = subdata(arg))
  end function
end

The function calls are losing the allocation for the subdata type itself. The b
component is correctly dealt with. I know what has to be done to fix this.

Arrays are a mess.

allocate (d(2), source = subdata ()) ICEs on a double free
allocate (d(2), source = [subdata (1), subdata (2)]) loses 8 bytes in two
chunks.

Watch this space.

Paul

[Bug fortran/66079] [6 Regression] memory leak with source allocation in internal subprogram

[damian at sourceryinstitute dot org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

--- Comment #3 from Damian Rouson <damian at sourceryinstitute dot org> ---
Thanks for working on this.

Damian

[Bug fortran/66079] [6 Regression] memory leak with source allocation in internal subprogram

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

--- Comment #4 from Paul Thomas <pault at gcc dot gnu.org> ---
Created attachment 35583
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35583&action=edit
Further advanced patch

This version fixes non-character allocatable scalar functions. Now, only
allocatable character functions leak memory.

Paul

[Bug fortran/66079] [6 Regression] memory leak with source allocation in internal subprogram

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

--- Comment #5 from Paul Thomas <pault at gcc dot gnu.org> ---
Created attachment 35603
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35603&action=edit
Nearly OK patch

The attached bootstraps and regtests on 6.0.0. The following runs without
memory leaks and fixes problems as indicated:

  type subdata
    integer, allocatable :: b
  endtype
!  block
    call newRealVec
!  end block
contains
  subroutine newRealVec
    type(subdata), allocatable :: d, e, f
    character(:), allocatable :: g, h, i
    allocate(d,source=subdata(1)) ! memory was lost, now OK
    allocate(e,source=d) ! OK
    allocate(f,source=create (99)) ! memory was lost, now OK
    if (d%b .ne. 1) call abort
    if (e%b .ne. 1) call abort
    if (f%b .ne. 99) call abort
    allocate (g, source = greeting1("good day"))
    if (g .ne. "good day") call abort
    allocate (h, source = greeting2("hello"))
    if (h .ne. "hello") call abort
    allocate (i, source = greeting3("hiya!"))
    if (i .ne. "hiya!") call abort
  end subroutine

  function create (arg) result(res)
    integer :: arg
    type(subdata), allocatable :: res
    allocate(res, source = subdata(arg))
  end function

  function greeting1 (arg) result(res) ! memory was lost, now OK
    character(*) :: arg
    Character(:), allocatable :: res
    allocate(res, source = arg)
  end function

  function greeting2 (arg) result(res)
    character(5) :: arg
    Character(:), allocatable :: res
    allocate(res, source = arg)
  end function

  function greeting3 (arg) result(res)
    character(5) :: arg
    Character(5), allocatable :: res, res1
    allocate(res, source = arg) ! Caused an ICE - now OK
    allocate(res1, source = arg) ! Caused an ICE - now OK
    if (res1 .ne. res) call abort
  end function
end

Just a small amount of cleaning up to do before submission.

Paul

[Bug fortran/66079] [6 Regression] memory leak with source allocation in internal subprogram

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

--- Comment #6 from Paul Thomas <pault at gcc dot gnu.org> ---
Author: pault
Date: Thu Jun 11 15:49:32 2015
New Revision: 224383

URL: https://gcc.gnu.org/viewcvs?rev=224383&root=gcc&view=rev
Log:
2015-06-11  Paul Thomas  <pault@gcc.gnu.org>

        PR fortran/66079
        * trans-expr.c (gfc_conv_procedure_call): Allocatable scalar
        function results must be freed and nullified after use. Create
        a temporary to hold the result to prevent duplicate calls.
        * trans-stmt.c (gfc_trans_allocate): Rename temporary variable
        as 'source'. Deallocate allocatable components of non-variable
        'source's.

