[Bug c++/48869] New: OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[Bug c++/48869] New: OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[bbenin at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

           Summary: OpenMP task construct fails to instantiate copy
                    constructor(same as Bug 36523)
           Product: gcc
           Version: 4.5.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: bbenin@gmail.com


The same bug like in gcc 4.4.0 Bug 36523.
The following test case fails with a linker error. It compiles okay with the
pragma omp task line removed.

$ cat task.cpp 
template<typename T>
struct A
{
  A() { }
  A(const A&) { }
  void foo() { }
};

int main()
{
  A<int> a;
  #pragma omp task
    a.foo();
  return 0;
}

$ g++ -fopenmp -Wall task.cpp 
/tmp/cc5epalM.o: In function `main.omp_cpyfn.1':
task.cpp:(.text+0x73): undefined reference to `A<int>::A(A<int> const&)'
collect2: ld returned 1 exit status

$ g++ -v
Using built-in specs.
COLLECT_GCC=g++
COLLECT_LTO_WRAPPER=/usr/lib/x86_64-linux-gnu/gcc/x86_64-linux-gnu/4.5.2/lto-wrapper
Target: x86_64-linux-gnu
.......
Thread model: posix
gcc version 4.5.2 (Ubuntu/Linaro 4.5.2-8ubuntu4)

[Bug c++/48869] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

Paolo Carlini <paolo.carlini at oracle dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011.05.04 13:06:58
                 CC|                            |jakub at gcc dot gnu.org
     Ever Confirmed|0                           |1

--- Comment #1 from Paolo Carlini <paolo.carlini at oracle dot com> 2011-05-04 13:06:58 UTC ---
Confirmed mainline and 4_6-branch.

[Bug c++/48869] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jason at gcc dot gnu.org,
                   |                            |rguenth at gcc dot gnu.org,
                   |                            |rth at gcc dot gnu.org

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-05-04 15:31:38 UTC ---
Broken by:
http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=149750
i.e. gimplification unit-a-time.  The problem is we won't know until
gimplification whether some non-POD vars will be firstprivate or not, and when
gimplification has been postponed so late, instantiate_pending_templates won't
be called any longer.
Unfortunately, I think duplicating most of the OpenMP related code from
gimplify.c to find out implicitly firstprivate vars in task constructs earlier
would be very hard and huge amount of code duplication, on the other side if we
wanted to instantiate the missing templates late, we'd have to basically repeat
the whole do { ... } while (reconsider); loop from
cp_write_global_declarations from within some langhook (I don't think we can do
that while in the middle of gimplifying something).
Richard/Richard/Jason, any ideas?

[Bug c++/48869] [4.5/4.6/4.7 Regression] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.4.6
   Target Milestone|---                         |4.5.4
            Summary|OpenMP task construct fails |[4.5/4.6/4.7 Regression]
                   |to instantiate copy         |OpenMP task construct fails
                   |constructor(same as Bug     |to instantiate copy
                   |36523)                      |constructor(same as Bug
                   |                            |36523)
      Known to fail|                            |4.5.3, 4.6.0, 4.7.0

[Bug c++/48869] [4.5/4.6/4.7 Regression] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[rguenther at suse dot de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

--- Comment #3 from rguenther at suse dot de <rguenther at suse dot de> 2011-05-04 15:46:47 UTC ---
On Wed, 4 May 2011, jakub at gcc dot gnu.org wrote:

> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869
> 
> Jakub Jelinek <jakub at gcc dot gnu.org> changed:
> 
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>                  CC|                            |jason at gcc dot gnu.org,
>                    |                            |rguenth at gcc dot gnu.org,
>                    |                            |rth at gcc dot gnu.org
> 
> --- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-05-04 15:31:38 UTC ---
> Broken by:
> http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=149750
> i.e. gimplification unit-a-time.  The problem is we won't know until
> gimplification whether some non-POD vars will be firstprivate or not, and when
> gimplification has been postponed so late, instantiate_pending_templates won't
> be called any longer.
> Unfortunately, I think duplicating most of the OpenMP related code from
> gimplify.c to find out implicitly firstprivate vars in task constructs earlier
> would be very hard and huge amount of code duplication, on the other side if we
> wanted to instantiate the missing templates late, we'd have to basically repeat
> the whole do { ... } while (reconsider); loop from
> cp_write_global_declarations from within some langhook (I don't think we can do
> that while in the middle of gimplifying something).
> Richard/Richard/Jason, any ideas?

Can we just always instantiate the templates (always assuming they will be
firstprivate) or are we strictly bound to only instantiate when we
know they will be used?  Thus, how does the OMP spec interact with C++
template instantiation rules?

I don't like the idea of instantiating templates when we started
gimplification ... can we move related analysis to the frontend
somewhere so we know whether things are firstprivate or not there?

Richard.

[Bug c++/48869] [4.5/4.6/4.7 Regression] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-05-04 16:25:46 UTC ---
Not sure if it would be ok to request instantiation of say copy ctor when we
are unsure if the var will be firstprivate or not.  The copy ctor can be marked
as =delete, or perhaps say the instantiation of the copy ctor could result in
static assertion failures or could use some =delete method...

