llvm/test/CodeGen/AArch64/aarch64-split-and-bitmask-immediate.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=aarch64-none-linux-gnu -verify-machineinstrs < %s | FileCheck %s

define i8 @test1(i32 %a) {
; CHECK-LABEL: test1:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    and w8, w0, #0x3ffc00
; CHECK-NEXT:    and w8, w8, #0xffe007ff
; CHECK-NEXT:    cmp w8, #1024
; CHECK-NEXT:    cset w0, eq
; CHECK-NEXT:    ret
entry:
  %and = and i32 %a, 2098176
  %cmp = icmp eq i32 %and, 1024
  %conv = zext i1 %cmp to i8
  ret i8 %conv
}

; This constant should not be split because it can be handled by one mov.
define i8 @test2(i32 %a) {
; CHECK-LABEL: test2:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov w8, #135
; CHECK-NEXT:    and w8, w0, w8
; CHECK-NEXT:    cmp w8, #1024
; CHECK-NEXT:    cset w0, eq
; CHECK-NEXT:    ret
entry:
  %and = and i32 %a, 135
  %cmp = icmp eq i32 %and, 1024
  %conv = zext i1 %cmp to i8
  ret i8 %conv
}

; This constant should not be split because the split immediate is not valid
; bitmask immediate.
define i8 @test3(i32 %a) {
; CHECK-LABEL: test3:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov w8, #1024
; CHECK-NEXT:    movk w8, #33, lsl #16
; CHECK-NEXT:    and w8, w0, w8
; CHECK-NEXT:    cmp w8, #1024
; CHECK-NEXT:    cset w0, eq
; CHECK-NEXT:    ret
entry:
  %and = and i32 %a, 2163712
  %cmp = icmp eq i32 %and, 1024
  %conv = zext i1 %cmp to i8
  ret i8 %conv
}

define i8 @test4(i64 %a) {
; CHECK-LABEL: test4:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    and x8, x0, #0x3ffc00
; CHECK-NEXT:    and x8, x8, #0xffffffffffe007ff
; CHECK-NEXT:    cmp x8, #1024
; CHECK-NEXT:    cset w0, eq
; CHECK-NEXT:    ret
entry:
  %and = and i64 %a, 2098176
  %cmp = icmp eq i64 %and, 1024
  %conv = zext i1 %cmp to i8
  ret i8 %conv
}

define i8 @test5(i64 %a) {
; CHECK-LABEL: test5:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    and x8, x0, #0x3ffffc000
; CHECK-NEXT:    and x8, x8, #0xfffffffe00007fff
; CHECK-NEXT:    cmp x8, #1024
; CHECK-NEXT:    cset w0, eq
; CHECK-NEXT:    ret
entry:
  %and = and i64 %a, 8589950976
  %cmp = icmp eq i64 %and, 1024
  %conv = zext i1 %cmp to i8
  ret i8 %conv
}

; This constant should not be split because it can be handled by one mov.
define i8 @test6(i64 %a) {
; CHECK-LABEL: test6:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov w8, #135
; CHECK-NEXT:    and x8, x0, x8
; CHECK-NEXT:    cmp x8, #1024
; CHECK-NEXT:    cset w0, eq
; CHECK-NEXT:    ret
entry:
  %and = and i64 %a, 135
  %cmp = icmp eq i64 %and, 1024
  %conv = zext i1 %cmp to i8
  ret i8 %conv
}

; This constant should not be split because the split immediate is not valid
; bitmask immediate.
define i8 @test7(i64 %a) {
; CHECK-LABEL: test7:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov w8, #1024
; CHECK-NEXT:    movk w8, #33, lsl #16
; CHECK-NEXT:    and x8, x0, x8
; CHECK-NEXT:    cmp x8, #1024
; CHECK-NEXT:    cset w0, eq
; CHECK-NEXT:    ret
entry:
  %and = and i64 %a, 2163712
  %cmp = icmp eq i64 %and, 1024
  %conv = zext i1 %cmp to i8
  ret i8 %conv
}

; The split bitmask immediates should be hoisted outside loop because they are
; loop invariant.
define void @test8(i64 %a, i64* noalias %src, i64* noalias %dst, i64 %n) {
; CHECK-LABEL: test8:
; CHECK:       // %bb.0: // %loop.ph
; CHECK-NEXT:    and x9, x0, #0x3ffc00
; CHECK-NEXT:    mov x8, xzr
; CHECK-NEXT:    and x9, x9, #0xffffffffffe007ff
; CHECK-NEXT:    b .LBB7_2
; CHECK-NEXT:  .LBB7_1: // %for.inc
; CHECK-NEXT:    // in Loop: Header=BB7_2 Depth=1
; CHECK-NEXT:    add x8, x8, #1
; CHECK-NEXT:    cmp x8, x3
; CHECK-NEXT:    b.gt .LBB7_4
; CHECK-NEXT:  .LBB7_2: // %loop
; CHECK-NEXT:    // =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    cmp x8, x9
; CHECK-NEXT:    b.hs .LBB7_1
; CHECK-NEXT:  // %bb.3: // %if.then
; CHECK-NEXT:    // in Loop: Header=BB7_2 Depth=1
; CHECK-NEXT:    lsl x10, x8, #3
; CHECK-NEXT:    ldr x11, [x1, x10]
; CHECK-NEXT:    str x11, [x2, x10]
; CHECK-NEXT:    b .LBB7_1
; CHECK-NEXT:  .LBB7_4: // %exit
; CHECK-NEXT:    ret
loop.ph:
  br label %loop

