[Bug target/109932] New: ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[Bug target/109932] New: ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[jamborm at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109932

            Bug ID: 109932
           Summary: ICE in in extract_insn, at recog.cc:2791 on ppc64le
                    with -mno-vsx
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: jamborm at gcc dot gnu.org
  Target Milestone: ---
              Host: x86_64-linux
            Target: ppc64le-linux-gnu

Using a cross compiler (revision ad0f80d945c) configured with

/home/worker/buildworker/tiber-gcc-trunk-ppc64le/build/configure
--enable-languages=c,c++,fortran,rust,m2 --disable-bootstrap
--disable-libsanitizer --disable-multilib --enable-checking=release
--prefix=/home/worker/cross --target=ppc64le-linux-gnu
--with-as=/usr/bin/powerpc64le-suse-linux-as

and running it on the following testcase (originally
CodeGen/PowerPC/builtins-ppc-int128.c /tmp/test.c from clang
testsuite)

-----------------------------------------------------------------
#include <altivec.h>

vector signed __int128 res_vslll;
unsigned long long aull[2] = { 1L, 2L };

void testVectorInt128Pack(){
  res_vslll = __builtin_pack_vector_int128(aull[0], aull[1]);
  __builtin_unpack_vector_int128(res_vslll, 0);
  __builtin_unpack_vector_int128(res_vslll, 1);
}
-----------------------------------------------------------------

with option -mno-vsx resuts in an ICE:

~/cross/bin/ppc64le-linux-gnu-gcc /tmp/test.c -mno-vsx
/tmp/test.c: In function ‘testVectorInt128Pack’:
/tmp/test.c:10:1: error: unrecognizable insn:
   10 | }
      | ^
(insn 9 8 12 2 (set (reg:V1TI 124)
        (unspec:V1TI [
                (reg:DI 117 [ _1 ])
                (reg:DI 118 [ _2 ])
            ] UNSPEC_PACK_128BIT)) "/tmp/test.c":7:15 -1
     (nil))
during RTL pass: vregs
/tmp/test.c:10:1: internal compiler error: in extract_insn, at recog.cc:2791
0x62c70b _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
       
/home/worker/buildworker/tiber-gcc-trunk-ppc64le/build/gcc/rtl-error.cc:108
0x62c727 _fatal_insn_not_found(rtx_def const*, char const*, int, char const*)
       
/home/worker/buildworker/tiber-gcc-trunk-ppc64le/build/gcc/rtl-error.cc:116
0x62bb8a extract_insn(rtx_insn*)
       
/home/worker/buildworker/tiber-gcc-trunk-ppc64le/build/gcc/recog.cc:2791
0x8dacb0 instantiate_virtual_regs_in_insn
       
/home/worker/buildworker/tiber-gcc-trunk-ppc64le/build/gcc/function.cc:1611
0x8dacb0 instantiate_virtual_regs
       
/home/worker/buildworker/tiber-gcc-trunk-ppc64le/build/gcc/function.cc:1984
0x8dacb0 execute
       
/home/worker/buildworker/tiber-gcc-trunk-ppc64le/build/gcc/function.cc:2033


I guess we should error out earlier and more user-friendly when
expanding the built-ins?

[Bug target/109932] ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[linkw at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109932

Kewen Lin <linkw at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2023-05-24
             Status|UNCONFIRMED                 |ASSIGNED
     Ever confirmed|0                           |1
                 CC|                            |linkw at gcc dot gnu.org
           Assignee|unassigned at gcc dot gnu.org      |linkw at gcc dot gnu.org
           Keywords|                            |ice-on-invalid-code

--- Comment #1 from Kewen Lin <linkw at gcc dot gnu.org> ---
Confirmed, thanks for reporting.

Both bif __builtin_pack_vector_int128 and __builtin_unpack_vector_int128 are
put in stanza power7 instead of vsx, so they miss to check vsx available or
not.

  const vsq __builtin_pack_vector_int128 (unsigned long long, \
                                          unsigned long long);
    PACK_V1TI packv1ti {}

  void __builtin_ppc_speculation_barrier ();
    SPECBARR speculation_barrier {}

  const unsigned long __builtin_unpack_vector_int128 (vsq, const int<1>);
    UNPACK_V1TI unpackv1ti {}

But the underlying insn patterns do need VSX support, see:

(define_insn "pack<mode>"
  [(set (match_operand:FMOVE128_VSX 0 "register_operand" "=wa")
        (unspec:FMOVE128_VSX
         [(match_operand:DI 1 "register_operand" "wa")
          (match_operand:DI 2 "register_operand" "wa")]
         UNSPEC_PACK_128BIT))]
  "TARGET_VSX"   // here
  "xxpermdi %x0,%x1,%x2,0"
  [(set_attr "type" "vecperm")])

(define_insn "unpack<mode>"
  [(set (match_operand:DI 0 "register_operand" "=wa,wa")
        (unspec:DI [(match_operand:FMOVE128_VSX 1 "register_operand" "0,wa")
                    (match_operand:QI 2 "const_0_to_1_operand" "O,i")]
         UNSPEC_UNPACK_128BIT))]
  "VECTOR_MEM_ALTIVEC_OR_VSX_P (<MODE>mode)"  // here
{
  if (REGNO (operands[0]) == REGNO (operands[1]) && INTVAL (operands[2]) == 0)
    return ASM_COMMENT_START " xxpermdi to same register";

  operands[3] = GEN_INT (INTVAL (operands[2]) == 0 ? 0 : 3);
  return "xxpermdi %x0,%x1,%x1,%3";
}
  [(set_attr "type" "vecperm")])

; Iterator for 128-bit VSX types for pack/unpack
(define_mode_iterator FMOVE128_VSX [V1TI KF])

both vector_mem for V1TI and KF are VSX.

So the fix would be to move these two bifs to be under stanza vsx.

[Bug target/109932] ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109932

--- Comment #2 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Kewen Lin <linkw@gcc.gnu.org>:

https://gcc.gnu.org/g:ff83d1b47aadcdaf80a4fda84b0dc00bb2cd3641

commit r14-1704-gff83d1b47aadcdaf80a4fda84b0dc00bb2cd3641
Author: Kewen Lin <linkw@linux.ibm.com>
Date:   Mon Jun 12 01:08:22 2023 -0500

    rs6000: Guard __builtin_{un,}pack_vector_int128 with vsx [PR109932]

    As PR109932 shows, builtins __builtin_{un,}pack_vector_int128
    should be guarded under vsx rather than power7, as their
    corresponding bif patterns have the conditions TARGET_VSX
    and VECTOR_MEM_ALTIVEC_OR_VSX_P (V1TImode).  This patch is to
    move __builtin_{un,}pack_vector_int128 to stanza vsx to ensure
    their supports.

            PR target/109932

    gcc/ChangeLog:

            * config/rs6000/rs6000-builtins.def (__builtin_pack_vector_int128,
            __builtin_unpack_vector_int128): Move from stanza power7 to vsx.

    gcc/testsuite/ChangeLog:

            * gcc.target/powerpc/pr109932-1.c: New test.
            * gcc.target/powerpc/pr109932-2.c: New test.

[Bug target/109932] ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109932

--- Comment #3 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Kewen Lin <linkw@gcc.gnu.org>:

https://gcc.gnu.org/g:16eb9d69079d769b2aa2c07ce54aca20f5547c14

commit r14-1776-g16eb9d69079d769b2aa2c07ce54aca20f5547c14
Author: Kewen Lin <linkw@linux.ibm.com>
Date:   Tue Jun 13 03:04:54 2023 -0500

    testsuite: Check int128 effective target for pr109932-{1,2}.c [PR110230]

    This patch is to make newly added test cases pr109932-{1,2}.c
    check int128 effective target to avoid unsupported type error
    on 32-bit.  I did hit this failure during testing and fixed
    it, but made a stupid mistake not updating the local formatted
    patch which was actually out of date.

            PR testsuite/110230
            PR target/109932

    gcc/testsuite/ChangeLog:

            * gcc.target/powerpc/pr109932-1.c: Adjust with int128 effective
target.
            * gcc.target/powerpc/pr109932-2.c: Ditto.

[Bug target/109932] ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109932

--- Comment #4 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-12 branch has been updated by Kewen Lin <linkw@gcc.gnu.org>:

https://gcc.gnu.org/g:31d88c795a0eb05df5a0684c34ec74116cce133f

commit r12-9713-g31d88c795a0eb05df5a0684c34ec74116cce133f
Author: Kewen Lin <linkw@linux.ibm.com>
Date:   Mon Jun 12 01:08:22 2023 -0500

    rs6000: Guard __builtin_{un,}pack_vector_int128 with vsx [PR109932]

    As PR109932 shows, builtins __builtin_{un,}pack_vector_int128
    should be guarded under vsx rather than power7, as their
    corresponding bif patterns have the conditions TARGET_VSX
    and VECTOR_MEM_ALTIVEC_OR_VSX_P (V1TImode).  This patch is to
    move __builtin_{un,}pack_vector_int128 to stanza vsx to ensure
    their supports.

            PR target/109932

    gcc/ChangeLog:

            * config/rs6000/rs6000-builtins.def (__builtin_pack_vector_int128,
            __builtin_unpack_vector_int128): Move from stanza power7 to vsx.

    gcc/testsuite/ChangeLog:

            * gcc.target/powerpc/pr109932-1.c: New test.
            * gcc.target/powerpc/pr109932-2.c: New test.

    (cherry picked from commit ff83d1b47aadcdaf80a4fda84b0dc00bb2cd3641)

[Bug target/109932] ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109932

--- Comment #5 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-12 branch has been updated by Kewen Lin <linkw@gcc.gnu.org>:

https://gcc.gnu.org/g:4591c2c8a6b15ca99ba049d84e0e694f12db4f60

commit r12-9714-g4591c2c8a6b15ca99ba049d84e0e694f12db4f60
Author: Kewen Lin <linkw@linux.ibm.com>
Date:   Tue Jun 13 03:04:54 2023 -0500

    testsuite: Check int128 effective target for pr109932-{1,2}.c [PR110230]

    This patch is to make newly added test cases pr109932-{1,2}.c
    check int128 effective target to avoid unsupported type error
    on 32-bit.  I did hit this failure during testing and fixed
    it, but made a stupid mistake not updating the local formatted
    patch which was actually out of date.

            PR testsuite/110230
            PR target/109932

    gcc/testsuite/ChangeLog:

            * gcc.target/powerpc/pr109932-1.c: Adjust with int128 effective
target.
            * gcc.target/powerpc/pr109932-2.c: Ditto.

    (cherry picked from commit 16eb9d69079d769b2aa2c07ce54aca20f5547c14)

[Bug target/109932] ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109932

--- Comment #6 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-13 branch has been updated by Kewen Lin <linkw@gcc.gnu.org>:

https://gcc.gnu.org/g:4e67d73ee5100c12993c79852e4ede13d2606cad

commit r13-7457-g4e67d73ee5100c12993c79852e4ede13d2606cad
Author: Kewen Lin <linkw@linux.ibm.com>
Date:   Mon Jun 12 01:08:22 2023 -0500

    rs6000: Guard __builtin_{un,}pack_vector_int128 with vsx [PR109932]

    As PR109932 shows, builtins __builtin_{un,}pack_vector_int128
    should be guarded under vsx rather than power7, as their
    corresponding bif patterns have the conditions TARGET_VSX
    and VECTOR_MEM_ALTIVEC_OR_VSX_P (V1TImode).  This patch is to
    move __builtin_{un,}pack_vector_int128 to stanza vsx to ensure
    their supports.

            PR target/109932

    gcc/ChangeLog:

            * config/rs6000/rs6000-builtins.def (__builtin_pack_vector_int128,
            __builtin_unpack_vector_int128): Move from stanza power7 to vsx.

    gcc/testsuite/ChangeLog:

            * gcc.target/powerpc/pr109932-1.c: New test.
            * gcc.target/powerpc/pr109932-2.c: New test.

