redundant-left-shift-input-masking-after-truncation-variant-a.ll 8 KB
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt %s -instcombine -S | FileCheck %s

; If we have some pattern that leaves only some low bits set, and then performs
; left-shift of those bits, we can combine those two shifts into a shift+mask.

; There are many variants to this pattern:
;   a)  (trunc (x & ((1 << maskNbits) - 1))) << shiftNbits
; simplify to:
;   (trunc(x)) << shiftNbits

; Simple tests.

declare void @use32(i32)
declare void @use64(i64)

define i32 @t0_basic(i64 %x, i32 %nbits) {
; CHECK-LABEL: @t0_basic(
; CHECK-NEXT:    [[T0:%.*]] = zext i32 [[NBITS:%.*]] to i64
; CHECK-NEXT:    [[T1:%.*]] = shl i64 1, [[T0]]
; CHECK-NEXT:    [[T2:%.*]] = add i64 [[T1]], -1
; CHECK-NEXT:    [[T3:%.*]] = sub i32 32, [[NBITS]]
; CHECK-NEXT:    [[T4:%.*]] = and i64 [[T2]], [[X:%.*]]
; CHECK-NEXT:    call void @use32(i32 [[NBITS]])
; CHECK-NEXT:    call void @use64(i64 [[T0]])
; CHECK-NEXT:    call void @use64(i64 [[T1]])
; CHECK-NEXT:    call void @use64(i64 [[T2]])
; CHECK-NEXT:    call void @use32(i32 [[T3]])
; CHECK-NEXT:    call void @use64(i64 [[T4]])
; CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[X]] to i32
; CHECK-NEXT:    [[T6:%.*]] = shl i32 [[TMP1]], [[T3]]
; CHECK-NEXT:    ret i32 [[T6]]
;
  %t0 = zext i32 %nbits to i64
  %t1 = shl i64 1, %t0
  %t2 = add i64 %t1, -1
  %t3 = sub i32 32, %nbits
  %t4 = and i64 %t2, %x

  call void @use32(i32 %nbits)
  call void @use64(i64 %t0)
  call void @use64(i64 %t1)
  call void @use64(i64 %t2)
  call void @use32(i32 %t3)
  call void @use64(i64 %t4)

  %t5 = trunc i64 %t4 to i32
  %t6 = shl i32 %t5, %t3
  ret i32 %t6
}

; Vectors

declare void @use8xi32(<8 x i32>)
declare void @use8xi64(<8 x i64>)

define <8 x i32> @t1_vec_splat(<8 x i64> %x, <8 x i32> %nbits) {
; CHECK-LABEL: @t1_vec_splat(
; CHECK-NEXT:    [[T0:%.*]] = zext <8 x i32> [[NBITS:%.*]] to <8 x i64>
; CHECK-NEXT:    [[T1:%.*]] = shl <8 x i64> <i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1>, [[T0]]
; CHECK-NEXT:    [[T2:%.*]] = add <8 x i64> [[T1]], <i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1>
; CHECK-NEXT:    [[T3:%.*]] = sub <8 x i32> <i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32>, [[NBITS]]
; CHECK-NEXT:    [[T4:%.*]] = and <8 x i64> [[T2]], [[X:%.*]]
; CHECK-NEXT:    call void @use8xi32(<8 x i32> [[NBITS]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T0]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T1]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T2]])
; CHECK-NEXT:    call void @use8xi32(<8 x i32> [[T3]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T4]])
; CHECK-NEXT:    [[TMP1:%.*]] = trunc <8 x i64> [[X]] to <8 x i32>
; CHECK-NEXT:    [[T6:%.*]] = shl <8 x i32> [[TMP1]], [[T3]]
; CHECK-NEXT:    ret <8 x i32> [[T6]]
;
  %t0 = zext <8 x i32> %nbits to <8 x i64>
  %t1 = shl <8 x i64> <i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1>, %t0
  %t2 = add <8 x i64> %t1, <i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1>
  %t3 = sub <8 x i32> <i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32>, %nbits
  %t4 = and <8 x i64> %t2, %x

  call void @use8xi32(<8 x i32> %nbits)
  call void @use8xi64(<8 x i64> %t0)
  call void @use8xi64(<8 x i64> %t1)
  call void @use8xi64(<8 x i64> %t2)
  call void @use8xi32(<8 x i32> %t3)
  call void @use8xi64(<8 x i64> %t4)

  %t5 = trunc <8 x i64> %t4 to <8 x i32>
  %t6 = shl <8 x i32> %t5, %t3
  ret <8 x i32> %t6
}

define <8 x i32> @t2_vec_splat_undef(<8 x i64> %x, <8 x i32> %nbits) {
; CHECK-LABEL: @t2_vec_splat_undef(
; CHECK-NEXT:    [[T0:%.*]] = zext <8 x i32> [[NBITS:%.*]] to <8 x i64>
; CHECK-NEXT:    [[T1:%.*]] = shl <8 x i64> <i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 undef, i64 1>, [[T0]]
; CHECK-NEXT:    [[T2:%.*]] = add <8 x i64> [[T1]], <i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 undef, i64 -1>
; CHECK-NEXT:    [[T3:%.*]] = sub <8 x i32> <i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 undef, i32 32>, [[NBITS]]
; CHECK-NEXT:    [[T4:%.*]] = and <8 x i64> [[T2]], [[X:%.*]]
; CHECK-NEXT:    call void @use8xi32(<8 x i32> [[NBITS]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T0]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T1]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T2]])
; CHECK-NEXT:    call void @use8xi32(<8 x i32> [[T3]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T4]])
; CHECK-NEXT:    [[TMP1:%.*]] = trunc <8 x i64> [[X]] to <8 x i32>
; CHECK-NEXT:    [[T6:%.*]] = shl <8 x i32> [[TMP1]], [[T3]]
; CHECK-NEXT:    ret <8 x i32> [[T6]]
;
  %t0 = zext <8 x i32> %nbits to <8 x i64>
  %t1 = shl <8 x i64> <i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 undef, i64 1>, %t0
  %t2 = add <8 x i64> %t1, <i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 undef, i64 -1>
  %t3 = sub <8 x i32> <i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 undef, i32 32>, %nbits
  %t4 = and <8 x i64> %t2, %x

  call void @use8xi32(<8 x i32> %nbits)
  call void @use8xi64(<8 x i64> %t0)
  call void @use8xi64(<8 x i64> %t1)
  call void @use8xi64(<8 x i64> %t2)
  call void @use8xi32(<8 x i32> %t3)
  call void @use8xi64(<8 x i64> %t4)

  %t5 = trunc <8 x i64> %t4 to <8 x i32>
  %t6 = shl <8 x i32> %t5, %t3
  ret <8 x i32> %t6
}

define <8 x i32> @t3_vec_nonsplat(<8 x i64> %x, <8 x i32> %nbits) {
; CHECK-LABEL: @t3_vec_nonsplat(
; CHECK-NEXT:    [[T0:%.*]] = zext <8 x i32> [[NBITS:%.*]] to <8 x i64>
; CHECK-NEXT:    [[T1:%.*]] = shl <8 x i64> <i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1>, [[T0]]
; CHECK-NEXT:    [[T2:%.*]] = add <8 x i64> [[T1]], <i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1>
; CHECK-NEXT:    [[T3:%.*]] = sub <8 x i32> <i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32>, [[NBITS]]
; CHECK-NEXT:    [[T4:%.*]] = and <8 x i64> [[T2]], [[X:%.*]]
; CHECK-NEXT:    call void @use8xi32(<8 x i32> [[NBITS]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T0]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T1]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T2]])
; CHECK-NEXT:    call void @use8xi32(<8 x i32> [[T3]])
; CHECK-NEXT:    call void @use8xi64(<8 x i64> [[T4]])
; CHECK-NEXT:    [[TMP1:%.*]] = trunc <8 x i64> [[X]] to <8 x i32>
; CHECK-NEXT:    [[T6:%.*]] = shl <8 x i32> [[TMP1]], [[T3]]
; CHECK-NEXT:    ret <8 x i32> [[T6]]
;
  %t0 = zext <8 x i32> %nbits to <8 x i64>
  %t1 = shl <8 x i64> <i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1>, %t0
  %t2 = add <8 x i64> %t1, <i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1>
  %t3 = sub <8 x i32> <i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32>, %nbits
  %t4 = and <8 x i64> %t2, %x

  call void @use8xi32(<8 x i32> %nbits)
  call void @use8xi64(<8 x i64> %t0)
  call void @use8xi64(<8 x i64> %t1)
  call void @use8xi64(<8 x i64> %t2)
  call void @use8xi32(<8 x i32> %t3)
  call void @use8xi64(<8 x i64> %t4)

  %t5 = trunc <8 x i64> %t4 to <8 x i32>
  %t6 = shl <8 x i32> %t5, %t3
  ret <8 x i32> %t6
}

; Extra uses

define i32 @n4_extrause(i64 %x, i32 %nbits) {
; CHECK-LABEL: @n4_extrause(
; CHECK-NEXT:    [[T0:%.*]] = zext i32 [[NBITS:%.*]] to i64
; CHECK-NEXT:    [[T1:%.*]] = shl i64 1, [[T0]]
; CHECK-NEXT:    [[T2:%.*]] = add i64 [[T1]], -1
; CHECK-NEXT:    [[T3:%.*]] = sub i32 32, [[NBITS]]
; CHECK-NEXT:    [[T4:%.*]] = and i64 [[T2]], [[X:%.*]]
; CHECK-NEXT:    call void @use32(i32 [[NBITS]])
; CHECK-NEXT:    call void @use64(i64 [[T0]])
; CHECK-NEXT:    call void @use64(i64 [[T1]])
; CHECK-NEXT:    call void @use64(i64 [[T2]])
; CHECK-NEXT:    call void @use32(i32 [[T3]])
; CHECK-NEXT:    call void @use64(i64 [[T4]])
; CHECK-NEXT:    [[T5:%.*]] = trunc i64 [[T4]] to i32
; CHECK-NEXT:    call void @use32(i32 [[T5]])
; CHECK-NEXT:    [[T6:%.*]] = shl i32 [[T5]], [[T3]]
; CHECK-NEXT:    ret i32 [[T6]]
;
  %t0 = zext i32 %nbits to i64
  %t1 = shl i64 1, %t0
  %t2 = add i64 %t1, -1
  %t3 = sub i32 32, %nbits
  %t4 = and i64 %t2, %x

  call void @use32(i32 %nbits)
  call void @use64(i64 %t0)
  call void @use64(i64 %t1)
  call void @use64(i64 %t2)
  call void @use32(i32 %t3)
  call void @use64(i64 %t4)

  %t5 = trunc i64 %t4 to i32
  call void @use32(i32 %t5)
  %t6 = shl i32 %t5, %t3
  ret i32 %t6
}