[Bug fortran/103898] New: [12 Regression] ICE: gimplification failed

[Bug fortran/103898] New: [12 Regression] ICE: gimplification failed

[antony at cosmologist dot info]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103898

            Bug ID: 103898
           Summary: [12 Regression] ICE: gimplification failed
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: antony at cosmologist dot info
  Target Milestone: ---

For full log see
https://app.travis-ci.com/github/cmbant/CosmoMC/jobs/552809686.
The relevant stack trace is

[ 94/125] Processing src/bflike/long_intrinsic_smw.f90
gimplification failed:
MAX_EXPR <_21, 0> <max_expr 0x7f1b5ff58f00
    type <integer_type 0x7f1b60196738 integer(kind=8) public DI
        size <integer_cst 0x7f1b6017fc00 constant 64>
        unit-size <integer_cst 0x7f1b6017fc18 constant 8>
        align:64 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type
0x7f1b60196738 precision:64 min <integer_cst 0x7f1b6017fe88
-9223372036854775808> max <integer_cst 0x7f1b6017fea0 9223372036854775807>
        pointer_to_this <pointer_type 0x7f1b601cb738>>

    arg:0 <ssa_name 0x7f1b5ff87d80 type <integer_type 0x7f1b60196738
integer(kind=8)>

        def_stmt _21 = _20 + 1;
        version:21>
    arg:1 <integer_cst 0x7f1b6019d2d0 type <integer_type 0x7f1b60196738
integer(kind=8)> constant 0>
    /cosmomc/code/plc_3.0/plc-3.01/src/bflike/long_intrinsic_smw.f90:155:36
start: /cosmomc/code/plc_3.0/plc-3.01/src/bflike/long_intrinsic_smw.f90:155:36
finish:
/cosmomc/code/plc_3.0/plc-3.01/src/bflike/long_intrinsic_smw.f90:155:36>
/cosmomc/code/plc_3.0/plc-3.01/src/bflike/long_intrinsic_smw.f90:155:36:
  155 |        mysize = size(array, dim=dim)
      |                                    ^
internal compiler error: gimplification failed
0x1daeb37 internal_error(char const*, ...)
        ???:0
0xc4a9b6 gimplify_expr(tree_node**, gimple**, gimple**, bool (*)(tree_node*),
int)
        ???:0
0xc4713f gimplify_expr(tree_node**, gimple**, gimple**, bool (*)(tree_node*),
int)
        ???:0
0xc4aa16 gimplify_stmt(tree_node**, gimple**)
        ???:0
0xc474e2 gimplify_expr(tree_node**, gimple**, gimple**, bool (*)(tree_node*),
int)
        ???:0
0xc4aa16 gimplify_stmt(tree_node**, gimple**)
        ???:0
0xc4848b gimplify_expr(tree_node**, gimple**, gimple**, bool (*)(tree_node*),
int)
        ???:0
0xc4aa16 gimplify_stmt(tree_node**, gimple**)
        ???:0
0xc479e2 gimplify_expr(tree_node**, gimple**, gimple**, bool (*)(tree_node*),
int)
        ???:0
0xc4aa16 gimplify_stmt(tree_node**, gimple**)
        ???:0
0xc474e2 gimplify_expr(tree_node**, gimple**, gimple**, bool (*)(tree_node*),
int)
        ???:0
0xc4aa16 gimplify_stmt(tree_node**, gimple**)
        ???:0
0xc4848b gimplify_expr(tree_node**, gimple**, gimple**, bool (*)(tree_node*),
int)
        ???:0
0xc4aa16 gimplify_stmt(tree_node**, gimple**)
        ???:0
0xc479e2 gimplify_expr(tree_node**, gimple**, gimple**, bool (*)(tree_node*),
int)
        ???:0
0xc58d19 gimplify_body(tree_node*, bool)
        ???:0
0xc5915d gimplify_function_tree(tree_node*)
        ???:0
0xa94487 cgraph_node::analyze()
        ???:0
0xa97c6d symbol_table::finalize_compilation_unit()
        ???:0
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.

The relevant source code is in
https://pla.esac.esa.int/pla/aio/product-action?COSMOLOGY.FILE_ID=COM_Likelihood_Code-v3.0_R3.01.tar.gz

This broke in trunk relatively recently, though at least a month ago.

I don't have a reduced test case (this is not my code).