    (cherry picked from commit ff83d1b47aadcdaf80a4fda84b0dc00bb2cd3641)

[Bug target/109932] ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109932

--- Comment #7 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-13 branch has been updated by Kewen Lin <linkw@gcc.gnu.org>:

https://gcc.gnu.org/g:4b4a21c93406aef276fbff00d3e9491285d7b4a9

commit r13-7458-g4b4a21c93406aef276fbff00d3e9491285d7b4a9
Author: Kewen Lin <linkw@linux.ibm.com>
Date:   Tue Jun 13 03:04:54 2023 -0500

    testsuite: Check int128 effective target for pr109932-{1,2}.c [PR110230]

    This patch is to make newly added test cases pr109932-{1,2}.c
    check int128 effective target to avoid unsupported type error
    on 32-bit.  I did hit this failure during testing and fixed
    it, but made a stupid mistake not updating the local formatted
    patch which was actually out of date.

            PR testsuite/110230
            PR target/109932

    gcc/testsuite/ChangeLog:

            * gcc.target/powerpc/pr109932-1.c: Adjust with int128 effective
target.
            * gcc.target/powerpc/pr109932-2.c: Ditto.

    (cherry picked from commit 16eb9d69079d769b2aa2c07ce54aca20f5547c14)

[Bug target/109932] ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109932

--- Comment #8 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-11 branch has been updated by Kewen Lin <linkw@gcc.gnu.org>:

https://gcc.gnu.org/g:db291447877aae67979ce3655fcc6fc877f57c6a

commit r11-10866-gdb291447877aae67979ce3655fcc6fc877f57c6a
Author: Kewen Lin <linkw@linux.ibm.com>
Date:   Tue Jun 20 01:40:52 2023 -0500

    rs6000: Guard __builtin_{un,}pack_vector_int128 with vsx [PR109932]

    As PR109932 shows, builtins __builtin_{un,}pack_vector_int128
    should be guarded under vsx rather than power7, as their
    corresponding bif patterns have the conditions TARGET_VSX
    and VECTOR_MEM_ALTIVEC_OR_VSX_P (V1TImode).  This patch is to
    ensure __builtin_{un,}pack_vector_int128 only available under
    vsx.

            PR target/109932

    gcc/ChangeLog:

            * config/rs6000/rs6000-builtin.def (BU_VSX_MISC_2): New macro.
            ({un,}pack_vector_int128): Use BU_VSX_MISC_2 instead of
            BU_P7_MISC_2.

    gcc/testsuite/ChangeLog:

            * gcc.target/powerpc/pr109932-1.c: New test.
            * gcc.target/powerpc/pr109932-2.c: New test.

[Bug target/109932] ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109932

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

https://gcc.gnu.org/g:98763feef24b58573fa9e6c6eedaccc1e932bb46

commit r10-11455-g98763feef24b58573fa9e6c6eedaccc1e932bb46
Author: Kewen Lin <linkw@linux.ibm.com>
Date:   Tue Jun 20 01:40:52 2023 -0500

    rs6000: Guard __builtin_{un,}pack_vector_int128 with vsx [PR109932]

    As PR109932 shows, builtins __builtin_{un,}pack_vector_int128
    should be guarded under vsx rather than power7, as their
    corresponding bif patterns have the conditions TARGET_VSX
    and VECTOR_MEM_ALTIVEC_OR_VSX_P (V1TImode).  This patch is to
    ensure __builtin_{un,}pack_vector_int128 only available under
    vsx.

            PR target/109932

    gcc/ChangeLog:

            * config/rs6000/rs6000-builtin.def (BU_VSX_MISC_2): New macro.
            ({un,}pack_vector_int128): Use BU_VSX_MISC_2 instead of
            BU_P7_MISC_2.

    gcc/testsuite/ChangeLog:

            * gcc.target/powerpc/pr109932-1.c: New test.
            * gcc.target/powerpc/pr109932-2.c: New test.

    (cherry picked from commit db291447877aae67979ce3655fcc6fc877f57c6a)

[Bug target/109932] ICE in in extract_insn, at recog.cc:2791 on ppc64le with -mno-vsx

[linkw at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109932

Kewen Lin <linkw at gcc dot gnu.org> changed:

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

--- Comment #10 from Kewen Lin <linkw at gcc dot gnu.org> ---
Should be fixed everywhere.