test/CodeGen/AArch64/sve-fold-loadext-and-splat-vector.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s | FileCheck %s

target triple = "aarch64-unknown-linux-gnu"

; Ensure that a no-op 'and' after an extending load gets removed when the and is
; constructed via a splat_vector node.
define <vscale x 2 x i64> @fold_loadext_and(ptr %ptr, i32 %needle, <vscale x 2 x i64> %b) #0 {
; CHECK-LABEL: fold_loadext_and:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    ld1w { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ret
  %load = load <vscale x 2 x i32>, ptr %ptr, align 4
  %ext = zext <vscale x 2 x i32> %load to <vscale x 2 x i64>
  %splatinsert = insertelement <vscale x 2 x i64> poison, i64 4294967295, i64 0
  %splat = shufflevector <vscale x 2 x i64> %splatinsert, <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
  %and = and <vscale x 2 x i64> %ext, %splat
  ret <vscale x 2 x i64> %and
}

; Same as above but testing the case we care about. Here the vscale x 2 x i32
; types get legalized into vscale x 2 x i64 types which introduces the extending
; load and 'and' nodes similar to the above case.
define <vscale x 2 x i1> @fold_loadext_and_legalize(ptr %ptr, <vscale x 2 x i32> %a) #0 {
; CHECK-LABEL: fold_loadext_and_legalize:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    and z0.d, z0.d, #0xffffffff
; CHECK-NEXT:    ld1w { z1.d }, p0/z, [x0]
; CHECK-NEXT:    cmpeq p0.d, p0/z, z1.d, z0.d
; CHECK-NEXT:    ret
  %load = load <vscale x 2 x i32>, ptr %ptr
  %cmp = icmp eq <vscale x 2 x i32> %load, %a
  ret <vscale x 2 x i1> %cmp
}

attributes #0 = { "target-features"="+sve" }