[Bug fortran/61138] New: Wrong code with pointer-bounds remapping

[Bug fortran/61138] New: Wrong code with pointer-bounds remapping

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

            Bug ID: 61138
           Summary: Wrong code with pointer-bounds remapping
           Product: gcc
           Version: 4.10.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: burnus at gcc dot gnu.org

Follow up to http://gcc.gnu.org/ml/fortran/2014-05/msg00023.html

The following code should print:
 6,  -5
 5,  6,  7,  8,  9,  10

and it does so with the Cray compiler. With GCC 4.10, however, it prints the
*whole* array:
  10  -5
   1   2   3   4   5   6   7   8   9  10



implicit none
integer, target :: tgt(10)
integer, target, allocatable :: tgt2(:)
integer, pointer :: ptr(:)

tgt = [1,2,3,4,5,6,7,8,9,10]
tgt2 = [1,2,3,4,5,6,7,8,9,10]


ptr(-5:) => tgt(5:)  ! Okay

if (size(ptr) /= 6 .or. lbound(ptr,1) /= -5) call abort()
if (any (ptr /= [5,6,7,8,9,10])) call abort()


ptr(-5:) => tgt2(5:)  ! wrongly associates the whole array

print '(*(i4))', size(ptr), lbound(ptr)
print '(*(i4))', ptr

if (size(ptr) /= 6 .or. lbound(ptr,1) /= -5) call abort()
if (any (ptr /= [5,6,7,8,9,10])) call abort()
end

[Bug fortran/61138] [4.8/4.9/4.10 Regression] Wrong code with pointer-bounds remapping

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

Dominique d'Humieres <dominiq at lps dot ens.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-05-10
            Summary|Wrong code with             |[4.8/4.9/4.10 Regression]
                   |pointer-bounds remapping    |Wrong code with
                   |                            |pointer-bounds remapping
     Ever confirmed|0                           |1

--- Comment #1 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
r190619 (2012-08-23) is OK, r190641 (2012-08-24) miscompiles the test. Likely
r190641.

[Bug fortran/61138] [4.8/4.9/4.10 Regression] Wrong code with pointer-bounds remapping

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P4
   Target Milestone|---                         |4.8.3

[Bug fortran/61138] [4.8/4.9/4.10 Regression] Wrong code with pointer-bounds remapping

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.8.3                       |4.8.4

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 4.8.3 is being released, adjusting target milestone.

[Bug fortran/61138] [4.8/4.9/4.10 Regression] Wrong code with pointer-bounds remapping

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

--- Comment #4 from Mikael Morin <mikael at gcc dot gnu.org> ---
(In reply to Mikael Morin from comment #2)
> gfc_trans_pointer_assignment sets lse.descriptor_only before calling
> gfc_conv_expr_descriptor (for the lhs),  and later on reuses lse for other
> things, without clearing descriptor_only.

The proper fix would avoid reusing any gfc_se struct entirely;
a more humble, ad hoc fix could look like this:

diff --git a/gcc/fortran/trans-expr.c b/gcc/fortran/trans-expr.c
index 5a50122..9748ade 100644
--- a/gcc/fortran/trans-expr.c
+++ b/gcc/fortran/trans-expr.c
@@ -6684,2 +6684,3 @@ gfc_trans_pointer_assignment (gfc_expr * expr1, gfc_expr
* expr2)
           lse.expr = tmp;
+          lse.descriptor_only = 0;
           lse.direct_byref = 1;
@@ -6699,2 +6700,3 @@ gfc_trans_pointer_assignment (gfc_expr * expr1, gfc_expr
* expr2)
       lse.direct_byref = 1;
+      lse.descriptor_only = 0;
       gfc_conv_expr_descriptor (&lse, expr2);
@@ -6750,2 +6752,3 @@ gfc_trans_pointer_assignment (gfc_expr * expr1, gfc_expr
* expr2)
       lse.direct_byref = 1;
+      lse.descriptor_only = 0;
       gfc_conv_expr_descriptor (&lse, expr2);

[Bug fortran/61138] [4.8/4.9/5 Regression] Wrong code with pointer-bounds remapping

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.8.4                       |4.8.5

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
GCC 4.8.4 has been released.

[Bug fortran/61138] [4.8/4.9/5 Regression] Wrong code with pointer-bounds remapping

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

--- Comment #6 from Mikael Morin <mikael at gcc dot gnu.org> ---
(In reply to Dominique d'Humieres from comment #1)
> Likely r190641.

Yes


(In reply to Mikael Morin from comment #2)
> gfc_trans_pointer_assignment sets lse.descriptor_only before calling
> gfc_conv_expr_descriptor (for the lhs),  and later on reuses lse for other
> things, without clearing descriptor_only.

Before r190641, gfc_conv_descriptor was not honoring lse.descriptor_only.
See https://gcc.gnu.org/ml/fortran/2012-08/msg00146.html

[Bug fortran/61138] [4.8/4.9/5 Regression] Wrong code with pointer-bounds remapping

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
                URL|                            |https://gcc.gnu.org/ml/gcc-
                   |                            |patches/2015-02/msg00531.ht
                   |                            |ml
           Assignee|unassigned at gcc dot gnu.org      |mikael at gcc dot gnu.org

--- Comment #7 from Mikael Morin <mikael at gcc dot gnu.org> ---
Patch:
https://gcc.gnu.org/ml/gcc-patches/2015-02/msg00531.html

[Bug fortran/61138] [4.8/4.9/5 Regression] Wrong code with pointer-bounds remapping

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

--- Comment #8 from Mikael Morin <mikael at gcc dot gnu.org> ---
Author: mikael
Date: Sat Mar 14 12:23:27 2015
New Revision: 221436

URL: https://gcc.gnu.org/viewcvs?rev=221436&root=gcc&view=rev
Log:
    PR fortran/61138
fortran/
    * trans-expr.c (gfc_trans_pointer_assignment): Clear DESCRIPTOR_ONLY
    field before reusing LSE.
testsuite/
    gfortran.dg/pointer_remapping_9.f90: New.


Added:
    trunk/gcc/testsuite/gfortran.dg/pointer_remapping_9.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/trans-expr.c
    trunk/gcc/testsuite/ChangeLog

[Bug fortran/61138] [4.8/4.9/5 Regression] Wrong code with pointer-bounds remapping

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

--- Comment #9 from Mikael Morin <mikael at gcc dot gnu.org> ---
Author: mikael
Date: Sat Mar 21 15:15:19 2015
New Revision: 221555

URL: https://gcc.gnu.org/viewcvs?rev=221555&root=gcc&view=rev
Log:
    PR fortran/61138
fortran/
    * trans-expr.c (gfc_trans_pointer_assignment): Clear DESCRIPTOR_ONLY
    field before reusing LSE.
testsuite/
    * gfortran.dg/pointer_remapping_9.f90: New.


Added:
    branches/gcc-4_9-branch/gcc/testsuite/gfortran.dg/pointer_remapping_9.f90
Modified:
    branches/gcc-4_9-branch/gcc/fortran/ChangeLog
    branches/gcc-4_9-branch/gcc/fortran/trans-expr.c
    branches/gcc-4_9-branch/gcc/testsuite/ChangeLog

[Bug fortran/61138] [4.8/4.9/5 Regression] Wrong code with pointer-bounds remapping

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

--- Comment #10 from Mikael Morin <mikael at gcc dot gnu.org> ---
Author: mikael
Date: Sat Mar 21 15:45:13 2015
New Revision: 221556

URL: https://gcc.gnu.org/viewcvs?rev=221556&root=gcc&view=rev
Log:
    Backport from trunk:
    2015-03-14  Mikael Morin  <mikael@gcc.gnu.org>

    PR fortran/61138
fortran/
    * trans-expr.c (gfc_trans_pointer_assignment): Clear DESCRIPTOR_ONLY
    field before reusing LSE.
testsuite/
    * gfortran.dg/pointer_remapping_9.f90: New.


Added:
    branches/gcc-4_8-branch/gcc/testsuite/gfortran.dg/pointer_remapping_9.f90
Modified:
    branches/gcc-4_8-branch/gcc/fortran/ChangeLog
    branches/gcc-4_8-branch/gcc/fortran/trans-expr.c
    branches/gcc-4_8-branch/gcc/testsuite/ChangeLog

[Bug fortran/61138] [4.8/4.9/5 Regression] Wrong code with pointer-bounds remapping

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61138

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

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

--- Comment #11 from Mikael Morin <mikael at gcc dot gnu.org> ---
Fixed for 4.8.5, 4.9.3 and 5.0.
Closing.