[Bug fortran/64757] New: internal compiler error: in fold_convert_loc, at fold-const.c:2353

[Bug fortran/64757] New: internal compiler error: in fold_convert_loc, at fold-const.c:2353

[mike at rilee dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

            Bug ID: 64757
           Summary: internal compiler error: in fold_convert_loc, at
                    fold-const.c:2353
           Product: gcc
           Version: 5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: mike at rilee dot net

Created attachment 34557
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=34557&action=edit
Internal compiler error driver for fold_convert_loc at fold-const.c:2353

The following yields an internal compiler error.  Driver attached with complete
information.

type(Test) :: aTest
type(TestReference), allocatable :: testList(:)

testList = [TestReference(aTest)]

The error:
!  testList = [TestReference(aTest)]
!  1
! internal compiler error: in fold_convert_loc, at fold-const.c:2353
! 
! t00.F90:27:0: internal compiler error: Abort trap: 6
! gfortran-mp-5: internal compiler error: Abort trap: 6 (program f951)
! Abort trap: 6

The system: gfortran-mp-5
bash-3.2$ uname -a
! Darwin rstllc5 13.4.0 Darwin Kernel Version 13.4.0: Sun Aug 17 19:50:11 PDT
2014; root:xnu-2422.115.4~1/RELEASE_X86_64 x86_64
! bash-3.2$ ${FC} --version
! GNU Fortran (MacPorts gcc5 5-20150118_0) 5.0.0 20150118 (experimental)
! [...]

[Bug fortran/64757] internal compiler error: in fold_convert_loc, at fold-const.c:2353

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

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

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

--- Comment #1 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
I get the same ICE with the change to

...
type(TestReference), allocatable :: testList!(:)
...
testList = TestReference(aTest)
...

Thus I think it is a duplicate of pr49213.

*** This bug has been marked as a duplicate of bug 49213 ***

[Bug fortran/64757] internal compiler error: in fold_convert_loc, at fold-const.c:2353

[janus at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

janus at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-valid-code
             Status|RESOLVED                    |NEW
   Last reconfirmed|                            |2015-01-24
                 CC|                            |janus at gcc dot gnu.org
         Resolution|DUPLICATE                   |---
     Ever confirmed|0                           |1

--- Comment #2 from janus at gcc dot gnu.org ---
(In reply to Dominique d'Humieres from comment #1
> Thus I think it is a duplicate of pr49213.

It it certainly related, but I'm not sure it is an exact duplicate. In
particular the above PR has quite a history and includes several test cases.
Let's keep it separate for now.

[Bug fortran/64757] [5 Regression] ICE in fold_convert_loc, at fold-const.c:2353

[janus at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

janus at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|internal compiler error: in |[5 Regression] ICE in
                   |fold_convert_loc, at        |fold_convert_loc, at
                   |fold-const.c:2353           |fold-const.c:2353

--- Comment #3 from janus at gcc dot gnu.org ---
Here is a reduction of the original test case which gives the ICE with trunk,
but compiles cleanly with 4.9:


  type :: Test
  end type

  type :: TestReference
     class(Test), allocatable :: test
  end type

  type(TestReference) :: testList

  testList = TestReference(Test())  ! ICE in fold_convert_loc

end



The backtrace is:

