[Bug target/112681] New: [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp()

[Bug target/112681] New: [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp()

[zsojka at seznam dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112681

            Bug ID: 112681
           Summary: [14 Regression] ICE: in extract_insn, at recog.cc:2804
                    (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and
                    memcmp()
           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: zsojka at seznam dot cz
  Target Milestone: ---
              Host: x86_64-pc-linux-gnu
            Target: x86_64-pc-linux-gnu

Created attachment 56673
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=56673&action=edit
reduced testcase

Compiler output:
$ x86_64-pc-linux-gnu-gcc -O2 -mfma -mno-sse4.2 testcase.C 
testcase.C: In function 'int foo()':
testcase.C:11:1: error: unrecognizable insn:
   11 | }
      | ^
(insn 8 7 9 2 (set (reg:V16QI 104)
        (xor:V16QI (reg:V16QI 103)
            (mem/c:V16QI (reg/f:DI 102) [0 MEM <char[1:16]> [(void *)&b + 8B]+0
S16 A64]))) "testcase.C":10:27 -1
     (nil))
during RTL pass: vregs
testcase.C:11:1: internal compiler error: in extract_insn, at recog.cc:2804
0x8d9b93 _fatal_insn(char const*, rtx_def const*, char const*, int, char
const*)
        /repo/gcc-trunk/gcc/rtl-error.cc:108
0x8d9c10 _fatal_insn_not_found(rtx_def const*, char const*, int, char const*)
        /repo/gcc-trunk/gcc/rtl-error.cc:116
0x8c8a7f extract_insn(rtx_insn*)
        /repo/gcc-trunk/gcc/recog.cc:2804
0x14032bd instantiate_virtual_regs_in_insn
        /repo/gcc-trunk/gcc/function.cc:1610
0x14032bd instantiate_virtual_regs
        /repo/gcc-trunk/gcc/function.cc:1993
0x14032bd execute
        /repo/gcc-trunk/gcc/function.cc:2040
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-r14-5779-20231123205631-g746344dd538-checking-yes-rtl-df-extra-nobootstrap-amd64/bin/../libexec/gcc/x86_64-pc-linux-gnu/14.0.0/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-r14-5779-20231123205631-g746344dd538-checking-yes-rtl-df-extra-nobootstrap-amd64
Thread model: posix
Supported LTO compression algorithms: zlib zstd
gcc version 14.0.0 20231123 (experimental) (GCC)

[Bug target/112681] [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp()

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112681

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |14.0

[Bug target/112681] [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp() since r14-5747

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112681

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
            Summary|[14 Regression] ICE: in     |[14 Regression] ICE: in
                   |extract_insn, at            |extract_insn, at
                   |recog.cc:2804               |recog.cc:2804
                   |(unrecognizable insn) with  |(unrecognizable insn) with
                   |-O2 -mfma -mno-sse4.2 and   |-O2 -mfma -mno-sse4.2 and
                   |memcmp()                    |memcmp() since r14-5747
                 CC|                            |jakub at gcc dot gnu.org
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2023-11-23

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Started with r14-5747-g2e51fff7ce02a6105aa1acff57cbbdd8a767a33f

[Bug target/112681] [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp() since r14-5747

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112681

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Doesn't need the virtual method nor C++:
struct S { void *c; char d[16]; } a, b;

int
foo (void)
{
  return __builtin_memcmp (a.d, b.d, sizeof (a.d)) != 0;
}

[Bug target/112681] [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp() since r14-5747

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112681

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |manuel.lauss at googlemail dot com

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
*** Bug 112676 has been marked as a duplicate of this bug. ***

[Bug target/112681] [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp() since r14-5747

