[Bug target/99542] New: [9/10/11 Regression] ICE in exact_div, at poly-int.h:2219

[Bug target/99542] New: [9/10/11 Regression] ICE in exact_div, at poly-int.h:2219

[asolokha at gmx dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

            Bug ID: 99542
           Summary: [9/10/11 Regression] ICE in exact_div, at
                    poly-int.h:2219
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code, openmp
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: asolokha at gmx dot com
  Target Milestone: ---
            Target: aarch64-linux-gnu

gcc-11.0.1-alpha20210307 snapshot (g:a18ebd6c439227b048a91fbfa66f5983f884c157)
ICEs when compiling gcc/testsuite/gcc.dg/gomp/pr89246-2.c w/ -fopenmp[-simd]:

% aarch64-linux-gnu-gcc-11.0.1 -fopenmp -c
gcc/testsuite/gcc.dg/gomp/pr89246-2.c
during IPA pass: simdclone
gcc/testsuite/gcc.dg/gomp/pr89246-2.c:13:1: internal compiler error: in
exact_div, at poly-int.h:2219
   13 | }
      | ^
0x8754e0 poly_int<2u, poly_result<unsigned long, if_nonpoly<unsigned short,
unsigned short, poly_int_traits<unsigned short>::is_poly>::type,
poly_coeff_pair_traits<unsigned long, if_nonpoly<unsigned short, unsigned
short, poly_int_traits<unsigned short>::is_poly>::type>::result_kind>::type>
exact_div<2u, unsigned long, unsigned short>(poly_int_pod<2u, unsigned long>
const&, unsigned short)
       
/var/tmp/portage/cross-aarch64-linux-gnu/gcc-11.0.1_alpha20210307/work/gcc-11-20210307/gcc/poly-int.h:2219
0x8760a5 poly_int<2u, poly_result<unsigned long, if_nonpoly<unsigned short,
unsigned short, poly_int_traits<unsigned short>::is_poly>::type,
poly_coeff_pair_traits<unsigned long, if_nonpoly<unsigned short, unsigned
short, poly_int_traits<unsigned short>::is_poly>::type>::result_kind>::type>
exact_div<2u, unsigned long, unsigned short>(poly_int_pod<2u, unsigned long>
const&, unsigned short)
       
/var/tmp/portage/cross-aarch64-linux-gnu/gcc-11.0.1_alpha20210307/work/gcc-11-20210307/gcc/tree.h:3467
0x8760a5 simd_clone_adjust_argument_types
       
/var/tmp/portage/cross-aarch64-linux-gnu/gcc-11.0.1_alpha20210307/work/gcc-11-20210307/gcc/omp-simd-clone.c:622
0x1aa6f5b expand_simd_clones(cgraph_node*)
       
/var/tmp/portage/cross-aarch64-linux-gnu/gcc-11.0.1_alpha20210307/work/gcc-11-20210307/gcc/omp-simd-clone.c:1784
0x1aa7287 ipa_omp_simd_clone
       
/var/tmp/portage/cross-aarch64-linux-gnu/gcc-11.0.1_alpha20210307/work/gcc-11-20210307/gcc/omp-simd-clone.c:1798
0x1aa7287 execute
       
/var/tmp/portage/cross-aarch64-linux-gnu/gcc-11.0.1_alpha20210307/work/gcc-11-20210307/gcc/omp-simd-clone.c:1826

(Is it the exact reason the testcase has /* { dg-options "-O0 -fno-openmp
-fno-openmp-simd" } */?)

