; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes --check-globals
; RUN: opt -aa-pipeline=basic-aa -passes=attributor -attributor-manifest-internal  -attributor-annotate-decl-cs  -S < %s | FileCheck %s --check-prefixes=TUNIT
; RUN: opt -aa-pipeline=basic-aa -passes=attributor-cgscc -attributor-manifest-internal  -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CGSCC
;

define internal i8 @read_arg(ptr %p) {
; CGSCC: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(argmem: read)
; CGSCC-LABEL: define {{[^@]+}}@read_arg
; CGSCC-SAME: (ptr nofree noundef nonnull readonly captures(none) dereferenceable(1022) [[P:%.*]]) #[[ATTR0:[0-9]+]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[L:%.*]] = load i8, ptr [[P]], align 1, !invariant.load [[META0:![0-9]+]]
; CGSCC-NEXT:    ret i8 [[L]]
;
entry:
  %l = load i8, ptr %p, align 1
  ret i8 %l
}

define internal i8 @read_arg_index(ptr %p, i64 %index) {
; CGSCC: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(argmem: read)
; CGSCC-LABEL: define {{[^@]+}}@read_arg_index
; CGSCC-SAME: (ptr nofree noundef nonnull readonly align 16 captures(none) dereferenceable(1024) [[P:%.*]]) #[[ATTR0]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[G:%.*]] = getelementptr inbounds i8, ptr [[P]], i64 2
; CGSCC-NEXT:    [[L:%.*]] = load i8, ptr [[G]], align 1, !invariant.load [[META0]]
; CGSCC-NEXT:    ret i8 [[L]]
;
entry:
  %g = getelementptr inbounds i8, ptr %p, i64 %index
  %l = load i8, ptr %g, align 1
  ret i8 %l
}

define i8 @call_simplifiable_1() {
; TUNIT: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(none)
; TUNIT-LABEL: define {{[^@]+}}@call_simplifiable_1
; TUNIT-SAME: () #[[ATTR0:[0-9]+]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[BYTES:%.*]] = alloca [1024 x i8], align 16
; TUNIT-NEXT:    [[I0:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 2
; TUNIT-NEXT:    ret i8 2
;
; CGSCC: Function Attrs: mustprogress nofree nosync nounwind willreturn memory(none)
; CGSCC-LABEL: define {{[^@]+}}@call_simplifiable_1
; CGSCC-SAME: () #[[ATTR1:[0-9]+]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[BYTES:%.*]] = alloca [1024 x i8], align 16
; CGSCC-NEXT:    [[I0:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 2
; CGSCC-NEXT:    store i8 2, ptr [[I0]], align 2
; CGSCC-NEXT:    [[R:%.*]] = call i8 @read_arg(ptr nofree noundef nonnull readonly align 2 captures(none) dereferenceable(1022) [[I0]]) #[[ATTR3:[0-9]+]]
; CGSCC-NEXT:    ret i8 [[R]]
;
entry:
  %Bytes = alloca [1024 x i8], align 16
  %i0 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 2
  store i8 2, ptr %i0, align 1
  %r = call i8 @read_arg(ptr %i0)
  ret i8 %r
}

;;; Same as read_arg, but we need a copy to form distinct leaves in the callgraph.

define internal i8 @read_arg_1(ptr %p) {
; CGSCC: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(argmem: read)
; CGSCC-LABEL: define {{[^@]+}}@read_arg_1
; CGSCC-SAME: (ptr nofree noundef nonnull readonly captures(none) dereferenceable(1) [[P:%.*]]) #[[ATTR0]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[L:%.*]] = load i8, ptr [[P]], align 1
; CGSCC-NEXT:    ret i8 [[L]]
;
entry:
  %l = load i8, ptr %p, align 1
  ret i8 %l
}

define internal i8 @sum_two_same_loads(ptr %p) {
; CGSCC: Function Attrs: mustprogress nofree nosync nounwind willreturn memory(argmem: read)
; CGSCC-LABEL: define {{[^@]+}}@sum_two_same_loads
; CGSCC-SAME: (ptr nofree noundef nonnull readonly captures(none) dereferenceable(1022) [[P:%.*]]) #[[ATTR2:[0-9]+]] {
; CGSCC-NEXT:    [[X:%.*]] = call i8 @read_arg_1(ptr nofree noundef nonnull readonly captures(none) dereferenceable(1022) [[P]]) #[[ATTR3]]
; CGSCC-NEXT:    [[Y:%.*]] = call i8 @read_arg_1(ptr nofree noundef nonnull readonly captures(none) dereferenceable(1022) [[P]]) #[[ATTR3]]
; CGSCC-NEXT:    [[Z:%.*]] = add nsw i8 [[X]], [[Y]]
; CGSCC-NEXT:    ret i8 [[Z]]
;
  %x = call i8 @read_arg_1(ptr %p)
  %y = call i8 @read_arg_1(ptr %p)
  %z = add nsw i8 %x, %y
  ret i8 %z
}

