[Bug target/104496] New: [12 Regression] ICE: in as_a, at machmode.h:365 with -O -mgeneral-regs-only -msse

[Bug target/104496] New: [12 Regression] ICE: in as_a, at machmode.h:365 with -O -mgeneral-regs-only -msse

[zsojka at seznam dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104496

            Bug ID: 104496
           Summary: [12 Regression] ICE: in as_a, at machmode.h:365 with
                    -O -mgeneral-regs-only -msse
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: zsojka at seznam dot cz
  Target Milestone: ---
              Host: x86_64-pc-linux-gnu
            Target: x86_64-pc-linux-gnu

Created attachment 52411
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52411&action=edit
reduced testcase

Compiler output:
$ x86_64-pc-linux-gnu-gcc -O -mgeneral-regs-only -msse testcase.c 
during GIMPLE pass: ccp
testcase.c: In function 'foo':
testcase.c:9:1: internal compiler error: in as_a, at machmode.h:365
    9 | }
      | ^
0x785e9a scalar_int_mode as_a<scalar_int_mode>(machine_mode)
        /repo/gcc-trunk/gcc/machmode.h:365
0x788612 scalar_mode as_a<scalar_mode>(machine_mode)
        /repo/gcc-trunk/gcc/tree.h:3568
0x788612 layout_type(tree_node*)
        /repo/gcc-trunk/gcc/stor-layout.cc:2402
0x16763a9 make_vector_type
        /repo/gcc-trunk/gcc/tree.cc:8945
0x16764dd make_vector_type
        /repo/gcc-trunk/gcc/tree.cc:8943
0x10cb3a0 vectorized_internal_fn_supported_p(internal_fn, tree_node*)
        /repo/gcc-trunk/gcc/internal-fn.cc:4401
0x186815f gimple_simplify_VEC_COND_EXPR
        /repo/build-gcc-trunk-amd64/gcc/gimple-match.cc:195745
0x187abb2 gimple_resimplify3
        /repo/gcc-trunk/gcc/gimple-match-head.cc:397
0x19184e5 gimple_simplify_BIT_IOR_EXPR
        /repo/build-gcc-trunk-amd64/gcc/gimple-match.cc:115459
0x185cd22 gimple_resimplify2
        /repo/gcc-trunk/gcc/gimple-match-head.cc:321
0x18bbe10 gimple_simplify(gimple*, gimple_match_op*, gimple**, tree_node*
(*)(tree_node*), tree_node* (*)(tree_node*))
        /repo/gcc-trunk/gcc/gimple-match-head.cc:1113
0x1044339 gimple_fold_stmt_to_constant_1(gimple*, tree_node* (*)(tree_node*),
tree_node* (*)(tree_node*))
        /repo/gcc-trunk/gcc/gimple-fold.cc:7328
0x14b5b6f ccp_fold
        /repo/gcc-trunk/gcc/tree-ssa-ccp.cc:1289
0x14b5b6f evaluate_stmt
        /repo/gcc-trunk/gcc/tree-ssa-ccp.cc:2208
0x14b7ebd visit_assignment
        /repo/gcc-trunk/gcc/tree-ssa-ccp.cc:2841
0x1556fb3 ssa_propagation_engine::simulate_stmt(gimple*)
        /repo/gcc-trunk/gcc/tree-ssa-propagate.cc:230
0x155728a ssa_propagation_engine::simulate_block(basic_block_def*)
        /repo/gcc-trunk/gcc/tree-ssa-propagate.cc:337
0x15576a3 ssa_propagation_engine::ssa_propagate()
        /repo/gcc-trunk/gcc/tree-ssa-propagate.cc:504
0x14aa200 do_ssa_ccp
        /repo/gcc-trunk/gcc/tree-ssa-ccp.cc:2960
0x14aa200 execute
        /repo/gcc-trunk/gcc/tree-ssa-ccp.cc:3004
Please submit a full bug report, with preprocessed source (by using
-freport-bug).
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.

$ x86_64-pc-linux-gnu-gcc -v
Using built-in specs.
COLLECT_GCC=/repo/gcc-trunk/binary-latest-amd64/bin/x86_64-pc-linux-gnu-gcc
COLLECT_LTO_WRAPPER=/repo/gcc-trunk/binary-trunk-r12-7186-20220210101436-gb25a7c05f12-checking-yes-rtl-df-extra-nobootstrap-amd64/bin/../libexec/gcc/x86_64-pc-linux-gnu/12.0.1/lto-wrapper
Target: x86_64-pc-linux-gnu
Configured with: /repo/gcc-trunk//configure --enable-languages=c,c++
--enable-valgrind-annotations --disable-nls --enable-checking=yes,rtl,df,extra
--disable-bootstrap --with-cloog --with-ppl --with-isl
--build=x86_64-pc-linux-gnu --host=x86_64-pc-linux-gnu
--target=x86_64-pc-linux-gnu --with-ld=/usr/bin/x86_64-pc-linux-gnu-ld
--with-as=/usr/bin/x86_64-pc-linux-gnu-as --disable-libstdcxx-pch
--prefix=/repo/gcc-trunk//binary-trunk-r12-7186-20220210101436-gb25a7c05f12-checking-yes-rtl-df-extra-nobootstrap-amd64
Thread model: posix
Supported LTO compression algorithms: zlib zstd
gcc version 12.0.1 20220210 (experimental) (GCC)

[Bug target/104496] [12 Regression] ICE: in as_a, at machmode.h:365 with -O -mgeneral-regs-only -msse

[crazylht at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104496

--- Comment #1 from Hongtao.liu <crazylht at gmail dot com> ---
w/ -mno-sse2 -mno-mmx, reproduce ICE.

