; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -passes=instcombine -S | FileCheck %s

declare void @use(i8)
declare void @use_i1(i1)

define i32 @select_0_or_1_from_bool(i1 %x) {
; CHECK-LABEL: @select_0_or_1_from_bool(
; CHECK-NEXT:    [[NOT_X:%.*]] = xor i1 [[X:%.*]], true
; CHECK-NEXT:    [[ADD:%.*]] = zext i1 [[NOT_X]] to i32
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %ext = sext i1 %x to i32
  %add = add i32 %ext, 1
  ret i32 %add
}

define <2 x i32> @select_0_or_1_from_bool_vec(<2 x i1> %x) {
; CHECK-LABEL: @select_0_or_1_from_bool_vec(
; CHECK-NEXT:    [[NOT_X:%.*]] = xor <2 x i1> [[X:%.*]], <i1 true, i1 true>
; CHECK-NEXT:    [[ADD:%.*]] = zext <2 x i1> [[NOT_X]] to <2 x i32>
; CHECK-NEXT:    ret <2 x i32> [[ADD]]
;
  %ext = sext <2 x i1> %x to <2 x i32>
  %add = add <2 x i32> %ext, <i32 1, i32 1>
  ret <2 x i32> %add
}

define i32 @select_C_minus_1_or_C_from_bool(i1 %x) {
; CHECK-LABEL: @select_C_minus_1_or_C_from_bool(
; CHECK-NEXT:    [[ADD:%.*]] = select i1 [[X:%.*]], i32 41, i32 42
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %ext = sext i1 %x to i32
  %add = add i32 %ext, 42
  ret i32 %add
}

define <2 x i32> @select_C_minus_1_or_C_from_bool_vec(<2 x i1> %x) {
; CHECK-LABEL: @select_C_minus_1_or_C_from_bool_vec(
; CHECK-NEXT:    [[ADD:%.*]] = select <2 x i1> [[X:%.*]], <2 x i32> <i32 41, i32 42>, <2 x i32> <i32 42, i32 43>
; CHECK-NEXT:    ret <2 x i32> [[ADD]]
;
  %ext = sext <2 x i1> %x to <2 x i32>
  %add = add <2 x i32> %ext, <i32 42, i32 43>
  ret <2 x i32> %add
}

; This is an 'andn' of the low bit.

define i32 @flip_and_mask(i32 %x) {
; CHECK-LABEL: @flip_and_mask(
; CHECK-NEXT:    [[TMP1:%.*]] = and i32 [[X:%.*]], 1
; CHECK-NEXT:    [[INC:%.*]] = xor i32 [[TMP1]], 1
; CHECK-NEXT:    ret i32 [[INC]]
;
  %shl = shl i32 %x, 31
  %shr = ashr i32 %shl, 31
  %inc = add i32 %shr, 1
  ret i32 %inc
}

define <2 x i8> @flip_and_mask_splat(<2 x i8> %x) {
; CHECK-LABEL: @flip_and_mask_splat(
; CHECK-NEXT:    [[TMP1:%.*]] = and <2 x i8> [[X:%.*]], <i8 1, i8 1>
; CHECK-NEXT:    [[INC:%.*]] = xor <2 x i8> [[TMP1]], <i8 1, i8 1>
; CHECK-NEXT:    ret <2 x i8> [[INC]]
;
  %shl = shl <2 x i8> %x, <i8 7, i8 7>
  %shr = ashr <2 x i8> %shl, <i8 7, i8 7>
  %inc = add <2 x i8> %shr, <i8 1, i8 1>
  ret <2 x i8> %inc
}

define i32 @test1(i32 %A) {
; CHECK-LABEL: @test1(
; CHECK-NEXT:    ret i32 [[A:%.*]]
;
  %B = add i32 %A, 0
  ret i32 %B
}

define i32 @test2(i32 %A) {
; CHECK-LABEL: @test2(
; CHECK-NEXT:    ret i32 [[A:%.*]]
;
  %B = add i32 %A, 5
  %C = add i32 %B, -5
  ret i32 %C
}

define i32 @test3(i32 %A) {
; CHECK-LABEL: @test3(
; CHECK-NEXT:    ret i32 [[A:%.*]]
;
  %B = add i32 %A, 5
  %C = sub i32 %B, 5
  ret i32 %C
}

; D = B + -A = B - A
define i32 @test4(i32 %A, i32 %B) {
; CHECK-LABEL: @test4(
; CHECK-NEXT:    [[D:%.*]] = sub i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    ret i32 [[D]]
;
  %C = sub i32 0, %A
  %D = add i32 %B, %C
  ret i32 %D
}

; D = -A + B = B - A
define i32 @test5(i32 %A, i32 %B) {
; CHECK-LABEL: @test5(
; CHECK-NEXT:    [[D:%.*]] = sub i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    ret i32 [[D]]
;
  %C = sub i32 0, %A
  %D = add i32 %C, %B
  ret i32 %D
}

define i32 @test5_both_nsw(i32 %A, i32 %B) {
; CHECK-LABEL: @test5_both_nsw(
; CHECK-NEXT:    [[D:%.*]] = sub nsw i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    ret i32 [[D]]
;
  %C = sub nsw i32 0, %A
  %D = add nsw i32 %C, %B
  ret i32 %D
}

define i32 @test5_neg_nsw(i32 %A, i32 %B) {
; CHECK-LABEL: @test5_neg_nsw(
; CHECK-NEXT:    [[D:%.*]] = sub i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    ret i32 [[D]]
;
  %C = sub nsw i32 0, %A
  %D = add i32 %C, %B
  ret i32 %D
}

define i32 @test5_add_nsw(i32 %A, i32 %B) {
; CHECK-LABEL: @test5_add_nsw(
; CHECK-NEXT:    [[D:%.*]] = sub i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    ret i32 [[D]]
;
  %C = sub i32 0, %A
  %D = add nsw i32 %C, %B
  ret i32 %D
}

define <2 x i8> @neg_op0_vec_undef_elt(<2 x i8> %a, <2 x i8> %b) {
; CHECK-LABEL: @neg_op0_vec_undef_elt(
; CHECK-NEXT:    [[R:%.*]] = sub <2 x i8> [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    ret <2 x i8> [[R]]
;
  %nega = sub <2 x i8> <i8 0, i8 undef>, %a
  %r = add <2 x i8> %nega, %b
  ret <2 x i8> %r
}

define <2 x i8> @neg_neg_vec_undef_elt(<2 x i8> %a, <2 x i8> %b) {
; CHECK-LABEL: @neg_neg_vec_undef_elt(
; CHECK-NEXT:    [[TMP1:%.*]] = add <2 x i8> [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[R:%.*]] = sub <2 x i8> zeroinitializer, [[TMP1]]
; CHECK-NEXT:    ret <2 x i8> [[R]]
;
  %nega = sub <2 x i8> <i8 undef, i8 0>, %a
  %negb = sub <2 x i8> <i8 undef, i8 0>, %b
  %r = add <2 x i8> %nega, %negb
  ret <2 x i8> %r
}

; C = 7*A+A == 8*A == A << 3
define i32 @test6(i32 %A) {
; CHECK-LABEL: @test6(
; CHECK-NEXT:    [[C:%.*]] = shl i32 [[A:%.*]], 3
; CHECK-NEXT:    ret i32 [[C]]
;
  %B = mul i32 7, %A
  %C = add i32 %B, %A
  ret i32 %C
}

; C = A+7*A == 8*A == A << 3
define i32 @test7(i32 %A) {
; CHECK-LABEL: @test7(
; CHECK-NEXT:    [[C:%.*]] = shl i32 [[A:%.*]], 3
; CHECK-NEXT:    ret i32 [[C]]
;
  %B = mul i32 7, %A
  %C = add i32 %A, %B
  ret i32 %C
}

; (A & C1)+(B & C2) -> (A & C1)|(B & C2) iff C1&C2 == 0
define i32 @test8(i32 %A, i32 %B) {
; CHECK-LABEL: @test8(
; CHECK-NEXT:    [[A1:%.*]] = and i32 [[A:%.*]], 7
; CHECK-NEXT:    [[B1:%.*]] = and i32 [[B:%.*]], 128
; CHECK-NEXT:    [[C:%.*]] = or disjoint i32 [[A1]], [[B1]]
; CHECK-NEXT:    ret i32 [[C]]
;
  %A1 = and i32 %A, 7
  %B1 = and i32 %B, 128
  %C = add i32 %A1, %B1
  ret i32 %C
}

define i32 @test9(i32 %A) {
; CHECK-LABEL: @test9(
; CHECK-NEXT:    [[C:%.*]] = shl i32 [[A:%.*]], 5
; CHECK-NEXT:    ret i32 [[C]]
;
  %B = shl i32 %A, 4
  %C = add i32 %B, %B
  ret i32 %C
}

; a != -b
define i1 @test10(i8 %a, i8 %b) {
; CHECK-LABEL: @test10(
; CHECK-NEXT:    [[ADD:%.*]] = sub i8 0, [[B:%.*]]
; CHECK-NEXT:    [[C:%.*]] = icmp ne i8 [[ADD]], [[A:%.*]]
; CHECK-NEXT:    ret i1 [[C]]
;
  %add = add i8 %a, %b
  %c = icmp ne i8 %add, 0
  ret i1 %c
}

define <2 x i1> @test10vec(<2 x i8> %a, <2 x i8> %b) {
; CHECK-LABEL: @test10vec(
; CHECK-NEXT:    [[C:%.*]] = sub <2 x i8> zeroinitializer, [[B:%.*]]
; CHECK-NEXT:    [[D:%.*]] = icmp ne <2 x i8> [[C]], [[A:%.*]]
; CHECK-NEXT:    ret <2 x i1> [[D]]
;
  %c = add <2 x i8> %a, %b
  %d = icmp ne <2 x i8> %c, zeroinitializer
  ret <2 x i1> %d
}

define i1 @test11(i8 %A) {
; CHECK-LABEL: @test11(
; CHECK-NEXT:    [[C:%.*]] = icmp ne i8 [[A:%.*]], 1
; CHECK-NEXT:    ret i1 [[C]]
;
  %B = add i8 %A, -1
  %c = icmp ne i8 %B, 0
  ret i1 %c
}

define <2 x i1> @test11vec(<2 x i8> %a) {
; CHECK-LABEL: @test11vec(
; CHECK-NEXT:    [[C:%.*]] = icmp ne <2 x i8> [[A:%.*]], <i8 1, i8 1>
; CHECK-NEXT:    ret <2 x i1> [[C]]
;
  %b = add <2 x i8> %a, <i8 -1, i8 -1>
  %c = icmp ne <2 x i8> %b, zeroinitializer
  ret <2 x i1> %c
}

define i8 @reassoc_shl1(i8 %x, i8 %y) {
; CHECK-LABEL: @reassoc_shl1(
; CHECK-NEXT:    [[REASS_ADD:%.*]] = shl i8 [[X:%.*]], 1
; CHECK-NEXT:    [[R:%.*]] = add i8 [[REASS_ADD]], [[Y:%.*]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %a = add i8 %y, %x
  %r = add i8 %a, %x
  ret i8 %r
}

define <2 x i8> @reassoc_shl1_commute1(<2 x i8> %x, <2 x i8> %y) {
; CHECK-LABEL: @reassoc_shl1_commute1(
; CHECK-NEXT:    [[REASS_ADD:%.*]] = shl <2 x i8> [[X:%.*]], <i8 1, i8 1>
; CHECK-NEXT:    [[R:%.*]] = add <2 x i8> [[REASS_ADD]], [[Y:%.*]]
; CHECK-NEXT:    ret <2 x i8> [[R]]
;
  %a = add <2 x i8> %x, %y
  %r = add <2 x i8> %a, %x
  ret <2 x i8> %r
}

define i8 @reassoc_shl1_commute2(i8 %px, i8 %py) {
; CHECK-LABEL: @reassoc_shl1_commute2(
; CHECK-NEXT:    [[X:%.*]] = sdiv i8 42, [[PX:%.*]]
; CHECK-NEXT:    [[Y:%.*]] = sdiv i8 43, [[PY:%.*]]
; CHECK-NEXT:    [[REASS_ADD:%.*]] = shl i8 [[X]], 1
; CHECK-NEXT:    [[R:%.*]] = add i8 [[Y]], [[REASS_ADD]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %x = sdiv i8 42, %px ; thwart complexity-based canonicalization
  %y = sdiv i8 43, %py ; thwart complexity-based canonicalization
  %a = add i8 %y, %x
  %r = add i8 %x, %a
  ret i8 %r
}

define i8 @reassoc_shl1_commute3(i8 %px, i8 %py) {
; CHECK-LABEL: @reassoc_shl1_commute3(
; CHECK-NEXT:    [[X:%.*]] = sdiv i8 42, [[PX:%.*]]
; CHECK-NEXT:    [[Y:%.*]] = sdiv i8 43, [[PY:%.*]]
; CHECK-NEXT:    [[REASS_ADD:%.*]] = shl i8 [[X]], 1
; CHECK-NEXT:    [[R:%.*]] = add i8 [[Y]], [[REASS_ADD]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %x = sdiv i8 42, %px ; thwart complexity-based canonicalization
  %y = sdiv i8 43, %py ; thwart complexity-based canonicalization
  %a = add i8 %x, %y
  %r = add i8 %x, %a
  ret i8 %r
}

define i8 @reassoc_shl1_extra_use(i8 %x, i8 %y) {
; CHECK-LABEL: @reassoc_shl1_extra_use(
; CHECK-NEXT:    [[A:%.*]] = add i8 [[Y:%.*]], [[X:%.*]]
; CHECK-NEXT:    call void @use(i8 [[A]])
; CHECK-NEXT:    [[R:%.*]] = add i8 [[A]], [[X]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %a = add i8 %y, %x
  call void @use(i8 %a)
  %r = add i8 %a, %x
  ret i8 %r
}

;; TODO: shl A, 1?
define i32 @test13(i32 %A, i32 %B, i32 %C) {
; CHECK-LABEL: @test13(
; CHECK-NEXT:    [[D_OK:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[E_OK:%.*]] = add i32 [[D_OK]], [[C:%.*]]
; CHECK-NEXT:    [[F:%.*]] = add i32 [[E_OK]], [[A]]
; CHECK-NEXT:    ret i32 [[F]]
;
  %D_OK = add i32 %A, %B
  %E_OK = add i32 %D_OK, %C
  %F = add i32 %E_OK, %A
  ret i32 %F
}

define i32 @test14(i32 %offset, i32 %difference) {
; CHECK-LABEL: @test14(
; CHECK-NEXT:    [[TMP_2:%.*]] = and i32 [[DIFFERENCE:%.*]], 3
; CHECK-NEXT:    [[TMP_3_OK:%.*]] = add i32 [[TMP_2]], [[OFFSET:%.*]]
; CHECK-NEXT:    [[TMP_5_MASK:%.*]] = and i32 [[DIFFERENCE]], -4
; CHECK-NEXT:    [[TMP_8:%.*]] = add i32 [[TMP_3_OK]], [[TMP_5_MASK]]
; CHECK-NEXT:    ret i32 [[TMP_8]]
;
  %tmp.2 = and i32 %difference, 3
  %tmp.3_OK = add i32 %tmp.2, %offset
  %tmp.5.mask = and i32 %difference, -4
  ; == add %offset, %difference
  %tmp.8 = add i32 %tmp.3_OK, %tmp.5.mask
  ret i32 %tmp.8
}

; Only one bit set
define i8 @test15(i8 %A) {
; CHECK-LABEL: @test15(
; CHECK-NEXT:    [[C:%.*]] = and i8 [[A:%.*]], 16
; CHECK-NEXT:    ret i8 [[C]]
;
  %B = add i8 %A, -64
  %C = and i8 %B, 16
  ret i8 %C
}

define i32 @test17(i32 %A) {
; CHECK-LABEL: @test17(
; CHECK-NEXT:    [[C:%.*]] = sub i32 0, [[A:%.*]]
; CHECK-NEXT:    ret i32 [[C]]
;
  %B = xor i32 %A, -1
  %C = add i32 %B, 1
  ret i32 %C
}

define i8 @test18(i8 %A) {
; CHECK-LABEL: @test18(
; CHECK-NEXT:    [[C:%.*]] = sub i8 16, [[A:%.*]]
; CHECK-NEXT:    ret i8 [[C]]
;
  %B = xor i8 %A, -1
  %C = add i8 %B, 17
  ret i8 %C
}

; ~X + -127 and (-128) - X with nsw are equally poisonous
define i8 @test18_nsw(i8 %A) {
; CHECK-LABEL: @test18_nsw(
; CHECK-NEXT:    [[C:%.*]] = sub nsw i8 -128, [[A:%.*]]
; CHECK-NEXT:    ret i8 [[C]]
;
  %B = xor i8 %A, -1
  %C = add nsw i8 %B, -127
  ret i8 %C
}

; nuw couldn't propagate as nsw is.
define i8 @test18_nuw(i8 %A) {
; CHECK-LABEL: @test18_nuw(
; CHECK-NEXT:    [[C:%.*]] = sub i8 -128, [[A:%.*]]
; CHECK-NEXT:    ret i8 [[C]]
;
  %B = xor i8 %A, -1
  %C = add nuw i8 %B, -127
  ret i8 %C
}

; 127 - X with nsw will be more poisonous than ~X + -128 with nsw. (see X = -1)
define i8 @test18_nsw_overflow(i8 %A) {
; CHECK-LABEL: @test18_nsw_overflow(
; CHECK-NEXT:    [[C:%.*]] = sub i8 127, [[A:%.*]]
; CHECK-NEXT:    ret i8 [[C]]
;
  %B = xor i8 %A, -1
  %C = add nsw i8 %B, -128
  ret i8 %C
}

define <2 x i64> @test18vec(<2 x i64> %A) {
; CHECK-LABEL: @test18vec(
; CHECK-NEXT:    [[ADD:%.*]] = sub <2 x i64> <i64 1, i64 2>, [[A:%.*]]
; CHECK-NEXT:    ret <2 x i64> [[ADD]]
;
  %xor = xor <2 x i64> %A, <i64 -1, i64 -1>
  %add = add <2 x i64> %xor, <i64 2, i64 3>
  ret <2 x i64> %add
}

define <2 x i8> @test18vec_nsw(<2 x i8> %A) {
; CHECK-LABEL: @test18vec_nsw(
; CHECK-NEXT:    [[C:%.*]] = sub nsw <2 x i8> <i8 -124, i8 -125>, [[A:%.*]]
; CHECK-NEXT:    ret <2 x i8> [[C]]
;
  %B = xor <2 x i8> %A, <i8 -1, i8 -1>
  %C = add nsw <2 x i8> %B, <i8 -123, i8 -124>
  ret <2 x i8> %C
}

define <2 x i8> @test18vec_nsw_false(<2 x i8> %A) {
; CHECK-LABEL: @test18vec_nsw_false(
; CHECK-NEXT:    [[C:%.*]] = sub nsw <2 x i8> <i8 -125, i8 -126>, [[A:%.*]]
; CHECK-NEXT:    ret <2 x i8> [[C]]
;
  %B = xor <2 x i8> %A, <i8 -1, i8 -1>
  %C = add nsw <2 x i8> %B, <i8 -124, i8 -125>
  ret <2 x i8> %C
}


define <2 x i8> @test18vec_nuw(<2 x i8> %A) {
; CHECK-LABEL: @test18vec_nuw(
; CHECK-NEXT:    [[C:%.*]] = sub <2 x i8> <i8 -128, i8 -127>, [[A:%.*]]
; CHECK-NEXT:    ret <2 x i8> [[C]]
;
  %B = xor <2 x i8> %A, <i8 -1, i8 -1>
  %C = add nuw <2 x i8> %B, <i8 -127, i8 -126>
  ret <2 x i8> %C
}

define <2 x i8> @test18vec_nsw_overflow(<2 x i8> %A) {
; CHECK-LABEL: @test18vec_nsw_overflow(
; CHECK-NEXT:    [[C:%.*]] = sub <2 x i8> <i8 -128, i8 127>, [[A:%.*]]
; CHECK-NEXT:    ret <2 x i8> [[C]]
;
  %B = xor <2 x i8> %A, <i8 -1, i8 -1>
  %C = add nsw <2 x i8> %B, <i8 -127, i8 -128>
  ret <2 x i8> %C
}

define i32 @test19(i1 %C) {
; CHECK-LABEL: @test19(
; CHECK-NEXT:    [[V:%.*]] = select i1 [[C:%.*]], i32 1123, i32 133
; CHECK-NEXT:    ret i32 [[V]]
;
  %A = select i1 %C, i32 1000, i32 10
  %V = add i32 %A, 123
  ret i32 %V
}

define <2 x i32> @test19vec(i1 %C) {
; CHECK-LABEL: @test19vec(
; CHECK-NEXT:    [[V:%.*]] = select i1 [[C:%.*]], <2 x i32> <i32 1123, i32 1123>, <2 x i32> <i32 133, i32 133>
; CHECK-NEXT:    ret <2 x i32> [[V]]
;
  %A = select i1 %C, <2 x i32> <i32 1000, i32 1000>, <2 x i32> <i32 10, i32 10>
  %V = add <2 x i32> %A, <i32 123, i32 123>
  ret <2 x i32> %V
}

