[Bug target/109733] New: ICE in extract_insn, at recog.cc:2791 since r14-475-g508f082829af68

[Bug target/109733] New: ICE in extract_insn, at recog.cc:2791 since r14-475-g508f082829af68

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109733

            Bug ID: 109733
           Summary: ICE in extract_insn, at recog.cc:2791 since
                    r14-475-g508f082829af68
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: marxin at gcc dot gnu.org
                CC: ubizjak at gmail dot com
  Target Milestone: ---
              Host: x86_64-linux-gnu

I noticed the following crash:

$ cat x.f90
  use iso_c_binding
  interface
    subroutine ctest_cont (a, is_cont) bind (c, name="ctest")
    end subroutine
    subroutine ctest_ar_cont (a, is_cont) bind (c, name="ctest")
      use iso_c_binding
      integer(C_INT), contiguous :: a(..)
      logical(C_Bool), value :: is_cont
    end subroutine
  end interface
  integer(C_INT), target :: aa(10,5)
  integer(C_INT), target :: bb(10,10)
  do j = 1, 5
    do i = 1, 10
    end do
  end do
  call ftest (transpose (aa), is_cont=.true._c_bool) ! Implementation choice:
copy in; hence, contiguous
  call ftest (bb(2:10:2, :), is_cont=.false._c_bool)
contains
  subroutine ftest (a, is_cont)
    integer(C_INT) :: a(:,:)
    logical(c_bool), value, intent(in) :: is_cont
    call ctest_ar_cont (a, is_cont=.true._c_bool)
  end subroutine
  subroutine ftest_ar (a, is_cont)
  end subroutine
end program

$ gfortran x.f90 -ftree-vrp -O1 -mavx512vbmi2 -m32 -march=i686 -c
x.f90:24:16:

   24 |   end subroutine
      |                ^
Error: unrecognizable insn:
(insn 429 416 379 18 (set (reg:SI 72 k4 [orig:121 _114 ] [121])
        (mult:SI (reg:SI 70 k2 [orig:112 ubound.0 ] [112])
            (const_int 4 [0x4]))) -1
     (nil))
during RTL pass: split2
x.f90:24:16: internal compiler error: in extract_insn, at recog.cc:2791
0x1f4c694 diagnostic_impl(rich_location*, diagnostic_metadata const*, int, char
const*, __va_list_tag (*) [1], diagnostic_t)
        ???:0
0x1f4d2e6 internal_error(char const*, ...)
        ???:0
0x899ba5 fancy_abort(char const*, int, char const*)
        ???:0
0x76b68a _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
        ???:0
0x76b6a6 _fatal_insn_not_found(rtx_def const*, char const*, int, char const*)
        ???:0
0x769bc3 extract_insn(rtx_insn*) [clone .cold]
        ???:0
0xef4694 extract_insn_cached(rtx_insn*)
        ???:0
0xbd7d24 cleanup_subreg_operands(rtx_insn*)
        ???:0
0xef2676 split_insn(rtx_insn*)
        ???:0
0xef7aa9 split_all_insns()
        ???:0
0xef7bb8 (anonymous namespace)::pass_split_all_insns::execute(function*)
        ???:0
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.

[Bug target/109733] [14 Regression] ICE in extract_insn, at recog.cc:2791 since r14-475-g508f082829af68

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109733

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2023-05-04
            Summary|ICE in extract_insn, at     |[14 Regression] ICE in
                   |recog.cc:2791 since         |extract_insn, at
                   |r14-475-g508f082829af68     |recog.cc:2791 since
                   |                            |r14-475-g508f082829af68
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW

[Bug target/109733] [14 Regression] ICE in extract_insn, at recog.cc:2791 since r14-475-g508f082829af68

[ubizjak at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109733

--- Comment #1 from Uroš Bizjak <ubizjak at gmail dot com> ---
The patched compiler just happens to trigger the existing problem where:

(insn 188 416 379 18 (parallel [
            (set (reg:SI 72 k4 [orig:121 _114 ] [121])
                (ashift:SI (reg:SI 70 k2 [orig:112 ubound.0 ] [112])
                    (const_int 2 [0x2])))
            (clobber (reg:CC 17 flags))
        ]) 811 {*ashlsi3_1}
     (nil))

gets split with a post-reload splitter a to non-existent:

(insn 429 416 379 18 (set (reg:SI 72 k4 [orig:121 _114 ] [121])
        (mult:SI (reg:SI 70 k2 [orig:112 ubound.0 ] [112])
            (const_int 4 [0x4]))) -1
     (nil))

Please note mask registers, which points to a post-reload splitter with too
broad predicates.

[Bug target/109733] [14 Regression] ICE in extract_insn, at recog.cc:2791 since r14-475-g508f082829af68

[ubizjak at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109733

Uroš Bizjak <ubizjak at gmail dot com> changed:

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

--- Comment #2 from Uroš Bizjak <ubizjak at gmail dot com> ---
Created attachment 54996
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=54996&action=edit
Patch in testing.

[Bug target/109733] [14 Regression] ICE in extract_insn, at recog.cc:2791 since r14-475-g508f082829af68

[ubizjak at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109733

Uroš Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #54996|0                           |1
        is obsolete|                            |

--- Comment #3 from Uroš Bizjak <ubizjak at gmail dot com> ---
Created attachment 54997
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=54997&action=edit
The correct patch.

[Bug target/109733] [14 Regression] ICE in extract_insn, at recog.cc:2791 since r14-475-g508f082829af68

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109733

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

https://gcc.gnu.org/g:8cac23781753bd8a016507dc9b21ec563e1d9b49

commit r14-485-g8cac23781753bd8a016507dc9b21ec563e1d9b49
Author: Uros Bizjak <ubizjak@gmail.com>
Date:   Thu May 4 20:26:12 2023 +0200

    i386: Tighten ashift to lea splitter operand predicates [PR109733]

    The predicates of ashift to lea post-reload splitter were too broad
    so the splitter tried to convert the mask shift instruction.  Tighten
    operand predicates to match only general registers.

    gcc/ChangeLog:

            PR target/109733
            * config/i386/predicates.md (index_reg_operand): New predicate.
            * config/i386/i386.md (ashift to lea spliter): Use
            general_reg_operand and index_reg_operand predicates.

[Bug target/109733] [14 Regression] ICE in extract_insn, at recog.cc:2791 since r14-475-g508f082829af68

[ubizjak at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109733

Uroš Bizjak <ubizjak at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |14.0
             Status|ASSIGNED                    |RESOLVED
         Resolution|---                         |FIXED

--- Comment #5 from Uroš Bizjak <ubizjak at gmail dot com> ---
Fixed.