[Bug fortran/49328] New: Internal compiler error due to explicit array bounds in contained routine argument

[Bug fortran/49328] New: Internal compiler error due to explicit array bounds in contained routine argument

[david.sagan at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49328

           Summary: Internal compiler error due to explicit array bounds
                    in contained routine argument
           Product: gcc
           Version: 4.5.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: david.sagan@gmail.com


Consider the following program:

!-----------------------

module test_mod
contains
subroutine twiss3_at_start ()

implicit none

type t_struct
  real beta, alpha, gamma, phi
end type

type (t_struct) b
real g(6,6), tune3

call mode1_calc (g(3:4, 3:4), tune3, b)

contains

subroutine mode1_calc (gg, tune, twiss)

type (t_struct) twiss
real gg(2,2), tune

twiss%beta = gg(2,2)**2

end subroutine
end subroutine
end module


!-----------------------

compiling gives:

lnx4103:~/dcs/bmad_distribution/bmad_dist/test> gfortran --version
GNU Fortran (GCC) 4.5.1
Copyright (C) 2010 Free Software Foundation, Inc.

lnx4103:~/dcs/bmad_distribution/bmad_dist/test> gfortran -c -O2 test.f90
test.f90: In function 'twiss3_at_start':
test.f90:14:0: internal compiler error: in build2_stat, at tree.c:3681
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.


Removing the "-O2" compile option fixes things. Also changing "real gg(2,2)" to
"real gg(:,:)" also fixes things.

[Bug fortran/49328] Internal compiler error due to explicit array bounds in contained routine argument

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49328

--- Comment #1 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2011-06-08 20:11:38 UTC ---
Confirmed on gcc4.5.2 and 4.5.3, but AFAICT this has been fixed on gcc4.6 and
4.7 (trunk) between revisions 170000 and 170140.

[Bug fortran/49328] Internal compiler error due to explicit array bounds in contained routine argument

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49328

--- Comment #2 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2011-06-08 20:22:40 UTC ---
> Confirmed on gcc4.5.2 and 4.5.3, but AFAICT this has been fixed on gcc4.6 and
> 4.7 (trunk) between revisions 170000 and 170140.

Sorry I did not read the tests correctly, the range is between r158253 and
r159105.

[Bug fortran/49328] [4.5 Regression] Internal compiler error due to explicit array bounds in contained routine argument

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49328

Thomas Koenig <tkoenig at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011-08-22
                 CC|                            |tkoenig at gcc dot gnu.org
   Target Milestone|---                         |4.5.4
            Summary|Internal compiler error due |[4.5 Regression] Internal
                   |to explicit array bounds in |compiler error due to
                   |contained routine argument  |explicit array bounds in
                   |                            |contained routine argument
     Ever Confirmed|0                           |1

--- Comment #3 from Thomas Koenig <tkoenig at gcc dot gnu.org> 2011-08-22 15:19:14 UTC ---
Test case also works on 4.4.

Since this has been fixed a long time ago, hunting down the fix
could be difficult.  If anybody could pinpoint the exact commit,
it might be doable, though.

[Bug fortran/49328] [4.5 Regression] Internal compiler error due to explicit array bounds in contained routine argument

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49328

Tobias Burnus <burnus at gcc dot gnu.org> changed:

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

--- Comment #4 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-08-22 16:35:02 UTC ---
Looks a bit similar to PR 47574 (= PR 32478), which has been fixed at:
  Date: Sat Feb  5 10:35:24 2011
  New Revision: 169850
That's a month before the 4.6.0 release and thus 4.6-Feb/4.6.0 and 4.7.0 should
be fine. However, it does not match the ranges given in comment 2.

[Bug fortran/49328] [4.5 Regression] Internal compiler error due to explicit array bounds in contained routine argument

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49328

--- Comment #5 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-08-22 16:45:22 UTC ---
(In reply to comment #4)
> Looks a bit similar to PR 47574 (= PR 32478)

Sorry for not looking properly. PR 32478 is wrong (bug was fixed 2007). And PR
47574's patch is actually in PR 46664 (Rev. 167173, Nov 26, 2010).

[Bug middle-end/49328] [4.5 Regression] Internal compiler error due to explicit array bounds in contained routine argument

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49328

Tobias Burnus <burnus at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-valid-code
          Component|fortran                     |middle-end

--- Comment #6 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-08-30 14:34:21 UTC ---
Move to component: middle-end.

Summary: Works with 4.4, fails with 4.5, works again since 4.6.
Only ICEs (cf. comment 0) with optimization.

I have not tried to trace down the regression-causing patch but only the one
which fixed the issue on the 4.6 trunk:

FAILING: 20100418  158491
OK:      20100419  158501

Thus, the commit fixing this issue was Rev. 158496:
  http://gcc.gnu.org/ml/gcc-cvs/2010-04/msg00602.html
see also:
  http://gcc.gnu.org/ml/gcc-patches/2010-04/msg01131.html

[Bug middle-end/49328] [4.5 Regression] Internal compiler error due to explicit array bounds in contained routine argument

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49328

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

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

[Bug middle-end/49328] [4.5 Regression] Internal compiler error due to explicit array bounds in contained routine argument

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49328

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

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

--- Comment #7 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-07-02 10:34:55 UTC ---
Fixed in 4.6.0, the 4.5 branch is being closed.