internal compiler error: in fold_convert_loc, bei fold-const.c:2357
0x814255 fold_convert_loc(unsigned int, tree_node*, tree_node*)
    /home/jweil/gcc/gcc50/trunk/gcc/fold-const.c:2357
0x6b16ea alloc_scalar_allocatable_for_subcomponent_assignment
    /home/jweil/gcc/gcc50/trunk/gcc/fortran/trans-expr.c:6403
0x6b16ea gfc_trans_subcomponent_assign
    /home/jweil/gcc/gcc50/trunk/gcc/fortran/trans-expr.c:6497
0x6b087b gfc_trans_structure_assign
    /home/jweil/gcc/gcc50/trunk/gcc/fortran/trans-expr.c:6649
0x6b2104 gfc_conv_structure(gfc_se*, gfc_expr*, int)
    /home/jweil/gcc/gcc50/trunk/gcc/fortran/trans-expr.c:6678
0x6ad46c gfc_conv_expr(gfc_se*, gfc_expr*)
    /home/jweil/gcc/gcc50/trunk/gcc/fortran/trans-expr.c:6848
0x6b32ff gfc_trans_assignment_1
    /home/jweil/gcc/gcc50/trunk/gcc/fortran/trans-expr.c:8553
0x682035 trans_code
    /home/jweil/gcc/gcc50/trunk/gcc/fortran/trans.c:1650
0x6a2e53 gfc_generate_function_code(gfc_namespace*)
    /home/jweil/gcc/gcc50/trunk/gcc/fortran/trans-decl.c:5844
0x63e470 translate_all_program_units
    /home/jweil/gcc/gcc50/trunk/gcc/fortran/parse.c:5341
0x63e470 gfc_parse_file()
    /home/jweil/gcc/gcc50/trunk/gcc/fortran/parse.c:5538
0x67de95 gfc_be_parse_file
    /home/jweil/gcc/gcc50/trunk/gcc/fortran/f95-lang.c:228

[Bug fortran/64757] [5 Regression] ICE in fold_convert_loc, at fold-const.c:2353

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

--- Comment #4 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> It it certainly related, but I'm not sure it is an exact duplicate.
> In particular the above PR has quite a history and includes several
> test cases. Let's keep it separate for now.

I disagree. AFACT this PR gives the same ICE as the test in pr49213 comment 8
or the following variant of the original code

program main

  type :: S
    integer :: n
  end type
  type(S) :: Sobj

  type, extends(S) :: S2
    integer :: m
  end type
  type(S2) :: S2obj

  type :: T
    class(S), allocatable :: x
  end type
  type(T) :: Tobj

  Sobj = S(1)
  Tobj = T(Sobj)

  S2obj = S2(1,2)
  print *, S2obj%n, S2obj%m

end program

Actually pr49213 is a collection of different bugs, one being this PR. IMO
pr49213 should be split along the different bugs.

[Bug fortran/64757] [5 Regression] ICE in fold_convert_loc, at fold-const.c:2353

[janus at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

--- Comment #5 from janus at gcc dot gnu.org ---
(In reply to Dominique d'Humieres from comment #4)
> Actually pr49213 is a collection of different bugs, one being this PR.

Right. Exactly for this reason I want to keep it separate. Makes it much easier
to keep track of the different problems separately.

[Bug fortran/64757] [5 Regression] ICE in fold_convert_loc, at fold-const.c:2353

[janus at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

--- Comment #6 from janus at gcc dot gnu.org ---
The following is sufficient to get rid of the ICE:


Index: gcc/fortran/trans-expr.c
===================================================================
--- gcc/fortran/trans-expr.c    (Revision 220084)
+++ gcc/fortran/trans-expr.c    (Arbeitskopie)
@@ -6490,6 +6490,8 @@ gfc_trans_subcomponent_assign (tree dest, gfc_comp
      routine below is motivated by the alloc_scalar_allocatable_for_
      assignment() routine, but with the realloc portions removed and
      different input.  */
+      if (cm->ts.type == BT_CLASS)
+    dest = gfc_class_data_get (dest);
       alloc_scalar_allocatable_for_subcomponent_assignment (&block,
                                 dest,
                                 cm,


However, one needs to do more to get the correct run-time behavior: The _vptr
needs to be set for the class component and one needs to use the correct size
for the memory allocation.

