[Bug target/126959] New: -march=znver4 prefers load from memory for mask register

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126959

            Bug ID: 126959
           Summary: -march=znver4 prefers load from memory for mask
                    register
           Product: gcc
           Version: 16.2.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: hjl.tools at gmail dot com
  Target Milestone: ---
            Target: x86

[hjl@gnu-tgl-3 tmp]$ cat x.c
void
foo (void)
{
  unsigned short int k = 0;
  __asm volatile ("" : : "k" (k));
  k = -1;
  __asm volatile ("" : : "k" (k));
}
[hjl@gnu-tgl-3 tmp]$ gcc -S -O2 -march=tigerlake x.c
[hjl@gnu-tgl-3 tmp]$ cat x.s
        .file   "x.c"
        .text
        .p2align 4
        .globl  foo
        .type   foo, @function
foo:
.LFB0:
        .cfi_startproc
        kxorw   %k0, %k0, %k0
        kxnorw  %k1, %k1, %k1
        ret
        .cfi_endproc
.LFE0:
        .size   foo, .-foo
        .ident  "GCC: (GNU) 16.2.1 20260807 (Red Hat 16.2.1-1)"
        .section        .note.GNU-stack,"",@progbits
[hjl@gnu-tgl-3 tmp]$ gcc -S -O2 -march=znver4 x.c
[hjl@gnu-tgl-3 tmp]$ cat x.s
        .file   "x.c"
        .text
        .p2align 4
        .globl  foo
        .type   foo, @function
foo:
.LFB0:
        .cfi_startproc
        kxorw   %k0, %k0, %k0
        kmovw   .LC0(%rip), %k1
        ret
        .cfi_endproc
.LFE0:
        .size   foo, .-foo
        .section        .rodata.cst2,"aM",@progbits,2
        .align 2
.LC0:
        .value  -1
        .ident  "GCC: (GNU) 16.2.1 20260807 (Red Hat 16.2.1-1)"
        .section        .note.GNU-stack,"",@progbits
[hjl@gnu-tgl-3 tmp]$