[Bug target/99542] [9/10/11 Regression] ICE in exact_div, at poly-int.h:2219

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |9.4

[Bug target/99542] [9/10/11 Regression] ICE in exact_div, at poly-int.h:2219

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org,
                   |                            |rsandifo at gcc dot gnu.org

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
This is aarch64 backend bug in aarch64_simd_clone_compute_vecsize_and_simdlen.
If you look at the i386 counterpart, for argument checking it does:
  tree type_arg_types = TYPE_ARG_TYPES (TREE_TYPE (node->decl));
  bool decl_arg_p = (node->definition || type_arg_types == NULL_TREE);

  for (t = (decl_arg_p ? DECL_ARGUMENTS (node->decl) : type_arg_types), i = 0;
       t && t != void_list_node; t = TREE_CHAIN (t), i++)
    {
      tree arg_type = decl_arg_p ? TREE_TYPE (t) : TREE_VALUE (t);
but aarch64 doesn't bother and instead it does just:
  for (t = DECL_ARGUMENTS (node->decl); t; t = DECL_CHAIN (t))
    {
      arg_type = TREE_TYPE (t);
The problem with that is that it doesn't check the arguments at all for
function declarations that aren't definitions.

[Bug target/99542] [9/10/11 Regression] ICE in exact_div, at poly-int.h:2219

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to Arseny Solokha from comment #0)
> (Is it the exact reason the testcase has /* { dg-options "-O0 -fno-openmp
> -fno-openmp-simd" } */?)

No, the reason for that is that it is the auxiliary source of another test that
already compiles it with -fopenmp-simd -O2.

[Bug target/99542] [9/10/11 Regression] ICE in exact_div, at poly-int.h:2219

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 50374
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=50374&action=edit
gcc11-pr99542.patch

Untested fix.  There is another difference from i386, for !TREE_PUBLIC
(node->decl) the i386 version never returns > 1, but instead just provides the
version which is the best for the current TU.  Wouldn't it make sense on
aarch64 too?  Or does it hope that the unused entrypoints will be optimized
away?
I'm not sure it happens, because simd cloning is fairly late IPA pass.

[Bug target/99542] [9/10/11 Regression] ICE in exact_div, at poly-int.h:2219

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

--- Comment #4 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jakub Jelinek <jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:fcefc59befd396267b824c170b6a37acaf10874e

commit r11-7683-gfcefc59befd396267b824c170b6a37acaf10874e
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Tue Mar 16 10:34:44 2021 +0100

    aarch64: Fix up aarch64_simd_clone_compute_vecsize_and_simdlen [PR99542]

    As the patch shows, there are several bugs in
    aarch64_simd_clone_compute_vecsize_and_simdlen.
    One is that unlike for function declarations that aren't definitions
    it completely ignores argument types.  Such decls don't have
DECL_ARGUMENTS,
    but we can walk TYPE_ARG_TYPES instead, like the i386 backend does or like
    the simd cloning code in the middle end does too.

    Another problem is that it checks types of uniform arguments.  That is
    unnecessary, uniform arguments are passed the way it normally is, it is
    a scalar argument rather than vector, so there is no reason not to support
    uniform argument of different size, or long double, structure etc.

    2021-03-16  Jakub Jelinek  <jakub@redhat.com>

            PR target/99542
            * config/aarch64/aarch64.c
            (aarch64_simd_clone_compute_vecsize_and_simdlen): If not a function
            definition, walk TYPE_ARG_TYPES list if non-NULL for argument types
            instead of DECL_ARGUMENTS.  Ignore types for uniform arguments.

            * gcc.dg/gomp/pr99542.c: New test.
            * gcc.dg/gomp/pr59669-2.c (bar): Don't expect a warning on aarch64.
            * gcc.dg/gomp/simd-clones-2.c (setArray): Likewise.
            * g++.dg/vect/simd-clone-7.cc (bar): Likewise.
            * g++.dg/gomp/declare-simd-1.C (f37): Expect a different warning
            on aarch64.
            * gcc.dg/declare-simd.c (fn2): Expect a new warning on aarch64.

[Bug target/99542] [9/10 Regression] ICE in exact_div, at poly-int.h:2219

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2021-03-16
           Assignee|unassigned at gcc dot gnu.org      |jakub at gcc dot gnu.org
            Summary|[9/10/11 Regression] ICE in |[9/10 Regression] ICE in
                   |exact_div, at               |exact_div, at
                   |poly-int.h:2219             |poly-int.h:2219
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |ASSIGNED

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed on the trunk so far.