define i8 @call_simplifiable_2() {
; TUNIT: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(none)
; TUNIT-LABEL: define {{[^@]+}}@call_simplifiable_2
; TUNIT-SAME: () #[[ATTR0]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[BYTES:%.*]] = alloca [1024 x i8], align 16
; TUNIT-NEXT:    [[I0:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 2
; TUNIT-NEXT:    [[I1:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 3
; TUNIT-NEXT:    ret i8 4
;
; CGSCC: Function Attrs: mustprogress nofree nosync nounwind willreturn memory(none)
; CGSCC-LABEL: define {{[^@]+}}@call_simplifiable_2
; CGSCC-SAME: () #[[ATTR1]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[BYTES:%.*]] = alloca [1024 x i8], align 16
; CGSCC-NEXT:    [[I0:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 2
; CGSCC-NEXT:    store i8 2, ptr [[I0]], align 2
; CGSCC-NEXT:    [[I1:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 3
; CGSCC-NEXT:    store i8 3, ptr [[I1]], align 1
; CGSCC-NEXT:    [[R:%.*]] = call i8 @sum_two_same_loads(ptr nofree noundef nonnull readonly align 2 captures(none) dereferenceable(1022) [[I0]]) #[[ATTR3]]
; CGSCC-NEXT:    ret i8 [[R]]
;
entry:
  %Bytes = alloca [1024 x i8], align 16
  %i0 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 2
  store i8 2, ptr %i0
  %i1 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 3
  store i8 3, ptr %i1
  %r = call i8 @sum_two_same_loads(ptr %i0)
  ret i8 %r
}

define i8 @call_simplifiable_3() {
; TUNIT: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(none)
; TUNIT-LABEL: define {{[^@]+}}@call_simplifiable_3
; TUNIT-SAME: () #[[ATTR0]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[BYTES:%.*]] = alloca [1024 x i8], align 16
; TUNIT-NEXT:    [[I2:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 2
; TUNIT-NEXT:    ret i8 2
;
; CGSCC: Function Attrs: mustprogress nofree nosync nounwind willreturn memory(none)
; CGSCC-LABEL: define {{[^@]+}}@call_simplifiable_3
; CGSCC-SAME: () #[[ATTR1]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[BYTES:%.*]] = alloca [1024 x i8], align 16
; CGSCC-NEXT:    [[I2:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 2
; CGSCC-NEXT:    store i8 2, ptr [[I2]], align 2
; CGSCC-NEXT:    [[R:%.*]] = call i8 @read_arg_index(ptr nofree noundef nonnull readonly align 16 captures(none) dereferenceable(1024) [[BYTES]]) #[[ATTR3]]
; CGSCC-NEXT:    ret i8 [[R]]
;
entry:
  %Bytes = alloca [1024 x i8], align 16
  %i0 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 0
  %i2 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 2
  store i8 2, ptr %i2, align 1
  %r = call i8 @read_arg_index(ptr %i0, i64 2)
  ret i8 %r
}

;;; Same as read_arg, but we need a copy to form distinct leaves in the callgraph.

define internal i8 @read_arg_2(ptr %p) {
; TUNIT: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(argmem: read)
; TUNIT-LABEL: define {{[^@]+}}@read_arg_2
; TUNIT-SAME: (ptr nofree noundef nonnull readonly captures(none) dereferenceable(971) [[P:%.*]]) #[[ATTR1:[0-9]+]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[L:%.*]] = load i8, ptr [[P]], align 1
; TUNIT-NEXT:    ret i8 [[L]]
;
; CGSCC: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(argmem: read)
; CGSCC-LABEL: define {{[^@]+}}@read_arg_2
; CGSCC-SAME: (ptr nofree noundef nonnull readonly captures(none) dereferenceable(1) [[P:%.*]]) #[[ATTR0]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[L:%.*]] = load i8, ptr [[P]], align 1
; CGSCC-NEXT:    ret i8 [[L]]
;
entry:
  %l = load i8, ptr %p, align 1
  ret i8 %l
}

