[Bug fortran/102787] New: [12 regression] ICE in new test case gfortran.dg/reshape_shape_2.f90

[Bug fortran/102787] New: [12 regression] ICE in new test case gfortran.dg/reshape_shape_2.f90

[seurer at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

            Bug ID: 102787
           Summary: [12 regression] ICE in new test case
                    gfortran.dg/reshape_shape_2.f90
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: seurer at gcc dot gnu.org
  Target Milestone: ---

g:b47490c572c5938f887b54240af6096a7c90f640, r12-4415
make  -k check-gcc-fortran
RUNTESTFLAGS="dg.exp=gfortran.dg/reshape_shape_2.f90"
FAIL: gfortran.dg/reshape_shape_2.f90   -O  (internal compiler error)
FAIL: gfortran.dg/reshape_shape_2.f90   -O   (test for errors, line 6)
FAIL: gfortran.dg/reshape_shape_2.f90   -O  (test for excess errors)
# of unexpected failures        3

This only seems to happen on BE systems from what I see.

spawn -ignore SIGHUP
/home/seurer/gcc/git/build/gcc-test/gcc/testsuite/gfortran/../../gfortran
-B/home/seurer/gcc/git/build/gcc-test/gcc/testsuite/gfortran/../../
-B/home/seurer/gcc/git/build/gcc-test/powerpc64-unknown-linux-gnu/./libgfortran/
/home/seurer/gcc/git/gcc-test/gcc/testsuite/gfortran.dg/reshape_shape_2.f90
-fdiagnostics-plain-output -fdiagnostics-plain-output -O -pedantic-errors -S -o
reshape_shape_2.s
f951: internal compiler error: Segmentation fault
0x10c4696b crash_signal
        /home/seurer/gcc/git/gcc-test/gcc/toplev.c:322
0x120494b0 __gmpn_mul_1
        /home/seurer/gcc/git/build/gcc-test/gmp/mpn/mul_1.c:50
0x12043433 __gmpz_mul
        /home/seurer/gcc/git/gcc-test/gmp/mpz/mul.c:89
0x1028bfaf gfc_check_reshape(gfc_expr*, gfc_expr*, gfc_expr*, gfc_expr*)
        /home/seurer/gcc/git/gcc-test/gcc/fortran/check.c:4874
0x102e76ab do_check
        /home/seurer/gcc/git/gcc-test/gcc/fortran/intrinsic.c:316
0x102e7bb3 check_specific
        /home/seurer/gcc/git/gcc-test/gcc/fortran/intrinsic.c:4799
0x102f5cd7 gfc_intrinsic_func_interface(gfc_expr*, int)
        /home/seurer/gcc/git/gcc-test/gcc/fortran/intrinsic.c:5036
0x10370dd3 resolve_unknown_f
        /home/seurer/gcc/git/gcc-test/gcc/fortran/resolve.c:2937
0x10370dd3 resolve_function
        /home/seurer/gcc/git/gcc-test/gcc/fortran/resolve.c:3281
0x10370dd3 gfc_resolve_expr(gfc_expr*)
        /home/seurer/gcc/git/gcc-test/gcc/fortran/resolve.c:7115
0x102cebeb gfc_reduce_init_expr(gfc_expr*)
        /home/seurer/gcc/git/gcc-test/gcc/fortran/expr.c:3125
0x102d332b gfc_match_init_expr(gfc_expr**)
        /home/seurer/gcc/git/gcc-test/gcc/fortran/expr.c:3173
0x102b2f17 variable_decl
        /home/seurer/gcc/git/gcc-test/gcc/fortran/decl.c:3016
0x102b2f17 gfc_match_data_decl()
        /home/seurer/gcc/git/gcc-test/gcc/fortran/decl.c:6325
0x103486a7 match_word
        /home/seurer/gcc/git/gcc-test/gcc/fortran/parse.c:67
0x1034950f decode_statement
        /home/seurer/gcc/git/gcc-test/gcc/fortran/parse.c:378
0x1034f847 next_free
        /home/seurer/gcc/git/gcc-test/gcc/fortran/parse.c:1388
0x1034f847 next_statement
        /home/seurer/gcc/git/gcc-test/gcc/fortran/parse.c:1620
0x10352fef parse_spec
        /home/seurer/gcc/git/gcc-test/gcc/fortran/parse.c:4159
0x10356c73 parse_progunit
        /home/seurer/gcc/git/gcc-test/gcc/fortran/parse.c:6125


commit b47490c572c5938f887b54240af6096a7c90f640
Author: Harald Anlauf <anlauf@gmx.de>
Date:   Thu Oct 14 20:19:50 2021 +0200

    Fortran: generate error message for negative elements in SHAPE array

[Bug fortran/102787] [12 regression] ICE in new test case gfortran.dg/reshape_shape_2.f90

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

--- Comment #1 from anlauf at gcc dot gnu.org ---
Interesting.  Cannot reproduce on x86_64-pc-linux-gnu.

