[Bug rtl-optimization/128119] New: [17 Regression] volatile and atomic loads removed at -O2 on x86_64

[yuki.tanaka at defaultcase dot de via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128119

            Bug ID: 128119
           Summary: [17 Regression] volatile and atomic loads removed at
                    -O2 on x86_64
           Product: gcc
           Version: 17.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: yuki.tanaka at defaultcase dot de
  Target Milestone: ---

typedef double v2df __attribute__ ((vector_size (16)));

unsigned long a(volatile unsigned long *p, v2df m)
{
  unsigned long c = (unsigned long) __builtin_ia32_comisdeq(m, m) | 1;
  unsigned long v = *p;
  return ((unsigned __int128) v * c) >> 64;
}

unsigned long b(unsigned long *p, v2df m)
{
  unsigned long c = (unsigned long) __builtin_ia32_comisdeq(m, m) | 1;
  unsigned long v = __atomic_load_n(p, __ATOMIC_RELAXED);
  return ((unsigned __int128) v * c) >> 64;
}

At -O2 GCC 17 removes the volatile load of *p in a and the atomic load in b.
GCC 16.2 keeps both.

Compiler Explorer, -O2 (GCC 16.2 vs. GCC 17):

x86-64 GCC 16.2:
"a":
        mov     eax, 1
        mul     QWORD PTR [rdi]
        mov     rax, rdx
        ret
"b":
        mov     eax, 1
        mul     QWORD PTR [rdi]
        mov     rax, rdx
        ret

x86-64 GCC 17.0.0:
"a":
        xor     eax, eax
        ret
"b":
        xor     eax, eax
        ret