[Bug target/110227] New: gcc generates invalid AVX-512 code

[Bug target/110227] New: gcc generates invalid AVX-512 code

[joseph.weening at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110227

            Bug ID: 110227
           Summary: gcc generates invalid AVX-512 code
           Product: gcc
           Version: 13.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: joseph.weening at gmail dot com
  Target Milestone: ---

gcc version 13.1.0 (GCC)
Target: x86_64-pc-linux-gnu
Configured with: ../gcc-13.1.0/configure --prefix=/usr/local/gcc/13.1.0
--disable-multilib --enable-languages=c,c++,fortran
--with-gmp=/usr/local/gmp/6.2.1 --with-mpc=/usr/local/mpc/1.3.1
--with-mpfr=/usr/local/mpfr/4.2.0 --with-isl=/usr/local/isl/0.24

The following program generates an error:
  /tmp/ccAuHFqz.s: Assembler messages:
  /tmp/ccAuHFqz.s:28: Error: unsupported instruction `vpcmpeqd'
The assembly code contains
        vpcmpeqd         %xmm16, %xmm16, %xmm16
which perhaps is invalid for xmm registers above 15.

#include <immintrin.h>

__attribute__((noinline))
static void vswap(int32_t *x) {
  __m256i x0 = _mm256_loadu_si256((__m256i *) (&x[0]));
  __m256i x1 = _mm256_loadu_si256((__m256i *) (&x[1]));
  _mm256_storeu_si256((__m256i *) (&x[0]),(x1));
  _mm256_storeu_si256((__m256i *) (&x[1]),(x0));
}

void vproc(int32_t *x) {
  for (int32_t p=4; p>=1; p>>=1) {
    if (p == 4) {
      __m256i mask = _mm256_set_epi32(0, 0, 0, 0, -1, -1, -1, -1);
      __m256i x0 = _mm256_loadu_si256((__m256i *) (&x[0]));
      x0 = _mm256_xor_si256(x0, mask);
      _mm256_storeu_si256((__m256i *) (&x[0]),(x0));
    }
    vswap(x);
  }
}

[Bug target/110227] gcc generates invalid AVX-512 code

[joseph.weening at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110227

--- Comment #1 from Joe Weening <joseph.weening at gmail dot com> ---
Sorry, forgot to include the command line:

$ gcc -march=cooperlake -O3 -c bug.c

[Bug target/110227] [13/14 Regression] gcc generates invalid AVX-512 code

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110227

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |13.2
   Last reconfirmed|                            |2023-06-12
     Ever confirmed|0                           |1
            Summary|gcc generates invalid       |[13/14 Regression] gcc
                   |AVX-512 code                |generates invalid AVX-512
                   |                            |code
             Status|UNCONFIRMED                 |NEW

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Semi-Reduced testcase:
```
#include <immintrin.h>

void f()
{
  __m256i mask = _mm256_set_epi32(0, 0, 0, 0, -1, -1, -1, -1);
  register __m256i reg asm("xmm16") = mask;
  asm(""::"v"(reg));
}
```

Most likely introduced by r13-2804-ga282f086ef26d9 .

[Bug target/110227] [13/14 Regression] gcc generates invalid AVX-512 code

[crazylht at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110227

--- Comment #3 from Hongtao.liu <crazylht at gmail dot com> ---
(In reply to Andrew Pinski from comment #2)
> Semi-Reduced testcase:
> ```
> #include <immintrin.h>
> 
> void f()
> {
>   __m256i mask = _mm256_set_epi32(0, 0, 0, 0, -1, -1, -1, -1);
>   register __m256i reg asm("xmm16") = mask;
>   asm(""::"v"(reg));
> }
> ```
> 
> Most likely introduced by r13-2804-ga282f086ef26d9 .

 (define_insn "mov<mode>_internal"
   [(set (match_operand:VMOVE 0 "nonimmediate_operand"
-        "=v,v ,v ,m")
+        "=v,v ,v,v ,m")
        (match_operand:VMOVE 1 "nonimmediate_or_sse_const_operand"
-        " C,<sseconstm1>,vm,v"))]
+        " C,<sseconstm1>,BH,vm,v"))]

We need to use "x" instead of "v".

[Bug target/110227] [13/14 Regression] gcc generates invalid AVX-512 code

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110227

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

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

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Looking at that commit, it also introduced weird formatting:
          return "vpcmpeqd \t %t0, %t0, %t0";
and
          return "vpcmpeqd \t %x0, %x0, %x0";
and
      return "vpcmpeqd \t %x0, %x0, %x0";
definitely shouldn't have the spaces around \t on either side.
But sure, agree the alternative with BH constraint on source should use x
rather than
v constraint on destination.

[Bug target/110227] [13/14 Regression] gcc generates invalid AVX-512 code

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110227

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2

[Bug target/110227] [13/14 Regression] gcc generates invalid AVX-512 code

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110227

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

https://gcc.gnu.org/g:b7e42b85212e03eb59193a712eb523f26911a581

commit r14-1802-gb7e42b85212e03eb59193a712eb523f26911a581
Author: liuhongt <hongtao.liu@intel.com>
Date:   Tue Jun 13 14:20:59 2023 +0800

    Use x instead of v for alternative 2 (v, BH) in mov<mode>_internal.

    Since there's no evex version for vpcmpeq ymm, ymm, ymm.

    gcc/ChangeLog:

            PR target/110227
            * config/i386/sse.md (mov<mode>_internal>): Use x instead of v
            for alternative 2 since there's no evex version for vpcmpeqd
            ymm, ymm, ymm.

    gcc/testsuite/ChangeLog:

            * gcc.target/i386/pr110227.c: New test.

[Bug target/110227] [13/14 Regression] gcc generates invalid AVX-512 code

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110227

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

https://gcc.gnu.org/g:66f8f9b35f0499cf54b8a432fbe5f645ad1c523a

commit r13-7445-g66f8f9b35f0499cf54b8a432fbe5f645ad1c523a
Author: liuhongt <hongtao.liu@intel.com>
Date:   Tue Jun 13 14:20:59 2023 +0800

    Use x instead of v for alternative 2 (v, BH) in mov<mode>_internal.

    Since there's no evex version for vpcmpeq ymm, ymm, ymm.

    gcc/ChangeLog:

            PR target/110227
            * config/i386/sse.md (mov<mode>_internal>): Use x instead of v
            for alternative 2 since there's no evex version for vpcmpeqd
            ymm, ymm, ymm.

    gcc/testsuite/ChangeLog:

            * gcc.target/i386/pr110227.c: New test.

[Bug target/110227] [13/14 Regression] gcc generates invalid AVX-512 code

[crazylht at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110227

--- Comment #7 from Hongtao.liu <crazylht at gmail dot com> ---
Fixed for GCC14 and GCC13.2

[Bug target/110227] [13/14 Regression] gcc generates invalid AVX-512 code

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110227

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|13.2                        |13.3

--- Comment #8 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 13.2 is being released, retargeting bugs to GCC 13.3.

[Bug target/110227] [13/14 Regression] gcc generates invalid AVX-512 code

[liuhongt at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110227

liuhongt at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |liuhongt at gcc dot gnu.org
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED

--- Comment #9 from liuhongt at gcc dot gnu.org ---
.