[Bug fortran/103898] [12 Regression] ICE: gimplification failed

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103898

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |12.0
           Keywords|                            |ice-on-valid-code

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
>From 005t.original:
            MAX_EXPR <(parm.24.dim[0].ubound - parm.24.dim[0].lbound) + 1, 0> =
MAX_EXPR <(parm.24.dim[(integer(kind=8)) *dim + -1].ubound -
parm.24.dim[(integer(kind=8)) *dim + -1].lbound) + 1, 0>;

yes that is definitely wrong.
Let me try to reduce it.

[Bug fortran/103898] [12 Regression] ICE: gimplification failed

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103898

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2022-01-03
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Reduced testcase:
Module g
    Integer, Parameter, Public :: i4b = Selected_int_kind(9)
contains
  function size_i_1(array, dim) result (mysize)
    integer(i4b), dimension(:), intent(in)   :: array
    integer(i4b), optional,     intent(in)   :: dim
    if (present(dim)) then
       mysize = size(array, dim=dim)
    endif
  end function size_i_1
end module

[Bug fortran/103898] [12 Regression] ICE: gimplification failed

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103898

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
In GCC 11.2.0, we had:
        D.3972 = dim != 0B;
        mysize = D.3972 ? (integer(kind=4)) _gfortran_size1 (D.3971,
(integer(kind=8)) *dim) : (integer(kind=4)) _gfortran_size0 (D.3971);

While on the trunk we have:

        D.4255 = dim != 0B;
        if (D.4255)
          {
            MAX_EXPR <(parm.4.dim[0].ubound - parm.4.dim[0].lbound) + 1, 0> =
MAX_EXPR <(parm.4.dim[(integer(kind=8)) *dim + -1].ubound -
parm.4.dim[(integer(kind=8)) *dim + -1].lbound) + 1, 0>;
          }
        mysize = (integer(kind=4)) MAX_EXPR <(parm.4.dim[0].ubound -
parm.4.dim[0].lbound) + 1, 0>;

[Bug fortran/103898] [12 Regression] ICE: gimplification failed

[sandra at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103898

--- Comment #4 from sandra at gcc dot gnu.org ---
This is probably related to my rewrite of the size/shape/ubound/lbound
intrinsics back in mid-November.  I can add this one to my queue, but I've
already got 3 or 4 other issues already waiting for me to do something, so if
anybody else wants to take this one meanwhile go ahead.

[Bug fortran/103898] [12 Regression] ICE: gimplification failed

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103898

--- Comment #5 from anlauf at gcc dot gnu.org ---
Started most likely with commit r12-3897 by Tobias.

Interestingly, the ICE occurs only for a rank-1 array, but not for rank-2 or
for assumed-rank.

[Bug fortran/103898] [12 Regression] ICE: gimplification failed

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103898

Martin Liška <marxin at gcc dot gnu.org> changed:

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

--- Comment #6 from Martin Liška <marxin at gcc dot gnu.org> ---
(In reply to anlauf from comment #5)
> Started most likely with commit r12-3897 by Tobias.

Confirmed it started with r12-3897-g00f6de9c69119594.

[Bug fortran/103898] [12 Regression] ICE: gimplification failed

[sandra at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103898

sandra at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |sandra at gcc dot gnu.org

--- Comment #7 from sandra at gcc dot gnu.org ---
Yeah, the implementation of the SIZE intrinsic is pretty borked.  I'm working
on it.

[Bug fortran/103898] [12 Regression] ICE: gimplification failed

[sandra at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103898

--- Comment #8 from sandra at gcc dot gnu.org ---
Patch posted:

https://gcc.gnu.org/pipermail/fortran/2022-January/057293.html

[Bug fortran/103898] [12 Regression] ICE: gimplification failed

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103898

--- Comment #9 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Sandra Loosemore <sandra@gcc.gnu.org>:

https://gcc.gnu.org/g:55385f12f604fe242030f0a8fb057598c0f187a6

commit r12-6336-g55385f12f604fe242030f0a8fb057598c0f187a6
Author: Sandra Loosemore <sandra@codesourcery.com>
Date:   Thu Jan 6 11:23:18 2022 -0800

    Fortran: Fix handling of optional argument to SIZE intrinsic [PR103898]

    This patch fixes a think-o in the code that triggered an ICE
    in the test case.

    2021-01-06  Sandra Loosemore  <sandra@codesourcery.com>

            PR fortran/103898

            gcc/fortran/
            * trans-intrinsic.c (gfc_conv_intrinsic_size): Make size_var
            actually be a variable and fix surrounding code.

            gcc/testsuite/
            * gfortran.dg/pr103898.f90: New test.

[Bug fortran/103898] [12 Regression] ICE: gimplification failed

[sandra at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103898

sandra at gcc dot gnu.org changed:

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

--- Comment #10 from sandra at gcc dot gnu.org ---
Fixed.