; This is an InstSimplify fold, but test it here to make sure that
; InstCombine does not prevent the fold.
; With NSW, add of sign bit -> or of sign bit.

define i32 @test20(i32 %x) {
; CHECK-LABEL: @test20(
; CHECK-NEXT:    ret i32 [[X:%.*]]
;
  %y = xor i32 %x, -2147483648
  %z = add nsw i32 %y, -2147483648
  ret i32 %z
}

define i32 @xor_sign_bit(i32 %x) {
; CHECK-LABEL: @xor_sign_bit(
; CHECK-NEXT:    [[ADD:%.*]] = add i32 [[X:%.*]], -2147483606
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %xor = xor i32 %x, 2147483648
  %add = add i32 %xor, 42
  ret i32 %add
}

define <2 x i32> @xor_sign_bit_vec_splat(<2 x i32> %x) {
; CHECK-LABEL: @xor_sign_bit_vec_splat(
; CHECK-NEXT:    [[ADD:%.*]] = add <2 x i32> [[X:%.*]], <i32 -2147483606, i32 -2147483606>
; CHECK-NEXT:    ret <2 x i32> [[ADD]]
;
  %xor = xor <2 x i32> %x, <i32 2147483648, i32 2147483648>
  %add = add <2 x i32> %xor, <i32 42, i32 42>
  ret <2 x i32> %add
}

; No-wrap info allows converting the add to 'or'.

define i8 @add_nsw_signbit(i8 %x) {
; CHECK-LABEL: @add_nsw_signbit(
; CHECK-NEXT:    [[Y:%.*]] = or i8 [[X:%.*]], -128
; CHECK-NEXT:    ret i8 [[Y]]
;
  %y = add nsw i8 %x, -128
  ret i8 %y
}

; No-wrap info allows converting the add to 'or'.

define i8 @add_nuw_signbit(i8 %x) {
; CHECK-LABEL: @add_nuw_signbit(
; CHECK-NEXT:    [[Y:%.*]] = or i8 [[X:%.*]], -128
; CHECK-NEXT:    ret i8 [[Y]]
;
  %y = add nuw i8 %x, 128
  ret i8 %y
}

define i32 @add_nsw_sext_add(i8 %x) {
; CHECK-LABEL: @add_nsw_sext_add(
; CHECK-NEXT:    [[TMP1:%.*]] = sext i8 [[X:%.*]] to i32
; CHECK-NEXT:    [[R:%.*]] = add nsw i32 [[TMP1]], 398
; CHECK-NEXT:    ret i32 [[R]]
;
  %add = add nsw i8 %x, 42
  %ext = sext i8 %add to i32
  %r = add i32 %ext, 356
  ret i32 %r
}

; Negative test - extra use of the sext means increase of instructions.

define i32 @add_nsw_sext_add_extra_use_1(i8 %x, ptr %p) {
; CHECK-LABEL: @add_nsw_sext_add_extra_use_1(
; CHECK-NEXT:    [[ADD:%.*]] = add nsw i8 [[X:%.*]], 42
; CHECK-NEXT:    [[EXT:%.*]] = sext i8 [[ADD]] to i32
; CHECK-NEXT:    store i32 [[EXT]], ptr [[P:%.*]], align 4
; CHECK-NEXT:    [[R:%.*]] = add nsw i32 [[EXT]], 356
; CHECK-NEXT:    ret i32 [[R]]
;
  %add = add nsw i8 %x, 42
  %ext = sext i8 %add to i32
  store i32 %ext, ptr %p
  %r = add i32 %ext, 356
  ret i32 %r
}

define <2 x i32> @add_nsw_sext_add_vec_extra_use_2(<2 x i8> %x, ptr %p) {
; CHECK-LABEL: @add_nsw_sext_add_vec_extra_use_2(
; CHECK-NEXT:    [[ADD:%.*]] = add nsw <2 x i8> [[X:%.*]], <i8 42, i8 -5>
; CHECK-NEXT:    store <2 x i8> [[ADD]], ptr [[P:%.*]], align 2
; CHECK-NEXT:    [[TMP1:%.*]] = sext <2 x i8> [[X]] to <2 x i32>
; CHECK-NEXT:    [[R:%.*]] = add nsw <2 x i32> [[TMP1]], <i32 398, i32 7>
; CHECK-NEXT:    ret <2 x i32> [[R]]
;
  %add = add nsw <2 x i8> %x, <i8 42, i8 -5>
  store <2 x i8> %add, ptr %p
  %ext = sext <2 x i8> %add to <2 x i32>
  %r = add <2 x i32> %ext, <i32 356, i32 12>
  ret <2 x i32> %r
}

define <2 x i32> @add_nuw_zext_add_vec(<2 x i16> %x) {
; CHECK-LABEL: @add_nuw_zext_add_vec(
; CHECK-NEXT:    [[TMP1:%.*]] = zext <2 x i16> [[X:%.*]] to <2 x i32>
; CHECK-NEXT:    [[R:%.*]] = add nsw <2 x i32> [[TMP1]], <i32 65850, i32 -7>
; CHECK-NEXT:    ret <2 x i32> [[R]]
;
  %add = add nuw <2 x i16> %x, <i16 -42, i16 5>
  %ext = zext <2 x i16> %add to <2 x i32>
  %r = add <2 x i32> %ext, <i32 356, i32 -12>
  ret <2 x i32> %r
}

; Negative test - extra use of the zext means increase of instructions.

define i64 @add_nuw_zext_add_extra_use_1(i8 %x, ptr %p) {
; CHECK-LABEL: @add_nuw_zext_add_extra_use_1(
; CHECK-NEXT:    [[ADD:%.*]] = add nuw i8 [[X:%.*]], 42
; CHECK-NEXT:    [[EXT:%.*]] = zext i8 [[ADD]] to i64
; CHECK-NEXT:    store i64 [[EXT]], ptr [[P:%.*]], align 4
; CHECK-NEXT:    [[R:%.*]] = add nuw nsw i64 [[EXT]], 356
; CHECK-NEXT:    ret i64 [[R]]
;
  %add = add nuw i8 %x, 42
  %ext = zext i8 %add to i64
  store i64 %ext, ptr %p
  %r = add i64 %ext, 356
  ret i64 %r
}

define i64 @add_nuw_zext_add_extra_use_2(i8 %x, ptr %p) {
; CHECK-LABEL: @add_nuw_zext_add_extra_use_2(
; CHECK-NEXT:    [[ADD:%.*]] = add nuw i8 [[X:%.*]], 42
; CHECK-NEXT:    store i8 [[ADD]], ptr [[P:%.*]], align 1
; CHECK-NEXT:    [[TMP1:%.*]] = zext i8 [[X]] to i64
; CHECK-NEXT:    [[R:%.*]] = add nuw nsw i64 [[TMP1]], -314
; CHECK-NEXT:    ret i64 [[R]]
;
  %add = add nuw i8 %x, 42
  store i8 %add, ptr %p
  %ext = zext i8 %add to i64
  %r = add i64 %ext, -356
  ret i64 %r
}

define i1 @test21(i32 %x) {
; CHECK-LABEL: @test21(
; CHECK-NEXT:    [[Y:%.*]] = icmp eq i32 [[X:%.*]], 119
; CHECK-NEXT:    ret i1 [[Y]]
;
  %t = add i32 %x, 4
  %y = icmp eq i32 %t, 123
  ret i1 %y
}

define <2 x i1> @test21vec(<2 x i32> %x) {
; CHECK-LABEL: @test21vec(
; CHECK-NEXT:    [[Y:%.*]] = icmp eq <2 x i32> [[X:%.*]], <i32 119, i32 119>
; CHECK-NEXT:    ret <2 x i1> [[Y]]
;
  %t = add <2 x i32> %x, <i32 4, i32 4>
  %y = icmp eq <2 x i32> %t, <i32 123, i32 123>
  ret <2 x i1> %y
}

define i32 @test22(i32 %V) {
; CHECK-LABEL: @test22(
; CHECK-NEXT:    switch i32 [[V:%.*]], label [[DEFAULT:%.*]] [
; CHECK-NEXT:      i32 10, label [[LAB1:%.*]]
; CHECK-NEXT:      i32 20, label [[LAB2:%.*]]
; CHECK-NEXT:    ]
; CHECK:       Default:
; CHECK-NEXT:    ret i32 123
; CHECK:       Lab1:
; CHECK-NEXT:    ret i32 12312
; CHECK:       Lab2:
; CHECK-NEXT:    ret i32 1231231
;
  %V2 = add i32 %V, 10
  switch i32 %V2, label %Default [
  i32 20, label %Lab1
  i32 30, label %Lab2
  ]

Default:                ; preds = %0
  ret i32 123

Lab1:           ; preds = %0
  ret i32 12312

Lab2:           ; preds = %0
  ret i32 1231231
}

define i32 @test23(i1 %C, i32 %a) {
; CHECK-LABEL: @test23(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br i1 [[C:%.*]], label [[ENDIF:%.*]], label [[ELSE:%.*]]
; CHECK:       else:
; CHECK-NEXT:    br label [[ENDIF]]
; CHECK:       endif:
; CHECK-NEXT:    [[B_0:%.*]] = phi i32 [ 1, [[ENTRY:%.*]] ], [ 2, [[ELSE]] ]
; CHECK-NEXT:    ret i32 [[B_0]]
;
entry:
  br i1 %C, label %endif, label %else

else:           ; preds = %entry
  br label %endif

endif:          ; preds = %else, %entry
  %b.0 = phi i32 [ 0, %entry ], [ 1, %else ]
  %tmp.4 = add i32 %b.0, 1
  ret i32 %tmp.4
}

define i32 @test24(i32 %A) {
; CHECK-LABEL: @test24(
; CHECK-NEXT:    [[B:%.*]] = shl i32 [[A:%.*]], 1
; CHECK-NEXT:    ret i32 [[B]]
;
  %B = add i32 %A, 1
  %C = shl i32 %B, 1
  %D = sub i32 %C, 2
  ret i32 %D
}

define i64 @test25(i64 %Y) {
; CHECK-LABEL: @test25(
; CHECK-NEXT:    [[TMP_8:%.*]] = shl i64 [[Y:%.*]], 3
; CHECK-NEXT:    ret i64 [[TMP_8]]
;
  %tmp.4 = shl i64 %Y, 2
  %tmp.12 = shl i64 %Y, 2
  %tmp.8 = add i64 %tmp.4, %tmp.12
  ret i64 %tmp.8
}

define i32 @test26(i32 %A, i32 %B) {
; CHECK-LABEL: @test26(
; CHECK-NEXT:    ret i32 [[A:%.*]]
;
  %C = add i32 %A, %B
  %D = sub i32 %C, %B
  ret i32 %D
}

; Fold add through select.
define i32 @test27(i1 %C, i32 %X, i32 %Y) {
; CHECK-LABEL: @test27(
; CHECK-NEXT:    [[C_UPGRD_1_V:%.*]] = select i1 [[C:%.*]], i32 [[X:%.*]], i32 123
; CHECK-NEXT:    ret i32 [[C_UPGRD_1_V]]
;
  %A = add i32 %X, %Y
  %B = add i32 %Y, 123
  %C.upgrd.1 = select i1 %C, i32 %A, i32 %B
  %D = sub i32 %C.upgrd.1, %Y
  ret i32 %D
}

define i32 @test28(i32 %X) {
; CHECK-LABEL: @test28(
; CHECK-NEXT:    [[Z:%.*]] = sub i32 -1192, [[X:%.*]]
; CHECK-NEXT:    ret i32 [[Z]]
;
  %Y = add i32 %X, 1234
  %Z = sub i32 42, %Y
  ret i32 %Z
}

define i32 @test29(i32 %x, i32 %y) {
; CHECK-LABEL: @test29(
; CHECK-NEXT:    [[TMP_2:%.*]] = sub i32 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    [[TMP_7:%.*]] = and i32 [[X]], 63
; CHECK-NEXT:    [[TMP_9:%.*]] = and i32 [[TMP_2]], -64
; CHECK-NEXT:    [[TMP_10:%.*]] = or disjoint i32 [[TMP_7]], [[TMP_9]]
; CHECK-NEXT:    ret i32 [[TMP_10]]
;
  %tmp.2 = sub i32 %x, %y
  %tmp.2.mask = and i32 %tmp.2, 63
  %tmp.6 = add i32 %tmp.2.mask, %y
  %tmp.7 = and i32 %tmp.6, 63
  %tmp.9 = and i32 %tmp.2, -64
  %tmp.10 = or i32 %tmp.7, %tmp.9
  ret i32 %tmp.10
}

; Add of sign bit -> xor of sign bit.
define i64 @test30(i64 %x) {
; CHECK-LABEL: @test30(
; CHECK-NEXT:    ret i64 [[X:%.*]]
;
  %tmp.2 = xor i64 %x, -9223372036854775808
  %tmp.4 = add i64 %tmp.2, -9223372036854775808
  ret i64 %tmp.4
}

define i32 @test31(i32 %A) {
; CHECK-LABEL: @test31(
; CHECK-NEXT:    [[TMP1:%.*]] = mul i32 [[A:%.*]], 5
; CHECK-NEXT:    ret i32 [[TMP1]]
;
  %B = add i32 %A, 4
  %C = mul i32 %B, 5
  %D = sub i32 %C, 20
  ret i32 %D
}

define i32 @test32(i32 %A) {
; CHECK-LABEL: @test32(
; CHECK-NEXT:    [[B:%.*]] = shl i32 [[A:%.*]], 2
; CHECK-NEXT:    ret i32 [[B]]
;
  %B = add i32 %A, 4
  %C = shl i32 %B, 2
  %D = sub i32 %C, 16
  ret i32 %D
}

define i8 @test33(i8 %A) {
; CHECK-LABEL: @test33(
; CHECK-NEXT:    [[C:%.*]] = or i8 [[A:%.*]], 1
; CHECK-NEXT:    ret i8 [[C]]
;
  %B = and i8 %A, -2
  %C = add i8 %B, 1
  ret i8 %C
}

define i8 @test34(i8 %A) {
; CHECK-LABEL: @test34(
; CHECK-NEXT:    [[C:%.*]] = and i8 [[A:%.*]], 12
; CHECK-NEXT:    ret i8 [[C]]
;
  %B = add i8 %A, 64
  %C = and i8 %B, 12
  ret i8 %C
}

; If all bits affected by the add are included
; in the mask, do the mask op before the add.

define i8 @masked_add(i8 %x) {
; CHECK-LABEL: @masked_add(
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X:%.*]], -16
; CHECK-NEXT:    [[R:%.*]] = add i8 [[AND]], 96
; CHECK-NEXT:    ret i8 [[R]]
;
  %and = and i8 %x, 240 ; 0xf0
  %r = add i8 %and, 96  ; 0x60
  ret i8 %r
}

define <2 x i8> @masked_add_splat(<2 x i8> %x) {
; CHECK-LABEL: @masked_add_splat(
; CHECK-NEXT:    [[AND:%.*]] = and <2 x i8> [[X:%.*]], <i8 -64, i8 -64>
; CHECK-NEXT:    [[R:%.*]] = add <2 x i8> [[AND]], <i8 64, i8 64>
; CHECK-NEXT:    ret <2 x i8> [[R]]
;
  %and = and <2 x i8> %x, <i8 192, i8 192> ; 0xc0
  %r = add <2 x i8> %and, <i8 64, i8 64>  ; 0x40
  ret <2 x i8> %r
}

define i8 @not_masked_add(i8 %x) {
; CHECK-LABEL: @not_masked_add(
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X:%.*]], 112
; CHECK-NEXT:    [[R:%.*]] = add nuw i8 [[AND]], 96
; CHECK-NEXT:    ret i8 [[R]]
;
  %and = and i8 %x, 112 ; 0x70
  %r = add i8 %and, 96  ; 0x60
  ret i8 %r
}

define i8 @masked_add_multi_use(i8 %x) {
; CHECK-LABEL: @masked_add_multi_use(
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X:%.*]], -16
; CHECK-NEXT:    [[R:%.*]] = add i8 [[AND]], 96
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    ret i8 [[R]]
;
  %and = and i8 %x, -16 ; 0xf0
  %r = add i8 %and, 96  ; 0x60
  call void @use(i8 %and) ; extra use
  ret i8 %r
}

define i32 @test35(i32 %a) {
; CHECK-LABEL: @test35(
; CHECK-NEXT:    ret i32 -1
;
  %tmpnot = xor i32 %a, -1
  %tmp2 = add i32 %tmpnot, %a
  ret i32 %tmp2
}

define i32 @test36(i32 %a) {
; CHECK-LABEL: @test36(
; CHECK-NEXT:    ret i32 0
;
  %x = and i32 %a, -2
  %y = and i32 %a, -126
  %z = add i32 %x, %y
  %q = and i32 %z, 1  ; always zero
  ret i32 %q
}

define i1 @test37(i32 %a, i32 %b) {
; CHECK-LABEL: @test37(
; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[B:%.*]], 0
; CHECK-NEXT:    ret i1 [[CMP]]
;
  %add = add i32 %a, %b
  %cmp = icmp eq i32 %add, %a
  ret i1 %cmp
}

define i1 @test38(i32 %a, i32 %b) {
; CHECK-LABEL: @test38(
; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[A:%.*]], 0
; CHECK-NEXT:    ret i1 [[CMP]]
;
  %add = add i32 %a, %b
  %cmp = icmp eq i32 %add, %b
  ret i1 %cmp
}

define i1 @test39(i32 %a, i32 %b) {
; CHECK-LABEL: @test39(
; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[B:%.*]], 0
; CHECK-NEXT:    ret i1 [[CMP]]
;
  %add = add i32 %b, %a
  %cmp = icmp eq i32 %add, %a
  ret i1 %cmp
}

define i1 @test40(i32 %a, i32 %b) {
; CHECK-LABEL: @test40(
; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[A:%.*]], 0
; CHECK-NEXT:    ret i1 [[CMP]]
;
  %add = add i32 %b, %a
  %cmp = icmp eq i32 %add, %b
  ret i1 %cmp
}

; (add (zext (add nuw X, C2)), C) --> (zext (add nuw X, C2 + C))

define i64 @test41(i32 %a) {
; CHECK-LABEL: @test41(
; CHECK-NEXT:    [[TMP1:%.*]] = add nuw i32 [[A:%.*]], 15
; CHECK-NEXT:    [[SUB:%.*]] = zext i32 [[TMP1]] to i64
; CHECK-NEXT:    ret i64 [[SUB]]
;
  %add = add nuw i32 %a, 16
  %zext = zext i32 %add to i64
  %sub = add i64 %zext, -1
  ret i64 %sub
}