However, changing the testcase as follows:

program p
  integer, parameter :: a(1) = 2
! integer, parameter :: b(2) = reshape([3,4], -[a]) ! { dg-error "negative" }
  integer, parameter :: b(2) = reshape([3,4], - a ) ! { dg-error "negative" }
end

thus removing the array constructor, I get an error pointing to the wrong line:

reshape_shape_2a.f90:5:30:

    5 |   integer, parameter :: a(1) = 2
      |                              1
Error: 'shape' argument of 'reshape' intrinsic at (1) has negative element (-2)

which is (correctly) coming from gfc_check_reshape that shows up in your
backtrace.

Possibly several latent issues uncovered by the testcase.
Need to dig deeper.

[Bug fortran/102787] [12 regression] ICE in new test case gfortran.dg/reshape_shape_2.f90

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

--- Comment #2 from anlauf at gcc dot gnu.org ---
The wrong locus for comment#1 is fixed by:

diff --git a/gcc/fortran/check.c b/gcc/fortran/check.c
index cfaf9d26bbc..bfb371b82c9 100644
--- a/gcc/fortran/check.c
+++ b/gcc/fortran/check.c
@@ -4748,7 +4748,7 @@ gfc_check_reshape (gfc_expr *source, gfc_expr *shape,
              gfc_error ("%qs argument of %qs intrinsic at %L has "
                         "negative element (%d)",
                         gfc_current_intrinsic_arg[1]->name,
-                        gfc_current_intrinsic, &e->where, extent);
+                        gfc_current_intrinsic, &shape->where, extent);
              return false;
            }
        }

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[12 regression] ICE in new  |ICE in new test case
                   |test case                   |gfortran.dg/reshape_shape_2
                   |gfortran.dg/reshape_shape_2 |.f90
                   |.f90                        |

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Note I don't think this is a regression as the testcase was ICEing before.

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2021-10-16
             Status|UNCONFIRMED                 |ASSIGNED

--- Comment #4 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
Confirmed on Darwin too.

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |NEW

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

--- Comment #5 from anlauf at gcc dot gnu.org ---
Looks like an issue with simplification of array within array constructor:

  print *, -   a
  print *, -  [a]
  print *, - [[a]]
end

-fdump-fortran-original produces:

  code:
  WRITE UNIT=6 FMT=-1
  TRANSFER (/ -2 /)
  DT_END
  WRITE UNIT=6 FMT=-1
  TRANSFER (/ (/ -2 /) /)
  DT_END
  WRITE UNIT=6 FMT=-1
  TRANSFER (/ -2 /)
  DT_END

-fdump-tree-original confirms this.

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

--- Comment #6 from anlauf at gcc dot gnu.org ---
WIP patch that fixes comment#5 as like comment#0 and regtests cleanly:

diff --git a/gcc/fortran/array.c b/gcc/fortran/array.c
index 6552eaf3b0c..e1ab2e06fdb 100644
--- a/gcc/fortran/array.c
+++ b/gcc/fortran/array.c
@@ -1812,6 +1812,29 @@ expand_constructor (gfc_constructor_base base)
          continue;
        }

+      /* Expand constant array within array constructor.  */
+      if (e->expr_type == EXPR_VARIABLE && e->rank && e->ref
+         && e->symtree && e->symtree->n.sym
+         && e->symtree->n.sym->attr.flavor == FL_PARAMETER
+         && e->symtree->n.sym->value
+         && e->symtree->n.sym->value->value.constructor)
+       {
+         gfc_ref *ref;
+         for (ref = e->ref; ref; ref = ref->next)
+           if (ref->type == REF_ARRAY)
+             break;
+
+         if (ref && ref->u.ar.as && ref->u.ar.as->type == AS_EXPLICIT
+             && ref->u.ar.type == AR_FULL)
+           {
+             gfc_expr *value = e->symtree->n.sym->value;
+             if (!expand_constructor (value->value.constructor))
+               return false;
+
+             continue;
+           }
+       }
+
       empty_constructor = false;
       e = gfc_copy_expr (e);
       if (!gfc_simplify_expr (e, 1))


It only lacks goodies such as array sections in the constructor.

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

--- Comment #7 from anlauf at gcc dot gnu.org ---
Slightly improved version of the patch of comment#6:

diff --git a/gcc/fortran/array.c b/gcc/fortran/array.c
index 6552eaf3b0c..a63a6631f59 100644
--- a/gcc/fortran/array.c
+++ b/gcc/fortran/array.c
@@ -1812,6 +1812,29 @@ expand_constructor (gfc_constructor_base base)
          continue;
        }

+      /* Expand constant array within array constructor.  */
+      if (e->expr_type == EXPR_VARIABLE && e->rank && e->ref
+         && e->symtree && e->symtree->n.sym
+         && e->symtree->n.sym->attr.flavor == FL_PARAMETER
+         && e->symtree->n.sym->value
+         && e->symtree->n.sym->value->value.constructor)
+       {
+         gfc_array_ref *ar;
+         ar = gfc_find_array_ref (e);
+         if (ar && ar->as && ar->as->type == AS_EXPLICIT)
+           {
+             if (ar->type == AR_FULL)
+               {
+                 gfc_expr *value = e->symtree->n.sym->value;
+                 if (!expand_constructor (value->value.constructor))
+                   return false;
+
+                 continue;
+               }
+             /* TODO: handle ar->type == AR_SECTION.  */
+           }
+       }
+
       empty_constructor = false;
       e = gfc_copy_expr (e);
       if (!gfc_simplify_expr (e, 1))

Still does not handle array sections.