[Bug target/99542] [9/10 Regression] ICE in exact_div, at poly-int.h:2219

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

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

https://gcc.gnu.org/g:a2a6e9214e27b32d4582c670faf9cdb74e54c2c6

commit r11-7698-ga2a6e9214e27b32d4582c670faf9cdb74e54c2c6
Author: Christophe Lyon <christophe.lyon@linaro.org>
Date:   Tue Mar 16 21:48:10 2021 +0000

    aarch64: Fix up aarch64_simd_clone_compute_vecsize_and_simdlen [PR99542]

    The gcc.dg/declare-simd.c test does not emit a warning with
    -mabi=ilp32.

    2021-03-16  Christophe Lyon  <christophe.lyon@linaro.org>

            PR target/99542
            gcc/testsuite/
            * gcc.dg/declare-simd.c (fn2): Expect a warning only under lp64.

[Bug target/99542] [9/10 Regression] ICE in exact_div, at poly-int.h:2219

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

--- Comment #7 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Tamar Christina <tnfchris@gcc.gnu.org>:

https://gcc.gnu.org/g:39916ceab4940315e84bcd966da2c1d4a8e1734b

commit r11-7700-g39916ceab4940315e84bcd966da2c1d4a8e1734b
Author: Tamar Christina <tamar.christina@arm.com>
Date:   Wed Mar 17 11:12:25 2021 +0000

    AArch64: Fix -Werror issue in
aarch64_simd_clone_compute_vecsize_and_simdlen

    g:fcefc59befd396267b824c170b6a37acaf10874e introduced a new variable named
    arg_type which shadows the function scoped one.

    The function scoped one is now unused and so causes bootstrap to fail due
to
    -Werror.

    This patch removes the unused variable.

    gcc/ChangeLog:

            PR target/99542
            * config/aarch64/aarch64.c
            (aarch64_simd_clone_compute_vecsize_and_simdlen): Remove unused
var.

[Bug target/99542] [9/10 Regression] ICE in exact_div, at poly-int.h:2219

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

--- Comment #8 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:06589d2232abc92ac9bcb43e4a4ec64ead627752

commit r10-9485-g06589d2232abc92ac9bcb43e4a4ec64ead627752
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Tue Mar 16 10:34:44 2021 +0100

    aarch64: Fix up aarch64_simd_clone_compute_vecsize_and_simdlen [PR99542]

    As the patch shows, there are several bugs in
    aarch64_simd_clone_compute_vecsize_and_simdlen.
    One is that unlike for function declarations that aren't definitions
    it completely ignores argument types.  Such decls don't have
DECL_ARGUMENTS,
    but we can walk TYPE_ARG_TYPES instead, like the i386 backend does or like
    the simd cloning code in the middle end does too.

    Another problem is that it checks types of uniform arguments.  That is
    unnecessary, uniform arguments are passed the way it normally is, it is
    a scalar argument rather than vector, so there is no reason not to support
    uniform argument of different size, or long double, structure etc.

    2021-03-16  Jakub Jelinek  <jakub@redhat.com>

            PR target/99542
            * config/aarch64/aarch64.c
            (aarch64_simd_clone_compute_vecsize_and_simdlen): If not a function
            definition, walk TYPE_ARG_TYPES list if non-NULL for argument types
            instead of DECL_ARGUMENTS.  Ignore types for uniform arguments.

            * gcc.dg/gomp/pr99542.c: New test.
            * gcc.dg/gomp/pr59669-2.c (bar): Don't expect a warning on aarch64.
            * gcc.dg/gomp/simd-clones-2.c (setArray): Likewise.
            * g++.dg/vect/simd-clone-7.cc (bar): Likewise.
            * g++.dg/gomp/declare-simd-1.C (f37): Expect a different warning
            on aarch64.
            * gcc.dg/declare-simd.c (fn2): Expect a new warning on aarch64.

    (cherry picked from commit fcefc59befd396267b824c170b6a37acaf10874e)

[Bug target/99542] [9/10 Regression] ICE in exact_div, at poly-int.h:2219

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

--- Comment #9 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:d6300df5f2b9fafa07be4f974fef1ed810d0e7fd

commit r10-9488-gd6300df5f2b9fafa07be4f974fef1ed810d0e7fd
Author: Christophe Lyon <christophe.lyon@linaro.org>
Date:   Tue Mar 16 21:48:10 2021 +0000

    aarch64: Fix up aarch64_simd_clone_compute_vecsize_and_simdlen [PR99542]

    The gcc.dg/declare-simd.c test does not emit a warning with
    -mabi=ilp32.