[Bug fortran/64757] [5 Regression] ICE in fold_convert_loc, at fold-const.c:2353

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P4
   Target Milestone|---                         |5.0

[Bug fortran/64757] [5 Regression] ICE in fold_convert_loc, at fold-const.c:2353

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

--- Comment #8 from Paul Thomas <pault at gcc dot gnu.org> ---
(In reply to Dominique d'Humieres from comment #4)
> > It it certainly related, but I'm not sure it is an exact duplicate.
> > In particular the above PR has quite a history and includes several
> > test cases. Let's keep it separate for now.
> 
> I disagree. AFACT this PR gives the same ICE as the test in pr49213 comment
> 8 or the following variant of the original code
> 
> program main
> 
>   type :: S
>     integer :: n
>   end type
>   type(S) :: Sobj
> 
>   type, extends(S) :: S2
>     integer :: m
>   end type
>   type(S2) :: S2obj
> 
>   type :: T
>     class(S), allocatable :: x
>   end type
>   type(T) :: Tobj
> 
>   Sobj = S(1)
>   Tobj = T(Sobj)
> 
>   S2obj = S2(1,2)
>   print *, S2obj%n, S2obj%m
> 
> end program
> 
> Actually pr49213 is a collection of different bugs, one being this PR. IMO
> pr49213 should be split along the different bugs.

The patch of comment #7 fixes the above as well.

Cheers

Paul

[Bug fortran/64757] [5 Regression] ICE in fold_convert_loc, at fold-const.c:2353

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

--- Comment #9 from Paul Thomas <pault at gcc dot gnu.org> ---
(In reply to Paul Thomas from comment #7)
> Created attachment 34607 [details]
> Draft patch for the PR
> 
> I believe that this regression was caused by the fix for pr60357.
> 
> The patch is regtesting right now.
> 
> Paul

It bootstraps and regtests just fine. I will submit to the list sometime today.

Paul

[Bug fortran/64757] [5 Regression] ICE in fold_convert_loc, at fold-const.c:2353

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

--- Comment #10 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> > Actually pr49213 is a collection of different bugs, one being this PR. IMO
> > pr49213 should be split along the different bugs.
>
> The patch of comment #7 fixes the above as well.

> It bootstraps and regtests just fine. 

Confirmed.

[Bug fortran/64757] [5 Regression] ICE in fold_convert_loc, at fold-const.c:2353

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

--- Comment #11 from Paul Thomas <pault at gcc dot gnu.org> ---
Author: pault
Date: Thu Feb  5 08:06:04 2015
New Revision: 220436

URL: https://gcc.gnu.org/viewcvs?rev=220436&root=gcc&view=rev
Log:
2015-02-05  Paul Thomas  <pault@gcc.gnu.org>

    PR fortran/64757
    * resolve.c (resolve_structure_cons): Obtain the rank of class
    components.
    * trans-expr.c (gfc_trans_alloc_subarray_assign): Do the
    assignment to allocatable class array components.
    (alloc_scalar_allocatable_for_subcomponent_assignment): If comp
    is a class component, allocate to the _data field.
    (gfc_trans_subcomponent_assign): If a class component with a
    derived type expression set the _vptr field and for array
    components, call gfc_trans_alloc_subarray_assign. For scalars,
    the assignment is performed here.

2015-02-05  Paul Thomas  <pault@gcc.gnu.org>

    PR fortran/64757
    * gfortran.dg/type_to_class_2.f90: New test
    * gfortran.dg/type_to_class_3.f90: New test

Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/testsuite/ChangeLog

[Bug fortran/64757] [5 Regression] ICE in fold_convert_loc, at fold-const.c:2353

[pault at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

Paul Thomas <pault at gcc dot gnu.org> changed:

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

--- Comment #12 from Paul Thomas <pault at gcc dot gnu.org> ---
I screwed up the PR number in the ChangeLogs by adding an extra zero :-) The
revision for the substance of the patch was 220435.

Patch OK'd by Tobias on #gfortran

Thanks for the report

Paul

[Bug fortran/64757] [5 Regression] ICE in fold_convert_loc, at fold-const.c:2353

[burnus at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64757

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

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

--- Comment #13 from Tobias Burnus <burnus at gcc dot gnu.org> ---
(In reply to Paul Thomas from comment #12)
> I screwed up the PR number in the ChangeLogs by adding an extra zero :-) The
> revision for the substance of the patch was 220435.
r220435  (<- is automatically turned into a link)

> Patch OK'd by Tobias on #gfortran

And at https://gcc.gnu.org/ml/fortran/2015-02/msg00022.html

Follow-up issues are tracked at PR64943, PR64944 and PR64945