[Bug fortran/17123] New: Assertion fail in trans-const.c

[Bug fortran/17123] New: Assertion fail in trans-const.c

[refson dot temp at ntlworld dot com]

This small fortran program gives an assertion fail in the compiler.

$ gfc g95bug.f90
f951: ../../gcc/gcc/fortran/trans-const.c:357: gfc_conv_constant: Assertion
`se->ss->type == GFC_SS_SCALAR' failed.
g95bug.f90: In function `g95sum':
g95bug.f90:12: internal compiler error: Aborted

$ gfc --version
GNU Fortran 95 (GCC 3.5.0 20040820 (experimental))
Copyright (C) 2003 Free Software Foundation, Inc.

$ cat g95bug.f90
module g95test
  type, public :: keyword
     character(len=3)            :: typ
  end type keyword

  type(keyword), dimension(:), allocatable :: key_temp

contains
  subroutine g95sum

    character(len=1) :: tp
    tp = achar(sum(ichar(key_temp%typ(1:1))))
  end subroutine g95sum

end module g95test

-- 
           Summary: Assertion fail in trans-const.c
           Product: gcc
           Version: tree-ssa
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: refson dot temp at ntlworld dot com
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123

[Bug fortran/17123] Assertion fail in trans-const.c

[refson dot temp at ntlworld dot com]

------- Additional Comments From refson dot temp at ntlworld dot com  2004-08-20 15:47 -------
Created an attachment (id=6963)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=6963&action=view)
Fortran 905 source test case


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123

[Bug fortran/17123] Assertion fail in trans-const.c

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-08-20 15:55 -------
Confirmed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
           Keywords|                            |ice-on-valid-code
   Last reconfirmed|0000-00-00 00:00:00         |2004-08-20 15:55:55
               date|                            |
            Version|tree-ssa                    |3.5.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123

[Bug fortran/17123] Assertion fail in trans-const.c

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2004-08-20 16:30 -------
Here's a reduced testcase.

======================================
program FOO
  type :: T
    character(len=1) :: C
  end type T

  type (T) :: X(1)
  character, dimension(1) :: A

  A = X%C(1:1)
end
======================================

This is probably related to PR 15196.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |reichelt at gcc dot gnu dot
                   |                            |org
  BugsThisDependsOn|                            |15196
           Keywords|                            |monitored


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123

[Bug fortran/17123] Assertion fail in trans-const.c

[tobi at gcc dot gnu dot org]

------- Additional Comments From tobi at gcc dot gnu dot org  2004-08-22 18:21 -------
The failure can be prevented by changing
(In reply to comment #3)
>   A = X%C(1:1)
to 
 A = X%C
in Volker's reduced testcase. In the original testcase se->ss->type ==
GFC_SS_SECTION.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123

[Bug fortran/17123] Assertion fail in trans-const.c

[tobi at gcc dot gnu dot org]

------- Additional Comments From tobi at gcc dot gnu dot org  2004-08-22 18:22 -------
(In reply to comment #4)
> In the original testcase se->ss->type == GFC_SS_SECTION.

I was referring to Volker's reduced testcase, if that wasn't clear :-/


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123

[Bug fortran/17123] Assertion fail in trans-const.c

[reichelt at gcc dot gnu dot org]

-- 
Bug 17123 depends on bug 15196, which changed state.

Bug 15196 Summary: ICE: gfc_trans_scalar_assign: Assertion `lse->string_length != NULL_TREE'
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15196

           What    |Old Value                   |New Value
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123

[Bug fortran/17123] Assertion fail in trans-const.c

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2004-08-26 09:13 -------
Although PR 15196 got fixed, this bug still remains.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123

[Bug fortran/17123] Assertion fail in trans-const.c

[refson dot temp at ntlworld dot com]

------- Additional Comments From refson dot temp at ntlworld dot com  2005-03-23 11:16 -------
This bug is a complete showstopper, but has been sitting unresolved
for 6 months.  Is there any way to increase the priority and actually
have it assigned to anyone?  There is no way to even test that the
rest of our code compiles until this is fixed.

 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123

[Bug fortran/17123] Assertion fail in trans-const.c

[pbrook at gcc dot gnu dot org]

------- Additional Comments From pbrook at gcc dot gnu dot org  2005-07-14 22:19 -------
There are many wrong-code and rejects-valid bugs effecting gfortran. This one is
no worse than the others. The only way to make sure a bug is fixed is to fix it
yourself or pay someone else  to fix it for you.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123

[Bug fortran/17123] Assertion fail in trans-const.c

[refson dot temp at ntlworld dot com]

------- Additional Comments From refson dot temp at ntlworld dot com  2005-07-24 10:36 -------
Please accept my apologies.  I had not realised that this was a
gcc-developers-only site and that bug reports from users are not
welcome.  I have no expertise in compiler development so the
likelyhood of my being able to fix this myself is small.  

Your suggestion of paying someone to fix it is interesting, but if, 
as you say there are many similar bugs I fear that this would be a 
poor investment compared to paying for one of the several available
commercial compilers (eg the Intel one which is very reasonably priced)
in which case I could be reasonably sure of being able to compile my code.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123

[Bug fortran/17123] Assertion fail in trans-const.c

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2005-09-13 20:11 -------
This has been fixed on mainline by the patch for PR 19928.
As noted there PR 17123 really seems to be a duplicate.

*** This bug has been marked as a duplicate of 19928 ***

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17123