define internal i8 @sum_two_different_loads(ptr %p, ptr %q) {
; TUNIT: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(argmem: read)
; TUNIT-LABEL: define {{[^@]+}}@sum_two_different_loads
; TUNIT-SAME: (ptr nofree nonnull readonly captures(none) dereferenceable(972) [[P:%.*]], ptr nofree noundef nonnull readonly captures(none) dereferenceable(971) [[Q:%.*]]) #[[ATTR1]] {
; TUNIT-NEXT:    [[X:%.*]] = call i8 @read_arg_2(ptr nofree noundef nonnull readonly captures(none) dereferenceable(972) [[P]]) #[[ATTR2:[0-9]+]]
; TUNIT-NEXT:    [[Y:%.*]] = call i8 @read_arg_2(ptr nofree noundef nonnull readonly captures(none) dereferenceable(971) [[Q]]) #[[ATTR2]]
; TUNIT-NEXT:    [[Z:%.*]] = add nsw i8 [[X]], [[Y]]
; TUNIT-NEXT:    ret i8 [[Z]]
;
; CGSCC: Function Attrs: mustprogress nofree nosync nounwind willreturn memory(argmem: read)
; CGSCC-LABEL: define {{[^@]+}}@sum_two_different_loads
; CGSCC-SAME: (ptr nofree noundef nonnull readonly captures(none) dereferenceable(972) [[P:%.*]], ptr nofree noundef nonnull readonly captures(none) dereferenceable(971) [[Q:%.*]]) #[[ATTR2]] {
; CGSCC-NEXT:    [[X:%.*]] = call i8 @read_arg_2(ptr nofree noundef nonnull readonly captures(none) dereferenceable(972) [[P]]) #[[ATTR3]]
; CGSCC-NEXT:    [[Y:%.*]] = call i8 @read_arg_2(ptr nofree noundef nonnull readonly captures(none) dereferenceable(971) [[Q]]) #[[ATTR3]]
; CGSCC-NEXT:    [[Z:%.*]] = add nsw i8 [[X]], [[Y]]
; CGSCC-NEXT:    ret i8 [[Z]]
;
  %x = call i8 @read_arg_2(ptr %p)
  %y = call i8 @read_arg_2(ptr %q)
  %z = add nsw i8 %x, %y
  ret i8 %z
}

define i8 @call_partially_simplifiable_1() {
; TUNIT: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(none)
; TUNIT-LABEL: define {{[^@]+}}@call_partially_simplifiable_1
; TUNIT-SAME: () #[[ATTR0]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[BYTES:%.*]] = alloca [1024 x i8], align 16
; TUNIT-NEXT:    [[I2:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 2
; TUNIT-NEXT:    store i8 2, ptr [[I2]], align 2
; TUNIT-NEXT:    [[I3:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 3
; TUNIT-NEXT:    store i8 3, ptr [[I3]], align 1
; TUNIT-NEXT:    [[I4:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 4
; TUNIT-NEXT:    [[R:%.*]] = call i8 @sum_two_different_loads(ptr nofree noundef nonnull readonly align 2 captures(none) dereferenceable(1022) [[I2]], ptr nofree noundef nonnull readonly captures(none) dereferenceable(1021) [[I3]]) #[[ATTR3:[0-9]+]]
; TUNIT-NEXT:    ret i8 [[R]]
;
; CGSCC: Function Attrs: mustprogress nofree nosync nounwind willreturn memory(none)
; CGSCC-LABEL: define {{[^@]+}}@call_partially_simplifiable_1
; CGSCC-SAME: () #[[ATTR1]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[BYTES:%.*]] = alloca [1024 x i8], align 16
; CGSCC-NEXT:    [[I2:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 2
; CGSCC-NEXT:    store i8 2, ptr [[I2]], align 2
; CGSCC-NEXT:    [[I3:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 3
; CGSCC-NEXT:    store i8 3, ptr [[I3]], align 1
; CGSCC-NEXT:    [[I4:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 4
; CGSCC-NEXT:    store i8 4, ptr [[I4]], align 4
; CGSCC-NEXT:    [[R:%.*]] = call i8 @sum_two_different_loads(ptr nofree noundef nonnull readonly align 2 captures(none) dereferenceable(1022) [[I2]], ptr nofree noundef nonnull readonly captures(none) dereferenceable(1021) [[I3]]) #[[ATTR3]]
; CGSCC-NEXT:    ret i8 [[R]]
;
entry:
  %Bytes = alloca [1024 x i8], align 16
  %i2 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 2
  store i8 2, ptr %i2
  %i3 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 3
  store i8 3, ptr %i3
  %i4 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 4
  ;;; This store is redundant, hence removed.
  store i8 4, ptr %i4
  %r = call i8 @sum_two_different_loads(ptr %i2, ptr %i3)
  ret i8 %r
}

define i8 @call_partially_simplifiable_2(i1 %cond) {
; TUNIT: Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(none)
; TUNIT-LABEL: define {{[^@]+}}@call_partially_simplifiable_2
; TUNIT-SAME: (i1 [[COND:%.*]]) #[[ATTR0]] {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[BYTES:%.*]] = alloca [1024 x i8], align 16
; TUNIT-NEXT:    [[I51:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 51
; TUNIT-NEXT:    [[I52:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 52
; TUNIT-NEXT:    store i8 2, ptr [[I52]], align 4
; TUNIT-NEXT:    [[I53:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 53
; TUNIT-NEXT:    store i8 3, ptr [[I53]], align 1
; TUNIT-NEXT:    [[I54:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 54
; TUNIT-NEXT:    [[SEL:%.*]] = select i1 [[COND]], ptr [[I51]], ptr [[I52]]
; TUNIT-NEXT:    [[R:%.*]] = call i8 @sum_two_different_loads(ptr nofree nonnull readonly captures(none) dereferenceable(972) [[SEL]], ptr nofree noundef nonnull readonly captures(none) dereferenceable(971) [[I53]]) #[[ATTR3]]
; TUNIT-NEXT:    ret i8 [[R]]
;
; CGSCC: Function Attrs: mustprogress nofree nosync nounwind willreturn memory(none)
; CGSCC-LABEL: define {{[^@]+}}@call_partially_simplifiable_2
; CGSCC-SAME: (i1 [[COND:%.*]]) #[[ATTR1]] {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[BYTES:%.*]] = alloca [1024 x i8], align 16
; CGSCC-NEXT:    [[I51:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 51
; CGSCC-NEXT:    [[I52:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 52
; CGSCC-NEXT:    store i8 2, ptr [[I52]], align 4
; CGSCC-NEXT:    [[I53:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 53
; CGSCC-NEXT:    store i8 3, ptr [[I53]], align 1
; CGSCC-NEXT:    [[I54:%.*]] = getelementptr inbounds [1024 x i8], ptr [[BYTES]], i64 0, i64 54
; CGSCC-NEXT:    store i8 4, ptr [[I54]], align 2
; CGSCC-NEXT:    [[SEL:%.*]] = select i1 [[COND]], ptr [[I51]], ptr [[I52]]
; CGSCC-NEXT:    [[R:%.*]] = call i8 @sum_two_different_loads(ptr nofree noundef nonnull readonly captures(none) dereferenceable(972) [[SEL]], ptr nofree noundef nonnull readonly captures(none) dereferenceable(971) [[I53]]) #[[ATTR3]]
; CGSCC-NEXT:    ret i8 [[R]]
;
entry:
  %Bytes = alloca [1024 x i8], align 16
  %i51 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 51
  %i52 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 52
  store i8 2, ptr %i52
  %i53 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 53
  store i8 3, ptr %i53
  %i54 = getelementptr inbounds [1024 x i8], ptr %Bytes, i64 0, i64 54
  ;;; This store is redundant, hence removed. Not affected by the select.
  store i8 4, ptr %i54
  %sel = select i1 %cond, ptr %i51, ptr %i52
  %r = call i8 @sum_two_different_loads(ptr %sel, ptr %i53)
  ret i8 %r
}

;.
; CGSCC: attributes #[[ATTR0]] = { mustprogress nofree norecurse nosync nounwind willreturn memory(argmem: read) }
; CGSCC: attributes #[[ATTR1]] = { mustprogress nofree nosync nounwind willreturn memory(none) }
; CGSCC: attributes #[[ATTR2]] = { mustprogress nofree nosync nounwind willreturn memory(argmem: read) }
; CGSCC: attributes #[[ATTR3]] = { nofree willreturn memory(read) }
;.
; TUNIT: attributes #[[ATTR0]] = { mustprogress nofree norecurse nosync nounwind willreturn memory(none) }
; TUNIT: attributes #[[ATTR1]] = { mustprogress nofree norecurse nosync nounwind willreturn memory(argmem: read) }
; TUNIT: attributes #[[ATTR2]] = { nofree nosync nounwind willreturn memory(read) }
; TUNIT: attributes #[[ATTR3]] = { nofree norecurse nosync nounwind willreturn memory(read) }
;.
; CGSCC: [[META0]] = !{}
;.