[Bug c++/48869] [4.5/4.6/4.7 Regression] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-05-04 16:37:40 UTC ---
Say:
template<typename T>
struct B
{
  B () {}
  B (const B&) = delete;
  ~B () {}
};

template<typename T>
struct A
{
  A () {}
  A (const A&) { B<T> b; B<T> c = b; }
  void foo () {}
  ~A() {}
};

int
main()
{
  A<int> a;
  #pragma omp task shared(a)
    a.foo ();
  #pragma omp task default(shared)
    a.foo ();
  #pragma omp parallel shared(a)
    #pragma omp task
      a.foo ();
#if 0
  #pragma omp task
    a.foo ();
  #pragma omp parallel private (a)
    #pragma omp task
      a.foo ();
#endif
  return 0;
}

With #if 0 all the a's are shared, even though it isn't immediately obvious in
all cases that it is that way without going through all the gimplify.c rules.
With #if 1 instead of #if 0 those two last tasks have implicit firstprivate(a)
and thus need to instantiate the copy ctor.

[Bug c++/48869] [4.5/4.6/4.7 Regression] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[jason at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

--- Comment #6 from Jason Merrill <jason at gcc dot gnu.org> 2011-05-04 17:46:33 UTC ---
I think it would be fine to mark_used the copy ctor in cases where it might or
might not be used.

[Bug c++/48869] [4.5/4.6/4.7 Regression] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
         AssignedTo|unassigned at gcc dot       |jakub at gcc dot gnu.org
                   |gnu.org                     |

--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-05-11 08:05:31 UTC ---
Created attachment 24221
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=24221
gcc46-pr48869.patch

Untested fix.

[Bug c++/48869] [4.5/4.6/4.7 Regression] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-05-19 07:44:35 UTC ---
Author: jakub
Date: Thu May 19 07:44:31 2011
New Revision: 173888

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=173888
Log:
    PR c++/48869
    * method.c (get_dtor, get_copy_ctor): Add COMPLAIN argument,
    pass it down to locate_fn_flags.
    * cp-tree.h (get_dtor, get_copy_ctor): Adjust prototypes.
    * semantics.c (cxx_omp_create_clause_info): Adjust callers.
    * cp-gimplify.c: Include splay-tree.h.
    (splay_tree_compare_decl_uid, omp_var_to_track,
    omp_cxx_notice_variable): New functions.
    (struct cp_genericize_omp_taskreg): New type.
    (struct cp_genericize_data): Add omp_ctx field.
    (cp_genericize_r): Attempt to determine implicitly determined
    firstprivate class type variables.
    (cp_genericize): Clear omp_ctx.
    * Make-lang.in (cp/cp-gimplify.o): Depend on $(SPLAY_TREE_H).

    * testsuite/libgomp.c++/pr48869.C: New test.

Added:
    trunk/libgomp/testsuite/libgomp.c++/pr48869.C
Modified:
    trunk/gcc/cp/ChangeLog
    trunk/gcc/cp/Make-lang.in
    trunk/gcc/cp/cp-gimplify.c
    trunk/gcc/cp/cp-tree.h
    trunk/gcc/cp/method.c
    trunk/gcc/cp/semantics.c
    trunk/libgomp/ChangeLog

[Bug c++/48869] [4.5 Regression] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.6.1, 4.7.0
            Summary|[4.5/4.6/4.7 Regression]    |[4.5 Regression] OpenMP
                   |OpenMP task construct fails |task construct fails to
                   |to instantiate copy         |instantiate copy
                   |constructor(same as Bug     |constructor(same as Bug
                   |36523)                      |36523)
      Known to fail|4.7.0                       |

--- Comment #10 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-05-19 08:03:37 UTC ---
Fixed for 4.6+ so far.

[Bug c++/48869] [4.5/4.6/4.7 Regression] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-05-19 08:00:56 UTC ---
Author: jakub
Date: Thu May 19 08:00:50 2011
New Revision: 173889

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=173889
Log:
    PR c++/48869
    * method.c (get_dtor, get_copy_ctor): Add COMPLAIN argument,
    pass it down to locate_fn_flags.
    * cp-tree.h (get_dtor, get_copy_ctor): Adjust prototypes.
    * semantics.c (cxx_omp_create_clause_info): Adjust callers.
    * cp-gimplify.c: Include splay-tree.h.
    (splay_tree_compare_decl_uid, omp_var_to_track,
    omp_cxx_notice_variable): New functions.
    (struct cp_genericize_omp_taskreg): New type.
    (struct cp_genericize_data): Add omp_ctx field.
    (cp_genericize_r): Attempt to determine implicitly determined
    firstprivate class type variables.
    (cp_genericize): Clear omp_ctx.
    * Make-lang.in (cp/cp-gimplify.o): Depend on $(SPLAY_TREE_H).

    * testsuite/libgomp.c++/pr48869.C: New test.

Added:
    branches/gcc-4_6-branch/libgomp/testsuite/libgomp.c++/pr48869.C
Modified:
    branches/gcc-4_6-branch/gcc/cp/ChangeLog
    branches/gcc-4_6-branch/gcc/cp/Make-lang.in
    branches/gcc-4_6-branch/gcc/cp/cp-gimplify.c
    branches/gcc-4_6-branch/gcc/cp/cp-tree.h
    branches/gcc-4_6-branch/gcc/cp/method.c
    branches/gcc-4_6-branch/gcc/cp/semantics.c
    branches/gcc-4_6-branch/libgomp/ChangeLog

[Bug c++/48869] [4.5 Regression] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2

[Bug c++/48869] [4.5 Regression] OpenMP task construct fails to instantiate copy constructor(same as Bug 36523)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48869

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|4.5.4                       |4.6.0

--- Comment #11 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-07-02 10:50:06 UTC ---
Fixed in 4.6.0, the 4.5 branch is being closed.