llvm/test/CodeGen/PowerPC/setcc-logic.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -ppc-gpr-icmps=all -verify-machineinstrs -mtriple=powerpc64le-unknown-unknown -ppc-convert-rr-to-ri=true \
; RUN:   -ppc-asm-full-reg-names -ppc-vsr-nums-as-vr | FileCheck %s

define zeroext i1 @all_bits_clear(i32 %P, i32 %Q)  {
; CHECK-LABEL: all_bits_clear:
; CHECK:       # %bb.0:
; CHECK-NEXT:    or r3, r3, r4
; CHECK-NEXT:    cntlzw r3, r3
; CHECK-NEXT:    srwi r3, r3, 5
; CHECK-NEXT:    blr
  %a = icmp eq i32 %P, 0
  %b = icmp eq i32 %Q, 0
  %c = and i1 %a, %b
  ret i1 %c
}

define zeroext i1 @all_sign_bits_clear(i32 %P, i32 %Q)  {
; CHECK-LABEL: all_sign_bits_clear:
; CHECK:       # %bb.0:
; CHECK-NEXT:    or r3, r3, r4
; CHECK-NEXT:    rlwinm r3, r3, 1, 31, 31
; CHECK-NEXT:    xori r3, r3, 1
; CHECK-NEXT:    blr
  %a = icmp sgt i32 %P, -1
  %b = icmp sgt i32 %Q, -1
  %c = and i1 %a, %b
  ret i1 %c
}

define zeroext i1 @all_bits_set(i32 %P, i32 %Q)  {
; CHECK-LABEL: all_bits_set:
; CHECK:       # %bb.0:
; CHECK-NEXT:    and r3, r3, r4
; CHECK-NEXT:    li r4, -1
; CHECK-NEXT:    xor r3, r3, r4
; CHECK-NEXT:    cntlzw r3, r3
; CHECK-NEXT:    srwi r3, r3, 5
; CHECK-NEXT:    blr
  %a = icmp eq i32 %P, -1
  %b = icmp eq i32 %Q, -1
  %c = and i1 %a, %b
  ret i1 %c
}

define zeroext i1 @all_sign_bits_set(i32 %P, i32 %Q)  {
; CHECK-LABEL: all_sign_bits_set:
; CHECK:       # %bb.0:
; CHECK-NEXT:    and r3, r3, r4
; CHECK-NEXT:    rlwinm r3, r3, 1, 31, 31
; CHECK-NEXT:    blr
  %a = icmp slt i32 %P, 0
  %b = icmp slt i32 %Q, 0
  %c = and i1 %a, %b
  ret i1 %c
}

define zeroext i1 @any_bits_set(i32 %P, i32 %Q)  {
; CHECK-LABEL: any_bits_set:
; CHECK:       # %bb.0:
; CHECK-NEXT:    or r3, r3, r4
; CHECK-NEXT:    cntlzw r3, r3
; CHECK-NEXT:    srwi r3, r3, 5
; CHECK-NEXT:    xori r3, r3, 1
; CHECK-NEXT:    blr
  %a = icmp ne i32 %P, 0
  %b = icmp ne i32 %Q, 0
  %c = or i1 %a, %b
  ret i1 %c
}

define zeroext i1 @any_sign_bits_set(i32 %P, i32 %Q)  {
; CHECK-LABEL: any_sign_bits_set:
; CHECK:       # %bb.0:
; CHECK-NEXT:    or r3, r3, r4
; CHECK-NEXT:    rlwinm r3, r3, 1, 31, 31
; CHECK-NEXT:    blr
  %a = icmp slt i32 %P, 0
  %b = icmp slt i32 %Q, 0
  %c = or i1 %a, %b
  ret i1 %c
}

define zeroext i1 @any_bits_clear(i32 %P, i32 %Q)  {
; CHECK-LABEL: any_bits_clear:
; CHECK:       # %bb.0:
; CHECK-NEXT:    and r3, r3, r4
; CHECK-NEXT:    li r4, -1
; CHECK-NEXT:    xor r3, r3, r4
; CHECK-NEXT:    cntlzw r3, r3
; CHECK-NEXT:    srwi r3, r3, 5
; CHECK-NEXT:    xori r3, r3, 1
; CHECK-NEXT:    blr
  %a = icmp ne i32 %P, -1
  %b = icmp ne i32 %Q, -1
  %c = or i1 %a, %b
  ret i1 %c
}

define zeroext i1 @any_sign_bits_clear(i32 %P, i32 %Q)  {
; CHECK-LABEL: any_sign_bits_clear:
; CHECK:       # %bb.0:
; CHECK-NEXT:    and r3, r3, r4
; CHECK-NEXT:    rlwinm r3, r3, 1, 31, 31
; CHECK-NEXT:    xori r3, r3, 1
; CHECK-NEXT:    blr
  %a = icmp sgt i32 %P, -1
  %b = icmp sgt i32 %Q, -1
  %c = or i1 %a, %b
  ret i1 %c
}

; PR3351 - (P == 0) & (Q == 0) -> (P|Q) == 0
define i32 @all_bits_clear_branch(ptr %P, ptr %Q)  {
; CHECK-LABEL: all_bits_clear_branch:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    or. r3, r3, r4
; CHECK-NEXT:    bne cr0, .LBB8_2
; CHECK-NEXT:  # %bb.1: # %bb1
; CHECK-NEXT:    li r3, 4
; CHECK-NEXT:    blr
; CHECK-NEXT:  .LBB8_2: # %return
; CHECK-NEXT:    li r3, 192
; CHECK-NEXT:    blr
entry:
  %a = icmp eq ptr %P, null
  %b = icmp eq ptr %Q, null
  %c = and i1 %a, %b
  br i1 %c, label %bb1, label %return

bb1:
  ret i32 4

return:
  ret i32 192
}

define i32 @all_sign_bits_clear_branch(i32 %P, i32 %Q)  {
; CHECK-LABEL: all_sign_bits_clear_branch:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    or r3, r3, r4
; CHECK-NEXT:    cmpwi r3, 0
; CHECK-NEXT:    blt cr0, .LBB9_2
; CHECK-NEXT:  # %bb.1: # %bb1
; CHECK-NEXT:    li r3, 4
; CHECK-NEXT:    blr
; CHECK-NEXT:  .LBB9_2: # %return
; CHECK-NEXT:    li r3, 192
; CHECK-NEXT:    blr
entry:
  %a = icmp sgt i32 %P, -1
  %b = icmp sgt i32 %Q, -1
  %c = and i1 %a, %b
  br i1 %c, label %bb1, label %return

bb1:
  ret i32 4

return:
  ret i32 192
}

define i32 @all_bits_set_branch(i32 %P, i32 %Q)  {
; CHECK-LABEL: all_bits_set_branch:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    and r3, r3, r4
; CHECK-NEXT:    cmpwi r3, -1
; CHECK-NEXT:    bne cr0, .LBB10_2
; CHECK-NEXT:  # %bb.1: # %bb1
; CHECK-NEXT:    li r3, 4
; CHECK-NEXT:    blr
; CHECK-NEXT:  .LBB10_2: # %return
; CHECK-NEXT:    li r3, 192
; CHECK-NEXT:    blr
entry:
  %a = icmp eq i32 %P, -1
  %b = icmp eq i32 %Q, -1
  %c = and i1 %a, %b
  br i1 %c, label %bb1, label %return

bb1:
  ret i32 4

return:
  ret i32 192
}

define i32 @all_sign_bits_set_branch(i32 %P, i32 %Q)  {
; CHECK-LABEL: all_sign_bits_set_branch:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    and r3, r3, r4
; CHECK-NEXT:    cmpwi r3, -1
; CHECK-NEXT:    bgt cr0, .LBB11_2
; CHECK-NEXT:  # %bb.1: # %bb1
; CHECK-NEXT:    li r3, 4
; CHECK-NEXT:    blr
; CHECK-NEXT:  .LBB11_2: # %return
; CHECK-NEXT:    li r3, 192
; CHECK-NEXT:    blr
entry:
  %a = icmp slt i32 %P, 0
  %b = icmp slt i32 %Q, 0
  %c = and i1 %a, %b
  br i1 %c, label %bb1, label %return

bb1:
  ret i32 4

return:
  ret i32 192
}

; PR3351 - (P != 0) | (Q != 0) -> (P|Q) != 0
define i32 @any_bits_set_branch(ptr %P, ptr %Q)  {
; CHECK-LABEL: any_bits_set_branch:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    or. r3, r3, r4
; CHECK-NEXT:    beq cr0, .LBB12_2
; CHECK-NEXT:  # %bb.1: # %bb1
; CHECK-NEXT:    li r3, 4
; CHECK-NEXT:    blr
; CHECK-NEXT:  .LBB12_2: # %return
; CHECK-NEXT:    li r3, 192
; CHECK-NEXT:    blr
entry:
  %a = icmp ne ptr %P, null
  %b = icmp ne ptr %Q, null
  %c = or i1 %a, %b
  br i1 %c, label %bb1, label %return

bb1:
  ret i32 4

return:
  ret i32 192
}

define i32 @any_sign_bits_set_branch(i32 %P, i32 %Q)  {
; CHECK-LABEL: any_sign_bits_set_branch:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    or r3, r3, r4
; CHECK-NEXT:    cmpwi r3, -1
; CHECK-NEXT:    bgt cr0, .LBB13_2
; CHECK-NEXT:  # %bb.1: # %bb1
; CHECK-NEXT:    li r3, 4
; CHECK-NEXT:    blr
; CHECK-NEXT:  .LBB13_2: # %return
; CHECK-NEXT:    li r3, 192
; CHECK-NEXT:    blr
entry:
  %a = icmp slt i32 %P, 0
  %b = icmp slt i32 %Q, 0
  %c = or i1 %a, %b
  br i1 %c, label %bb1, label %return

bb1:
  ret i32 4

return:
  ret i32 192
}

define i32 @any_bits_clear_branch(i32 %P, i32 %Q)  {
; CHECK-LABEL: any_bits_clear_branch:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    and r3, r3, r4
; CHECK-NEXT:    cmpwi r3, -1
; CHECK-NEXT:    beq cr0, .LBB14_2
; CHECK-NEXT:  # %bb.1: # %bb1
; CHECK-NEXT:    li r3, 4
; CHECK-NEXT:    blr
; CHECK-NEXT:  .LBB14_2: # %return
; CHECK-NEXT:    li r3, 192
; CHECK-NEXT:    blr
entry:
  %a = icmp ne i32 %P, -1
  %b = icmp ne i32 %Q, -1
  %c = or i1 %a, %b
  br i1 %c, label %bb1, label %return

bb1:
  ret i32 4

return:
  ret i32 192
}

define i32 @any_sign_bits_clear_branch(i32 %P, i32 %Q)  {
; CHECK-LABEL: any_sign_bits_clear_branch:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    and r3, r3, r4
; CHECK-NEXT:    cmpwi r3, 0
; CHECK-NEXT:    blt cr0, .LBB15_2
; CHECK-NEXT:  # %bb.1: # %bb1
; CHECK-NEXT:    li r3, 4
; CHECK-NEXT:    blr
; CHECK-NEXT:  .LBB15_2: # %return
; CHECK-NEXT:    li r3, 192
; CHECK-NEXT:    blr
entry:
  %a = icmp sgt i32 %P, -1
  %b = icmp sgt i32 %Q, -1
  %c = or i1 %a, %b
  br i1 %c, label %bb1, label %return

bb1:
  ret i32 4

return:
  ret i32 192
}

define <4 x i1> @all_bits_clear_vec(<4 x i32> %P, <4 x i32> %Q) {
; CHECK-LABEL: all_bits_clear_vec:
; CHECK:       # %bb.0:
; CHECK-NEXT:    xxlor v2, v2, v3
; CHECK-NEXT:    xxlxor v3, v3, v3
; CHECK-NEXT:    vcmpequw v2, v2, v3
; CHECK-NEXT:    blr
  %a = icmp eq <4 x i32> %P, zeroinitializer
  %b = icmp eq <4 x i32> %Q, zeroinitializer
  %c = and <4 x i1> %a, %b
  ret <4 x i1> %c
}

define <4 x i1> @all_sign_bits_clear_vec(<4 x i32> %P, <4 x i32> %Q) {
; CHECK-LABEL: all_sign_bits_clear_vec:
; CHECK:       # %bb.0:
; CHECK-NEXT:    xxlor v2, v2, v3
; CHECK-NEXT:    xxleqv v3, v3, v3
; CHECK-NEXT:    vcmpgtsw v2, v2, v3
; CHECK-NEXT:    blr
  %a = icmp sgt <4 x i32> %P, <i32 -1, i32 -1, i32 -1, i32 -1>
  %b = icmp sgt <4 x i32> %Q, <i32 -1, i32 -1, i32 -1, i32 -1>
  %c = and <4 x i1> %a, %b
  ret <4 x i1> %c
}

define <4 x i1> @all_bits_set_vec(<4 x i32> %P, <4 x i32> %Q) {
; CHECK-LABEL: all_bits_set_vec:
; CHECK:       # %bb.0:
; CHECK-NEXT:    xxland v2, v2, v3
; CHECK-NEXT:    xxleqv v3, v3, v3
; CHECK-NEXT:    vcmpequw v2, v2, v3
; CHECK-NEXT:    blr
  %a = icmp eq <4 x i32> %P, <i32 -1, i32 -1, i32 -1, i32 -1>
  %b = icmp eq <4 x i32> %Q, <i32 -1, i32 -1, i32 -1, i32 -1>
  %c = and <4 x i1> %a, %b
  ret <4 x i1> %c
}

define <4 x i1> @all_sign_bits_set_vec(<4 x i32> %P, <4 x i32> %Q) {
; CHECK-LABEL: all_sign_bits_set_vec:
; CHECK:       # %bb.0:
; CHECK-NEXT:    xxland v2, v2, v3
; CHECK-NEXT:    xxlxor v3, v3, v3
; CHECK-NEXT:    vcmpgtsw v2, v3, v2
; CHECK-NEXT:    blr
  %a = icmp slt <4 x i32> %P, zeroinitializer
  %b = icmp slt <4 x i32> %Q, zeroinitializer
  %c = and <4 x i1> %a, %b
  ret <4 x i1> %c
}

define <4 x i1> @any_bits_set_vec(<4 x i32> %P, <4 x i32> %Q) {
; CHECK-LABEL: any_bits_set_vec:
; CHECK:       # %bb.0:
; CHECK-NEXT:    xxlor v2, v2, v3
; CHECK-NEXT:    xxlxor v3, v3, v3
; CHECK-NEXT:    vcmpgtuw v2, v2, v3
; CHECK-NEXT:    blr
  %a = icmp ne <4 x i32> %P, zeroinitializer
  %b = icmp ne <4 x i32> %Q, zeroinitializer
  %c = or <4 x i1> %a, %b
  ret <4 x i1> %c
}

define <4 x i1> @any_sign_bits_set_vec(<4 x i32> %P, <4 x i32> %Q) {
; CHECK-LABEL: any_sign_bits_set_vec:
; CHECK:       # %bb.0:
; CHECK-NEXT:    xxlor v2, v2, v3
; CHECK-NEXT:    xxlxor v3, v3, v3
; CHECK-NEXT:    vcmpgtsw v2, v3, v2
; CHECK-NEXT:    blr
  %a = icmp slt <4 x i32> %P, zeroinitializer
  %b = icmp slt <4 x i32> %Q, zeroinitializer
  %c = or <4 x i1> %a, %b
  ret <4 x i1> %c
}

define <4 x i1> @any_bits_clear_vec(<4 x i32> %P, <4 x i32> %Q) {
; CHECK-LABEL: any_bits_clear_vec:
; CHECK:       # %bb.0:
; CHECK-NEXT:    xxland v2, v2, v3
; CHECK-NEXT:    xxleqv v3, v3, v3
; CHECK-NEXT:    vcmpequw v2, v2, v3
; CHECK-NEXT:    xxlnor v2, v2, v2
; CHECK-NEXT:    blr
  %a = icmp ne <4 x i32> %P, <i32 -1, i32 -1, i32 -1, i32 -1>
  %b = icmp ne <4 x i32> %Q, <i32 -1, i32 -1, i32 -1, i32 -1>
  %c = or <4 x i1> %a, %b
  ret <4 x i1> %c
}

define <4 x i1> @any_sign_bits_clear_vec(<4 x i32> %P, <4 x i32> %Q) {
; CHECK-LABEL: any_sign_bits_clear_vec:
; CHECK:       # %bb.0:
; CHECK-NEXT:    xxland v2, v2, v3
; CHECK-NEXT:    xxleqv v3, v3, v3
; CHECK-NEXT:    vcmpgtsw v2, v2, v3
; CHECK-NEXT:    blr
  %a = icmp sgt <4 x i32> %P, <i32 -1, i32 -1, i32 -1, i32 -1>
  %b = icmp sgt <4 x i32> %Q, <i32 -1, i32 -1, i32 -1, i32 -1>
  %c = or <4 x i1> %a, %b
  ret <4 x i1> %c
}

define zeroext i1 @ne_neg1_and_ne_zero(i64 %x) {
; CHECK-LABEL: ne_neg1_and_ne_zero:
; CHECK:       # %bb.0:
; CHECK-NEXT:    addi r3, r3, 1
; CHECK-NEXT:    li r4, 1
; CHECK-NEXT:    subfic r3, r3, 1
; CHECK-NEXT:    subfe r3, r4, r4
; CHECK-NEXT:    neg r3, r3
; CHECK-NEXT:    blr
  %cmp1 = icmp ne i64 %x, -1
  %cmp2 = icmp ne i64 %x, 0
  %and = and i1 %cmp1, %cmp2
  ret i1 %and
}

; PR32401 - https://bugs.llvm.org/show_bug.cgi?id=32401

define zeroext i1 @and_eq(i16 zeroext  %a, i16 zeroext %b, i16 zeroext %c, i16 zeroext %d) {
; CHECK-LABEL: and_eq:
; CHECK:       # %bb.0:
; CHECK-NEXT:    xor r3, r3, r4
; CHECK-NEXT:    xor r4, r5, r6
; CHECK-NEXT:    or r3, r3, r4
; CHECK-NEXT:    cntlzw r3, r3
; CHECK-NEXT:    srwi r3, r3, 5
; CHECK-NEXT:    blr
  %cmp1 = icmp eq i16 %a, %b
  %cmp2 = icmp eq i16 %c, %d
  %and = and i1 %cmp1, %cmp2
  ret i1 %and
}

define zeroext i1 @or_ne(i32 %a, i32 %b, i32 %c, i32 %d) {
; CHECK-LABEL: or_ne:
; CHECK:       # %bb.0:
; CHECK-NEXT:    xor r3, r3, r4
; CHECK-NEXT:    xor r4, r5, r6
; CHECK-NEXT:    or r3, r3, r4
; CHECK-NEXT:    cntlzw r3, r3
; CHECK-NEXT:    srwi r3, r3, 5
; CHECK-NEXT:    xori r3, r3, 1
; CHECK-NEXT:    blr
  %cmp1 = icmp ne i32 %a, %b
  %cmp2 = icmp ne i32 %c, %d
  %or = or i1 %cmp1, %cmp2
  ret i1 %or
}

; This should not be transformed because vector compares + bitwise logic are faster.

define <4 x i1> @and_eq_vec(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c, <4 x i32> %d) {
; CHECK-LABEL: and_eq_vec:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vcmpequw v2, v2, v3
; CHECK-NEXT:    vcmpequw v3, v4, v5
; CHECK-NEXT:    xxland v2, v2, v3
; CHECK-NEXT:    blr
  %cmp1 = icmp eq <4 x i32> %a, %b
  %cmp2 = icmp eq <4 x i32> %c, %d
  %and = and <4 x i1> %cmp1, %cmp2
  ret <4 x i1> %and
}

define i1 @or_icmps_const_1bit_diff(i64 %x) {
; CHECK-LABEL: or_icmps_const_1bit_diff:
; CHECK:       # %bb.0:
; CHECK-NEXT:    addi r3, r3, -13
; CHECK-NEXT:    rldicl r3, r3, 61, 1
; CHECK-NEXT:    rotldi r3, r3, 3
; CHECK-NEXT:    cntlzd r3, r3
; CHECK-NEXT:    rldicl r3, r3, 58, 63
; CHECK-NEXT:    blr
  %a = icmp eq i64 %x, 17
  %b = icmp eq i64 %x, 13
  %r = or i1 %a, %b
  ret i1 %r
}

define i1 @and_icmps_const_1bit_diff(i32 %x) {
; CHECK-LABEL: and_icmps_const_1bit_diff:
; CHECK:       # %bb.0:
; CHECK-NEXT:    addi r3, r3, -4625
; CHECK-NEXT:    rlwinm r3, r3, 0, 28, 26
; CHECK-NEXT:    cntlzw r3, r3
; CHECK-NEXT:    not r3, r3
; CHECK-NEXT:    rlwinm r3, r3, 27, 31, 31
; CHECK-NEXT:    blr
  %a = icmp ne i32 %x, 4625
  %b = icmp ne i32 %x, 4641
  %r = and i1 %a, %b
  ret i1 %r
}