loop:
  %iv = phi i64 [ %inc, %for.inc ], [ 0, %loop.ph ]
  %and = and i64 %a, 2098176
  %cmp = icmp ult i64 %iv, %and
  br i1 %cmp, label %if.then, label %if.else

if.then:
  %src.arrayidx = getelementptr inbounds i64, i64* %src, i64 %iv
  %val = load i64, i64* %src.arrayidx
  %dst.arrayidx = getelementptr inbounds i64, i64* %dst, i64 %iv
  store i64 %val, i64* %dst.arrayidx
  br label %for.inc

if.else:
  br label %for.inc

for.inc:
  %inc = add nuw nsw i64 %iv, 1
  %cond = icmp sgt i64 %inc, %n
  br i1 %cond, label %exit, label %loop

exit:
  ret void
}

; This constant should not be split because the `and` is not loop invariant.
define i32 @test9(i32* nocapture %x, i32* nocapture readonly %y, i32 %n) {
; CHECK-LABEL: test9:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    cmp w2, #1
; CHECK-NEXT:    b.lt .LBB8_3
; CHECK-NEXT:  // %bb.1: // %for.body.preheader
; CHECK-NEXT:    mov w9, #1024
; CHECK-NEXT:    mov w8, w2
; CHECK-NEXT:    movk w9, #32, lsl #16
; CHECK-NEXT:  .LBB8_2: // %for.body
; CHECK-NEXT:    // =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    ldr w10, [x1], #4
; CHECK-NEXT:    subs x8, x8, #1
; CHECK-NEXT:    and w10, w10, w9
; CHECK-NEXT:    str w10, [x0], #4
; CHECK-NEXT:    b.ne .LBB8_2
; CHECK-NEXT:  .LBB8_3: // %for.cond.cleanup
; CHECK-NEXT:    mov w0, wzr
; CHECK-NEXT:    ret
entry:
  %cmp8 = icmp sgt i32 %n, 0
  br i1 %cmp8, label %for.body.preheader, label %for.cond.cleanup

for.body.preheader:                               ; preds = %entry
  %wide.trip.count = zext i32 %n to i64
  br label %for.body

for.cond.cleanup:                                 ; preds = %for.body, %entry
  ret i32 0

for.body:                                         ; preds = %for.body.preheader, %for.body
  %indvars.iv = phi i64 [ 0, %for.body.preheader ], [ %indvars.iv.next, %for.body ]
  %arrayidx = getelementptr inbounds i32, i32* %y, i64 %indvars.iv
  %0 = load i32, i32* %arrayidx, align 4
  %and = and i32 %0, 2098176
  %arrayidx2 = getelementptr inbounds i32, i32* %x, i64 %indvars.iv
  store i32 %and, i32* %arrayidx2, align 4
  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
  %exitcond.not = icmp eq i64 %indvars.iv.next, %wide.trip.count
  br i1 %exitcond.not, label %for.cond.cleanup, label %for.body
}

; After instruction selection end, we can see the `and` and `or` share the
; constant as below.
;
; %4:gpr32 = MOVi32imm 2098176
; %5:gpr32 = ANDWrr killed %3:gpr32, %4:gpr32
; STRWui killed %5:gpr32, %0:gpr64common, 0 :: (store (s32) into %ir.x, !tbaa !8)
; %6:gpr32 = LDRWui %1:gpr64common, 0 :: (load (s32) from %ir.y, !tbaa !8)
; %7:gpr32 = ORRWrr killed %6:gpr32, %4:gpr32
;
; In this case, the constant should not be split because it causes more
; instructions.
define void @test10(i32* nocapture %x, i32* nocapture readonly %y, i32* nocapture %z) {
; CHECK-LABEL: test10:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    ldr w8, [x1]
; CHECK-NEXT:    mov w9, #1024
; CHECK-NEXT:    movk w9, #32, lsl #16
; CHECK-NEXT:    and w8, w8, w9
; CHECK-NEXT:    str w8, [x0]
; CHECK-NEXT:    ldr w8, [x1]
; CHECK-NEXT:    orr w8, w8, w9
; CHECK-NEXT:    str w8, [x2]
; CHECK-NEXT:    ret
entry:
  %0 = load i32, i32* %y, align 4
  %and = and i32 %0, 2098176
  store i32 %and, i32* %x, align 4
  %1 = load i32, i32* %y, align 4
  %or = or i32 %1, 2098176
  store i32 %or, i32* %z, align 4
  ret void
}

; This test genereates below MIs.
;
; MOVi32imm -1610612736
; SUBREG_TO_REG
;
; The constant should be zero-extended to 64 bit and it should not be split.
define i8 @test11(i64 %a) {
; CHECK-LABEL: test11:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov w8, #-1610612736
; CHECK-NEXT:    and x8, x0, x8
; CHECK-NEXT:    cmp x8, #1024
; CHECK-NEXT:    cset w0, eq
; CHECK-NEXT:    ret
entry:
  %and = and i64 %a, 2684354560
  %cmp = icmp eq i64 %and, 1024
  %conv = zext i1 %cmp to i8
  ret i8 %conv
}