2015-06-11  Paul Thomas  <pault@gcc.gnu.org>

        PR fortran/66079
        * gfortran.dg/allocatable_scalar_13.f90: New test


Added:
    trunk/gcc/testsuite/gfortran.dg/allocatable_scalar_13.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/trans-expr.c
    trunk/gcc/fortran/trans-stmt.c
    trunk/gcc/testsuite/ChangeLog

[Bug fortran/66079] [6 Regression] memory leak with source allocation in internal subprogram

[damian at sourceryinstitute dot org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

--- Comment #7 from Damian Rouson <damian at sourceryinstitute dot org> ---
Thanks for the fix!  I'll update my source, rebuild, and test the fix.

Damian

[Bug fortran/66079] [6 Regression] memory leak with source allocation in internal subprogram

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

Mikael Morin <mikael at gcc dot gnu.org> changed:

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

--- Comment #10 from Mikael Morin <mikael at gcc dot gnu.org> ---
(In reply to Paul Thomas from comment #8)
> Created attachment 36113 [details]
> Patch for th 5 branch
> 
Doesn't look like a patch. :-/

[Bug fortran/66079] [6 Regression] memory leak with source allocation in internal subprogram

[paul.richard.thomas at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

--- Comment #11 from paul.richard.thomas at gmail dot com <paul.richard.thomas at gmail dot com> ---
Hi Mikael,

It looks like my finger slipped on the mouse wheel - I'll put it right tonight.

Thanks

Paul

On 4 August 2015 at 11:27, mikael at gcc dot gnu.org
<gcc-bugzilla@gcc.gnu.org> wrote:
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079
>
> Mikael Morin <mikael at gcc dot gnu.org> changed:
>
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>                  CC|                            |mikael at gcc dot gnu.org
>
> --- Comment #10 from Mikael Morin <mikael at gcc dot gnu.org> ---
> (In reply to Paul Thomas from comment #8)
>> Created attachment 36113 [details]
>> Patch for th 5 branch
>>
> Doesn't look like a patch. :-/
>
> --
> You are receiving this mail because:
> You are on the CC list for the bug.
> You are the assignee for the bug.

[Bug fortran/66079] [5 Regression] memory leak with source allocation in internal subprogram

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|6.0                         |5.3

[Bug fortran/66079] [5 Regression] memory leak with source allocation in internal subprogram

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P4

[Bug fortran/66079] [5 Regression] memory leak with source allocation in internal subprogram

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

--- Comment #12 from Paul Thomas <pault at gcc dot gnu.org> ---
Author: pault
Date: Sun Oct 18 18:36:52 2015
New Revision: 228952

URL: https://gcc.gnu.org/viewcvs?rev=228952&root=gcc&view=rev
Log:
2015-10-18  Paul Thomas  <pault@gcc.gnu.org>

        PR fortran/66079
        * trans-expr.c (gfc_conv_procedure_call): Allocatable scalar
        function results must be freed and nullified after use. Create
        a temporary to hold the result to prevent duplicate calls.
        * trans-stmt.c (gfc_trans_allocate): Rename temporary variable
        as 'source'. Deallocate allocatable components of non-variable
        'source's.

2015-10-18  Paul Thomas  <pault@gcc.gnu.org>

        PR fortran/66079
        * gfortran.dg/allocatable_scalar_13.f90: New test


Added:
    branches/gcc-5-branch/gcc/testsuite/gfortran.dg/allocatable_scalar_13.f90
Modified:
    branches/gcc-5-branch/gcc/fortran/ChangeLog
    branches/gcc-5-branch/gcc/fortran/trans-expr.c
    branches/gcc-5-branch/gcc/fortran/trans-stmt.c
    branches/gcc-5-branch/gcc/testsuite/ChangeLog

[Bug fortran/66079] [5 Regression] memory leak with source allocation in internal subprogram

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66079

Paul Thomas <pault at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #13 from Paul Thomas <pault at gcc dot gnu.org> ---
Closing after commits to trunk and 5 branch.

Thanks for the report.

Paul