; (add (zext (add nuw X, C2)), C) --> (zext (add nuw X, C2 + C))

define <2 x i64> @test41vec(<2 x i32> %a) {
; CHECK-LABEL: @test41vec(
; CHECK-NEXT:    [[TMP1:%.*]] = add nuw <2 x i32> [[A:%.*]], <i32 15, i32 15>
; CHECK-NEXT:    [[SUB:%.*]] = zext <2 x i32> [[TMP1]] to <2 x i64>
; CHECK-NEXT:    ret <2 x i64> [[SUB]]
;
  %add = add nuw <2 x i32> %a, <i32 16, i32 16>
  %zext = zext <2 x i32> %add to <2 x i64>
  %sub = add <2 x i64> %zext, <i64 -1, i64 -1>
  ret <2 x i64> %sub
}

define <2 x i64> @test41vec_and_multiuse(<2 x i32> %a) {
; CHECK-LABEL: @test41vec_and_multiuse(
; CHECK-NEXT:    [[ADD:%.*]] = add nuw <2 x i32> [[A:%.*]], <i32 16, i32 16>
; CHECK-NEXT:    [[ZEXT:%.*]] = zext <2 x i32> [[ADD]] to <2 x i64>
; CHECK-NEXT:    [[REASS_ADD:%.*]] = shl nuw nsw <2 x i64> [[ZEXT]], <i64 1, i64 1>
; CHECK-NEXT:    [[EXTRAUSE:%.*]] = add nsw <2 x i64> [[REASS_ADD]], <i64 -1, i64 -1>
; CHECK-NEXT:    ret <2 x i64> [[EXTRAUSE]]
;
  %add = add nuw <2 x i32> %a, <i32 16, i32 16>
  %zext = zext <2 x i32> %add to <2 x i64>
  %sub = add <2 x i64> %zext, <i64 -1, i64 -1>
  %extrause = add <2 x i64> %zext, %sub
  ret <2 x i64> %extrause
}

define i32 @test42(i1 %C) {
; CHECK-LABEL: @test42(
; CHECK-NEXT:    [[V:%.*]] = select i1 [[C:%.*]], i32 1123, i32 133
; CHECK-NEXT:    ret i32 [[V]]
;
  %A = select i1 %C, i32 1000, i32 10
  %V = add i32 123, %A
  ret i32 %V
}

define <2 x i32> @test42vec(i1 %C) {
; CHECK-LABEL: @test42vec(
; CHECK-NEXT:    [[V:%.*]] = select i1 [[C:%.*]], <2 x i32> <i32 1123, i32 1123>, <2 x i32> <i32 133, i32 133>
; CHECK-NEXT:    ret <2 x i32> [[V]]
;
  %A = select i1 %C, <2 x i32> <i32 1000, i32 1000>, <2 x i32> <i32 10, i32 10>
  %V = add <2 x i32> <i32 123, i32 123>, %A
  ret <2 x i32> %V
}

define <2 x i32> @test42vec2(i1 %C) {
; CHECK-LABEL: @test42vec2(
; CHECK-NEXT:    [[V:%.*]] = select i1 [[C:%.*]], <2 x i32> <i32 1123, i32 2833>, <2 x i32> <i32 133, i32 363>
; CHECK-NEXT:    ret <2 x i32> [[V]]
;
  %A = select i1 %C, <2 x i32> <i32 1000, i32 2500>, <2 x i32> <i32 10, i32 30>
  %V = add <2 x i32> <i32 123, i32 333>, %A
  ret <2 x i32> %V
}

define i32 @test55(i1 %which) {
; CHECK-LABEL: @test55(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br i1 [[WHICH:%.*]], label [[FINAL:%.*]], label [[DELAY:%.*]]
; CHECK:       delay:
; CHECK-NEXT:    br label [[FINAL]]
; CHECK:       final:
; CHECK-NEXT:    [[A:%.*]] = phi i32 [ 1123, [[ENTRY:%.*]] ], [ 133, [[DELAY]] ]
; CHECK-NEXT:    ret i32 [[A]]
;
entry:
  br i1 %which, label %final, label %delay

delay:
  br label %final

final:
  %A = phi i32 [ 1000, %entry ], [ 10, %delay ]
  %value = add i32 123, %A
  ret i32 %value
}

define <2 x i32> @test43vec(i1 %which) {
; CHECK-LABEL: @test43vec(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br i1 [[WHICH:%.*]], label [[FINAL:%.*]], label [[DELAY:%.*]]
; CHECK:       delay:
; CHECK-NEXT:    br label [[FINAL]]
; CHECK:       final:
; CHECK-NEXT:    [[A:%.*]] = phi <2 x i32> [ <i32 1123, i32 1123>, [[ENTRY:%.*]] ], [ <i32 133, i32 133>, [[DELAY]] ]
; CHECK-NEXT:    ret <2 x i32> [[A]]
;
entry:
  br i1 %which, label %final, label %delay

delay:
  br label %final

final:
  %A = phi <2 x i32> [ <i32 1000, i32 1000>, %entry ], [ <i32 10, i32 10>, %delay ]
  %value = add <2 x i32> <i32 123, i32 123>, %A
  ret <2 x i32> %value
}

define <2 x i32> @test43vec2(i1 %which) {
; CHECK-LABEL: @test43vec2(
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br i1 [[WHICH:%.*]], label [[FINAL:%.*]], label [[DELAY:%.*]]
; CHECK:       delay:
; CHECK-NEXT:    br label [[FINAL]]
; CHECK:       final:
; CHECK-NEXT:    [[A:%.*]] = phi <2 x i32> [ <i32 1123, i32 2833>, [[ENTRY:%.*]] ], [ <i32 133, i32 363>, [[DELAY]] ]
; CHECK-NEXT:    ret <2 x i32> [[A]]
;
entry:
  br i1 %which, label %final, label %delay

delay:
  br label %final

final:
  %A = phi <2 x i32> [ <i32 1000, i32 2500>, %entry ], [ <i32 10, i32 30>, %delay ]
  %value = add <2 x i32> <i32 123, i32 333>, %A
  ret <2 x i32> %value
}

; E = (A + 1) + ~B = A - B
define i32 @add_not_increment(i32 %A, i32 %B) {
; CHECK-LABEL: @add_not_increment(
; CHECK-NEXT:    [[E:%.*]] = sub i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    ret i32 [[E]]
;
  %C = xor i32 %B, -1
  %D = add i32 %A, 1
  %E = add i32 %D, %C
  ret i32 %E
}

; E = (A + 1) + ~B = A - B
define <2 x i32> @add_not_increment_vec(<2 x i32> %A, <2 x i32> %B) {
; CHECK-LABEL: @add_not_increment_vec(
; CHECK-NEXT:    [[E:%.*]] = sub <2 x i32> [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    ret <2 x i32> [[E]]
;
  %C = xor <2 x i32> %B, <i32 -1, i32 -1>
  %D = add <2 x i32> %A, <i32 1, i32 1>
  %E = add <2 x i32> %D, %C
  ret <2 x i32> %E
}

; E = ~B + (1 + A) = A - B
define i32 @add_not_increment_commuted(i32 %A, i32 %B) {
; CHECK-LABEL: @add_not_increment_commuted(
; CHECK-NEXT:    [[E:%.*]] = sub i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    ret i32 [[E]]
;
  %C = xor i32 %B, -1
  %D = add i32 %A, 1
  %E = add i32 %C, %D
  ret i32 %E
}

; E = (A + ~B) + 1 = A - B
define i32 @add_to_sub(i32 %M, i32 %B) {
; CHECK-LABEL: @add_to_sub(
; CHECK-NEXT:    [[A:%.*]] = mul i32 [[M:%.*]], 42
; CHECK-NEXT:    [[E:%.*]] = sub i32 [[A]], [[B:%.*]]
; CHECK-NEXT:    ret i32 [[E]]
;
  %A = mul i32 %M, 42          ; thwart complexity-based ordering
  %C = xor i32 %B, -1
  %D = add i32 %A, %C
  %E = add i32 %D, 1
  ret i32 %E
}

; E = (~B + A) + 1 = A - B
define i32 @add_to_sub2(i32 %A, i32 %M) {
; CHECK-LABEL: @add_to_sub2(
; CHECK-NEXT:    [[B_NEG:%.*]] = mul i32 [[M:%.*]], -42
; CHECK-NEXT:    [[E:%.*]] = add i32 [[B_NEG]], [[A:%.*]]
; CHECK-NEXT:    ret i32 [[E]]
;
  %B = mul i32 %M, 42          ; thwart complexity-based ordering
  %C = xor i32 %B, -1
  %D = add i32 %C, %A
  %E = add i32 %D, 1
  ret i32 %E
}

; (X | C1) + C2 --> (X | C1) ^ C1 iff (C1 == -C2)
define i32 @test44(i32 %A) {
; CHECK-LABEL: @test44(
; CHECK-NEXT:    [[C:%.*]] = and i32 [[A:%.*]], -124
; CHECK-NEXT:    ret i32 [[C]]
;
  %B = or i32 %A, 123
  %C = add i32 %B, -123
  ret i32 %C
}

define i32 @test44_extra_use(i32 %A) {
; CHECK-LABEL: @test44_extra_use(
; CHECK-NEXT:    [[B:%.*]] = or i32 [[A:%.*]], 123
; CHECK-NEXT:    [[C:%.*]] = and i32 [[A]], -124
; CHECK-NEXT:    [[D:%.*]] = mul i32 [[B]], [[C]]
; CHECK-NEXT:    ret i32 [[D]]
;
  %B = or i32 %A, 123
  %C = add i32 %B, -123
  %D = mul i32 %B, %C
  ret i32 %D
}

define i32 @test44_non_matching(i32 %A) {
; CHECK-LABEL: @test44_non_matching(
; CHECK-NEXT:    [[B:%.*]] = or i32 [[A:%.*]], 123
; CHECK-NEXT:    [[C:%.*]] = add i32 [[B]], -321
; CHECK-NEXT:    ret i32 [[C]]
;
  %B = or i32 %A, 123
  %C = add i32 %B, -321
  ret i32 %C
}

define <2 x i32> @test44_vec(<2 x i32> %A) {
; CHECK-LABEL: @test44_vec(
; CHECK-NEXT:    [[C:%.*]] = and <2 x i32> [[A:%.*]], <i32 -124, i32 -124>
; CHECK-NEXT:    ret <2 x i32> [[C]]
;
  %B = or <2 x i32> %A, <i32 123, i32 123>
  %C = add <2 x i32> %B, <i32 -123, i32 -123>
  ret <2 x i32> %C
}

define <2 x i32> @test44_vec_non_matching(<2 x i32> %A) {
; CHECK-LABEL: @test44_vec_non_matching(
; CHECK-NEXT:    [[B:%.*]] = or <2 x i32> [[A:%.*]], <i32 123, i32 123>
; CHECK-NEXT:    [[C:%.*]] = add <2 x i32> [[B]], <i32 -321, i32 -321>
; CHECK-NEXT:    ret <2 x i32> [[C]]
;
  %B = or <2 x i32> %A, <i32 123, i32 123>
  %C = add <2 x i32> %B, <i32 -321, i32 -321>
  ret <2 x i32> %C
}

define <2 x i32> @test44_vec_undef(<2 x i32> %A) {
; CHECK-LABEL: @test44_vec_undef(
; CHECK-NEXT:    [[B:%.*]] = or <2 x i32> [[A:%.*]], <i32 123, i32 undef>
; CHECK-NEXT:    [[C:%.*]] = add <2 x i32> [[B]], <i32 -123, i32 undef>
; CHECK-NEXT:    ret <2 x i32> [[C]]
;
  %B = or <2 x i32> %A, <i32 123, i32 undef>
  %C = add <2 x i32> %B, <i32 -123, i32 undef>
  ret <2 x i32> %C
}

define <2 x i32> @test44_vec_non_splat(<2 x i32> %A) {
; CHECK-LABEL: @test44_vec_non_splat(
; CHECK-NEXT:    [[B:%.*]] = or <2 x i32> [[A:%.*]], <i32 123, i32 456>
; CHECK-NEXT:    [[C:%.*]] = add <2 x i32> [[B]], <i32 -123, i32 -456>
; CHECK-NEXT:    ret <2 x i32> [[C]]
;
  %B = or <2 x i32> %A, <i32 123, i32 456>
  %C = add <2 x i32> %B, <i32 -123, i32 -456>
  ret <2 x i32> %C
}

define i32 @lshr_add(i1 %x, i1 %y) {
; CHECK-LABEL: @lshr_add(
; CHECK-NEXT:    [[TMP1:%.*]] = xor i1 [[X:%.*]], true
; CHECK-NEXT:    [[TMP2:%.*]] = and i1 [[TMP1]], [[Y:%.*]]
; CHECK-NEXT:    [[R:%.*]] = zext i1 [[TMP2]] to i32
; CHECK-NEXT:    ret i32 [[R]]
;
  %xz = zext i1 %x to i32
  %ys = sext i1 %y to i32
  %sub = add i32 %xz, %ys
  %r = lshr i32 %sub, 31
  ret i32 %r
}

define i5 @and_add(i1 %x, i1 %y) {
; CHECK-LABEL: @and_add(
; CHECK-NEXT:    [[TMP1:%.*]] = xor i1 [[X:%.*]], true
; CHECK-NEXT:    [[TMP2:%.*]] = and i1 [[TMP1]], [[Y:%.*]]
; CHECK-NEXT:    [[R:%.*]] = select i1 [[TMP2]], i5 -2, i5 0
; CHECK-NEXT:    ret i5 [[R]]
;
  %xz = zext i1 %x to i5
  %ys = sext i1 %y to i5
  %sub = add i5 %xz, %ys
  %r = and i5 %sub, 30
  ret i5 %r
}

define <2 x i8> @ashr_add_commute(<2 x i1> %x, <2 x i1> %y) {
; CHECK-LABEL: @ashr_add_commute(
; CHECK-NEXT:    [[TMP1:%.*]] = xor <2 x i1> [[X:%.*]], <i1 true, i1 true>
; CHECK-NEXT:    [[TMP2:%.*]] = and <2 x i1> [[TMP1]], [[Y:%.*]]
; CHECK-NEXT:    [[TMP3:%.*]] = sext <2 x i1> [[TMP2]] to <2 x i8>
; CHECK-NEXT:    ret <2 x i8> [[TMP3]]
;
  %xz = zext <2 x i1> %x to <2 x i8>
  %ys = sext <2 x i1> %y to <2 x i8>
  %sub = add nsw <2 x i8> %ys, %xz
  %r = ashr <2 x i8> %sub, <i8 1, i8 1>
  ret <2 x i8> %r
}

define i32 @cmp_math(i32 %x, i32 %y) {
; CHECK-LABEL: @cmp_math(
; CHECK-NEXT:    [[LT:%.*]] = icmp ult i32 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    [[R:%.*]] = zext i1 [[LT]] to i32
; CHECK-NEXT:    ret i32 [[R]]
;
  %gt = icmp ugt i32 %x, %y
  %lt = icmp ult i32 %x, %y
  %xz = zext i1 %gt to i32
  %yz = zext i1 %lt to i32
  %s = sub i32 %xz, %yz
  %r = lshr i32 %s, 31
  ret i32 %r
}

; Negative test - wrong type

define i32 @lshr_add_nonbool(i2 %x, i1 %y) {
; CHECK-LABEL: @lshr_add_nonbool(
; CHECK-NEXT:    [[XZ:%.*]] = zext i2 [[X:%.*]] to i32
; CHECK-NEXT:    [[YS:%.*]] = sext i1 [[Y:%.*]] to i32
; CHECK-NEXT:    [[SUB:%.*]] = add nsw i32 [[XZ]], [[YS]]
; CHECK-NEXT:    [[R:%.*]] = lshr i32 [[SUB]], 31
; CHECK-NEXT:    ret i32 [[R]]
;
  %xz = zext i2 %x to i32
  %ys = sext i1 %y to i32
  %sub = add i32 %xz, %ys
  %r = lshr i32 %sub, 31
  ret i32 %r
}

; Negative test - wrong demand

define i32 @and31_add(i1 %x, i1 %y) {
; CHECK-LABEL: @and31_add(
; CHECK-NEXT:    [[XZ:%.*]] = zext i1 [[X:%.*]] to i32
; CHECK-NEXT:    [[YS:%.*]] = sext i1 [[Y:%.*]] to i32
; CHECK-NEXT:    [[SUB:%.*]] = add nsw i32 [[XZ]], [[YS]]
; CHECK-NEXT:    [[R:%.*]] = and i32 [[SUB]], 31
; CHECK-NEXT:    ret i32 [[R]]
;
  %xz = zext i1 %x to i32
  %ys = sext i1 %y to i32
  %sub = add i32 %xz, %ys
  %r = and i32 %sub, 31
  ret i32 %r
}

; Negative test - extra use

define i32 @lshr_add_use(i1 %x, i1 %y, ptr %p) {
; CHECK-LABEL: @lshr_add_use(
; CHECK-NEXT:    [[XZ:%.*]] = zext i1 [[X:%.*]] to i32
; CHECK-NEXT:    store i32 [[XZ]], ptr [[P:%.*]], align 4
; CHECK-NEXT:    [[YS:%.*]] = sext i1 [[Y:%.*]] to i32
; CHECK-NEXT:    [[SUB:%.*]] = add nsw i32 [[XZ]], [[YS]]
; CHECK-NEXT:    [[R:%.*]] = lshr i32 [[SUB]], 31
; CHECK-NEXT:    ret i32 [[R]]
;
  %xz = zext i1 %x to i32
  store i32 %xz, ptr %p
  %ys = sext i1 %y to i32
  %sub = add i32 %xz, %ys
  %r = lshr i32 %sub, 31
  ret i32 %r
}

; Negative test - extra use

define i32 @lshr_add_use2(i1 %x, i1 %y, ptr %p) {
; CHECK-LABEL: @lshr_add_use2(
; CHECK-NEXT:    [[XZ:%.*]] = zext i1 [[X:%.*]] to i32
; CHECK-NEXT:    [[YS:%.*]] = sext i1 [[Y:%.*]] to i32
; CHECK-NEXT:    store i32 [[YS]], ptr [[P:%.*]], align 4
; CHECK-NEXT:    [[SUB:%.*]] = add nsw i32 [[XZ]], [[YS]]
; CHECK-NEXT:    [[R:%.*]] = lshr i32 [[SUB]], 31
; CHECK-NEXT:    ret i32 [[R]]
;
  %xz = zext i1 %x to i32
  %ys = sext i1 %y to i32
  store i32 %ys, ptr %p
  %sub = add i32 %xz, %ys
  %r = lshr i32 %sub, 31
  ret i32 %r
}

define i32 @lshr_add_sexts(i1 %x, i1 %y) {
; CHECK-LABEL: @lshr_add_sexts(
; CHECK-NEXT:    [[TMP1:%.*]] = or i1 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    [[R:%.*]] = zext i1 [[TMP1]] to i32
; CHECK-NEXT:    ret i32 [[R]]
;
  %xs = sext i1 %x to i32
  %ys = sext i1 %y to i32
  %sub = add i32 %xs, %ys
  %r = lshr i32 %sub, 31
  ret i32 %r
}

define i5 @and_add_sexts(i1 %x, i1 %y) {
; CHECK-LABEL: @and_add_sexts(
; CHECK-NEXT:    [[TMP1:%.*]] = or i1 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    [[R:%.*]] = select i1 [[TMP1]], i5 -2, i5 0
; CHECK-NEXT:    ret i5 [[R]]
;
  %xs = sext i1 %x to i5
  %ys = sext i1 %y to i5
  %sub = add i5 %xs, %ys
  %r = and i5 %sub, 30
  ret i5 %r
}

define <2 x i8> @ashr_add_sexts(<2 x i1> %x, <2 x i1> %y) {
; CHECK-LABEL: @ashr_add_sexts(
; CHECK-NEXT:    [[TMP1:%.*]] = or <2 x i1> [[Y:%.*]], [[X:%.*]]
; CHECK-NEXT:    [[TMP2:%.*]] = sext <2 x i1> [[TMP1]] to <2 x i8>
; CHECK-NEXT:    ret <2 x i8> [[TMP2]]
;
  %xs = sext <2 x i1> %x to <2 x i8>
  %ys = sext <2 x i1> %y to <2 x i8>
  %sub = add nsw <2 x i8> %ys, %xs
  %r = ashr <2 x i8> %sub, <i8 1, i8 1>
  ret <2 x i8> %r
}

define i32 @cmp_math_sexts(i32 %x, i32 %y) {
; CHECK-LABEL: @cmp_math_sexts(
; CHECK-NEXT:    [[DOTNOT:%.*]] = icmp ne i32 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    [[R:%.*]] = zext i1 [[DOTNOT]] to i32
; CHECK-NEXT:    ret i32 [[R]]
;
  %gt = icmp ugt i32 %x, %y
  %lt = icmp ult i32 %x, %y
  %xz = sext i1 %gt to i32
  %yz = zext i1 %lt to i32
  %s = sub i32 %xz, %yz
  %r = lshr i32 %s, 31
  ret i32 %r
}

; Negative test - wrong type

define i32 @lshr_add_nonbool_sexts(i2 %x, i1 %y) {
; CHECK-LABEL: @lshr_add_nonbool_sexts(
; CHECK-NEXT:    [[XS:%.*]] = sext i2 [[X:%.*]] to i32
; CHECK-NEXT:    [[YS:%.*]] = sext i1 [[Y:%.*]] to i32
; CHECK-NEXT:    [[SUB:%.*]] = add nsw i32 [[XS]], [[YS]]
; CHECK-NEXT:    [[R:%.*]] = lshr i32 [[SUB]], 31
; CHECK-NEXT:    ret i32 [[R]]
;
  %xs = sext i2 %x to i32
  %ys = sext i1 %y to i32
  %sub = add i32 %xs, %ys
  %r = lshr i32 %sub, 31
  ret i32 %r
}

; Negative test - wrong demand

define i32 @and31_add_sexts(i1 %x, i1 %y) {
; CHECK-LABEL: @and31_add_sexts(
; CHECK-NEXT:    [[XS:%.*]] = sext i1 [[X:%.*]] to i32
; CHECK-NEXT:    [[YS:%.*]] = sext i1 [[Y:%.*]] to i32
; CHECK-NEXT:    [[SUB:%.*]] = add nsw i32 [[XS]], [[YS]]
; CHECK-NEXT:    [[R:%.*]] = and i32 [[SUB]], 31
; CHECK-NEXT:    ret i32 [[R]]
;
  %xs = sext i1 %x to i32
  %ys = sext i1 %y to i32
  %sub = add i32 %xs, %ys
  %r = and i32 %sub, 31
  ret i32 %r
}

; Negative test - extra use

define i32 @lshr_add_use_sexts(i1 %x, i1 %y, ptr %p) {
; CHECK-LABEL: @lshr_add_use_sexts(
; CHECK-NEXT:    [[XS:%.*]] = sext i1 [[X:%.*]] to i32
; CHECK-NEXT:    store i32 [[XS]], ptr [[P:%.*]], align 4
; CHECK-NEXT:    [[YS:%.*]] = sext i1 [[Y:%.*]] to i32
; CHECK-NEXT:    [[SUB:%.*]] = add nsw i32 [[XS]], [[YS]]
; CHECK-NEXT:    [[R:%.*]] = lshr i32 [[SUB]], 31
; CHECK-NEXT:    ret i32 [[R]]
;
  %xs = sext i1 %x to i32
  store i32 %xs, ptr %p
  %ys = sext i1 %y to i32
  %sub = add i32 %xs, %ys
  %r = lshr i32 %sub, 31
  ret i32 %r
}

; Negative test - extra use

define i32 @lshr_add_use2_sexts(i1 %x, i1 %y, ptr %p) {
; CHECK-LABEL: @lshr_add_use2_sexts(
; CHECK-NEXT:    [[XS:%.*]] = sext i1 [[X:%.*]] to i32
; CHECK-NEXT:    [[YS:%.*]] = sext i1 [[Y:%.*]] to i32
; CHECK-NEXT:    store i32 [[YS]], ptr [[P:%.*]], align 4
; CHECK-NEXT:    [[SUB:%.*]] = add nsw i32 [[XS]], [[YS]]
; CHECK-NEXT:    [[R:%.*]] = lshr i32 [[SUB]], 31
; CHECK-NEXT:    ret i32 [[R]]
;
  %xs = sext i1 %x to i32
  %ys = sext i1 %y to i32
  store i32 %ys, ptr %p
  %sub = add i32 %xs, %ys
  %r = lshr i32 %sub, 31
  ret i32 %r
}

define i8 @add_like_or_t0(i8 %x) {
; CHECK-LABEL: @add_like_or_t0(
; CHECK-NEXT:    [[I0:%.*]] = shl i8 [[X:%.*]], 4
; CHECK-NEXT:    [[R:%.*]] = add i8 [[I0]], 57
; CHECK-NEXT:    ret i8 [[R]]
;
  %i0 = shl i8 %x, 4
  %i1 = or i8 %i0, 15 ; no common bits
  %r = add i8 %i1, 42
  ret i8 %r
}
define i8 @add_like_or_n1(i8 %x) {
; CHECK-LABEL: @add_like_or_n1(
; CHECK-NEXT:    [[I0:%.*]] = shl i8 [[X:%.*]], 4
; CHECK-NEXT:    [[I1:%.*]] = or i8 [[I0]], 31
; CHECK-NEXT:    [[R:%.*]] = add i8 [[I1]], 42
; CHECK-NEXT:    ret i8 [[R]]
;
  %i0 = shl i8 %x, 4
  %i1 = or i8 %i0, 31 ; 4'th bit might be common-set
  %r = add i8 %i1, 42
  ret i8 %r
}
define i8 @add_like_or_t2_extrause(i8 %x) {
; CHECK-LABEL: @add_like_or_t2_extrause(
; CHECK-NEXT:    [[I0:%.*]] = shl i8 [[X:%.*]], 4
; CHECK-NEXT:    [[I1:%.*]] = or disjoint i8 [[I0]], 15
; CHECK-NEXT:    call void @use(i8 [[I1]])
; CHECK-NEXT:    [[R:%.*]] = add i8 [[I0]], 57
; CHECK-NEXT:    ret i8 [[R]]
;
  %i0 = shl i8 %x, 4
  %i1 = or i8 %i0, 15 ; no common bits
  call void @use(i8 %i1) ; extra use
  %r = add i8 %i1, 42
  ret i8 %r
}

define i8 @add_like_or_disjoint(i8 %x) {
; CHECK-LABEL: @add_like_or_disjoint(
; CHECK-NEXT:    [[R:%.*]] = add i8 [[X:%.*]], 57
; CHECK-NEXT:    ret i8 [[R]]
;
  %i1 = or disjoint i8 %x, 15
  %r = add i8 %i1, 42
  ret i8 %r
}

define i8 @add_and_xor(i8 noundef %x, i8 %y) {
; CHECK-LABEL: @add_and_xor(
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[Y:%.*]], [[X:%.*]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %xor = xor i8 %x, -1
  %and = and i8 %xor, %y
  %add = add i8 %and, %x
  ret i8 %add
}

define i8 @add_and_xor_wrong_const(i8 %x, i8 %y) {
; CHECK-LABEL: @add_and_xor_wrong_const(
; CHECK-NEXT:    [[XOR:%.*]] = xor i8 [[X:%.*]], -2
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[XOR]], [[Y:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = add i8 [[AND]], [[X]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %xor = xor i8 %x, -2
  %and = and i8 %xor, %y
  %add = add i8 %and, %x
  ret i8 %add
}

define i8 @add_and_xor_wrong_op(i8 %x, i8 %y, i8 %z) {
; CHECK-LABEL: @add_and_xor_wrong_op(
; CHECK-NEXT:    [[XOR:%.*]] = xor i8 [[Z:%.*]], -1
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[XOR]], [[Y:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = add i8 [[AND]], [[X:%.*]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %xor = xor i8 %z, -1
  %and = and i8 %xor, %y
  %add = add i8 %and, %x
  ret i8 %add
}

define i8 @add_and_xor_commuted1(i8 noundef %x, i8 %_y) {
; CHECK-LABEL: @add_and_xor_commuted1(
; CHECK-NEXT:    [[Y:%.*]] = udiv i8 42, [[_Y:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[Y]], [[X:%.*]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %y = udiv i8 42, %_y ; thwart complexity-based canonicalization
  %xor = xor i8 %x, -1
  %and = and i8 %y, %xor
  %add = add i8 %and, %x
  ret i8 %add
}

define i8 @add_and_xor_commuted2(i8 noundef %_x, i8 %y) {
; CHECK-LABEL: @add_and_xor_commuted2(
; CHECK-NEXT:    [[X:%.*]] = udiv i8 42, [[_X:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[X]], [[Y:%.*]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %x = udiv i8 42, %_x ; thwart complexity-based canonicalization
  %xor = xor i8 %x, -1
  %and = and i8 %xor, %y
  %add = add i8 %x, %and
  ret i8 %add
}

define i8 @add_and_xor_commuted3(i8 noundef %_x, i8 %_y) {
; CHECK-LABEL: @add_and_xor_commuted3(
; CHECK-NEXT:    [[X:%.*]] = udiv i8 42, [[_X:%.*]]
; CHECK-NEXT:    [[Y:%.*]] = udiv i8 42, [[_Y:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[X]], [[Y]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %x = udiv i8 42, %_x ; thwart complexity-based canonicalization
  %y = udiv i8 42, %_y ; thwart complexity-based canonicalization
  %xor = xor i8 %x, -1
  %and = and i8 %y, %xor
  %add = add i8 %x, %and
  ret i8 %add
}

define i8 @add_and_xor_extra_use(i8 noundef %x, i8 %y) {
; CHECK-LABEL: @add_and_xor_extra_use(
; CHECK-NEXT:    [[XOR:%.*]] = xor i8 [[X:%.*]], -1
; CHECK-NEXT:    call void @use(i8 [[XOR]])
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[XOR]], [[Y:%.*]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[Y]], [[X]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %xor = xor i8 %x, -1
  call void @use(i8 %xor)
  %and = and i8 %xor, %y
  call void @use(i8 %and)
  %add = add i8 %and, %x
  ret i8 %add
}

define i8 @add_xor_and_const(i8 noundef %x) {
; CHECK-LABEL: @add_xor_and_const(
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[X:%.*]], 42
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %and = and i8 %x, 42
  %xor = xor i8 %and, 42
  %add = add i8 %xor, %x
  ret i8 %add
}

define i8 @add_xor_and_const_wrong_const(i8 %x) {
; CHECK-LABEL: @add_xor_and_const_wrong_const(
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X:%.*]], 42
; CHECK-NEXT:    [[XOR:%.*]] = xor i8 [[AND]], 88
; CHECK-NEXT:    [[ADD:%.*]] = add i8 [[XOR]], [[X]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %and = and i8 %x, 42
  %xor = xor i8 %and, 88
  %add = add i8 %xor, %x
  ret i8 %add
}

define i8 @add_xor_and_var(i8 noundef %x, i8 noundef %y) {
; CHECK-LABEL: @add_xor_and_var(
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[Y]], [[X]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %and = and i8 %x, %y
  call void @use(i8 %and)
  %xor = xor i8 %and, %y
  %add = add i8 %xor, %x
  ret i8 %add
}

define i8 @add_xor_and_var_wrong_op1(i8 %x, i8 %y, i8 %z) {
; CHECK-LABEL: @add_xor_and_var_wrong_op1(
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[XOR:%.*]] = xor i8 [[AND]], [[Z:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = add i8 [[XOR]], [[X]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %and = and i8 %x, %y
  call void @use(i8 %and)
  %xor = xor i8 %and, %z
  %add = add i8 %xor, %x
  ret i8 %add
}

define i8 @add_xor_and_var_wrong_op2(i8 %x, i8 %y, i8 %z) {
; CHECK-LABEL: @add_xor_and_var_wrong_op2(
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[XOR:%.*]] = xor i8 [[AND]], [[Y]]
; CHECK-NEXT:    [[ADD:%.*]] = add i8 [[XOR]], [[Z:%.*]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %and = and i8 %x, %y
  call void @use(i8 %and)
  %xor = xor i8 %and, %y
  %add = add i8 %xor, %z
  ret i8 %add
}

define i8 @add_xor_and_var_commuted1(i8 noundef %x, i8 noundef %y) {
; CHECK-LABEL: @add_xor_and_var_commuted1(
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[Y:%.*]], [[X:%.*]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[Y]], [[X]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %and = and i8 %y, %x
  call void @use(i8 %and)
  %xor = xor i8 %and, %y
  %add = add i8 %xor, %x
  ret i8 %add
}

define i8 @add_xor_and_var_commuted2(i8 noundef %_x, i8 noundef %_y) {
; CHECK-LABEL: @add_xor_and_var_commuted2(
; CHECK-NEXT:    [[X:%.*]] = udiv i8 42, [[_X:%.*]]
; CHECK-NEXT:    [[Y:%.*]] = udiv i8 42, [[_Y:%.*]]
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[Y]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[Y]], [[X]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %x = udiv i8 42, %_x ; thwart complexity-based canonicalization
  %y = udiv i8 42, %_y ; thwart complexity-based canonicalization
  %and = and i8 %x, %y
  call void @use(i8 %and)
  %xor = xor i8 %y, %and
  %add = add i8 %xor, %x
  ret i8 %add
}

define i8 @add_xor_and_var_commuted3(i8 noundef %x, i8 noundef %_y) {
; CHECK-LABEL: @add_xor_and_var_commuted3(
; CHECK-NEXT:    [[Y:%.*]] = udiv i8 42, [[_Y:%.*]]
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[Y]], [[X:%.*]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[Y]], [[X]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %y = udiv i8 42, %_y ; thwart complexity-based canonicalization
  %and = and i8 %y, %x
  call void @use(i8 %and)
  %xor = xor i8 %y, %and
  %add = add i8 %xor, %x
  ret i8 %add
}

define i8 @add_xor_and_var_commuted4(i8 noundef %_x, i8 noundef %y) {
; CHECK-LABEL: @add_xor_and_var_commuted4(
; CHECK-NEXT:    [[X:%.*]] = udiv i8 42, [[_X:%.*]]
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[Y:%.*]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[X]], [[Y]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %x = udiv i8 42, %_x ; thwart complexity-based canonicalization
  %and = and i8 %x, %y
  call void @use(i8 %and)
  %xor = xor i8 %and, %y
  %add = add i8 %x, %xor
  ret i8 %add
}

define i8 @add_xor_and_var_commuted5(i8 noundef %_x, i8 noundef %_y) {
; CHECK-LABEL: @add_xor_and_var_commuted5(
; CHECK-NEXT:    [[X:%.*]] = udiv i8 42, [[_X:%.*]]
; CHECK-NEXT:    [[Y:%.*]] = udiv i8 42, [[_Y:%.*]]
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[Y]], [[X]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[X]], [[Y]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %x = udiv i8 42, %_x ; thwart complexity-based canonicalization
  %y = udiv i8 42, %_y ; thwart complexity-based canonicalization
  %and = and i8 %y, %x
  call void @use(i8 %and)
  %xor = xor i8 %and, %y
  %add = add i8 %x, %xor
  ret i8 %add
}

define i8 @add_xor_and_var_commuted6(i8 noundef %_x, i8 noundef %_y) {
; CHECK-LABEL: @add_xor_and_var_commuted6(
; CHECK-NEXT:    [[X:%.*]] = udiv i8 42, [[_X:%.*]]
; CHECK-NEXT:    [[Y:%.*]] = udiv i8 42, [[_Y:%.*]]
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], [[Y]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[X]], [[Y]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %x = udiv i8 42, %_x ; thwart complexity-based canonicalization
  %y = udiv i8 42, %_y ; thwart complexity-based canonicalization
  %and = and i8 %x, %y
  call void @use(i8 %and)
  %xor = xor i8 %y, %and
  %add = add i8 %x, %xor
  ret i8 %add
}

define i8 @add_xor_and_var_commuted7(i8 noundef %_x, i8 noundef %_y) {
; CHECK-LABEL: @add_xor_and_var_commuted7(
; CHECK-NEXT:    [[X:%.*]] = udiv i8 42, [[_X:%.*]]
; CHECK-NEXT:    [[Y:%.*]] = udiv i8 42, [[_Y:%.*]]
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[Y]], [[X]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[X]], [[Y]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %x = udiv i8 42, %_x ; thwart complexity-based canonicalization
  %y = udiv i8 42, %_y ; thwart complexity-based canonicalization
  %and = and i8 %y, %x
  call void @use(i8 %and)
  %xor = xor i8 %y, %and
  %add = add i8 %x, %xor
  ret i8 %add
}

define i8 @add_xor_and_var_extra_use(i8 noundef %x, i8 noundef %y) {
; CHECK-LABEL: @add_xor_and_var_extra_use(
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    call void @use(i8 [[AND]])
; CHECK-NEXT:    [[XOR:%.*]] = xor i8 [[AND]], [[Y]]
; CHECK-NEXT:    call void @use(i8 [[XOR]])
; CHECK-NEXT:    [[ADD:%.*]] = or i8 [[Y]], [[X]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %and = and i8 %x, %y
  call void @use(i8 %and)
  %xor = xor i8 %and, %y
  call void @use(i8 %xor)
  %add = add i8 %xor, %x
  ret i8 %add
}

define i32 @add_add_add(i32 %A, i32 %B, i32 %C, i32 %D) {
; CHECK-LABEL: @add_add_add(
; CHECK-NEXT:    [[E:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[F:%.*]] = add i32 [[E]], [[C:%.*]]
; CHECK-NEXT:    [[G:%.*]] = add i32 [[F]], [[D:%.*]]
; CHECK-NEXT:    ret i32 [[G]]
;
  %E = add i32 %A, %B
  %F = add i32 %E, %C
  %G = add i32 %F, %D
  ret i32 %G
}

define i32 @add_add_add_commute1(i32 %A, i32 %B, i32 %C, i32 %D) {
; CHECK-LABEL: @add_add_add_commute1(
; CHECK-NEXT:    [[E:%.*]] = add i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    [[F:%.*]] = add i32 [[E]], [[C:%.*]]
; CHECK-NEXT:    [[G:%.*]] = add i32 [[F]], [[D:%.*]]
; CHECK-NEXT:    ret i32 [[G]]
;
  %E = add i32 %B, %A
  %F = add i32 %E, %C
  %G = add i32 %F, %D
  ret i32 %G
}

define i32 @add_add_add_commute2(i32 %A, i32 %B, i32 %C, i32 %D) {
; CHECK-LABEL: @add_add_add_commute2(
; CHECK-NEXT:    [[E:%.*]] = add i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    [[F:%.*]] = add i32 [[E]], [[C:%.*]]
; CHECK-NEXT:    [[G:%.*]] = add i32 [[F]], [[D:%.*]]
; CHECK-NEXT:    ret i32 [[G]]
;
  %E = add i32 %B, %A
  %F = add i32 %C, %E
  %G = add i32 %F, %D
  ret i32 %G
}

define i32 @add_add_add_commute3(i32 %A, i32 %B, i32 %C, i32 %D) {
; CHECK-LABEL: @add_add_add_commute3(
; CHECK-NEXT:    [[E:%.*]] = add i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    [[F:%.*]] = add i32 [[E]], [[C:%.*]]
; CHECK-NEXT:    [[G:%.*]] = add i32 [[F]], [[D:%.*]]
; CHECK-NEXT:    ret i32 [[G]]
;
  %E = add i32 %B, %A
  %F = add i32 %C, %E
  %G = add i32 %D, %F
  ret i32 %G
}

; x * y + x --> (y + 1) * x

define i8 @mul_add_common_factor_commute1(i8 %x, i8 %y) {
; CHECK-LABEL: @mul_add_common_factor_commute1(
; CHECK-NEXT:    [[X1:%.*]] = add i8 [[Y:%.*]], 1
; CHECK-NEXT:    [[A:%.*]] = mul i8 [[X1]], [[X:%.*]]
; CHECK-NEXT:    ret i8 [[A]]
;
  %m = mul nsw i8 %x, %y
  %a = add nsw i8 %m, %x
  ret i8 %a
}

define <2 x i8> @mul_add_common_factor_commute2(<2 x i8> %x, <2 x i8> %y) {
; CHECK-LABEL: @mul_add_common_factor_commute2(
; CHECK-NEXT:    [[M1:%.*]] = add <2 x i8> [[Y:%.*]], <i8 1, i8 1>
; CHECK-NEXT:    [[A:%.*]] = mul nuw <2 x i8> [[M1]], [[X:%.*]]
; CHECK-NEXT:    ret <2 x i8> [[A]]
;
  %m = mul nuw <2 x i8> %y, %x
  %a = add nuw <2 x i8> %m, %x
  ret <2 x i8> %a
}

define i8 @mul_add_common_factor_commute3(i8 %p, i8 %y) {
; CHECK-LABEL: @mul_add_common_factor_commute3(
; CHECK-NEXT:    [[X:%.*]] = mul i8 [[P:%.*]], [[P]]
; CHECK-NEXT:    [[M1:%.*]] = add i8 [[Y:%.*]], 1
; CHECK-NEXT:    [[A:%.*]] = mul i8 [[X]], [[M1]]
; CHECK-NEXT:    ret i8 [[A]]
;
  %x = mul i8 %p, %p ; thwart complexity-based canonicalization
  %m = mul nuw i8 %x, %y
  %a = add nsw i8 %x, %m
  ret i8 %a
}

define i8 @mul_add_common_factor_commute4(i8 %p, i8 %q) {
; CHECK-LABEL: @mul_add_common_factor_commute4(
; CHECK-NEXT:    [[X:%.*]] = mul i8 [[P:%.*]], [[P]]
; CHECK-NEXT:    [[Y:%.*]] = mul i8 [[Q:%.*]], [[Q]]
; CHECK-NEXT:    [[M1:%.*]] = add i8 [[Y]], 1
; CHECK-NEXT:    [[A:%.*]] = mul i8 [[X]], [[M1]]
; CHECK-NEXT:    ret i8 [[A]]
;
  %x = mul i8 %p, %p ; thwart complexity-based canonicalization
  %y = mul i8 %q, %q ; thwart complexity-based canonicalization
  %m = mul nsw i8 %y, %x
  %a = add nuw i8 %x, %m
  ret i8 %a
}

; negative test - uses

define i8 @mul_add_common_factor_use(i8 %x, i8 %y) {
; CHECK-LABEL: @mul_add_common_factor_use(
; CHECK-NEXT:    [[M:%.*]] = mul i8 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    call void @use(i8 [[M]])
; CHECK-NEXT:    [[A:%.*]] = add i8 [[M]], [[X]]
; CHECK-NEXT:    ret i8 [[A]]
;
  %m = mul i8 %x, %y
  call void @use(i8 %m)
  %a = add i8 %m, %x
  ret i8 %a
}

define i8 @not_mul(i8 %x) {
; CHECK-LABEL: @not_mul(
; CHECK-NEXT:    [[TMP1:%.*]] = mul i8 [[X:%.*]], -41
; CHECK-NEXT:    [[PLUSX:%.*]] = add i8 [[TMP1]], -1
; CHECK-NEXT:    ret i8 [[PLUSX]]
;
  %mul = mul nsw i8 %x, 42
  %not = xor i8 %mul, -1
  %plusx = add nsw i8 %not, %x
  ret i8 %plusx
}

define <2 x i8> @not_mul_commute(<2 x i8> %p) {
; CHECK-LABEL: @not_mul_commute(
; CHECK-NEXT:    [[X:%.*]] = mul <2 x i8> [[P:%.*]], [[P]]
; CHECK-NEXT:    [[TMP1:%.*]] = mul <2 x i8> [[X]], <i8 43, i8 43>
; CHECK-NEXT:    [[PLUSX:%.*]] = add <2 x i8> [[TMP1]], <i8 -1, i8 -1>
; CHECK-NEXT:    ret <2 x i8> [[PLUSX]]
;
  %x = mul <2 x i8> %p, %p ; thwart complexity-based canonicalization
  %mul = mul nuw <2 x i8> %x, <i8 -42, i8 -42>
  %not = xor <2 x i8> %mul, <i8 -1, i8 -1>
  %plusx = add nuw <2 x i8> %x, %not
  ret <2 x i8> %plusx
}

; negative test - need common operand

define i8 @not_mul_wrong_op(i8 %x, i8 %y) {
; CHECK-LABEL: @not_mul_wrong_op(
; CHECK-NEXT:    [[MUL:%.*]] = mul i8 [[X:%.*]], 42
; CHECK-NEXT:    [[NOT:%.*]] = xor i8 [[MUL]], -1
; CHECK-NEXT:    [[PLUSX:%.*]] = add i8 [[NOT]], [[Y:%.*]]
; CHECK-NEXT:    ret i8 [[PLUSX]]
;
  %mul = mul i8 %x, 42
  %not = xor i8 %mul, -1
  %plusx = add i8 %not, %y
  ret i8 %plusx
}

; negative test - avoid creating an extra mul

define i8 @not_mul_use1(i8 %x) {
; CHECK-LABEL: @not_mul_use1(
; CHECK-NEXT:    [[MUL:%.*]] = mul nsw i8 [[X:%.*]], 42
; CHECK-NEXT:    call void @use(i8 [[MUL]])
; CHECK-NEXT:    [[NOT:%.*]] = xor i8 [[MUL]], -1
; CHECK-NEXT:    [[PLUSX:%.*]] = add nsw i8 [[NOT]], [[X]]
; CHECK-NEXT:    ret i8 [[PLUSX]]
;
  %mul = mul nsw i8 %x, 42
  call void @use(i8 %mul)
  %not = xor i8 %mul, -1
  %plusx = add nsw i8 %not, %x
  ret i8 %plusx
}

; negative test - too many instructions

define i8 @not_mul_use2(i8 %x) {
; CHECK-LABEL: @not_mul_use2(
; CHECK-NEXT:    [[MUL:%.*]] = mul i8 [[X:%.*]], 42
; CHECK-NEXT:    [[NOT:%.*]] = xor i8 [[MUL]], -1
; CHECK-NEXT:    call void @use(i8 [[NOT]])
; CHECK-NEXT:    [[PLUSX:%.*]] = add i8 [[NOT]], [[X]]
; CHECK-NEXT:    ret i8 [[PLUSX]]
;
  %mul = mul i8 %x, 42
  %not = xor i8 %mul, -1
  call void @use(i8 %not)
  %plusx = add i8 %not, %x
  ret i8 %plusx
}

define i8 @full_ashr_inc(i8 %x) {
; CHECK-LABEL: @full_ashr_inc(
; CHECK-NEXT:    [[ISNOTNEG:%.*]] = icmp sgt i8 [[X:%.*]], -1
; CHECK-NEXT:    [[R:%.*]] = zext i1 [[ISNOTNEG]] to i8
; CHECK-NEXT:    ret i8 [[R]]
;
  %a = ashr i8 %x, 7
  %r = add i8 %a, 1
  ret i8 %r
}

define <2 x i6> @full_ashr_inc_vec(<2 x i6> %x) {
; CHECK-LABEL: @full_ashr_inc_vec(
; CHECK-NEXT:    [[ISNOTNEG:%.*]] = icmp sgt <2 x i6> [[X:%.*]], <i6 -1, i6 -1>
; CHECK-NEXT:    [[R:%.*]] = zext <2 x i1> [[ISNOTNEG]] to <2 x i6>
; CHECK-NEXT:    ret <2 x i6> [[R]]
;
  %a = ashr <2 x i6> %x, <i6 5, i6 poison>
  %r = add <2 x i6> %a, <i6 1, i6 1>
  ret <2 x i6> %r
}

; negative test - extra use

define i8 @full_ashr_inc_use(i8 %x) {
; CHECK-LABEL: @full_ashr_inc_use(
; CHECK-NEXT:    [[A:%.*]] = ashr i8 [[X:%.*]], 7
; CHECK-NEXT:    call void @use(i8 [[A]])
; CHECK-NEXT:    [[R:%.*]] = add nsw i8 [[A]], 1
; CHECK-NEXT:    ret i8 [[R]]
;
  %a = ashr i8 %x, 7
  call void @use(i8 %a)
  %r = add i8 %a, 1
  ret i8 %r
}

; negative test - wrong shift amount

define i8 @not_full_ashr_inc(i8 %x) {
; CHECK-LABEL: @not_full_ashr_inc(
; CHECK-NEXT:    [[A:%.*]] = ashr i8 [[X:%.*]], 6
; CHECK-NEXT:    [[R:%.*]] = add nsw i8 [[A]], 1
; CHECK-NEXT:    ret i8 [[R]]
;
  %a = ashr i8 %x, 6
  %r = add i8 %a, 1
  ret i8 %r
}

; negative test - wrong add amount

define i8 @full_ashr_not_inc(i8 %x) {
; CHECK-LABEL: @full_ashr_not_inc(
; CHECK-NEXT:    [[A:%.*]] = ashr i8 [[X:%.*]], 7
; CHECK-NEXT:    [[R:%.*]] = add nsw i8 [[A]], 2
; CHECK-NEXT:    ret i8 [[R]]
;
  %a = ashr i8 %x, 7
  %r = add i8 %a, 2
  ret i8 %r
}

define i8 @select_negate_or_zero(i1 %b, i8 %x, i8 %y) {
; CHECK-LABEL: @select_negate_or_zero(
; CHECK-NEXT:    [[TMP1:%.*]] = select i1 [[B:%.*]], i8 0, i8 [[X:%.*]]
; CHECK-NEXT:    [[ADD1:%.*]] = sub i8 [[Y:%.*]], [[TMP1]]
; CHECK-NEXT:    ret i8 [[ADD1]]
;
  %negx = sub i8 0, %x
  %sel = select i1 %b, i8 0, i8 %negx
  %add = add i8 %sel, %y
  ret i8 %add
}

; commuted add operands - same result

define <2 x i8> @select_negate_or_zero_commute(<2 x i1> %b, <2 x i8> %x, <2 x i8> %p) {
; CHECK-LABEL: @select_negate_or_zero_commute(
; CHECK-NEXT:    [[Y:%.*]] = mul <2 x i8> [[P:%.*]], [[P]]
; CHECK-NEXT:    [[TMP1:%.*]] = select <2 x i1> [[B:%.*]], <2 x i8> zeroinitializer, <2 x i8> [[X:%.*]]
; CHECK-NEXT:    [[ADD1:%.*]] = sub <2 x i8> [[Y]], [[TMP1]]
; CHECK-NEXT:    ret <2 x i8> [[ADD1]]
;
  %y = mul <2 x i8> %p, %p ; thwart complexity-based canonicalization
  %negx = sub <2 x i8> <i8 poison, i8 0>, %x
  %sel = select <2 x i1> %b, <2 x i8> <i8 poison, i8 0>, <2 x i8> %negx
  %add = add <2 x i8> %y, %sel
  ret <2 x i8> %add
}

; swapped select operands and extra use are ok

define i8 @select_negate_or_zero_swap(i1 %b, i8 %x, i8 %y) {
; CHECK-LABEL: @select_negate_or_zero_swap(
; CHECK-NEXT:    [[NEGX:%.*]] = sub i8 0, [[X:%.*]]
; CHECK-NEXT:    call void @use(i8 [[NEGX]])
; CHECK-NEXT:    [[TMP1:%.*]] = select i1 [[B:%.*]], i8 [[X]], i8 0
; CHECK-NEXT:    [[ADD1:%.*]] = sub i8 [[Y:%.*]], [[TMP1]]
; CHECK-NEXT:    ret i8 [[ADD1]]
;
  %negx = sub i8 0, %x
  call void @use(i8 %negx)
  %sel = select i1 %b, i8 %negx, i8 0
  %add = add i8 %sel, %y
  ret i8 %add
}

; commuted add operands - same result

define i8 @select_negate_or_zero_swap_commute(i1 %b, i8 %x, i8 %p) {
; CHECK-LABEL: @select_negate_or_zero_swap_commute(
; CHECK-NEXT:    [[Y:%.*]] = mul i8 [[P:%.*]], [[P]]
; CHECK-NEXT:    [[TMP1:%.*]] = select i1 [[B:%.*]], i8 [[X:%.*]], i8 0
; CHECK-NEXT:    [[ADD1:%.*]] = sub i8 [[Y]], [[TMP1]]
; CHECK-NEXT:    ret i8 [[ADD1]]
;
  %y = mul i8 %p, %p ; thwart complexity-based canonicalization
  %negx = sub i8 0, %x
  %sel = select i1 %b, i8 %negx, i8 0
  %add = add i8 %y, %sel
  ret i8 %add
}

; negative test - one arm of the select must simplify

define i8 @select_negate_or_nonzero(i1 %b, i8 %x, i8 %y) {
; CHECK-LABEL: @select_negate_or_nonzero(
; CHECK-NEXT:    [[NEGX:%.*]] = sub i8 0, [[X:%.*]]
; CHECK-NEXT:    [[SEL:%.*]] = select i1 [[B:%.*]], i8 42, i8 [[NEGX]]
; CHECK-NEXT:    [[ADD:%.*]] = add i8 [[SEL]], [[Y:%.*]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %negx = sub i8 0, %x
  %sel = select i1 %b, i8 42, i8 %negx
  %add = add i8 %sel, %y
  ret i8 %add
}

; negative test - must have a negate, not any subtract

define i8 @select_nonnegate_or_zero(i1 %b, i8 %x, i8 %y) {
; CHECK-LABEL: @select_nonnegate_or_zero(
; CHECK-NEXT:    [[NEGX:%.*]] = sub i8 42, [[X:%.*]]
; CHECK-NEXT:    [[SEL:%.*]] = select i1 [[B:%.*]], i8 0, i8 [[NEGX]]
; CHECK-NEXT:    [[ADD:%.*]] = add i8 [[SEL]], [[Y:%.*]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %negx = sub i8 42, %x
  %sel = select i1 %b, i8 0, i8 %negx
  %add = add i8 %sel, %y
  ret i8 %add
}

; negative test - don't create an extra instruction

define i8 @select_negate_or_nonzero_use(i1 %b, i8 %x, i8 %y) {
; CHECK-LABEL: @select_negate_or_nonzero_use(
; CHECK-NEXT:    [[NEGX:%.*]] = sub i8 0, [[X:%.*]]
; CHECK-NEXT:    [[SEL:%.*]] = select i1 [[B:%.*]], i8 0, i8 [[NEGX]]
; CHECK-NEXT:    call void @use(i8 [[SEL]])
; CHECK-NEXT:    [[ADD:%.*]] = add i8 [[SEL]], [[Y:%.*]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %negx = sub i8 0, %x
  %sel = select i1 %b, i8 0, i8 %negx
  call void @use(i8 %sel)
  %add = add i8 %sel, %y
  ret i8 %add
}

; extra reduction because y + ~y -> -1

define i5 @select_negate_not(i1 %b, i5 %x, i5 %y) {
; CHECK-LABEL: @select_negate_not(
; CHECK-NEXT:    [[TMP1:%.*]] = sub i5 [[Y:%.*]], [[X:%.*]]
; CHECK-NEXT:    [[ADD1:%.*]] = select i1 [[B:%.*]], i5 -1, i5 [[TMP1]]
; CHECK-NEXT:    ret i5 [[ADD1]]
;
  %negx = sub i5 0, %x
  %noty = xor i5 %y, -1
  %sel = select i1 %b, i5 %noty, i5 %negx
  %add = add i5 %sel, %y
  ret i5 %add
}

define i5 @select_negate_not_commute(i1 %b, i5 %x, i5 %p) {
; CHECK-LABEL: @select_negate_not_commute(
; CHECK-NEXT:    [[Y:%.*]] = mul i5 [[P:%.*]], [[P]]
; CHECK-NEXT:    [[TMP1:%.*]] = sub i5 [[Y]], [[X:%.*]]
; CHECK-NEXT:    [[ADD1:%.*]] = select i1 [[B:%.*]], i5 -1, i5 [[TMP1]]
; CHECK-NEXT:    ret i5 [[ADD1]]
;
  %y = mul i5 %p, %p ; thwart complexity-based canonicalization
  %negx = sub i5 0, %x
  %noty = xor i5 %y, -1
  %sel = select i1 %b, i5 %noty, i5 %negx
  %add = add i5 %y, %sel
  ret i5 %add
}

define i5 @select_negate_not_swap(i1 %b, i5 %x, i5 %y) {
; CHECK-LABEL: @select_negate_not_swap(
; CHECK-NEXT:    [[TMP1:%.*]] = sub i5 [[Y:%.*]], [[X:%.*]]
; CHECK-NEXT:    [[ADD1:%.*]] = select i1 [[B:%.*]], i5 [[TMP1]], i5 -1
; CHECK-NEXT:    ret i5 [[ADD1]]
;
  %negx = sub i5 0, %x
  %noty = xor i5 %y, -1
  %sel = select i1 %b, i5 %negx, i5 %noty
  %add = add i5 %sel, %y
  ret i5 %add
}

define i5 @select_negate_not_swap_commute(i1 %b, i5 %x, i5 %p) {
; CHECK-LABEL: @select_negate_not_swap_commute(
; CHECK-NEXT:    [[Y:%.*]] = mul i5 [[P:%.*]], [[P]]
; CHECK-NEXT:    [[TMP1:%.*]] = sub i5 [[Y]], [[X:%.*]]
; CHECK-NEXT:    [[ADD1:%.*]] = select i1 [[B:%.*]], i5 [[TMP1]], i5 -1
; CHECK-NEXT:    ret i5 [[ADD1]]
;
  %y = mul i5 %p, %p ; thwart complexity-based canonicalization
  %negx = sub i5 0, %x
  %noty = xor i5 %y, -1
  %sel = select i1 %b, i5 %negx, i5 %noty
  %add = add i5 %y, %sel
  ret i5 %add
}

define i32 @add_select_sub_both_arms_simplify(i1 %b, i32 %a) {
; CHECK-LABEL: @add_select_sub_both_arms_simplify(
; CHECK-NEXT:    [[ADD:%.*]] = select i1 [[B:%.*]], i32 [[A:%.*]], i32 99
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %sub = sub i32 99, %a
  %sel = select i1 %b, i32 0, i32 %sub
  %add = add i32 %sel, %a
  ret i32 %add
}

define <2 x i8> @add_select_sub_both_arms_simplify_swap(<2 x i1> %b, <2 x i8> %a) {
; CHECK-LABEL: @add_select_sub_both_arms_simplify_swap(
; CHECK-NEXT:    [[ADD:%.*]] = select <2 x i1> [[B:%.*]], <2 x i8> <i8 42, i8 99>, <2 x i8> [[A:%.*]]
; CHECK-NEXT:    ret <2 x i8> [[ADD]]
;
  %sub = sub <2 x i8> <i8 42, i8 99>, %a
  %sel = select <2 x i1> %b, <2 x i8> %sub, <2 x i8> zeroinitializer
  %add = add <2 x i8> %sel, %a
  ret <2 x i8> %add
}

define i8 @add_select_sub_both_arms_simplify_use1(i1 %b, i8 %a) {
; CHECK-LABEL: @add_select_sub_both_arms_simplify_use1(
; CHECK-NEXT:    [[SUB:%.*]] = sub i8 42, [[A:%.*]]
; CHECK-NEXT:    call void @use(i8 [[SUB]])
; CHECK-NEXT:    [[ADD:%.*]] = select i1 [[B:%.*]], i8 [[A]], i8 42
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %sub = sub i8 42, %a
  call void @use(i8 %sub)
  %sel = select i1 %b, i8 0, i8 %sub
  %add = add i8 %sel, %a
  ret i8 %add
}

define i8 @add_select_sub_both_arms_simplify_use2(i1 %b, i8 %a) {
; CHECK-LABEL: @add_select_sub_both_arms_simplify_use2(
; CHECK-NEXT:    [[SUB:%.*]] = sub i8 42, [[A:%.*]]
; CHECK-NEXT:    [[SEL:%.*]] = select i1 [[B:%.*]], i8 0, i8 [[SUB]]
; CHECK-NEXT:    call void @use(i8 [[SEL]])
; CHECK-NEXT:    [[ADD:%.*]] = add i8 [[SEL]], [[A]]
; CHECK-NEXT:    ret i8 [[ADD]]
;
  %sub = sub i8 42, %a
  %sel = select i1 %b, i8 0, i8 %sub
  call void @use(i8 %sel)
  %add = add i8 %sel, %a
  ret i8 %add
}

define i5 @demand_low_bits_uses(i8 %x, i8 %y) {
; CHECK-LABEL: @demand_low_bits_uses(
; CHECK-NEXT:    [[TMP1:%.*]] = shl i8 [[X:%.*]], 5
; CHECK-NEXT:    [[A:%.*]] = sub i8 [[Y:%.*]], [[TMP1]]
; CHECK-NEXT:    call void @use(i8 [[A]])
; CHECK-NEXT:    [[R:%.*]] = trunc i8 [[Y]] to i5
; CHECK-NEXT:    ret i5 [[R]]
;
  %m = mul i8 %x, -32 ; 0xE0
  %a = add i8 %m, %y
  call void @use(i8 %a)
  %r = trunc i8 %a to i5
  ret i5 %r
}

; negative test - demands one more bit

define i6 @demand_low_bits_uses_extra_bit(i8 %x, i8 %y) {
; CHECK-LABEL: @demand_low_bits_uses_extra_bit(
; CHECK-NEXT:    [[TMP1:%.*]] = shl i8 [[X:%.*]], 5
; CHECK-NEXT:    [[A:%.*]] = sub i8 [[Y:%.*]], [[TMP1]]
; CHECK-NEXT:    call void @use(i8 [[A]])
; CHECK-NEXT:    [[R:%.*]] = trunc i8 [[A]] to i6
; CHECK-NEXT:    ret i6 [[R]]
;
  %m = mul i8 %x, -32 ; 0xE0
  %a = add i8 %m, %y
  call void @use(i8 %a)
  %r = trunc i8 %a to i6
  ret i6 %r
}

define i8 @demand_low_bits_uses_commute(i8 %x, i8 %p, i8 %z) {
; CHECK-LABEL: @demand_low_bits_uses_commute(
; CHECK-NEXT:    [[Y:%.*]] = mul i8 [[P:%.*]], [[P]]
; CHECK-NEXT:    [[M:%.*]] = and i8 [[X:%.*]], -64
; CHECK-NEXT:    [[A:%.*]] = add i8 [[Y]], [[M]]
; CHECK-NEXT:    call void @use(i8 [[A]])
; CHECK-NEXT:    [[S:%.*]] = sub i8 [[Y]], [[Z:%.*]]
; CHECK-NEXT:    [[R:%.*]] = shl i8 [[S]], 2
; CHECK-NEXT:    ret i8 [[R]]
;
  %y = mul i8 %p, %p ; thwart complexity-based canonicalization
  %m = and i8 %x, -64 ; 0xC0
  %a = add i8 %y, %m
  call void @use(i8 %a)
  %s = sub i8 %a, %z
  %r = shl i8 %s, 2
  ret i8 %r
}

; negative test - demands one more bit

define i8 @demand_low_bits_uses_commute_extra_bit(i8 %x, i8 %p, i8 %z) {
; CHECK-LABEL: @demand_low_bits_uses_commute_extra_bit(
; CHECK-NEXT:    [[Y:%.*]] = mul i8 [[P:%.*]], [[P]]
; CHECK-NEXT:    [[M:%.*]] = and i8 [[X:%.*]], -64
; CHECK-NEXT:    [[A:%.*]] = add i8 [[Y]], [[M]]
; CHECK-NEXT:    call void @use(i8 [[A]])
; CHECK-NEXT:    [[S:%.*]] = sub i8 [[A]], [[Z:%.*]]
; CHECK-NEXT:    [[R:%.*]] = shl i8 [[S]], 1
; CHECK-NEXT:    ret i8 [[R]]
;
  %y = mul i8 %p, %p ; thwart complexity-based canonicalization
  %m = and i8 %x, -64 ; 0xC0
  %a = add i8 %y, %m
  call void @use(i8 %a)
  %s = sub i8 %a, %z
  %r = shl i8 %s, 1
  ret i8 %r
}

define { i64, i64 } @PR57576(i64 noundef %x, i64 noundef %y, i64 noundef %z, i64 noundef %w) {
; CHECK-LABEL: @PR57576(
; CHECK-NEXT:    [[ZX:%.*]] = zext i64 [[X:%.*]] to i128
; CHECK-NEXT:    [[ZY:%.*]] = zext i64 [[Y:%.*]] to i128
; CHECK-NEXT:    [[ZZ:%.*]] = zext i64 [[Z:%.*]] to i128
; CHECK-NEXT:    [[SHY:%.*]] = shl nuw i128 [[ZY]], 64
; CHECK-NEXT:    [[XY:%.*]] = or disjoint i128 [[SHY]], [[ZX]]
; CHECK-NEXT:    [[SUB:%.*]] = sub i128 [[XY]], [[ZZ]]
; CHECK-NEXT:    [[T:%.*]] = trunc i128 [[SUB]] to i64
; CHECK-NEXT:    [[TMP1:%.*]] = lshr i128 [[SUB]], 64
; CHECK-NEXT:    [[DOTTR:%.*]] = trunc i128 [[TMP1]] to i64
; CHECK-NEXT:    [[DOTNARROW:%.*]] = sub i64 [[DOTTR]], [[W:%.*]]
; CHECK-NEXT:    [[R1:%.*]] = insertvalue { i64, i64 } poison, i64 [[T]], 0
; CHECK-NEXT:    [[R2:%.*]] = insertvalue { i64, i64 } [[R1]], i64 [[DOTNARROW]], 1
; CHECK-NEXT:    ret { i64, i64 } [[R2]]
;
  %zx = zext i64 %x to i128
  %zy = zext i64 %y to i128
  %zw = zext i64 %w to i128
  %zz = zext i64 %z to i128
  %shy = shl nuw i128 %zy, 64
  %mw = mul i128 %zw, -18446744073709551616
  %xy = or i128 %shy, %zx
  %sub = sub i128 %xy, %zz
  %add = add i128 %sub, %mw
  %t = trunc i128 %add to i64
  %h = lshr i128 %add, 64
  %t2 = trunc i128 %h to i64
  %r1 = insertvalue { i64, i64 } poison, i64 %t, 0
  %r2 = insertvalue { i64, i64 } %r1, i64 %t2, 1
  ret { i64, i64 } %r2
}

define i8 @mul_negpow2(i8 %x, i8 %y) {
; CHECK-LABEL: @mul_negpow2(
; CHECK-NEXT:    [[TMP1:%.*]] = shl i8 [[X:%.*]], 1
; CHECK-NEXT:    [[A:%.*]] = sub i8 [[Y:%.*]], [[TMP1]]
; CHECK-NEXT:    ret i8 [[A]]
;
  %m = mul i8 %x, -2
  %a = add i8 %m, %y
  ret i8 %a
}

define <2 x i8> @mul_negpow2_commute_vec(<2 x i8> %x, <2 x i8> %p) {
; CHECK-LABEL: @mul_negpow2_commute_vec(
; CHECK-NEXT:    [[Y:%.*]] = mul <2 x i8> [[P:%.*]], [[P]]
; CHECK-NEXT:    [[TMP1:%.*]] = shl <2 x i8> [[X:%.*]], <i8 3, i8 3>
; CHECK-NEXT:    [[A:%.*]] = sub <2 x i8> [[Y]], [[TMP1]]
; CHECK-NEXT:    ret <2 x i8> [[A]]
;
  %y = mul <2 x i8> %p, %p ; thwart complexity-based canonicalization
  %m = mul <2 x i8> %x, <i8 -8, i8 -8>
  %a = add <2 x i8> %y, %m
  ret <2 x i8> %a
}

; negative test - extra use

define i8 @mul_negpow2_use(i8 %x) {
; CHECK-LABEL: @mul_negpow2_use(
; CHECK-NEXT:    [[M:%.*]] = mul i8 [[X:%.*]], -2
; CHECK-NEXT:    call void @use(i8 [[M]])
; CHECK-NEXT:    [[A:%.*]] = add i8 [[M]], 42
; CHECK-NEXT:    ret i8 [[A]]
;
  %m = mul i8 %x, -2
  call void @use(i8 %m)
  %a = add i8 %m, 42
  ret i8 %a
}

; negative test - not negative-power-of-2 multiplier

define i8 @mul_not_negpow2(i8 %x) {
; CHECK-LABEL: @mul_not_negpow2(
; CHECK-NEXT:    [[M:%.*]] = mul i8 [[X:%.*]], -3
; CHECK-NEXT:    [[A:%.*]] = add i8 [[M]], 42
; CHECK-NEXT:    ret i8 [[A]]
;
  %m = mul i8 %x, -3
  %a = add i8 %m, 42
  ret i8 %a
}

define i16 @add_sub_zext(i8 %x, i8 %y) {
; CHECK-LABEL: @add_sub_zext(
; CHECK-NEXT:    [[TMP1:%.*]] = zext i8 [[Y:%.*]] to i16
; CHECK-NEXT:    ret i16 [[TMP1]]
;
  %1 = sub nuw i8 %y, %x
  %2 = zext i8 %1 to i16
  %3 = zext i8 %x to i16
  %4 = add i16 %2, %3
  ret i16 %4
}

define i16 @add_commute_sub_zext(i8 %x, i8 %y) {
; CHECK-LABEL: @add_commute_sub_zext(
; CHECK-NEXT:    [[TMP1:%.*]] = zext i8 [[Y:%.*]] to i16
; CHECK-NEXT:    ret i16 [[TMP1]]
;
  %1 = sub nuw i8 %y, %x
  %2 = zext i8 %1 to i16
  %3 = zext i8 %x to i16
  %4 = add i16 %3, %2
  ret i16 %4
}

define <2 x i8> @add_sub_2xi5_zext(<2 x i5> %x, <2 x i5> %y) {
; CHECK-LABEL: @add_sub_2xi5_zext(
; CHECK-NEXT:    [[TMP1:%.*]] = zext <2 x i5> [[Y:%.*]] to <2 x i8>
; CHECK-NEXT:    ret <2 x i8> [[TMP1]]
;
  %1 = sub nuw <2 x i5> %y, %x
  %2 = zext <2 x i5> %1 to <2 x i8>
  %3 = zext <2 x i5> %x to <2 x i8>
  %4 = add <2 x i8> %3, %2
  ret <2 x i8> %4
}


define i3 @add_commute_sub_i2_zext_i3(i2 %x, i2 %y) {
; CHECK-LABEL: @add_commute_sub_i2_zext_i3(
; CHECK-NEXT:    [[TMP1:%.*]] = zext i2 [[Y:%.*]] to i3
; CHECK-NEXT:    ret i3 [[TMP1]]
;
  %1 = sub nuw i2 %y, %x
  %2 = zext i2 %1 to i3
  %3 = zext i2 %x to i3
  %4 = add i3 %3, %2
  ret i3 %4
}

define i16 @add_sub_use_zext(i8 %x, i8 %y) {
; CHECK-LABEL: @add_sub_use_zext(
; CHECK-NEXT:    [[TMP1:%.*]] = sub nuw i8 [[Y:%.*]], [[X:%.*]]
; CHECK-NEXT:    call void @use(i8 [[TMP1]])
; CHECK-NEXT:    [[TMP2:%.*]] = zext i8 [[Y]] to i16
; CHECK-NEXT:    ret i16 [[TMP2]]
;
  %1 = sub nuw i8 %y, %x
  call void @use(i8 %1)
  %2 = zext i8 %1 to i16
  %3 = zext i8 %x to i16
  %4 = add i16 %2, %3
  ret i16 %4
}

; Negative test: x - y + x  != y
define i16 @add_sub_commute_zext(i8 %x, i8 %y) {
; CHECK-LABEL: @add_sub_commute_zext(
; CHECK-NEXT:    [[TMP1:%.*]] = sub nuw i8 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    [[TMP2:%.*]] = zext i8 [[TMP1]] to i16
; CHECK-NEXT:    [[TMP3:%.*]] = zext i8 [[X]] to i16
; CHECK-NEXT:    [[TMP4:%.*]] = add nuw nsw i16 [[TMP2]], [[TMP3]]
; CHECK-NEXT:    ret i16 [[TMP4]]
;
  %1 = sub nuw i8 %x, %y
  %2 = zext i8 %1 to i16
  %3 = zext i8 %x to i16
  %4 = add i16 %2, %3
  ret i16 %4
}

; Negative test: no nuw flags
define i16 @add_no_nuw_sub_zext(i8 %x, i8 %y) {
; CHECK-LABEL: @add_no_nuw_sub_zext(
; CHECK-NEXT:    [[TMP1:%.*]] = sub i8 [[Y:%.*]], [[X:%.*]]
; CHECK-NEXT:    [[TMP2:%.*]] = zext i8 [[TMP1]] to i16
; CHECK-NEXT:    [[TMP3:%.*]] = zext i8 [[X]] to i16
; CHECK-NEXT:    [[TMP4:%.*]] = add nuw nsw i16 [[TMP3]], [[TMP2]]
; CHECK-NEXT:    ret i16 [[TMP4]]
;
  %1 = sub i8 %y, %x
  %2 = zext i8 %1 to i16
  %3 = zext i8 %x to i16
  %4 = add i16 %3, %2
  ret i16 %4
}

define i16 @add_no_nuw_sub_commute_zext(i8 %x, i8 %y) {
; CHECK-LABEL: @add_no_nuw_sub_commute_zext(
; CHECK-NEXT:    [[TMP1:%.*]] = sub i8 [[X:%.*]], [[Y:%.*]]
; CHECK-NEXT:    [[TMP2:%.*]] = zext i8 [[TMP1]] to i16
; CHECK-NEXT:    [[TMP3:%.*]] = zext i8 [[X]] to i16
; CHECK-NEXT:    [[TMP4:%.*]] = add nuw nsw i16 [[TMP3]], [[TMP2]]
; CHECK-NEXT:    ret i16 [[TMP4]]
;
  %1 = sub i8 %x, %y
  %2 = zext i8 %1 to i16
  %3 = zext i8 %x to i16
  %4 = add i16 %3, %2
  ret i16 %4
}

define i16 @add_sub_zext_constant(i8 %x) {
; CHECK-LABEL: @add_sub_zext_constant(
; CHECK-NEXT:    ret i16 254
;
  %1 = sub nuw i8 254, %x
  %2 = zext i8 %1 to i16
  %3 = zext i8 %x to i16
  %4 = add i16 %2, %3
  ret i16 %4
}

define <vscale x 1 x i32> @add_to_or_scalable(<vscale x 1 x i32> %in) {
; CHECK-LABEL: @add_to_or_scalable(
; CHECK-NEXT:    [[SHL:%.*]] = shl <vscale x 1 x i32> [[IN:%.*]], shufflevector (<vscale x 1 x i32> insertelement (<vscale x 1 x i32> poison, i32 1, i32 0), <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT:    [[ADD:%.*]] = or disjoint <vscale x 1 x i32> [[SHL]], shufflevector (<vscale x 1 x i32> insertelement (<vscale x 1 x i32> poison, i32 1, i32 0), <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer)
; CHECK-NEXT:    ret <vscale x 1 x i32> [[ADD]]
;
  %shl = shl <vscale x 1 x i32> %in, shufflevector (<vscale x 1 x i32> insertelement (<vscale x 1 x i32> poison, i32 1, i32 0), <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer)
  %add = add <vscale x 1 x i32> %shl, shufflevector (<vscale x 1 x i32> insertelement (<vscale x 1 x i32> poison, i32 1, i32 0), <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer)
  ret <vscale x 1 x i32> %add
}

define i5 @zext_zext_not(i3 noundef %x) {
; CHECK-LABEL: @zext_zext_not(
; CHECK-NEXT:    ret i5 7
;
  %zx = zext i3 %x to i5
  %notx = xor i3 %x, -1
  %znotx = zext i3 %notx to i5
  %r = add i5 %zx, %znotx
  ret i5 %r
}

define <2 x i5> @zext_zext_not_commute(<2 x i3> noundef %x) {
; CHECK-LABEL: @zext_zext_not_commute(
; CHECK-NEXT:    ret <2 x i5> <i5 7, i5 7>
;
  %zx = zext <2 x i3> %x to <2 x i5>
  %notx = xor <2 x i3> %x, <i3 -1, i3 poison>
  %znotx = zext <2 x i3> %notx to <2 x i5>
  %r = add <2 x i5> %znotx, %zx
  ret <2 x i5> %r
}

define i9 @sext_sext_not(i3 noundef %x) {
; CHECK-LABEL: @sext_sext_not(
; CHECK-NEXT:    ret i9 -1
;
  %sx = sext i3 %x to i9
  %notx = xor i3 %x, -1
  %snotx = sext i3 %notx to i9
  %r = add i9 %sx, %snotx
  ret i9 %r
}

define i8 @sext_sext_not_commute(i3 noundef %x) {
; CHECK-LABEL: @sext_sext_not_commute(
; CHECK-NEXT:    [[SX:%.*]] = sext i3 [[X:%.*]] to i8
; CHECK-NEXT:    call void @use(i8 [[SX]])
; CHECK-NEXT:    ret i8 -1
;

  %sx = sext i3 %x to i8
  call void @use(i8 %sx)
  %notx = xor i3 %x, -1
  %snotx = sext i3 %notx to i8
  %r = add i8 %snotx, %sx
  ret i8 %r
}

define i5 @zext_sext_not(i4 noundef %x) {
; CHECK-LABEL: @zext_sext_not(
; CHECK-NEXT:    [[ZX:%.*]] = zext i4 [[X:%.*]] to i5
; CHECK-NEXT:    [[NOTX:%.*]] = xor i4 [[X]], -1
; CHECK-NEXT:    [[SNOTX:%.*]] = sext i4 [[NOTX]] to i5
; CHECK-NEXT:    [[R:%.*]] = or disjoint i5 [[ZX]], [[SNOTX]]
; CHECK-NEXT:    ret i5 [[R]]
;
  %zx = zext i4 %x to i5
  %notx = xor i4 %x, -1
  %snotx = sext i4 %notx to i5
  %r = add i5 %zx, %snotx
  ret i5 %r
}

define i8 @zext_sext_not_commute(i4 noundef %x) {
; CHECK-LABEL: @zext_sext_not_commute(
; CHECK-NEXT:    [[ZX:%.*]] = zext i4 [[X:%.*]] to i8
; CHECK-NEXT:    call void @use(i8 [[ZX]])
; CHECK-NEXT:    [[NOTX:%.*]] = xor i4 [[X]], -1
; CHECK-NEXT:    [[SNOTX:%.*]] = sext i4 [[NOTX]] to i8
; CHECK-NEXT:    call void @use(i8 [[SNOTX]])
; CHECK-NEXT:    [[R:%.*]] = or disjoint i8 [[SNOTX]], [[ZX]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %zx = zext i4 %x to i8
  call void @use(i8 %zx)
  %notx = xor i4 %x, -1
  %snotx = sext i4 %notx to i8
  call void @use(i8 %snotx)
  %r = add i8 %snotx, %zx
  ret i8 %r
}

define i9 @sext_zext_not(i4 noundef %x) {
; CHECK-LABEL: @sext_zext_not(
; CHECK-NEXT:    [[SX:%.*]] = sext i4 [[X:%.*]] to i9
; CHECK-NEXT:    [[NOTX:%.*]] = xor i4 [[X]], -1
; CHECK-NEXT:    [[ZNOTX:%.*]] = zext i4 [[NOTX]] to i9
; CHECK-NEXT:    [[R:%.*]] = or disjoint i9 [[SX]], [[ZNOTX]]
; CHECK-NEXT:    ret i9 [[R]]
;
  %sx = sext i4 %x to i9
  %notx = xor i4 %x, -1
  %znotx = zext i4 %notx to i9
  %r = add i9 %sx, %znotx
  ret i9 %r
}

define i9 @sext_zext_not_commute(i4 noundef %x) {
; CHECK-LABEL: @sext_zext_not_commute(
; CHECK-NEXT:    [[SX:%.*]] = sext i4 [[X:%.*]] to i9
; CHECK-NEXT:    [[NOTX:%.*]] = xor i4 [[X]], -1
; CHECK-NEXT:    [[ZNOTX:%.*]] = zext i4 [[NOTX]] to i9
; CHECK-NEXT:    [[R:%.*]] = or disjoint i9 [[ZNOTX]], [[SX]]
; CHECK-NEXT:    ret i9 [[R]]
;
  %sx = sext i4 %x to i9
  %notx = xor i4 %x, -1
  %znotx = zext i4 %notx to i9
  %r = add i9 %znotx, %sx
  ret i9 %r
}

; PR57741

define i32 @floor_sdiv(i32 %x) {
; CHECK-LABEL: @floor_sdiv(
; CHECK-NEXT:    [[R:%.*]] = ashr i32 [[X:%.*]], 2
; CHECK-NEXT:    ret i32 [[R]]
;
  %d = sdiv i32 %x, 4
  %a = and i32 %x, -2147483645
  %i = icmp ugt i32 %a, -2147483648
  %s = sext i1 %i to i32
  %r = add i32 %d, %s
  ret i32 %r
}

define i8 @floor_sdiv_by_2(i8 %x) {
; CHECK-LABEL: @floor_sdiv_by_2(
; CHECK-NEXT:    [[RV:%.*]] = ashr i8 [[X:%.*]], 1
; CHECK-NEXT:    ret i8 [[RV]]
;
  %div = sdiv i8 %x, 2
  %and = and i8 %x, -127
  %icmp = icmp eq i8 %and, -127
  %sext = sext i1 %icmp to i8
  %rv = add nsw i8 %div, %sext
  ret i8 %rv
}

define i8 @floor_sdiv_by_2_wrong_mask(i8 %x) {
; CHECK-LABEL: @floor_sdiv_by_2_wrong_mask(
; CHECK-NEXT:    [[DIV:%.*]] = sdiv i8 [[X:%.*]], 2
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], 127
; CHECK-NEXT:    [[ICMP:%.*]] = icmp eq i8 [[AND]], 127
; CHECK-NEXT:    [[SEXT:%.*]] = sext i1 [[ICMP]] to i8
; CHECK-NEXT:    [[RV:%.*]] = add nsw i8 [[DIV]], [[SEXT]]
; CHECK-NEXT:    ret i8 [[RV]]
;
  %div = sdiv i8 %x, 2
  %and = and i8 %x, 127
  %icmp = icmp eq i8 %and, 127
  %sext = sext i1 %icmp to i8
  %rv = add nsw i8 %div, %sext
  ret i8 %rv
}

define i8 @floor_sdiv_by_2_wrong_constant(i8 %x) {
; CHECK-LABEL: @floor_sdiv_by_2_wrong_constant(
; CHECK-NEXT:    [[DIV:%.*]] = sdiv i8 [[X:%.*]], 4
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], -125
; CHECK-NEXT:    [[ICMP:%.*]] = icmp eq i8 [[AND]], -125
; CHECK-NEXT:    [[SEXT:%.*]] = sext i1 [[ICMP]] to i8
; CHECK-NEXT:    [[RV:%.*]] = add nsw i8 [[DIV]], [[SEXT]]
; CHECK-NEXT:    ret i8 [[RV]]
;
  %div = sdiv i8 %x, 4
  %and = and i8 %x, -125
  %icmp = icmp eq i8 %and, -125
  %sext = sext i1 %icmp to i8
  %rv = add nsw i8 %div, %sext
  ret i8 %rv
}

define i8 @floor_sdiv_by_2_wrong_cast(i8 %x) {
; CHECK-LABEL: @floor_sdiv_by_2_wrong_cast(
; CHECK-NEXT:    [[DIV:%.*]] = sdiv i8 [[X:%.*]], 2
; CHECK-NEXT:    [[AND:%.*]] = and i8 [[X]], -127
; CHECK-NEXT:    [[ICMP:%.*]] = icmp eq i8 [[AND]], -127
; CHECK-NEXT:    [[SEXT:%.*]] = zext i1 [[ICMP]] to i8
; CHECK-NEXT:    [[RV:%.*]] = add nsw i8 [[DIV]], [[SEXT]]
; CHECK-NEXT:    ret i8 [[RV]]
;
  %div = sdiv i8 %x, 2
  %and = and i8 %x, -127
  %icmp = icmp eq i8 %and, -127
  %sext = zext i1 %icmp to i8
  %rv = add nsw i8 %div, %sext
  ret i8 %rv
}

; vectors work too and commute is handled by complexity-based canonicalization

define <2 x i32> @floor_sdiv_vec_commute(<2 x i32> %x) {
; CHECK-LABEL: @floor_sdiv_vec_commute(
; CHECK-NEXT:    [[R:%.*]] = ashr <2 x i32> [[X:%.*]], <i32 2, i32 2>
; CHECK-NEXT:    ret <2 x i32> [[R]]
;
  %d = sdiv <2 x i32> %x, <i32 4, i32 4>
  %a = and <2 x i32> %x, <i32 -2147483645, i32 -2147483645>
  %i = icmp ugt <2 x i32> %a, <i32 -2147483648, i32 -2147483648>
  %s = sext <2 x i1> %i to <2 x i32>
  %r = add <2 x i32> %s, %d
  ret <2 x i32> %r
}

; extra uses are ok

define i8 @floor_sdiv_uses(i8 %x) {
; CHECK-LABEL: @floor_sdiv_uses(
; CHECK-NEXT:    [[D:%.*]] = sdiv i8 [[X:%.*]], 16
; CHECK-NEXT:    call void @use(i8 [[D]])
; CHECK-NEXT:    [[A:%.*]] = and i8 [[X]], -113
; CHECK-NEXT:    call void @use(i8 [[A]])
; CHECK-NEXT:    [[I:%.*]] = icmp ugt i8 [[A]], -128
; CHECK-NEXT:    [[S:%.*]] = sext i1 [[I]] to i8
; CHECK-NEXT:    call void @use(i8 [[S]])
; CHECK-NEXT:    [[R:%.*]] = ashr i8 [[X]], 4
; CHECK-NEXT:    ret i8 [[R]]
;
  %d = sdiv i8 %x, 16
  call void @use(i8 %d)
  %a = and i8 %x, 143 ; 128 + 15
  call void @use(i8 %a)
  %i = icmp ugt i8 %a, 128
  %s = sext i1 %i to i8
  call void @use(i8 %s)
  %r = add i8 %d, %s
  ret i8 %r
}

; negative test

define i32 @floor_sdiv_wrong_div(i32 %x) {
; CHECK-LABEL: @floor_sdiv_wrong_div(
; CHECK-NEXT:    [[D:%.*]] = sdiv i32 [[X:%.*]], 8
; CHECK-NEXT:    [[A:%.*]] = and i32 [[X]], -2147483645
; CHECK-NEXT:    [[I:%.*]] = icmp ugt i32 [[A]], -2147483648
; CHECK-NEXT:    [[S:%.*]] = sext i1 [[I]] to i32
; CHECK-NEXT:    [[R:%.*]] = add nsw i32 [[D]], [[S]]
; CHECK-NEXT:    ret i32 [[R]]
;
  %d = sdiv i32 %x, 8
  %a = and i32 %x, -2147483645
  %i = icmp ugt i32 %a, -2147483648
  %s = sext i1 %i to i32
  %r = add i32 %d, %s
  ret i32 %r
}

; negative test

define i32 @floor_sdiv_wrong_mask(i32 %x) {
; CHECK-LABEL: @floor_sdiv_wrong_mask(
; CHECK-NEXT:    [[D:%.*]] = sdiv i32 [[X:%.*]], 4
; CHECK-NEXT:    [[A:%.*]] = and i32 [[X]], -2147483644
; CHECK-NEXT:    [[I:%.*]] = icmp ugt i32 [[A]], -2147483648
; CHECK-NEXT:    [[S:%.*]] = sext i1 [[I]] to i32
; CHECK-NEXT:    [[R:%.*]] = add nsw i32 [[D]], [[S]]
; CHECK-NEXT:    ret i32 [[R]]
;
  %d = sdiv i32 %x, 4
  %a = and i32 %x, -2147483644
  %i = icmp ugt i32 %a, -2147483648
  %s = sext i1 %i to i32
  %r = add i32 %d, %s
  ret i32 %r
}

; negative test

define i32 @floor_sdiv_wrong_cmp(i32 %x) {
; CHECK-LABEL: @floor_sdiv_wrong_cmp(
; CHECK-NEXT:    [[D:%.*]] = sdiv i32 [[X:%.*]], 4
; CHECK-NEXT:    [[A:%.*]] = and i32 [[X]], -2147483646
; CHECK-NEXT:    [[I:%.*]] = icmp eq i32 [[A]], -2147483646
; CHECK-NEXT:    [[S:%.*]] = sext i1 [[I]] to i32
; CHECK-NEXT:    [[R:%.*]] = add nsw i32 [[D]], [[S]]
; CHECK-NEXT:    ret i32 [[R]]
;
  %d = sdiv i32 %x, 4
  %a = and i32 %x, -2147483645
  %i = icmp ugt i32 %a, -2147483647
  %s = sext i1 %i to i32
  %r = add i32 %d, %s
  ret i32 %r
}

; negative test

define i32 @floor_sdiv_wrong_ext(i32 %x) {
; CHECK-LABEL: @floor_sdiv_wrong_ext(
; CHECK-NEXT:    [[D:%.*]] = sdiv i32 [[X:%.*]], 4
; CHECK-NEXT:    [[A:%.*]] = and i32 [[X]], -2147483645
; CHECK-NEXT:    [[I:%.*]] = icmp ugt i32 [[A]], -2147483648
; CHECK-NEXT:    [[S:%.*]] = zext i1 [[I]] to i32
; CHECK-NEXT:    [[R:%.*]] = add nsw i32 [[D]], [[S]]
; CHECK-NEXT:    ret i32 [[R]]
;
  %d = sdiv i32 %x, 4
  %a = and i32 %x, -2147483645
  %i = icmp ugt i32 %a, -2147483648
  %s = zext i1 %i to i32
  %r = add i32 %d, %s
  ret i32 %r
}

; negative test

define i32 @floor_sdiv_wrong_op(i32 %x, i32 %y) {
; CHECK-LABEL: @floor_sdiv_wrong_op(
; CHECK-NEXT:    [[D:%.*]] = sdiv i32 [[X:%.*]], 4
; CHECK-NEXT:    [[A:%.*]] = and i32 [[Y:%.*]], -2147483645
; CHECK-NEXT:    [[I:%.*]] = icmp ugt i32 [[A]], -2147483648
; CHECK-NEXT:    [[S:%.*]] = zext i1 [[I]] to i32
; CHECK-NEXT:    [[R:%.*]] = add nsw i32 [[D]], [[S]]
; CHECK-NEXT:    ret i32 [[R]]
;
  %d = sdiv i32 %x, 4
  %a = and i32 %y, -2147483645
  %i = icmp ugt i32 %a, -2147483648
  %s = zext i1 %i to i32
  %r = add i32 %d, %s
  ret i32 %r
}

; (X s>> (BW - 1)) + (zext (X s> 0)) --> (X s>> (BW - 1)) | (zext (X != 0))

define i8 @signum_i8_i8(i8 %x) {
; CHECK-LABEL: @signum_i8_i8(
; CHECK-NEXT:    [[SIGNBIT:%.*]] = ashr i8 [[X:%.*]], 7
; CHECK-NEXT:    [[ISNOTNULL:%.*]] = icmp ne i8 [[X]], 0
; CHECK-NEXT:    [[ISNOTNULL_ZEXT:%.*]] = zext i1 [[ISNOTNULL]] to i8
; CHECK-NEXT:    [[R:%.*]] = or i8 [[SIGNBIT]], [[ISNOTNULL_ZEXT]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %sgt0 = icmp sgt i8 %x, 0
  %zgt0 = zext i1 %sgt0 to i8
  %signbit = ashr i8 %x, 7
  %r = add i8 %zgt0, %signbit
  ret i8 %r
}

; extra use of shift is ok

define i8 @signum_i8_i8_use1(i8 %x) {
; CHECK-LABEL: @signum_i8_i8_use1(
; CHECK-NEXT:    [[SIGNBIT:%.*]] = ashr i8 [[X:%.*]], 7
; CHECK-NEXT:    call void @use(i8 [[SIGNBIT]])
; CHECK-NEXT:    [[ISNOTNULL:%.*]] = icmp ne i8 [[X]], 0
; CHECK-NEXT:    [[ISNOTNULL_ZEXT:%.*]] = zext i1 [[ISNOTNULL]] to i8
; CHECK-NEXT:    [[R:%.*]] = or i8 [[SIGNBIT]], [[ISNOTNULL_ZEXT]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %sgt0 = icmp sgt i8 %x, 0
  %zgt0 = zext i1 %sgt0 to i8
  %signbit = ashr i8 %x, 7
  call void @use(i8 %signbit)
  %r = add i8 %zgt0, %signbit
  ret i8 %r
}

; negative test

define i8 @signum_i8_i8_use2(i8 %x) {
; CHECK-LABEL: @signum_i8_i8_use2(
; CHECK-NEXT:    [[SGT0:%.*]] = icmp sgt i8 [[X:%.*]], 0
; CHECK-NEXT:    [[ZGT0:%.*]] = zext i1 [[SGT0]] to i8
; CHECK-NEXT:    call void @use(i8 [[ZGT0]])
; CHECK-NEXT:    [[SIGNBIT:%.*]] = ashr i8 [[X]], 7
; CHECK-NEXT:    [[R:%.*]] = add nsw i8 [[SIGNBIT]], [[ZGT0]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %sgt0 = icmp sgt i8 %x, 0
  %zgt0 = zext i1 %sgt0 to i8
  call void @use(i8 %zgt0)
  %signbit = ashr i8 %x, 7
  %r = add i8 %zgt0, %signbit
  ret i8 %r
}

; negative test

define i8 @signum_i8_i8_use3(i8 %x) {
; CHECK-LABEL: @signum_i8_i8_use3(
; CHECK-NEXT:    [[SGT0:%.*]] = icmp sgt i8 [[X:%.*]], 0
; CHECK-NEXT:    call void @use_i1(i1 [[SGT0]])
; CHECK-NEXT:    [[ZGT0:%.*]] = zext i1 [[SGT0]] to i8
; CHECK-NEXT:    [[SIGNBIT:%.*]] = ashr i8 [[X]], 7
; CHECK-NEXT:    [[R:%.*]] = add nsw i8 [[SIGNBIT]], [[ZGT0]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %sgt0 = icmp sgt i8 %x, 0
  call void @use_i1(i1 %sgt0)
  %zgt0 = zext i1 %sgt0 to i8
  %signbit = ashr i8 %x, 7
  %r = add i8 %zgt0, %signbit
  ret i8 %r
}

; poison/undef is ok to propagate in shift amount
; complexity canonicalization guarantees that shift is op0 of add

define <2 x i5> @signum_v2i5_v2i5(<2 x i5> %x) {
; CHECK-LABEL: @signum_v2i5_v2i5(
; CHECK-NEXT:    [[SIGNBIT:%.*]] = ashr <2 x i5> [[X:%.*]], <i5 4, i5 poison>
; CHECK-NEXT:    [[ISNOTNULL:%.*]] = icmp ne <2 x i5> [[X]], zeroinitializer
; CHECK-NEXT:    [[ISNOTNULL_ZEXT:%.*]] = zext <2 x i1> [[ISNOTNULL]] to <2 x i5>
; CHECK-NEXT:    [[R:%.*]] = or <2 x i5> [[SIGNBIT]], [[ISNOTNULL_ZEXT]]
; CHECK-NEXT:    ret <2 x i5> [[R]]
;
  %sgt0 = icmp sgt <2 x i5> %x, zeroinitializer
  %zgt0 = zext <2 x i1> %sgt0 to <2 x i5>
  %signbit = ashr <2 x i5> %x, <i5 4, i5 poison>
  %r = add <2 x i5> %signbit, %zgt0
  ret <2 x i5> %r
}

; negative test

define i8 @signum_i8_i8_wrong_sh_amt(i8 %x) {
; CHECK-LABEL: @signum_i8_i8_wrong_sh_amt(
; CHECK-NEXT:    [[SGT0:%.*]] = icmp sgt i8 [[X:%.*]], 0
; CHECK-NEXT:    [[ZGT0:%.*]] = zext i1 [[SGT0]] to i8
; CHECK-NEXT:    [[SIGNBIT:%.*]] = ashr i8 [[X]], 6
; CHECK-NEXT:    [[R:%.*]] = add nsw i8 [[SIGNBIT]], [[ZGT0]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %sgt0 = icmp sgt i8 %x, 0
  %zgt0 = zext i1 %sgt0 to i8
  %signbit = ashr i8 %x, 6
  %r = add i8 %zgt0, %signbit
  ret i8 %r
}

; negative test

define i8 @signum_i8_i8_wrong_ext(i8 %x) {
; CHECK-LABEL: @signum_i8_i8_wrong_ext(
; CHECK-NEXT:    [[SGT0:%.*]] = icmp sgt i8 [[X:%.*]], 0
; CHECK-NEXT:    [[ZGT0:%.*]] = sext i1 [[SGT0]] to i8
; CHECK-NEXT:    [[SIGNBIT:%.*]] = ashr i8 [[X]], 7
; CHECK-NEXT:    [[R:%.*]] = add nsw i8 [[SIGNBIT]], [[ZGT0]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %sgt0 = icmp sgt i8 %x, 0
  %zgt0 = sext i1 %sgt0 to i8
  %signbit = ashr i8 %x, 7
  %r = add i8 %zgt0, %signbit
  ret i8 %r
}

; negative test

define i8 @signum_i8_i8_wrong_pred(i8 %x) {
; CHECK-LABEL: @signum_i8_i8_wrong_pred(
; CHECK-NEXT:    [[SGT0:%.*]] = icmp sgt i8 [[X:%.*]], -1
; CHECK-NEXT:    [[ZGT0:%.*]] = zext i1 [[SGT0]] to i8
; CHECK-NEXT:    [[SIGNBIT:%.*]] = ashr i8 [[X]], 7
; CHECK-NEXT:    [[R:%.*]] = add nsw i8 [[SIGNBIT]], [[ZGT0]]
; CHECK-NEXT:    ret i8 [[R]]
;
  %sgt0 = icmp sge i8 %x, 0
  %zgt0 = zext i1 %sgt0 to i8
  %signbit = ashr i8 %x, 7
  %r = add i8 %zgt0, %signbit
  ret i8 %r
}

define i32 @dec_zext_add_assume_nonzero(i8 %x) {
; CHECK-LABEL: @dec_zext_add_assume_nonzero(
; CHECK-NEXT:    [[Z:%.*]] = icmp ne i8 [[X:%.*]], 0
; CHECK-NEXT:    call void @llvm.assume(i1 [[Z]])
; CHECK-NEXT:    [[C:%.*]] = zext i8 [[X]] to i32
; CHECK-NEXT:    ret i32 [[C]]
;
  %z = icmp ne i8 %x, 0
  call void @llvm.assume(i1 %z)
  %a = add i8 %x, -1
  %b = zext i8 %a to i32
  %c = add i32 %b, 1
  ret i32 %c
}

define i32 @dec_zext_add_nonzero(i8 %x) {
; CHECK-LABEL: @dec_zext_add_nonzero(
; CHECK-NEXT:    [[O:%.*]] = or i8 [[X:%.*]], 4
; CHECK-NEXT:    [[C:%.*]] = zext i8 [[O]] to i32
; CHECK-NEXT:    ret i32 [[C]]
;
  %o = or i8 %x, 4
  %a = add i8 %o, -1
  %b = zext i8 %a to i32
  %c = add i32 %b, 1
  ret i32 %c
}

define <2 x i32> @dec_zext_add_nonzero_vec(<2 x i8> %x) {
; CHECK-LABEL: @dec_zext_add_nonzero_vec(
; CHECK-NEXT:    [[O:%.*]] = or <2 x i8> [[X:%.*]], <i8 8, i8 8>
; CHECK-NEXT:    [[C:%.*]] = zext <2 x i8> [[O]] to <2 x i32>
; CHECK-NEXT:    ret <2 x i32> [[C]]
;
  %o = or <2 x i8> %x, <i8 8, i8 8>
  %a = add <2 x i8> %o, <i8 -1, i8 -1>
  %b = zext <2 x i8> %a to <2 x i32>
  %c = add <2 x i32> %b, <i32 1, i32 1>
  ret <2 x i32> %c
}

define <2 x i32> @dec_zext_add_nonzero_vec_poison1(<2 x i8> %x) {
; CHECK-LABEL: @dec_zext_add_nonzero_vec_poison1(
; CHECK-NEXT:    [[O:%.*]] = or <2 x i8> [[X:%.*]], <i8 8, i8 8>
; CHECK-NEXT:    [[C:%.*]] = zext <2 x i8> [[O]] to <2 x i32>
; CHECK-NEXT:    ret <2 x i32> [[C]]
;
  %o = or <2 x i8> %x, <i8 8, i8 8>
  %a = add <2 x i8> %o, <i8 -1, i8 poison>
  %b = zext <2 x i8> %a to <2 x i32>
  %c = add <2 x i32> %b, <i32 1, i32 1>
  ret <2 x i32> %c
}

define <2 x i32> @dec_zext_add_nonzero_vec_poison2(<2 x i8> %x) {
; CHECK-LABEL: @dec_zext_add_nonzero_vec_poison2(
; CHECK-NEXT:    [[O:%.*]] = or <2 x i8> [[X:%.*]], <i8 8, i8 8>
; CHECK-NEXT:    [[A:%.*]] = add nsw <2 x i8> [[O]], <i8 -1, i8 -1>
; CHECK-NEXT:    [[B:%.*]] = zext <2 x i8> [[A]] to <2 x i32>
; CHECK-NEXT:    [[C:%.*]] = add nuw nsw <2 x i32> [[B]], <i32 1, i32 poison>
; CHECK-NEXT:    ret <2 x i32> [[C]]
;
  %o = or <2 x i8> %x, <i8 8, i8 8>
  %a = add <2 x i8> %o, <i8 -1, i8 -1>
  %b = zext <2 x i8> %a to <2 x i32>
  %c = add <2 x i32> %b, <i32 1, i32 poison>
  ret <2 x i32> %c
}

define i32 @add_zext_sext_i1(i1 %a) {
; CHECK-LABEL: @add_zext_sext_i1(
; CHECK-NEXT:    ret i32 0
;
  %zext = zext i1 %a to i32
  %sext = sext i1 %a to i32
  %add = add i32 %zext, %sext
  ret i32 %add
}

define i32 @add_sext_zext_i1(i1 %a) {
; CHECK-LABEL: @add_sext_zext_i1(
; CHECK-NEXT:    ret i32 0
;
  %zext = zext i1 %a to i32
  %sext = sext i1 %a to i32
  %add = add i32 %sext, %zext
  ret i32 %add
}

define <2 x i32> @add_zext_sext_i1_vec(<2 x i1> %a) {
; CHECK-LABEL: @add_zext_sext_i1_vec(
; CHECK-NEXT:    ret <2 x i32> zeroinitializer
;
  %zext = zext <2 x i1> %a to <2 x i32>
  %sext = sext <2 x i1> %a to <2 x i32>
  %add = add <2 x i32> %zext, %sext
  ret <2 x i32> %add
}

define i32 @add_zext_zext_i1(i1 %a) {
; CHECK-LABEL: @add_zext_zext_i1(
; CHECK-NEXT:    [[ADD:%.*]] = select i1 [[A:%.*]], i32 2, i32 0
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %zext = zext i1 %a to i32
  %add = add i32 %zext, %zext
  ret i32 %add
}

define i32 @add_sext_sext_i1(i1 %a) {
; CHECK-LABEL: @add_sext_sext_i1(
; CHECK-NEXT:    [[SEXT:%.*]] = sext i1 [[A:%.*]] to i32
; CHECK-NEXT:    [[ADD:%.*]] = shl nsw i32 [[SEXT]], 1
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %sext = sext i1 %a to i32
  %add = add i32 %sext, %sext
  ret i32 %add
}

define i32 @add_zext_sext_not_i1(i8 %a) {
; CHECK-LABEL: @add_zext_sext_not_i1(
; CHECK-NEXT:    [[ZEXT:%.*]] = zext i8 [[A:%.*]] to i32
; CHECK-NEXT:    [[SEXT:%.*]] = sext i8 [[A]] to i32
; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[ZEXT]], [[SEXT]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %zext = zext i8 %a to i32
  %sext = sext i8 %a to i32
  %add = add i32 %zext, %sext
  ret i32 %add
}

define i32 @add_zext_sext_i1_different_values(i1 %a, i1 %b) {
; CHECK-LABEL: @add_zext_sext_i1_different_values(
; CHECK-NEXT:    [[ZEXT:%.*]] = zext i1 [[A:%.*]] to i32
; CHECK-NEXT:    [[SEXT:%.*]] = sext i1 [[B:%.*]] to i32
; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[ZEXT]], [[SEXT]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %zext = zext i1 %a to i32
  %sext = sext i1 %b to i32
  %add = add i32 %zext, %sext
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_nsw(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_nsw(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul nsw i32 %a, %a
  %two_a = shl i32 %a, 1
  %two_a_plus_b = add i32 %two_a, %b
  %mul = mul i32 %two_a_plus_b, %b
  %add = add i32 %mul, %a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_u(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_u(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul i32 %a, %a
  %two_a = shl i32 %a, 1
  %two_a_plus_b = add i32 %two_a, %b
  %mul = mul i32 %two_a_plus_b, %b
  %add = add i32 %mul, %a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_nuw(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_nuw(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul nuw i32 %a, %a
  %two_a = mul i32 %a, 2
  %two_a_plus_b = add i32 %two_a, %b
  %mul = mul nuw i32 %two_a_plus_b, %b
  %add = add i32 %mul, %a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_flipped(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_flipped(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul nsw i32 %a, %a
  %two_a = shl i32 %a, 1
  %two_a_plus_b = add i32 %two_a, %b
  %mul = mul i32 %two_a_plus_b, %b
  %add = add i32 %a_sq, %mul
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_flipped2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_flipped2(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul nsw i32 %a, %a
  %two_a = shl i32 %a, 1
  %two_a_plus_b = add i32 %two_a, %b
  %mul = mul i32 %b, %two_a_plus_b
  %add = add i32 %mul, %a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_flipped3(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_flipped3(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul nsw i32 %a, %a
  %two_a = shl i32 %a, 1
  %two_a_plus_b = add i32 %b, %two_a
  %mul = mul i32 %two_a_plus_b, %b
  %add = add i32 %mul, %a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_order2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order2(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %twoa, %b
  %b_sq = mul i32 %b, %b
  %twoab_b2 = add i32 %twoab, %b_sq
  %ab2 = add i32 %a_sq, %twoab_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order2_flipped(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order2_flipped(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %twoa, %b
  %b_sq = mul i32 %b, %b
  %twoab_b2 = add i32 %twoab, %b_sq
  %ab2 = add i32 %twoab_b2, %a_sq
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order2_flipped2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order2_flipped2(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %twoa, %b
  %b_sq = mul i32 %b, %b
  %twoab_b2 = add i32 %b_sq, %twoab
  %ab2 = add i32 %a_sq, %twoab_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order2_flipped3(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order2_flipped3(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %b, %twoa
  %b_sq = mul i32 %b, %b
  %twoab_b2 = add i32 %twoab, %b_sq
  %ab2 = add i32 %a_sq, %twoab_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order3(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order3(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %twoa, %b
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order3_flipped(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order3_flipped(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %twoa, %b
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %a2_b2, %twoab
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order3_flipped2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order3_flipped2(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %twoa, %b
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %b_sq, %a_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order3_flipped3(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order3_flipped3(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %b, %twoa
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order4(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order4(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %ab = mul i32 %a, %b
  %twoab = mul i32 %ab, 2
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order4_flipped(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order4_flipped(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %ab = mul i32 %a, %b
  %twoab = mul i32 %ab, 2
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %a2_b2, %twoab
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order4_flipped2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order4_flipped2(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %ab = mul i32 %a, %b
  %twoab = mul i32 %ab, 2
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %b_sq, %a_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order4_flipped3(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order4_flipped3(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %ab = mul i32 %a, %b
  %twoab = mul i32 2, %ab
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order4_flipped4(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order4_flipped4(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %ab = mul i32 %b, %a
  %twoab = mul i32 %ab, 2
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order5(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order5(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twob = mul i32 %b, 2
  %twoab = mul i32 %twob, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order5_flipped(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order5_flipped(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twob = mul i32 %b, 2
  %twoab = mul i32 %twob, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %a2_b2, %twoab
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order5_flipped2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order5_flipped2(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twob = mul i32 %b, 2
  %twoab = mul i32 %twob, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %b_sq, %a_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order5_flipped3(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order5_flipped3(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twob = mul i32 %b, 2
  %twoab = mul i32 %a, %twob
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order5_flipped4(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order5_flipped4(
; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[B:%.*]], [[A:%.*]]
; CHECK-NEXT:    [[AB2:%.*]] = mul i32 [[TMP1]], [[TMP1]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twob = mul i32 2, %b
  %twoab = mul i32 %twob, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_not_one_use(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_not_one_use(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[TWO_A:%.*]] = shl i32 [[A]], 1
; CHECK-NEXT:    [[TWO_A_PLUS_B:%.*]] = add i32 [[TWO_A]], [[B:%.*]]
; CHECK-NEXT:    [[MUL:%.*]] = mul i32 [[TWO_A_PLUS_B]], [[B]]
; CHECK-NEXT:    tail call void @fake_func(i32 [[MUL]])
; CHECK-NEXT:    [[ADD:%.*]] = add i32 [[MUL]], [[A_SQ]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul nsw i32 %a, %a
  %two_a = shl i32 %a, 1
  %two_a_plus_b = add i32 %two_a, %b
  %mul = mul i32 %two_a_plus_b, %b
  tail call void @fake_func (i32 %mul)
  %add = add i32 %mul, %a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_not_one_use2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_not_one_use2(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[TWO_A:%.*]] = shl i32 [[A]], 1
; CHECK-NEXT:    [[TWO_A_PLUS_B:%.*]] = add i32 [[TWO_A]], [[B:%.*]]
; CHECK-NEXT:    [[MUL:%.*]] = mul i32 [[TWO_A_PLUS_B]], [[B]]
; CHECK-NEXT:    tail call void @fake_func(i32 [[A_SQ]])
; CHECK-NEXT:    [[ADD:%.*]] = add i32 [[MUL]], [[A_SQ]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul nsw i32 %a, %a
  %two_a = shl i32 %a, 1
  %two_a_plus_b = add i32 %two_a, %b
  %mul = mul i32 %two_a_plus_b, %b
  tail call void @fake_func (i32 %a_sq)
  %add = add i32 %mul, %a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_order2_not_one_use(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order2_not_one_use(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[TWOA:%.*]] = shl i32 [[A]], 1
; CHECK-NEXT:    [[TWOAB1:%.*]] = add i32 [[TWOA]], [[B:%.*]]
; CHECK-NEXT:    [[TWOAB_B2:%.*]] = mul i32 [[TWOAB1]], [[B]]
; CHECK-NEXT:    tail call void @fake_func(i32 [[TWOAB_B2]])
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[A_SQ]], [[TWOAB_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %twoa, %b
  %b_sq = mul i32 %b, %b
  %twoab_b2 = add i32 %twoab, %b_sq
  tail call void @fake_func (i32 %twoab_b2)
  %ab2 = add i32 %a_sq, %twoab_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order2_not_one_use2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order2_not_one_use2(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[TWOA:%.*]] = shl i32 [[A]], 1
; CHECK-NEXT:    [[TWOAB1:%.*]] = add i32 [[TWOA]], [[B:%.*]]
; CHECK-NEXT:    [[TWOAB_B2:%.*]] = mul i32 [[TWOAB1]], [[B]]
; CHECK-NEXT:    tail call void @fake_func(i32 [[A_SQ]])
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[A_SQ]], [[TWOAB_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %twoa, %b
  %b_sq = mul i32 %b, %b
  %twoab_b2 = add i32 %twoab, %b_sq
  tail call void @fake_func (i32 %a_sq)
  %ab2 = add i32 %a_sq, %twoab_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order3_not_one_use(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order3_not_one_use(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[TWOA:%.*]] = shl i32 [[A]], 1
; CHECK-NEXT:    [[TWOAB:%.*]] = mul i32 [[TWOA]], [[B:%.*]]
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    tail call void @fake_func(i32 [[TWOAB]])
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %twoa, %b
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  tail call void @fake_func (i32 %twoab)
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order3_not_one_use2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order3_not_one_use2(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[TWOA:%.*]] = shl i32 [[A]], 1
; CHECK-NEXT:    [[TWOAB:%.*]] = mul i32 [[TWOA]], [[B:%.*]]
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    tail call void @fake_func(i32 [[A2_B2]])
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twoa = mul i32 %a, 2
  %twoab = mul i32 %twoa, %b
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  tail call void @fake_func (i32 %a2_b2)
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order4_not_one_use(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order4_not_one_use(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[AB:%.*]] = mul i32 [[A]], [[B:%.*]]
; CHECK-NEXT:    [[TWOAB:%.*]] = shl i32 [[AB]], 1
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    tail call void @fake_func(i32 [[TWOAB]])
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %ab = mul i32 %a, %b
  %twoab = mul i32 %ab, 2
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  tail call void @fake_func (i32 %twoab)
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order4_not_one_use2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order4_not_one_use2(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[AB:%.*]] = mul i32 [[A]], [[B:%.*]]
; CHECK-NEXT:    [[TWOAB:%.*]] = shl i32 [[AB]], 1
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    tail call void @fake_func(i32 [[A2_B2]])
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %ab = mul i32 %a, %b
  %twoab = mul i32 %ab, 2
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  tail call void @fake_func (i32 %a2_b2)
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order5_not_one_use(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order5_not_one_use(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[TWOB:%.*]] = shl i32 [[B:%.*]], 1
; CHECK-NEXT:    [[TWOAB:%.*]] = mul i32 [[TWOB]], [[A]]
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    tail call void @fake_func(i32 [[TWOAB]])
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twob = mul i32 %b, 2
  %twoab = mul i32 %twob, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  tail call void @fake_func (i32 %twoab)
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_order5_not_one_use2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_order5_not_one_use2(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[TWOB:%.*]] = shl i32 [[B:%.*]], 1
; CHECK-NEXT:    [[TWOAB:%.*]] = mul i32 [[TWOB]], [[A]]
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    tail call void @fake_func(i32 [[A2_B2]])
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_sq = mul nsw i32 %a, %a
  %twob = mul i32 %b, 2
  %twoab = mul i32 %twob, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  tail call void @fake_func (i32 %a2_b2)
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_invalid0(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_invalid0(
; CHECK-NEXT:    [[TWO_A:%.*]] = shl i32 [[A:%.*]], 1
; CHECK-NEXT:    [[TWO_A_PLUS_B:%.*]] = add i32 [[TWO_A]], [[B:%.*]]
; CHECK-NEXT:    [[MUL1:%.*]] = add i32 [[TWO_A_PLUS_B]], [[A]]
; CHECK-NEXT:    [[ADD:%.*]] = mul i32 [[MUL1]], [[B]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %not_a_sq = mul nsw i32 %a, %b
  %two_a = shl i32 %a, 1
  %two_a_plus_b = add i32 %two_a, %b
  %mul = mul i32 %two_a_plus_b, %b
  %add = add i32 %mul, %not_a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_invalid1(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_invalid1(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[NOT_TWO_A_PLUS_B:%.*]] = mul i32 [[A]], 3
; CHECK-NEXT:    [[MUL:%.*]] = mul i32 [[NOT_TWO_A_PLUS_B]], [[B:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = add i32 [[MUL]], [[A_SQ]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul nsw i32 %a, %a
  %two_a = shl i32 %a, 1
  %not_two_a_plus_b = add i32 %two_a, %a
  %mul = mul i32 %not_two_a_plus_b, %b
  %add = add i32 %mul, %a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_invalid2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_invalid2(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[NOT_TWO_A:%.*]] = shl i32 [[A]], 2
; CHECK-NEXT:    [[TWO_A_PLUS_B:%.*]] = add i32 [[NOT_TWO_A]], [[B:%.*]]
; CHECK-NEXT:    [[MUL:%.*]] = mul i32 [[TWO_A_PLUS_B]], [[B]]
; CHECK-NEXT:    [[ADD:%.*]] = add i32 [[MUL]], [[A_SQ]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul nsw i32 %a, %a
  %not_two_a = shl i32 %a, 2
  %two_a_plus_b = add i32 %not_two_a, %b
  %mul = mul i32 %two_a_plus_b, %b
  %add = add i32 %mul, %a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_invalid3(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_invalid3(
; CHECK-NEXT:    [[TWO_A_PLUS_B:%.*]] = mul i32 [[B:%.*]], 3
; CHECK-NEXT:    [[MUL1:%.*]] = add i32 [[TWO_A_PLUS_B]], [[A:%.*]]
; CHECK-NEXT:    [[ADD:%.*]] = mul i32 [[MUL1]], [[A]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul nsw i32 %a, %a
  %not_two_a = shl i32 %b, 1
  %two_a_plus_b = add i32 %not_two_a, %b
  %mul = mul i32 %two_a_plus_b, %a
  %add = add i32 %mul, %a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_invalid4(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_invalid4(
; CHECK-NEXT:    [[A_SQ:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[TWO_A_PLUS_B:%.*]] = mul i32 [[B:%.*]], 3
; CHECK-NEXT:    [[MUL:%.*]] = mul i32 [[TWO_A_PLUS_B]], [[B]]
; CHECK-NEXT:    [[ADD:%.*]] = add i32 [[MUL]], [[A_SQ]]
; CHECK-NEXT:    ret i32 [[ADD]]
;
  %a_sq = mul nsw i32 %a, %a
  %not_two_a = shl i32 %b, 1
  %two_a_plus_b = add i32 %not_two_a, %b
  %mul = mul i32 %two_a_plus_b, %b
  %add = add i32 %mul, %a_sq
  ret i32 %add
}

define i32 @add_reduce_sqr_sum_varB_invalid0(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_varB_invalid0(
; CHECK-NEXT:    [[NOT_A_B:%.*]] = mul nsw i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[TWOAB:%.*]] = shl nuw i32 [[NOT_A_B]], 1
; CHECK-NEXT:    [[A_SQ:%.*]] = mul i32 [[A]], [[A]]
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B:%.*]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %not_a_b = mul nsw i32 %a, %a
  %twoab = mul i32 %not_a_b, 2
  %a_sq = mul i32 %a, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_varB_invalid1(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_varB_invalid1(
; CHECK-NEXT:    [[NOT_A_B:%.*]] = mul nsw i32 [[B:%.*]], [[B]]
; CHECK-NEXT:    [[TWOAB:%.*]] = shl nuw i32 [[NOT_A_B]], 1
; CHECK-NEXT:    [[A_SQ:%.*]] = mul i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %not_a_b = mul nsw i32 %b, %b
  %twoab = mul i32 %not_a_b, 2
  %a_sq = mul i32 %a, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_varB_invalid2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_varB_invalid2(
; CHECK-NEXT:    [[A_B:%.*]] = mul nsw i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[NOT_TWOAB:%.*]] = shl i32 [[A_B]], 2
; CHECK-NEXT:    [[A_SQ:%.*]] = mul i32 [[A]], [[A]]
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[NOT_TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_b = mul nsw i32 %a, %b
  %not_twoab = mul i32 %a_b, 4
  %a_sq = mul i32 %a, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %not_twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_varB_invalid3(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_varB_invalid3(
; CHECK-NEXT:    [[A_B:%.*]] = mul nsw i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[TWOAB:%.*]] = shl i32 [[A_B]], 1
; CHECK-NEXT:    [[B_SQ1:%.*]] = add i32 [[A]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = mul i32 [[B_SQ1]], [[B]]
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_b = mul nsw i32 %a, %b
  %twoab = mul i32 %a_b, 2
  %not_a_sq = mul i32 %b, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %not_a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_varB_invalid4(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_varB_invalid4(
; CHECK-NEXT:    [[A_B:%.*]] = mul nsw i32 [[A:%.*]], [[B:%.*]]
; CHECK-NEXT:    [[TWOAB:%.*]] = shl i32 [[A_B]], 1
; CHECK-NEXT:    [[NOT_B_SQ1:%.*]] = add i32 [[A]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = mul i32 [[NOT_B_SQ1]], [[A]]
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %a_b = mul nsw i32 %a, %b
  %twoab = mul i32 %a_b, 2
  %a_sq = mul i32 %a, %a
  %not_b_sq = mul i32 %b, %a
  %a2_b2 = add i32 %a_sq, %not_b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_varC_invalid0(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_varC_invalid0(
; CHECK-NEXT:    [[NOT_TWOA:%.*]] = shl nsw i32 [[B:%.*]], 1
; CHECK-NEXT:    [[TWOAB:%.*]] = mul i32 [[NOT_TWOA]], [[B]]
; CHECK-NEXT:    [[A_SQ:%.*]] = mul i32 [[A:%.*]], [[A]]
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %not_twoa = mul nsw i32 %b, 2
  %twoab = mul i32 %not_twoa, %b
  %a_sq = mul i32 %a, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_varC_invalid1(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_varC_invalid1(
; CHECK-NEXT:    [[NOT_TWOA:%.*]] = shl nsw i32 [[A:%.*]], 2
; CHECK-NEXT:    [[TWOAB:%.*]] = mul i32 [[NOT_TWOA]], [[B:%.*]]
; CHECK-NEXT:    [[A_SQ:%.*]] = mul i32 [[A]], [[A]]
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %not_twoa = mul nsw i32 %a, 4
  %twoab = mul i32 %not_twoa, %b
  %a_sq = mul i32 %a, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %twoab, %a2_b2
  ret i32 %ab2
}

define i32 @add_reduce_sqr_sum_varC_invalid2(i32 %a, i32 %b) {
; CHECK-LABEL: @add_reduce_sqr_sum_varC_invalid2(
; CHECK-NEXT:    [[TWOA:%.*]] = shl nsw i32 [[A:%.*]], 1
; CHECK-NEXT:    [[NOT_TWOAB:%.*]] = mul i32 [[TWOA]], [[A]]
; CHECK-NEXT:    [[A_SQ:%.*]] = mul i32 [[A]], [[A]]
; CHECK-NEXT:    [[B_SQ:%.*]] = mul i32 [[B:%.*]], [[B]]
; CHECK-NEXT:    [[A2_B2:%.*]] = add i32 [[A_SQ]], [[B_SQ]]
; CHECK-NEXT:    [[AB2:%.*]] = add i32 [[NOT_TWOAB]], [[A2_B2]]
; CHECK-NEXT:    ret i32 [[AB2]]
;
  %twoa = mul nsw i32 %a, 2
  %not_twoab = mul i32 %twoa, %a
  %a_sq = mul i32 %a, %a
  %b_sq = mul i32 %b, %b
  %a2_b2 = add i32 %a_sq, %b_sq
  %ab2 = add i32 %not_twoab, %a2_b2
  ret i32 %ab2
}

declare void @llvm.assume(i1)
declare void @fake_func(i32)