TheRock/compiler/amd-llvm/llvm/test/CodeGen/RISCV/GlobalISel/rv32p.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc -mtriple=riscv32 -global-isel -mattr=+experimental-p,+zbb -verify-machineinstrs \
; RUN:   < %s | FileCheck %s

define i8 @cls_i8(i8 %x) {
; CHECK-LABEL: cls_i8:
; CHECK:       # %bb.0:
; CHECK-NEXT:    sext.b a0, a0
; CHECK-NEXT:    cls a0, a0
; CHECK-NEXT:    addi a0, a0, -24
; CHECK-NEXT:    ret
  %a = ashr i8 %x, 7
  %b = xor i8 %x, %a
  %c = call i8 @llvm.ctlz.i8(i8 %b, i1 false)
  %d = sub i8 %c, 1
  ret i8 %d
}

define i8 @cls_i8_2(i8 %x) {
; CHECK-LABEL: cls_i8_2:
; CHECK:       # %bb.0:
; CHECK-NEXT:    sext.b a0, a0
; CHECK-NEXT:    cls a0, a0
; CHECK-NEXT:    addi a0, a0, -24
; CHECK-NEXT:    ret
  %a = ashr i8 %x, 7
  %b = xor i8 %x, %a
  %c = shl i8 %b, 1
  %d = or i8 %c, 1
  %e = call i8 @llvm.ctlz.i8(i8 %d, i1 true)
  ret i8 %e
}

define i16 @cls_i16(i16 %x) {
; CHECK-LABEL: cls_i16:
; CHECK:       # %bb.0:
; CHECK-NEXT:    sext.h a0, a0
; CHECK-NEXT:    cls a0, a0
; CHECK-NEXT:    addi a0, a0, -16
; CHECK-NEXT:    ret
  %a = ashr i16 %x, 15
  %b = xor i16 %x, %a
  %c = call i16 @llvm.ctlz.i16(i16 %b, i1 false)
  %d = sub i16 %c, 1
  ret i16 %d
}

define i16 @cls_i16_2(i16 %x) {
; CHECK-LABEL: cls_i16_2:
; CHECK:       # %bb.0:
; CHECK-NEXT:    sext.h a0, a0
; CHECK-NEXT:    cls a0, a0
; CHECK-NEXT:    addi a0, a0, -16
; CHECK-NEXT:    ret
  %a = ashr i16 %x, 15
  %b = xor i16 %x, %a
  %c = shl i16 %b, 1
  %d = or i16 %c, 1
  %e = call i16 @llvm.ctlz.i16(i16 %d, i1 true)
  ret i16 %e
}

define i32 @cls_i32(i32 %x) {
; CHECK-LABEL: cls_i32:
; CHECK:       # %bb.0:
; CHECK-NEXT:    cls a0, a0
; CHECK-NEXT:    ret
  %a = ashr i32 %x, 31
  %b = xor i32 %x, %a
  %c = call i32 @llvm.ctlz.i32(i32 %b, i1 false)
  %d = sub i32 %c, 1
  ret i32 %d
}

define i32 @cls_i32_2(i32 %x) {
; CHECK-LABEL: cls_i32_2:
; CHECK:       # %bb.0:
; CHECK-NEXT:    cls a0, a0
; CHECK-NEXT:    ret
  %a = ashr i32 %x, 31
  %b = xor i32 %x, %a
  %c = shl i32 %b, 1
  %d = or i32 %c, 1
  %e = call i32 @llvm.ctlz.i32(i32 %d, i1 true)
  ret i32 %e
}

define i64 @cls_i64(i64 %x) {
; CHECK-LABEL: cls_i64:
; CHECK:       # %bb.0:
; CHECK-NEXT:    srai a2, a1, 31
; CHECK-NEXT:    xor a1, a1, a2
; CHECK-NEXT:    beqz a1, .LBB6_2
; CHECK-NEXT:  # %bb.1:
; CHECK-NEXT:    clz a0, a1
; CHECK-NEXT:    j .LBB6_3
; CHECK-NEXT:  .LBB6_2:
; CHECK-NEXT:    xor a0, a0, a2
; CHECK-NEXT:    clz a0, a0
; CHECK-NEXT:    addi a0, a0, 32
; CHECK-NEXT:  .LBB6_3:
; CHECK-NEXT:    addi a0, a0, -1
; CHECK-NEXT:    sltiu a1, a0, -1
; CHECK-NEXT:    li a2, -1
; CHECK-NEXT:    add a1, a2, a1
; CHECK-NEXT:    ret
  %a = ashr i64 %x, 63
  %b = xor i64 %x, %a
  %c = call i64 @llvm.ctlz.i64(i64 %b, i1 false)
  %d = sub i64 %c, 1
  ret i64 %d
}

define i64 @cls_i64_2(i64 %x) {
; CHECK-LABEL: cls_i64_2:
; CHECK:       # %bb.0:
; CHECK-NEXT:    srai a2, a1, 31
; CHECK-NEXT:    xor a0, a0, a2
; CHECK-NEXT:    xor a1, a1, a2
; CHECK-NEXT:    slli a1, a1, 1
; CHECK-NEXT:    srli a2, a0, 31
; CHECK-NEXT:    or a1, a1, a2
; CHECK-NEXT:    beqz a1, .LBB7_2
; CHECK-NEXT:  # %bb.1:
; CHECK-NEXT:    clz a0, a1
; CHECK-NEXT:    li a1, 0
; CHECK-NEXT:    ret
; CHECK-NEXT:  .LBB7_2:
; CHECK-NEXT:    slli a0, a0, 1
; CHECK-NEXT:    ori a0, a0, 1
; CHECK-NEXT:    clz a0, a0
; CHECK-NEXT:    addi a0, a0, 32
; CHECK-NEXT:    li a1, 0
; CHECK-NEXT:    ret
  %a = ashr i64 %x, 63
  %b = xor i64 %x, %a
  %c = shl i64 %b, 1
  %d = or i64 %c, 1
  %e = call i64 @llvm.ctlz.i64(i64 %d, i1 true)
  ret i64 %e
}

; The result is in the range [1-31], so we don't need an andi after the cls.
define i32 @cls_i32_knownbits(i32 %x) {
; CHECK-LABEL: cls_i32_knownbits:
; CHECK:       # %bb.0:
; CHECK-NEXT:    cls a0, a0
; CHECK-NEXT:    ret
  %a = ashr i32 %x, 31
  %b = xor i32 %x, %a
  %c = call i32 @llvm.ctlz.i32(i32 %b, i1 false)
  %d = sub i32 %c, 1
  %e = and i32 %d, 31
  ret i32 %e
}

; There are at least 16 redundant sign bits so we don't need an ori after the clsw.
define i32 @cls_i32_knownbits_2(i16 signext %x) {
; CHECK-LABEL: cls_i32_knownbits_2:
; CHECK:       # %bb.0:
; CHECK-NEXT:    cls a0, a0
; CHECK-NEXT:    ret
  %sext = sext i16 %x to i32
  %a = ashr i32 %sext, 31
  %b = xor i32 %sext, %a
  %c = call i32 @llvm.ctlz.i32(i32 %b, i1 false)
  %d = sub i32 %c, 1
  %e = or i32 %d, 16
  ret i32 %e
}

; There are at least 24 redundant sign bits so we don't need an ori after the clsw.
define i32 @cls_i32_knownbits_3(i8 signext %x) {
; CHECK-LABEL: cls_i32_knownbits_3:
; CHECK:       # %bb.0:
; CHECK-NEXT:    cls a0, a0
; CHECK-NEXT:    ret
  %sext = sext i8 %x to i32
  %a = ashr i32 %sext, 31
  %b = xor i32 %sext, %a
  %c = call i32 @llvm.ctlz.i32(i32 %b, i1 false)
  %d = sub i32 %c, 1
  %e = or i32 %d, 24
  ret i32 %e
}

; Negative test. We only know there is at least 1 redundant sign bit. We can't
; remove the ori.
define i32 @cls_i32_knownbits_4(i32 signext %x) {
; CHECK-LABEL: cls_i32_knownbits_4:
; CHECK:       # %bb.0:
; CHECK-NEXT:    slli a0, a0, 1
; CHECK-NEXT:    srai a0, a0, 1
; CHECK-NEXT:    cls a0, a0
; CHECK-NEXT:    ori a0, a0, 1
; CHECK-NEXT:    ret
  %shl = shl i32 %x, 1
  %ashr = ashr i32 %shl, 1
  %a = ashr i32 %ashr, 31
  %b = xor i32 %ashr, %a
  %c = call i32 @llvm.ctlz.i32(i32 %b, i1 false)
  %d = sub i32 %c, 1
  %e = or i32 %d, 1
  ret i32 %e
 }

; Negative test. Check that the number of sign bits is not
; overestimated. If it is, the ori disappears.
define i32 @cls_i32_knownbits_no_overestimate(i32 signext %x) {
; CHECK-LABEL: cls_i32_knownbits_no_overestimate:
; CHECK:       # %bb.0:
; CHECK-NEXT:    srai a1, a0, 15
; CHECK-NEXT:    srai a0, a0, 31
; CHECK-NEXT:    xor a0, a1, a0
; CHECK-NEXT:    clz a0, a0
; CHECK-NEXT:    addi a0, a0, -1
; CHECK-NEXT:    ori a0, a0, 16
; CHECK-NEXT:    ret
  %ashr = ashr i32 %x, 15
  %a = ashr i32 %ashr, 31
  %b = xor i32 %ashr, %a
  %c = call i32 @llvm.ctlz.i32(i32 %b, i1 false)
  %d = sub i32 %c, 1
  %e = or i32 %d, 16
  ret i32 %e
 }