[Bug fortran/34848] New: internal compiler error: Segmentation fault/optional arguments

[Bug fortran/34848] New: internal compiler error: Segmentation fault/optional arguments

[krefson at googlemail dot com]

Gfortran version 4.3.0 20080117 (experimental) [trunk revision 131592] (GCC)
gives an ICE on the following program.

$ gfortran -c ia.f90
ia.f90: In function 'doit':
ia.f90:12: internal compiler error: Segmentation fault

---------------------------------------------------------------------------
module krmod
contains
  function doit(genloc,vloc,scheme)
    implicit none
    character(len=6), optional,               intent(in)  :: genloc
    real,    optional, dimension(:), intent(in)  :: vloc
    character(len=2), optional,               intent(out) :: scheme
    real :: doit(100)
    !-------------------------------------------------------------------------!
    if(present(genloc)) then
       doit = tm_doit(genloc=genloc)
       if(present(scheme)) scheme = 'tm'
    else
       doit = tm_doit(vloc=vloc)
       if(present(scheme)) scheme = 'tm'
    endif
    return
  end function doit
  function tm_doit(genloc,vloc)
    implicit none
    character(len=6), optional,               intent(in)  :: genloc
    real,    optional, dimension(:), intent(in)  :: vloc
    real :: tm_doit(100)
    tm_doit = 1.0
  end function tm_doit
end module krmod


-- 
           Summary: internal compiler error: Segmentation fault/optional
                    arguments
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: major
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: krefson at googlemail dot com
  GCC host triplet: x86_64-unknown-linux-gnu
GCC target triplet: x86_64-unknown-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] internal compiler error with optional argument of character type and array return type

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2008-01-18 11:48 -------
Reduced testcase:
module krmod
contains
 function doit()
   implicit none
    real :: doit1(100)
    real :: doit
    doit1 = tm_doit()
   return
 end function doit
 function tm_doit(genloc)
   implicit none
   character, optional  :: genloc
   real :: tm_doit(100)
 end function tm_doit
end module krmod


----
The optional argument has to be a character type and the return type has to be
an array


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|major                       |normal
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   GCC host triplet|x86_64-unknown-linux-gnu    |
 GCC target triplet|x86_64-unknown-linux-gnu    |
           Keywords|                            |ice-on-invalid-code
   Last reconfirmed|0000-00-00 00:00:00         |2008-01-18 11:48:37
               date|                            |
            Summary|internal compiler error:    |internal compiler error with
                   |Segmentation fault/optional |optional argument of
                   |arguments                   |character type and array
                   |                            |return type


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[burnus at gcc dot gnu dot org]

------- Comment #2 from burnus at gcc dot gnu dot org  2008-01-18 12:45 -------
It's a regression - and I might be guilty of it with my Bind(C) patches...


-- 

burnus at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |burnus at gcc dot gnu dot
                   |                            |org
            Summary|internal compiler error with|[4.3 Regression] internal
                   |optional argument of        |compiler error with optional
                   |character type and array    |argument of character type
                   |return type                 |and array return type


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
   Target Milestone|---                         |4.3.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[tkoenig at gcc dot gnu dot org]

------- Comment #3 from tkoenig at gcc dot gnu dot org  2008-01-20 09:59 -------
(In reply to comment #2)
> It's a regression - and I might be guilty of it with my Bind(C) patches...

Well, if it's a regression, there's a bigger chance for it to
be fixed before 4.3 comes out :-)


-- 

tkoenig at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |tkoenig at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[krefson at googlemail dot com]

------- Comment #4 from krefson at googlemail dot com  2008-01-20 13:53 -------
(In reply to comment #3)
> (In reply to comment #2)
> > It's a regression - and I might be guilty of it with my Bind(C) patches...
> 
> Well, if it's a regression, there's a bigger chance for it to
> be fixed before 4.3 comes out :-)
> 

It would be a milestone if this was fixed, as this would be the first version
of gfortran I will be able to recommend to our users for compiling the CASTEP
electronic structure code. No previous release version has worked - 4.2.3
and earlier all fail to compile CASTEP due to bug # 30095, but snapshots of
4.3.0 from late Nov 2007 appear to work very well indeed.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[burnus at gcc dot gnu dot org]

------- Comment #5 from burnus at gcc dot gnu dot org  2008-01-20 14:34 -------
I have a patch. Actually, I do not understand why it worked before.