    2021-03-16  Christophe Lyon  <christophe.lyon@linaro.org>

            PR target/99542
            gcc/testsuite/
            * gcc.dg/declare-simd.c (fn2): Expect a warning only under lp64.

    (cherry picked from commit a2a6e9214e27b32d4582c670faf9cdb74e54c2c6)

[Bug target/99542] [9 Regression] ICE in exact_div, at poly-int.h:2219

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[9/10 Regression] ICE in    |[9 Regression] ICE in
                   |exact_div, at               |exact_div, at
                   |poly-int.h:2219             |poly-int.h:2219

--- Comment #10 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed for 10.3 too.

[Bug target/99542] [9 Regression] ICE in exact_div, at poly-int.h:2219

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

--- Comment #11 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-9 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:0fe231e4dde882cd56029e122abc4fa940ad4ac5

commit r9-9431-g0fe231e4dde882cd56029e122abc4fa940ad4ac5
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Tue Mar 16 10:34:44 2021 +0100

    aarch64: Fix up aarch64_simd_clone_compute_vecsize_and_simdlen [PR99542]

    As the patch shows, there are several bugs in
    aarch64_simd_clone_compute_vecsize_and_simdlen.
    One is that unlike for function declarations that aren't definitions
    it completely ignores argument types.  Such decls don't have
DECL_ARGUMENTS,
    but we can walk TYPE_ARG_TYPES instead, like the i386 backend does or like
    the simd cloning code in the middle end does too.

    Another problem is that it checks types of uniform arguments.  That is
    unnecessary, uniform arguments are passed the way it normally is, it is
    a scalar argument rather than vector, so there is no reason not to support
    uniform argument of different size, or long double, structure etc.

    2021-03-16  Jakub Jelinek  <jakub@redhat.com>

            PR target/99542
            * config/aarch64/aarch64.c
            (aarch64_simd_clone_compute_vecsize_and_simdlen): If not a function
            definition, walk TYPE_ARG_TYPES list if non-NULL for argument types
            instead of DECL_ARGUMENTS.  Ignore types for uniform arguments.

            * gcc.dg/gomp/pr99542.c: New test.
            * gcc.dg/gomp/pr59669-2.c (bar): Don't expect a warning on aarch64.
            * gcc.dg/gomp/simd-clones-2.c (setArray): Likewise.
            * g++.dg/vect/simd-clone-7.cc (bar): Likewise.
            * g++.dg/gomp/declare-simd-1.C (f37): Expect a different warning
            on aarch64.
            * gcc.dg/declare-simd.c (fn2): Expect a new warning on aarch64.

    (cherry picked from commit 06589d2232abc92ac9bcb43e4a4ec64ead627752)

[Bug target/99542] [9 Regression] ICE in exact_div, at poly-int.h:2219

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

--- Comment #12 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-9 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:4eb2e3eb0f4a28fade00c1dca626cc947d20e7c4

commit r9-9433-g4eb2e3eb0f4a28fade00c1dca626cc947d20e7c4
Author: Christophe Lyon <christophe.lyon@linaro.org>
Date:   Tue Mar 16 21:48:10 2021 +0000

    aarch64: Fix up aarch64_simd_clone_compute_vecsize_and_simdlen [PR99542]

    The gcc.dg/declare-simd.c test does not emit a warning with
    -mabi=ilp32.

    2021-03-16  Christophe Lyon  <christophe.lyon@linaro.org>

            PR target/99542
            gcc/testsuite/
            * gcc.dg/declare-simd.c (fn2): Expect a warning only under lp64.

    (cherry picked from commit d6300df5f2b9fafa07be4f974fef1ed810d0e7fd)

[Bug target/99542] [9 Regression] ICE in exact_div, at poly-int.h:2219

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99542

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

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

--- Comment #13 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed.