[Bug fortran/107680] New: ICE in arith_power, at fortran/arith.cc:989 and :1006

[Bug fortran/107680] New: ICE in arith_power, at fortran/arith.cc:989 and :1006

[gscfq@t-online.de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107680

            Bug ID: 107680
           Summary: ICE in arith_power, at fortran/arith.cc:989 and :1006
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: gscfq@t-online.de
  Target Milestone: ---

Affects versions down to at least r5 :


$ cat z1.f90
program p
   print *, [real :: ([1])] ** 2.0
end

$ cat z2.f90
program p
   print *, [complex :: ([1])] ** 2.0
end

$ cat z3.f90
program p
   print *, [complex :: ([1])] ** (1.0,2.0)
end


$ gfortran-13-20221106 -c z1.f90
f951: internal compiler error: Segmentation fault
0xf4697f crash_signal
        ../../gcc/toplev.cc:314
0x7c4b87 arith_power
        ../../gcc/fortran/arith.cc:989
0x7c2ce9 reduce_binary_ac
        ../../gcc/fortran/arith.cc:1365
0x7c2d3e reduce_binary_ac
        ../../gcc/fortran/arith.cc:1369
0x7c2f9f reduce_binary
        ../../gcc/fortran/arith.cc:1520
0x7c33d2 eval_intrinsic
        ../../gcc/fortran/arith.cc:1701
0x835df6 match_mult_operand
        ../../gcc/fortran/matchexp.cc:288
0x835f98 match_add_operand
        ../../gcc/fortran/matchexp.cc:356
0x8361ec match_level_2
        ../../gcc/fortran/matchexp.cc:480
0x836342 match_level_3
        ../../gcc/fortran/matchexp.cc:551
0x836434 match_level_4
        ../../gcc/fortran/matchexp.cc:599
0x836434 match_and_operand
        ../../gcc/fortran/matchexp.cc:693
0x836622 match_or_operand
        ../../gcc/fortran/matchexp.cc:722
0x8366f2 match_equiv_operand
        ../../gcc/fortran/matchexp.cc:765
0x8367c4 match_level_5
        ../../gcc/fortran/matchexp.cc:811
0x835b91 gfc_match_expr(gfc_expr**)
        ../../gcc/fortran/matchexp.cc:870
0x81d289 match_io_element
        ../../gcc/fortran/io.cc:3668
0x81fbba match_io_list
        ../../gcc/fortran/io.cc:3716
0x81ffbe match_io
        ../../gcc/fortran/io.cc:4394
0x823aba gfc_match_print()
        ../../gcc/fortran/io.cc:4450

[Bug fortran/107680] ICE in arith_power, at fortran/arith.cc:989 and :1006

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107680

anlauf at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |anlauf at gcc dot gnu.org
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2022-11-14
             Status|UNCONFIRMED                 |NEW

--- Comment #1 from anlauf at gcc dot gnu.org ---
Confirmed.

[Bug fortran/107680] ICE in arith_power, at fortran/arith.cc:989 and :1006

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107680

--- Comment #2 from anlauf at gcc dot gnu.org ---
We could prohibit the simplification of ** like in the following:

diff --git a/gcc/fortran/arith.cc b/gcc/fortran/arith.cc
index fc9224ebc5c..540b8e2b945 100644
--- a/gcc/fortran/arith.cc
+++ b/gcc/fortran/arith.cc
@@ -845,6 +845,12 @@ arith_power (gfc_expr *op1, gfc_expr *op2, gfc_expr
**resultp)
   if (!gfc_numeric_ts (&op1->ts) || !gfc_numeric_ts (&op2->ts))
     return ARITH_INVALID_TYPE;

+  /* The result type shall accomodate the result of the simplification.  */
+  if (op1->ts.type == BT_INTEGER && op2->ts.type != BT_INTEGER)
+    return ARITH_NOT_REDUCED;
+  if (op1->ts.type == BT_REAL && op2->ts.type == BT_COMPLEX)
+    return ARITH_NOT_REDUCED;
+
   rc = ARITH_OK;
   result = gfc_get_constant_expr (op1->ts.type, op1->ts.kind, &op1->where);

[Bug fortran/107680] ICE in arith_power, at fortran/arith.cc:989 and :1006

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107680

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

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

--- Comment #3 from Mikael Morin <mikael at gcc dot gnu.org> ---
I thought the call to gfc_type_convert_binary in eval_intrinsic was taking care
of mismatching types, doesn't it?

[Bug fortran/107680] ICE in arith_power, at fortran/arith.cc:989 and :1006

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107680

--- Comment #4 from anlauf at gcc dot gnu.org ---
(In reply to Mikael Morin from comment #3)
> I thought the call to gfc_type_convert_binary in eval_intrinsic was taking
> care of mismatching types, doesn't it?

It does, and then it doesn't do it sometimes...

For

  real, parameter :: x(*) = [real :: ([1])]   **  2.0

after adding breakpoints in gfc_type_convert_binary and arith_power,
I see correct types in the former (e->value.op.op2->ts.type = BT_REAL)
but incorrect types in the latter (op1->ts.type = BT_INTEGER).

It seems to be the ratatouille of parentheses, array constructors,
typespec, and arithmetic operation that is crucial to get this type
of error.

In 12-branch, I also see other bad things happening, which I believe have
partially been fixed by the series of patches for pr107000 and friends.
Try:

  print *, [integer :: ([3.0])] **  2.0

This gives

0.00000000

for all versions <= 12, and

9.00000000

for mainline.

Given these observations, I suspect that typespecs are not consistently
handled...  See also the previous discussions.

[Bug fortran/107680] ICE in arith_power, at fortran/arith.cc:989 and :1006

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107680

anlauf at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |anlauf at gcc dot gnu.org
             Status|NEW                         |ASSIGNED

--- Comment #5 from anlauf at gcc dot gnu.org ---
Taking: https://gcc.gnu.org/pipermail/fortran/2022-November/058509.html

[Bug fortran/107680] ICE in arith_power, at fortran/arith.cc:989 and :1006

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107680

--- Comment #6 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Harald Anlauf <anlauf@gcc.gnu.org>:

https://gcc.gnu.org/g:713dcfc85ebbabaf74a1bcbac4ba1143519b31d6

commit r13-4106-g713dcfc85ebbabaf74a1bcbac4ba1143519b31d6
Author: Harald Anlauf <anlauf@gmx.de>
Date:   Tue Nov 15 21:20:20 2022 +0100

    Fortran: ICE in simplification of array expression involving power
[PR107680]

    gcc/fortran/ChangeLog:

            PR fortran/107680
            * arith.cc (arith_power): Check that operands are properly
converted
            before attempting to simplify.

    gcc/testsuite/ChangeLog:

            PR fortran/107680
            * gfortran.dg/pr107680.f90: New test.

[Bug fortran/107680] ICE in arith_power, at fortran/arith.cc:989 and :1006

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107680

anlauf at gcc dot gnu.org changed:

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

--- Comment #7 from anlauf at gcc dot gnu.org ---
Fixed on mainline.

The open issue of the fate(?) of the typespec is tracked in pr107721.

Thanks for the report!

[Bug fortran/107680] ICE in arith_power, at fortran/arith.cc:989 and :1006

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107680

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |13.0