[Bug target/104496] [12 Regression] ICE: in as_a, at machmode.h:365 with -O -mgeneral-regs-only -msse

[crazylht at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104496

--- Comment #2 from Hongtao.liu <crazylht at gmail dot com> ---
here is type V, a vector type with DImode and hit ICE in build 
build_vector_type_for_mode (type, mode) which take type as inner_type.

 <vector_type 0x7fffea3ec498
    type <integer_type 0x7fffea2fe5e8 int public SI
        size <integer_cst 0x7fffea2e2f90 constant 32>
        unit-size <integer_cst 0x7fffea2e2fa8 constant 4>
        align:32 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type
0x7fffea2fe5e8 precision:32 min <integer_cst 0x7fffea2e2f48 -2147483648> max
<integer_cst 0x7fffea2e2f60 2147483647>
        pointer_to_this <pointer_type 0x7fffea3069d8>>
    DI
    size <integer_cst 0x7fffea2e2d50 type <integer_type 0x7fffea2fe0a8
bitsizetype> constant 64>
    unit-size <integer_cst 0x7fffea2e2d68 type <integer_type 0x7fffea2fe000
sizetype> constant 8>

[Bug target/104496] [12 Regression] ICE: in as_a, at machmode.h:365 with -O -mgeneral-regs-only -msse

[crazylht at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104496

--- Comment #3 from Hongtao.liu <crazylht at gmail dot com> ---
(In reply to Hongtao.liu from comment #2)
> here is type V, a vector type with DImode and hit ICE in build 
> build_vector_type_for_mode (type, mode) which take type as inner_type.
> 
>  <vector_type 0x7fffea3ec498
>     type <integer_type 0x7fffea2fe5e8 int public SI
>         size <integer_cst 0x7fffea2e2f90 constant 32>
>         unit-size <integer_cst 0x7fffea2e2fa8 constant 4>
>         align:32 warn_if_not_align:0 symtab:0 alias-set -1 canonical-type
> 0x7fffea2fe5e8 precision:32 min <integer_cst 0x7fffea2e2f48 -2147483648> max
> <integer_cst 0x7fffea2e2f60 2147483647>
>         pointer_to_this <pointer_type 0x7fffea3069d8>>
>     DI
>     size <integer_cst 0x7fffea2e2d50 type <integer_type 0x7fffea2fe0a8
> bitsizetype> constant 64>
>     unit-size <integer_cst 0x7fffea2e2d68 type <integer_type 0x7fffea2fe000
> sizetype> constant 8>

/* If TYPE is a vector type, return true if IFN is a direct internal
   function that is supported for that type.  If TYPE is a scalar type,
   return true if IFN is a direct internal function that is supported for
   the target's preferred vector version of TYPE.  */

Hmm, TYPE is vector type, but mode is scalar mode.

[Bug target/104496] [12 Regression] ICE: in as_a, at machmode.h:365 with -O -mgeneral-regs-only -msse

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104496

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |12.0
             Status|UNCONFIRMED                 |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org
   Last reconfirmed|                            |2022-02-11
     Ever confirmed|0                           |1

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
So the issue is that for GPR mode vectors vectorized_internal_fn_supported_p
isn't careful enough.  Let me fix that.

[Bug target/104496] [12 Regression] ICE: in as_a, at machmode.h:365 with -O -mgeneral-regs-only -msse

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104496

--- Comment #5 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Richard Biener <rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:72f8d228aff80ede554b1735533af7eb0bd4a912

commit r12-7195-g72f8d228aff80ede554b1735533af7eb0bd4a912
Author: Richard Biener <rguenther@suse.de>
Date:   Fri Feb 11 10:27:20 2022 +0100

    middle-end/104496 - fix vectorized_internal_fn_supported_p

    This fixes vectorized_internal_fn_supported_p behavior when
    facing vector types with an integer mode.

    2022-02-11  Richard Biener  <rguenther@suse.de>

            PR middle-end/104496
            * internal-fn.cc (vectorized_internal_fn_supported_p):
            Bail out for integer mode vector types.

            * gcc.target/i386/pr104496.c: New testcase.

[Bug target/104496] [12 Regression] ICE: in as_a, at machmode.h:365 with -O -mgeneral-regs-only -msse

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104496

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

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

--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.