[Bug target/120926] New: Optimise away SVE PTRUE when applied to CNTP instruction

[ktkachov at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120926

            Bug ID: 120926
           Summary: Optimise away SVE PTRUE when applied to CNTP
                    instruction
           Product: gcc
           Version: 15.0
            Status: UNCONFIRMED
          Keywords: aarch64-sve, missed-optimization
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: ktkachov at gcc dot gnu.org
  Target Milestone: ---
            Target: aarch64

#include <arm_sve.h>

uint64_t foo(svbool_t p) {
  return svcntp_b64(svptrue_b64(), p);
}

generates:
foo:
        ptrue   p3.b, all
        cntp    x0, p3, p0.d
        ret

But I think it could generate:
foo:
        cntp    x0, p0, p0.d
        ret

As the CNTP instruction effectively performs an AND of the source and governing
predicates.