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112681

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Flags can be just -O2 -msse4.1 -mno-sse4.2 as well or even -O2 -msse4.2
-mno-avx.
--- gcc/config/i386/i386-expand.cc.jj   2023-11-21 09:31:35.792395304 +0100
+++ gcc/config/i386/i386-expand.cc      2023-11-23 20:43:51.675586034 +0100
@@ -2453,6 +2453,8 @@ ix86_expand_branch (enum rtx_code code,
          /* Generate XOR since we can't check that one operand is zero
             vector.  */
          tmp = gen_reg_rtx (mode);
+         if (!vector_operand (op1, mode))
+           op1 = force_reg (mode, op1);
          emit_insn (gen_rtx_SET (tmp, gen_rtx_XOR (mode, op0, op1)));
          tmp = gen_lowpart (p_mode, tmp);
          emit_insn (gen_rtx_SET (gen_rtx_REG (CCZmode, FLAGS_REG),
seems to fix it.

[Bug target/112681] [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp() since r14-5747

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112681

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Or maybe better
--- gcc/config/i386/i386-expand.cc.jj   2023-11-21 09:31:35.792395304 +0100
+++ gcc/config/i386/i386-expand.cc      2023-11-23 20:57:57.128721762 +0100
@@ -2453,7 +2453,8 @@ ix86_expand_branch (enum rtx_code code,
          /* Generate XOR since we can't check that one operand is zero
             vector.  */
          tmp = gen_reg_rtx (mode);
-         emit_insn (gen_rtx_SET (tmp, gen_rtx_XOR (mode, op0, op1)));
+         rtx ops[3] = { tmp, op0, op1 };
+         ix86_expand_vector_logical_operator (XOR, mode, ops);
          tmp = gen_lowpart (p_mode, tmp);
          emit_insn (gen_rtx_SET (gen_rtx_REG (CCZmode, FLAGS_REG),
                                  gen_rtx_UNSPEC (CCZmode,

[Bug target/112681] [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp() since r14-5747

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112681

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

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

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 56676
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=56676&action=edit
gcc14-pr112681.patch

Full untested patch.

[Bug target/112681] [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp() since r14-5747

[manuel.lauss at googlemail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112681

--- Comment #7 from Manuel Lauss <manuel.lauss at googlemail dot com> ---
(In reply to Jakub Jelinek from comment #6)
> Created attachment 56676 [details]
> gcc14-pr112681.patch
> 
> Full untested patch.

Fixes all ICEs I've seen today.
Thank you!

[Bug target/112681] [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp() since r14-5747

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112681

--- Comment #8 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:3eb9cae6d375d222787498b15ac87f383b3834fe

commit r14-5822-g3eb9cae6d375d222787498b15ac87f383b3834fe
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Fri Nov 24 12:12:20 2023 +0100

    i386: Fix ICE during cbranchv16qi4 expansion [PR112681]

    The following testcase ICEs, because cbranchv16qi4 expansion calls
    ix86_expand_branch with op1 being a pre-AVX unaligned memory and
    ix86_expand_branch emits a xorv16qi3 instruction without making sure
    the operand predicates are satisfied.
    While I could manually check if the argument (or both?) doesn't
    match vector_operand predicate (apparently this one or bcst_vector_operand
    is used in all integral 16+ bytes *xorv*3 instructions) force it into a
    register, but as all gen_xorv*3 expanders call
    ix86_expand_vector_logical_operator, it seems easier to just call that
    function which ensures the right thing happens.  Calling the individual
    gen_xorv*3 functions would mean ugly switch on the modes and using high
    level expand_simple_binop here seems too high level to me.

    2023-11-24  Jakub Jelinek  <jakub@redhat.com>

            PR target/112681
            * config/i386/i386-expand.cc (ix86_expand_branch): Use
            ix86_expand_vector_logical_operator to expand vector XOR rather
than
            gen_rtx_SET on gen_rtx_XOR.

            * gcc.target/i386/sse4-pr112681.c: New test.

[Bug target/112681] [14 Regression] ICE: in extract_insn, at recog.cc:2804 (unrecognizable insn) with -O2 -mfma -mno-sse4.2 and memcmp() since r14-5747

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112681

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

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

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed.