llvm/test/CodeGen/Xtensa/rotl-rotr.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc -mtriple=xtensa -verify-machineinstrs < %s \
; RUN:   | FileCheck -check-prefix=XTENSA %s

define i32 @rotl_32(i32 %x, i32 %y) nounwind {
; XTENSA-LABEL: rotl_32:
; XTENSA:         ssl a3
; XTENSA-NEXT:    sll a8, a2
; XTENSA-NEXT:    movi a9, 32
; XTENSA-NEXT:    sub a9, a9, a3
; XTENSA-NEXT:    ssr a9
; XTENSA-NEXT:    srl a9, a2
; XTENSA-NEXT:    or a2, a8, a9
; XTENSA-NEXT:    ret
  %z = sub i32 32, %y
  %b = shl i32 %x, %y
  %c = lshr i32 %x, %z
  %d = or i32 %b, %c
  ret i32 %d
}

define i32 @rotr_32(i32 %x, i32 %y) nounwind {
; XTENSA-LABEL: rotr_32:
; XTENSA:         ssr a3
; XTENSA-NEXT:    srl a8, a2
; XTENSA-NEXT:    movi a9, 32
; XTENSA-NEXT:    sub a9, a9, a3
; XTENSA-NEXT:    ssl a9
; XTENSA-NEXT:    sll a9, a2
; XTENSA-NEXT:    or a2, a8, a9
; XTENSA-NEXT:    ret
  %z = sub i32 32, %y
  %b = lshr i32 %x, %y
  %c = shl i32 %x, %z
  %d = or i32 %b, %c
  ret i32 %d
}

define i64 @rotl_64(i64 %x, i64 %y) nounwind {
; XTENSA-LABEL: rotl_64:
; XTENSA:         movi a8, 64
; XTENSA-NEXT:    sub a8, a8, a4
; XTENSA-NEXT:    ssr a8
; XTENSA-NEXT:    src a11, a3, a2
; XTENSA-NEXT:    movi a9, 32
; XTENSA-NEXT:    sub a9, a9, a4
; XTENSA-NEXT:    ssr a9
; XTENSA-NEXT:    srl a7, a3
; XTENSA-NEXT:    movi a10, 0
; XTENSA-NEXT:    blt a9, a10, .LBB2_2
; XTENSA-NEXT:  # %bb.1:
; XTENSA-NEXT:    or a11, a7, a7
; XTENSA-NEXT:  .LBB2_2:
; XTENSA-NEXT:    ssl a4
; XTENSA-NEXT:    sll a7, a2
; XTENSA-NEXT:    addi a5, a4, -32
; XTENSA-NEXT:    blt a5, a10, .LBB2_4
; XTENSA-NEXT:  # %bb.3:
; XTENSA-NEXT:    or a7, a10, a10
; XTENSA-NEXT:  .LBB2_4:
; XTENSA-NEXT:    ssl a4
; XTENSA-NEXT:    src a6, a3, a2
; XTENSA-NEXT:    ssl a5
; XTENSA-NEXT:    sll a4, a2
; XTENSA-NEXT:    blt a5, a10, .LBB2_6
; XTENSA-NEXT:  # %bb.5:
; XTENSA-NEXT:    or a6, a4, a4
; XTENSA-NEXT:  .LBB2_6:
; XTENSA-NEXT:    or a2, a7, a11
; XTENSA-NEXT:    ssr a8
; XTENSA-NEXT:    srl a8, a3
; XTENSA-NEXT:    blt a9, a10, .LBB2_8
; XTENSA-NEXT:  # %bb.7:
; XTENSA-NEXT:    or a8, a10, a10
; XTENSA-NEXT:  .LBB2_8:
; XTENSA-NEXT:    or a3, a6, a8
; XTENSA-NEXT:    ret
  %z = sub i64 64, %y
  %b = shl i64 %x, %y
  %c = lshr i64 %x, %z
  %d = or i64 %b, %c
  ret i64 %d
}

define i64 @rotr_64(i64 %x, i64 %y) nounwind {
; XTENSA-LABEL: rotr_64:
; XTENSA:         ssr a4
; XTENSA-NEXT:    src a10, a3, a2
; XTENSA-NEXT:    addi a8, a4, -32
; XTENSA-NEXT:    ssr a8
; XTENSA-NEXT:    srl a11, a3
; XTENSA-NEXT:    movi a9, 0
; XTENSA-NEXT:    blt a8, a9, .LBB3_2
; XTENSA-NEXT:  # %bb.1:
; XTENSA-NEXT:    or a10, a11, a11
; XTENSA-NEXT:  .LBB3_2:
; XTENSA-NEXT:    movi a11, 32
; XTENSA-NEXT:    sub a7, a11, a4
; XTENSA-NEXT:    movi a11, 64
; XTENSA-NEXT:    sub a11, a11, a4
; XTENSA-NEXT:    ssl a11
; XTENSA-NEXT:    sll a6, a2
; XTENSA-NEXT:    blt a7, a9, .LBB3_4
; XTENSA-NEXT:  # %bb.3:
; XTENSA-NEXT:    or a6, a9, a9
; XTENSA-NEXT:  .LBB3_4:
; XTENSA-NEXT:    ssl a11
; XTENSA-NEXT:    src a11, a3, a2
; XTENSA-NEXT:    ssl a7
; XTENSA-NEXT:    sll a5, a2
; XTENSA-NEXT:    blt a7, a9, .LBB3_6
; XTENSA-NEXT:  # %bb.5:
; XTENSA-NEXT:    or a11, a5, a5
; XTENSA-NEXT:  .LBB3_6:
; XTENSA-NEXT:    or a2, a10, a6
; XTENSA-NEXT:    ssr a4
; XTENSA-NEXT:    srl a10, a3
; XTENSA-NEXT:    blt a8, a9, .LBB3_8
; XTENSA-NEXT:  # %bb.7:
; XTENSA-NEXT:    or a10, a9, a9
; XTENSA-NEXT:  .LBB3_8:
; XTENSA-NEXT:    or a3, a10, a11
; XTENSA-NEXT:    ret
  %z = sub i64 64, %y
  %b = lshr i64 %x, %y
  %c = shl i64 %x, %z
  %d = or i64 %b, %c
  ret i64 %d
}

define i32 @rotl_32_mask(i32 %x, i32 %y) nounwind {
; XTENSA-LABEL: rotl_32_mask:
; XTENSA:         ssl a3
; XTENSA-NEXT:    sll a8, a2
; XTENSA-NEXT:    neg a9, a3
; XTENSA-NEXT:    movi a10, 31
; XTENSA-NEXT:    and a9, a9, a10
; XTENSA-NEXT:    ssr a9
; XTENSA-NEXT:    srl a9, a2
; XTENSA-NEXT:    or a2, a8, a9
; XTENSA-NEXT:    ret
  %z = sub i32 0, %y
  %and = and i32 %z, 31
  %b = shl i32 %x, %y
  %c = lshr i32 %x, %and
  %d = or i32 %b, %c
  ret i32 %d
}

define i32 @rotl_32_mask_and_63_and_31(i32 %x, i32 %y) nounwind {
; XTENSA-LABEL: rotl_32_mask_and_63_and_31:
; XTENSA:         movi a8, 63
; XTENSA-NEXT:    and a8, a3, a8
; XTENSA-NEXT:    ssl a8
; XTENSA-NEXT:    sll a8, a2
; XTENSA-NEXT:    neg a9, a3
; XTENSA-NEXT:    movi a10, 31
; XTENSA-NEXT:    and a9, a9, a10
; XTENSA-NEXT:    ssr a9
; XTENSA-NEXT:    srl a9, a2
; XTENSA-NEXT:    or a2, a8, a9
; XTENSA-NEXT:    ret
  %a = and i32 %y, 63
  %b = shl i32 %x, %a
  %c = sub i32 0, %y
  %d = and i32 %c, 31
  %e = lshr i32 %x, %d
  %f = or i32 %b, %e
  ret i32 %f
}

define i32 @rotr_32_mask(i32 %x, i32 %y) nounwind {
; XTENSA-LABEL: rotr_32_mask:
; XTENSA:         ssr a3
; XTENSA-NEXT:    srl a8, a2
; XTENSA-NEXT:    neg a9, a3
; XTENSA-NEXT:    movi a10, 31
; XTENSA-NEXT:    and a9, a9, a10
; XTENSA-NEXT:    ssl a9
; XTENSA-NEXT:    sll a9, a2
; XTENSA-NEXT:    or a2, a8, a9
; XTENSA-NEXT:    ret
  %z = sub i32 0, %y
  %and = and i32 %z, 31
  %b = lshr i32 %x, %y
  %c = shl i32 %x, %and
  %d = or i32 %b, %c
  ret i32 %d
}

define i32 @rotr_32_mask_and_63_and_31(i32 %x, i32 %y) nounwind {
; XTENSA-LABEL: rotr_32_mask_and_63_and_31:
; XTENSA:         movi a8, 63
; XTENSA-NEXT:    and a8, a3, a8
; XTENSA-NEXT:    ssr a8
; XTENSA-NEXT:    srl a8, a2
; XTENSA-NEXT:    neg a9, a3
; XTENSA-NEXT:    movi a10, 31
; XTENSA-NEXT:    and a9, a9, a10
; XTENSA-NEXT:    ssl a9
; XTENSA-NEXT:    sll a9, a2
; XTENSA-NEXT:    or a2, a8, a9
; XTENSA-NEXT:    ret
  %a = and i32 %y, 63
  %b = lshr i32 %x, %a
  %c = sub i32 0, %y
  %d = and i32 %c, 31
  %e = shl i32 %x, %d
  %f = or i32 %b, %e
  ret i32 %f
}

define i64 @rotl_64_mask(i64 %x, i64 %y) nounwind {
; XTENSA-LABEL: rotl_64_mask:
; XTENSA:         ssl a4
; XTENSA-NEXT:    src a10, a3, a2
; XTENSA-NEXT:    addi a8, a4, -32
; XTENSA-NEXT:    ssl a8
; XTENSA-NEXT:    sll a11, a2
; XTENSA-NEXT:    movi a9, 0
; XTENSA-NEXT:    blt a8, a9, .LBB8_2
; XTENSA-NEXT:  # %bb.1:
; XTENSA-NEXT:    or a10, a11, a11
; XTENSA-NEXT:  .LBB8_2:
; XTENSA-NEXT:    neg a11, a4
; XTENSA-NEXT:    movi a7, 63
; XTENSA-NEXT:    and a7, a11, a7
; XTENSA-NEXT:    ssr a7
; XTENSA-NEXT:    srl a11, a3
; XTENSA-NEXT:    addi a6, a7, -32
; XTENSA-NEXT:    blt a6, a9, .LBB8_4
; XTENSA-NEXT:  # %bb.3:
; XTENSA-NEXT:    or a11, a9, a9
; XTENSA-NEXT:  .LBB8_4:
; XTENSA-NEXT:    ssr a7
; XTENSA-NEXT:    src a7, a3, a2
; XTENSA-NEXT:    ssr a6
; XTENSA-NEXT:    srl a5, a3
; XTENSA-NEXT:    blt a6, a9, .LBB8_6
; XTENSA-NEXT:  # %bb.5:
; XTENSA-NEXT:    or a7, a5, a5
; XTENSA-NEXT:  .LBB8_6:
; XTENSA-NEXT:    or a3, a10, a11
; XTENSA-NEXT:    ssl a4
; XTENSA-NEXT:    sll a10, a2
; XTENSA-NEXT:    blt a8, a9, .LBB8_8
; XTENSA-NEXT:  # %bb.7:
; XTENSA-NEXT:    or a10, a9, a9
; XTENSA-NEXT:  .LBB8_8:
; XTENSA-NEXT:    or a2, a10, a7
; XTENSA-NEXT:    ret
  %z = sub i64 0, %y
  %and = and i64 %z, 63
  %b = shl i64 %x, %y
  %c = lshr i64 %x, %and
  %d = or i64 %b, %c
  ret i64 %d
}

define i64 @rotl_64_mask_and_127_and_63(i64 %x, i64 %y) nounwind {
; XTENSA-LABEL: rotl_64_mask_and_127_and_63:
; XTENSA:         movi a8, 127
; XTENSA-NEXT:    and a8, a4, a8
; XTENSA-NEXT:    ssl a8
; XTENSA-NEXT:    src a11, a3, a2
; XTENSA-NEXT:    addi a9, a8, -32
; XTENSA-NEXT:    ssl a9
; XTENSA-NEXT:    sll a7, a2
; XTENSA-NEXT:    movi a10, 0
; XTENSA-NEXT:    blt a9, a10, .LBB9_2
; XTENSA-NEXT:  # %bb.1:
; XTENSA-NEXT:    or a11, a7, a7
; XTENSA-NEXT:  .LBB9_2:
; XTENSA-NEXT:    neg a7, a4
; XTENSA-NEXT:    movi a6, 63
; XTENSA-NEXT:    and a6, a7, a6
; XTENSA-NEXT:    ssr a6
; XTENSA-NEXT:    srl a7, a3
; XTENSA-NEXT:    addi a5, a6, -32
; XTENSA-NEXT:    blt a5, a10, .LBB9_4
; XTENSA-NEXT:  # %bb.3:
; XTENSA-NEXT:    or a7, a10, a10
; XTENSA-NEXT:  .LBB9_4:
; XTENSA-NEXT:    ssr a6
; XTENSA-NEXT:    src a6, a3, a2
; XTENSA-NEXT:    ssr a5
; XTENSA-NEXT:    srl a4, a3
; XTENSA-NEXT:    blt a5, a10, .LBB9_6
; XTENSA-NEXT:  # %bb.5:
; XTENSA-NEXT:    or a6, a4, a4
; XTENSA-NEXT:  .LBB9_6:
; XTENSA-NEXT:    or a3, a11, a7
; XTENSA-NEXT:    ssl a8
; XTENSA-NEXT:    sll a8, a2
; XTENSA-NEXT:    blt a9, a10, .LBB9_8
; XTENSA-NEXT:  # %bb.7:
; XTENSA-NEXT:    or a8, a10, a10
; XTENSA-NEXT:  .LBB9_8:
; XTENSA-NEXT:    or a2, a8, a6
; XTENSA-NEXT:    ret
  %a = and i64 %y, 127
  %b = shl i64 %x, %a
  %c = sub i64 0, %y
  %d = and i64 %c, 63
  %e = lshr i64 %x, %d
  %f = or i64 %b, %e
  ret i64 %f
}

define i64 @rotr_64_mask(i64 %x, i64 %y) nounwind {
; XTENSA-LABEL: rotr_64_mask:
; XTENSA:         ssr a4
; XTENSA-NEXT:    src a10, a3, a2
; XTENSA-NEXT:    addi a8, a4, -32
; XTENSA-NEXT:    ssr a8
; XTENSA-NEXT:    srl a11, a3
; XTENSA-NEXT:    movi a9, 0
; XTENSA-NEXT:    blt a8, a9, .LBB10_2
; XTENSA-NEXT:  # %bb.1:
; XTENSA-NEXT:    or a10, a11, a11
; XTENSA-NEXT:  .LBB10_2:
; XTENSA-NEXT:    neg a11, a4
; XTENSA-NEXT:    movi a7, 63
; XTENSA-NEXT:    and a7, a11, a7
; XTENSA-NEXT:    ssl a7
; XTENSA-NEXT:    sll a11, a2
; XTENSA-NEXT:    addi a6, a7, -32
; XTENSA-NEXT:    blt a6, a9, .LBB10_4
; XTENSA-NEXT:  # %bb.3:
; XTENSA-NEXT:    or a11, a9, a9
; XTENSA-NEXT:  .LBB10_4:
; XTENSA-NEXT:    ssl a7
; XTENSA-NEXT:    src a7, a3, a2
; XTENSA-NEXT:    ssl a6
; XTENSA-NEXT:    sll a5, a2
; XTENSA-NEXT:    blt a6, a9, .LBB10_6
; XTENSA-NEXT:  # %bb.5:
; XTENSA-NEXT:    or a7, a5, a5
; XTENSA-NEXT:  .LBB10_6:
; XTENSA-NEXT:    or a2, a10, a11
; XTENSA-NEXT:    ssr a4
; XTENSA-NEXT:    srl a10, a3
; XTENSA-NEXT:    blt a8, a9, .LBB10_8
; XTENSA-NEXT:  # %bb.7:
; XTENSA-NEXT:    or a10, a9, a9
; XTENSA-NEXT:  .LBB10_8:
; XTENSA-NEXT:    or a3, a10, a7
; XTENSA-NEXT:    ret
  %z = sub i64 0, %y
  %and = and i64 %z, 63
  %b = lshr i64 %x, %y
  %c = shl i64 %x, %and
  %d = or i64 %b, %c
  ret i64 %d
}

define i64 @rotr_64_mask_and_127_and_63(i64 %x, i64 %y) nounwind {
; XTENSA-LABEL: rotr_64_mask_and_127_and_63:
; XTENSA:         movi a8, 127
; XTENSA-NEXT:    and a8, a4, a8
; XTENSA-NEXT:    ssr a8
; XTENSA-NEXT:    src a11, a3, a2
; XTENSA-NEXT:    addi a9, a8, -32
; XTENSA-NEXT:    ssr a9
; XTENSA-NEXT:    srl a7, a3
; XTENSA-NEXT:    movi a10, 0
; XTENSA-NEXT:    blt a9, a10, .LBB11_2
; XTENSA-NEXT:  # %bb.1:
; XTENSA-NEXT:    or a11, a7, a7
; XTENSA-NEXT:  .LBB11_2:
; XTENSA-NEXT:    neg a7, a4
; XTENSA-NEXT:    movi a6, 63
; XTENSA-NEXT:    and a6, a7, a6
; XTENSA-NEXT:    ssl a6
; XTENSA-NEXT:    sll a7, a2
; XTENSA-NEXT:    addi a5, a6, -32
; XTENSA-NEXT:    blt a5, a10, .LBB11_4
; XTENSA-NEXT:  # %bb.3:
; XTENSA-NEXT:    or a7, a10, a10
; XTENSA-NEXT:  .LBB11_4:
; XTENSA-NEXT:    ssl a6
; XTENSA-NEXT:    src a6, a3, a2
; XTENSA-NEXT:    ssl a5
; XTENSA-NEXT:    sll a4, a2
; XTENSA-NEXT:    blt a5, a10, .LBB11_6
; XTENSA-NEXT:  # %bb.5:
; XTENSA-NEXT:    or a6, a4, a4
; XTENSA-NEXT:  .LBB11_6:
; XTENSA-NEXT:    or a2, a11, a7
; XTENSA-NEXT:    ssr a8
; XTENSA-NEXT:    srl a8, a3
; XTENSA-NEXT:    blt a9, a10, .LBB11_8
; XTENSA-NEXT:  # %bb.7:
; XTENSA-NEXT:    or a8, a10, a10
; XTENSA-NEXT:  .LBB11_8:
; XTENSA-NEXT:    or a3, a8, a6
; XTENSA-NEXT:    ret
  %a = and i64 %y, 127
  %b = lshr i64 %x, %a
  %c = sub i64 0, %y
  %d = and i64 %c, 63
  %e = shl i64 %x, %d
  %f = or i64 %b, %e
  ret i64 %f
}

define signext i32 @rotl_32_mask_shared(i32 signext %a, i32 signext %b, i32 signext %amt) nounwind {
; XTENSA-LABEL: rotl_32_mask_shared:
; XTENSA:         movi a8, 31
; XTENSA-NEXT:    and a9, a4, a8
; XTENSA-NEXT:    ssl a9
; XTENSA-NEXT:    sll a10, a2
; XTENSA-NEXT:    neg a11, a4
; XTENSA-NEXT:    and a8, a11, a8
; XTENSA-NEXT:    ssr a8
; XTENSA-NEXT:    srl a8, a2
; XTENSA-NEXT:    or a8, a10, a8
; XTENSA-NEXT:    ssl a9
; XTENSA-NEXT:    sll a9, a3
; XTENSA-NEXT:    add a2, a8, a9
; XTENSA-NEXT:    ret
  %maskedamt = and i32 %amt, 31
  %1 = tail call i32 @llvm.fshl.i32(i32 %a, i32 %a, i32 %maskedamt)
  %2 = shl i32 %b, %maskedamt
  %3 = add i32 %1, %2
  ret i32 %3
}
declare i32 @llvm.fshl.i32(i32, i32, i32)

define signext i32 @rotr_32_mask_shared(i32 signext %a, i32 signext %b, i32 signext %amt) nounwind {
; XTENSA-LABEL: rotr_32_mask_shared:
; XTENSA:         movi a8, 31
; XTENSA-NEXT:    and a9, a4, a8
; XTENSA-NEXT:    ssr a9
; XTENSA-NEXT:    srl a10, a2
; XTENSA-NEXT:    neg a11, a4
; XTENSA-NEXT:    and a8, a11, a8
; XTENSA-NEXT:    ssl a8
; XTENSA-NEXT:    sll a8, a2
; XTENSA-NEXT:    or a8, a10, a8
; XTENSA-NEXT:    ssl a9
; XTENSA-NEXT:    sll a9, a3
; XTENSA-NEXT:    add a2, a8, a9
; XTENSA-NEXT:    ret
  %maskedamt = and i32 %amt, 31
  %1 = tail call i32 @llvm.fshr.i32(i32 %a, i32 %a, i32 %maskedamt)
  %2 = shl i32 %b, %maskedamt
  %3 = add i32 %1, %2
  ret i32 %3
}
declare i32 @llvm.fshr.i32(i32, i32, i32)

define signext i32 @rotl_32_mask_multiple(i32 signext %a, i32 signext %b, i32 signext %amt) nounwind {
; XTENSA-LABEL: rotl_32_mask_multiple:
; XTENSA:         movi a8, 31
; XTENSA-NEXT:    and a9, a4, a8
; XTENSA-NEXT:    ssl a9
; XTENSA-NEXT:    sll a10, a3
; XTENSA-NEXT:    neg a11, a4
; XTENSA-NEXT:    and a8, a11, a8
; XTENSA-NEXT:    ssr a8
; XTENSA-NEXT:    srl a11, a3
; XTENSA-NEXT:    or a10, a10, a11
; XTENSA-NEXT:    ssl a9
; XTENSA-NEXT:    sll a9, a2
; XTENSA-NEXT:    ssr a8
; XTENSA-NEXT:    srl a8, a2
; XTENSA-NEXT:    or a8, a9, a8
; XTENSA-NEXT:    add a2, a8, a10
; XTENSA-NEXT:    ret
  %maskedamt = and i32 %amt, 31
  %1 = tail call i32 @llvm.fshl.i32(i32 %a, i32 %a, i32 %maskedamt)
  %2 = tail call i32 @llvm.fshl.i32(i32 %b, i32 %b, i32 %maskedamt)
  %3 = add i32 %1, %2
  ret i32 %3
}

define signext i32 @rotr_32_mask_multiple(i32 signext %a, i32 signext %b, i32 signext %amt) nounwind {
; XTENSA-LABEL: rotr_32_mask_multiple:
; XTENSA:         movi a8, 31
; XTENSA-NEXT:    and a9, a4, a8
; XTENSA-NEXT:    ssr a9
; XTENSA-NEXT:    srl a10, a3
; XTENSA-NEXT:    neg a11, a4
; XTENSA-NEXT:    and a8, a11, a8
; XTENSA-NEXT:    ssl a8
; XTENSA-NEXT:    sll a11, a3
; XTENSA-NEXT:    or a10, a10, a11
; XTENSA-NEXT:    ssr a9
; XTENSA-NEXT:    srl a9, a2
; XTENSA-NEXT:    ssl a8
; XTENSA-NEXT:    sll a8, a2
; XTENSA-NEXT:    or a8, a9, a8
; XTENSA-NEXT:    add a2, a8, a10
; XTENSA-NEXT:    ret
  %maskedamt = and i32 %amt, 31
  %1 = tail call i32 @llvm.fshr.i32(i32 %a, i32 %a, i32 %maskedamt)
  %2 = tail call i32 @llvm.fshr.i32(i32 %b, i32 %b, i32 %maskedamt)
  %3 = add i32 %1, %2
  ret i32 %3
}