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

--- Comment #8 from anlauf at gcc dot gnu.org ---
Simpler and better patch which handles array sections as well as vector
subscripts:

diff --git a/gcc/fortran/array.c b/gcc/fortran/array.c
index 6552eaf3b0c..f870c225195 100644
--- a/gcc/fortran/array.c
+++ b/gcc/fortran/array.c
@@ -1804,6 +1804,12 @@ expand_constructor (gfc_constructor_base base)
       if (empty_constructor)
        empty_ts = e->ts;

+      /* Expand constant array within array constructor.  */
+      if (e->expr_type == EXPR_VARIABLE && e->rank > 0 && e->ref
+         && e->symtree && e->symtree->n.sym
+         && e->symtree->n.sym->attr.flavor == FL_PARAMETER)
+       gfc_simplify_expr (e, 0);
+
       if (e->expr_type == EXPR_ARRAY)
        {
          if (!expand_constructor (e->value.constructor))

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

--- Comment #9 from anlauf at gcc dot gnu.org ---
(In reply to anlauf from comment #8)
> Simpler and better patch which handles array sections as well as vector
> subscripts:

@Bill: does this patch or the submitted one in

https://gcc.gnu.org/pipermail/fortran/2021-November/057066.html

fix the regression for you?

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

--- Comment #10 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:6b8ecbc6d6652d061d7c72c64352d51eca2df6ca

commit r12-5639-g6b8ecbc6d6652d061d7c72c64352d51eca2df6ca
Author: Harald Anlauf <anlauf@gmx.de>
Date:   Sat Nov 27 21:43:52 2021 +0100

    Fortran: improve expansion of constant array expressions within
constructors

    gcc/fortran/ChangeLog:

            PR fortran/102787
            * array.c (expand_constructor): When encountering a constant array
            expression or array section within a constructor, simplify it to
            enable better expansion.

    gcc/testsuite/ChangeLog:

            * gfortran.dg/array_constructor_54.f90: New test.

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

anlauf at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |gscfq@t-online.de

--- Comment #11 from anlauf at gcc dot gnu.org ---
*** Bug 103261 has been marked as a duplicate of this bug. ***

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

--- Comment #12 from anlauf at gcc dot gnu.org ---
*** Bug 103263 has been marked as a duplicate of this bug. ***

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

--- Comment #13 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-11 branch has been updated by Harald Anlauf
<anlauf@gcc.gnu.org>:

https://gcc.gnu.org/g:217d8bf22b148ebe52456a4bccfe9d725e7c68e9

commit r11-9359-g217d8bf22b148ebe52456a4bccfe9d725e7c68e9
Author: Harald Anlauf <anlauf@gmx.de>
Date:   Sat Nov 27 21:43:52 2021 +0100

    Fortran: improve expansion of constant array expressions within
constructors

    gcc/fortran/ChangeLog:

            PR fortran/102787
            * array.c (expand_constructor): When encountering a constant array
            expression or array section within a constructor, simplify it to
            enable better expansion.

    gcc/testsuite/ChangeLog:

            * gfortran.dg/array_constructor_54.f90: New test.

    (cherry picked from commit 6b8ecbc6d6652d061d7c72c64352d51eca2df6ca)

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

--- Comment #14 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Harald Anlauf
<anlauf@gcc.gnu.org>:

https://gcc.gnu.org/g:1f49f2ad78360ea6634deca9a8cc35f9c4f1051d

commit r10-10334-g1f49f2ad78360ea6634deca9a8cc35f9c4f1051d
Author: Harald Anlauf <anlauf@gmx.de>
Date:   Sat Nov 27 21:43:52 2021 +0100

    Fortran: improve expansion of constant array expressions within
constructors

    gcc/fortran/ChangeLog:

            PR fortran/102787
            * array.c (expand_constructor): When encountering a constant array
            expression or array section within a constructor, simplify it to
            enable better expansion.

    gcc/testsuite/ChangeLog:

            * gfortran.dg/array_constructor_54.f90: New test.

    (cherry picked from commit 6b8ecbc6d6652d061d7c72c64352d51eca2df6ca)

[Bug fortran/102787] ICE in new test case gfortran.dg/reshape_shape_2.f90

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102787

anlauf at gcc dot gnu.org changed:

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

--- Comment #15 from anlauf at gcc dot gnu.org ---
Fixed on mainline for gcc-12, and on 11- and 10-branch.  Closing.

Thanks for the report!