-- 

burnus at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |burnus at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2008-01-18 11:48:37         |2008-01-20 14:34:44
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[burnus at gcc dot gnu dot org]

------- Comment #6 from burnus at gcc dot gnu dot org  2008-01-20 15:36 -------
Created an attachment (id=14978)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=14978&action=view)
First draft of the patch

The patch works for the test case, but it fails for auto_char_dummy_array_1.f90
(ICE) and for bind_c_usage_13.f03 (string length is passed, but should not for
Bind(C)).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[jakub at gcc dot gnu dot org]

-- 

jakub at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[burnus at gcc dot gnu dot org]

------- Comment #7 from burnus at gcc dot gnu dot org  2008-01-22 19:36 -------
Subject: Bug 34848

Author: burnus
Date: Tue Jan 22 19:35:49 2008
New Revision: 131738

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=131738
Log:
2008-01-22  Tobias Burnus  <burnus@net-b.de>

       PR fortran/34848
       * interface.c (compare_actual_formal): Fix adding type
       to missing_arg_type for absent optional arguments.

2008-01-22  Tobias Burnus  <burnus@net-b.de>

       PR fortran/34848
       * gfortran.dg/missing_optional_dummy_4.f90


Added:
    trunk/gcc/testsuite/gfortran.dg/missing_optional_dummy_4.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/interface.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[burnus at gcc dot gnu dot org]

------- Comment #8 from burnus at gcc dot gnu dot org  2008-01-22 19:59 -------
I only want to point out that the patch above only fixes a minor problem (which
was part of the draft patch): Before, for absent optional character strings no
string length was passed. Now "0" is passed. I don't know whether this could
causes problems in practice or not.
Patch: See above or http://gcc.gnu.org/ml/fortran/2008-01/msg00276.html

Next task: To fix the actual problem of this PR.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[burnus at gcc dot gnu dot org]

------- Comment #9 from burnus at gcc dot gnu dot org  2008-01-26 20:06 -------
The following does no longer fix PR34868 but it seems to have no regressions.
"e" is the expression holding the actual argument; if it is NULL it means the
argument is absent. (This is handled before, including passing then the string
length "0".)

Index: trans-expr.c
===================================================================
--- trans-expr.c        (Revision 131874)
+++ trans-expr.c        (Arbeitskopie)
@@ -2506,7 +2506,7 @@ gfc_conv_function_call (gfc_se * se, gfc
            }
        }

-      if (fsym && need_interface_mapping)
+      if (fsym && need_interface_mapping && e)
        gfc_add_interface_mapping (&mapping, fsym, &parmse, e);

       gfc_add_block_to_block (&se->pre, &parmse.pre);


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[burnus at gcc dot gnu dot org]

------- Comment #10 from burnus at gcc dot gnu dot org  2008-01-26 20:45 -------
Subject: Bug 34848

Author: burnus
Date: Sat Jan 26 20:44:49 2008
New Revision: 131876

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=131876
Log:
2008-01-26  Tobias Burnus  <burnus@net-b.de>

        PR fortran/34848
        * trans-expr.c (gfc_conv_function_call): Don't call
        gfc_add_interface_mapping if the expression is NULL.

2008-01-26  Tobias Burnus  <burnus@net-b.de>

        PR fortran/34848
        * gfortran.dg/missing_optional_dummy_5.f90: New.


Added:
    trunk/gcc/testsuite/gfortran.dg/missing_optional_dummy_5.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/trans-expr.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[burnus at gcc dot gnu dot org]

------- Comment #11 from burnus at gcc dot gnu dot org  2008-01-26 20:47 -------
FIXED on the trunk (4.3.0).

Hooray - we managed to fix this before 4.3.0 was released :-)

Thanks for the bug report!


-- 

burnus at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[dominiq at lps dot ens dot fr]

------- Comment #12 from dominiq at lps dot ens dot fr  2008-01-27 10:32 -------
It seems that this patch breaks the original test case of pr33998:

pr33998.f90: In function 'my_string':
pr33998.f90:7: internal compiler error: in gfc_typenode_for_spec, at
fortran/trans-types.c:842

but not the reduced ones of comment #1.

Sould I reopen pr33998?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[dominiq at lps dot ens dot fr]

------- Comment #13 from dominiq at lps dot ens dot fr  2008-01-27 10:41 -------
Following comment #12, I also see an ICE for pr34897:

pr34897.f90: In function 'my_string':
pr34897.f90:1: warning: Function does not return a value
pr34897.f90:4: internal compiler error: in gfc_typenode_for_spec, at
fortran/trans-types.c:842


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[dominiq at lps dot ens dot fr]

------- Comment #14 from dominiq at lps dot ens dot fr  2008-01-27 11:26 -------
Note that I am not sure to blame the right patch. What I can tell for sure is
that pr33998.f90 started to fail between the 19th (working, rev. 131656) and
the 20th (ICE, rev. 131679), and pr34897.f90 between the 20th (working, rev.
131679) and the 21st (ICE, rev. 131700).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848

[Bug fortran/34848] [4.3 Regression] internal compiler error with optional argument of character type and array return type

[dominiq at lps dot ens dot fr]

------- Comment #15 from dominiq at lps dot ens dot fr  2008-01-27 12:02 -------
> Note that I am not sure to blame the right patch. 

I was correct, pr33998.f90 started to fail after rev. 131676 and pr34897.f90
after rev. 131679.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34848