xref: /llvm-project/llvm/test/Transforms/InstCombine/sadd-with-overflow.ll (revision 8670faf9394ed34ff894b0e784cdee4ba70d7837) 
1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2; RUN: opt < %s -instcombine -S | FileCheck %s
3
4declare { <2 x i32>, <2 x i1> } @llvm.sadd.with.overflow.v2i32(<2 x i32>, <2 x i32>)
5
6declare { i32, i1 } @llvm.sadd.with.overflow.i32(i32, i32)
7
8declare { i8, i1 } @llvm.sadd.with.overflow.i8(i8, i8)
9
10define { i32, i1 } @simple_fold(i32) {
11; CHECK-LABEL: @simple_fold(
12; CHECK-NEXT:    [[TMP2:%.*]] = add nsw i32 [[TMP0:%.*]], 7
13; CHECK-NEXT:    [[TMP3:%.*]] = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 [[TMP2]], i32 13)
14; CHECK-NEXT:    ret { i32, i1 } [[TMP3]]
15;
16  %2 = add nsw i32 %0, 7
17  %3 = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 %2, i32 13)
18  ret { i32, i1 } %3
19}
20
21define { i32, i1 } @fold_mixed_signs(i32) {
22; CHECK-LABEL: @fold_mixed_signs(
23; CHECK-NEXT:    [[TMP2:%.*]] = add nsw i32 [[TMP0:%.*]], 13
24; CHECK-NEXT:    [[TMP3:%.*]] = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 [[TMP2]], i32 -7)
25; CHECK-NEXT:    ret { i32, i1 } [[TMP3]]
26;
27  %2 = add nsw i32 %0, 13
28  %3 = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 %2, i32 -7)
29  ret { i32, i1 } %3
30}
31
32define { i8, i1 } @no_fold_on_constant_add_overflow(i8) {
33; CHECK-LABEL: @no_fold_on_constant_add_overflow(
34; CHECK-NEXT:    [[TMP2:%.*]] = add nsw i8 [[TMP0:%.*]], 127
35; CHECK-NEXT:    [[TMP3:%.*]] = tail call { i8, i1 } @llvm.sadd.with.overflow.i8(i8 [[TMP2]], i8 127)
36; CHECK-NEXT:    ret { i8, i1 } [[TMP3]]
37;
38  %2 = add nsw i8 %0, 127
39  %3 = tail call { i8, i1 } @llvm.sadd.with.overflow.i8(i8 %2, i8 127)
40  ret { i8, i1 } %3
41}
42
43define { <2 x i32>, <2 x i1> } @fold_simple_splat_constant(<2 x i32>) {
44; CHECK-LABEL: @fold_simple_splat_constant(
45; CHECK-NEXT:    [[TMP2:%.*]] = add nsw <2 x i32> [[TMP0:%.*]], <i32 12, i32 12>
46; CHECK-NEXT:    [[TMP3:%.*]] = tail call { <2 x i32>, <2 x i1> } @llvm.sadd.with.overflow.v2i32(<2 x i32> [[TMP2]], <2 x i32> <i32 30, i32 30>)
47; CHECK-NEXT:    ret { <2 x i32>, <2 x i1> } [[TMP3]]
48;
49  %2 = add nsw <2 x i32> %0, <i32 12, i32 12>
50  %3 = tail call { <2 x i32>, <2 x i1> } @llvm.sadd.with.overflow.v2i32(<2 x i32> %2, <2 x i32> <i32 30, i32 30>)
51  ret { <2 x i32>, <2 x i1> } %3
52}
53
54define { <2 x i32>, <2 x i1> } @no_fold_splat_undef_constant(<2 x i32>) {
55; CHECK-LABEL: @no_fold_splat_undef_constant(
56; CHECK-NEXT:    [[TMP2:%.*]] = add nsw <2 x i32> [[TMP0:%.*]], <i32 12, i32 undef>
57; CHECK-NEXT:    [[TMP3:%.*]] = tail call { <2 x i32>, <2 x i1> } @llvm.sadd.with.overflow.v2i32(<2 x i32> [[TMP2]], <2 x i32> <i32 30, i32 30>)
58; CHECK-NEXT:    ret { <2 x i32>, <2 x i1> } [[TMP3]]
59;
60  %2 = add nsw <2 x i32> %0, <i32 12, i32 undef>
61  %3 = tail call { <2 x i32>, <2 x i1> } @llvm.sadd.with.overflow.v2i32(<2 x i32> %2, <2 x i32> <i32 30, i32 30>)
62  ret { <2 x i32>, <2 x i1> } %3
63}
64
65define { <2 x i32>, <2 x i1> } @no_fold_splat_not_constant(<2 x i32>, <2 x i32>) {
66; CHECK-LABEL: @no_fold_splat_not_constant(
67; CHECK-NEXT:    [[TMP3:%.*]] = add nsw <2 x i32> [[TMP0:%.*]], [[TMP1:%.*]]
68; CHECK-NEXT:    [[TMP4:%.*]] = tail call { <2 x i32>, <2 x i1> } @llvm.sadd.with.overflow.v2i32(<2 x i32> [[TMP3]], <2 x i32> <i32 30, i32 30>)
69; CHECK-NEXT:    ret { <2 x i32>, <2 x i1> } [[TMP4]]
70;
71  %3 = add nsw <2 x i32> %0, %1
72  %4 = tail call { <2 x i32>, <2 x i1> } @llvm.sadd.with.overflow.v2i32(<2 x i32> %3, <2 x i32> <i32 30, i32 30>)
73  ret { <2 x i32>, <2 x i1> } %4
74}
75
76define { i32, i1 } @fold_nuwnsw(i32) {
77; CHECK-LABEL: @fold_nuwnsw(
78; CHECK-NEXT:    [[TMP2:%.*]] = add nuw nsw i32 [[TMP0:%.*]], 12
79; CHECK-NEXT:    [[TMP3:%.*]] = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 [[TMP2]], i32 30)
80; CHECK-NEXT:    ret { i32, i1 } [[TMP3]]
81;
82  %2 = add nuw nsw i32 %0, 12
83  %3 = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 %2, i32 30)
84  ret { i32, i1 } %3
85}
86
87define { i32, i1 } @no_fold_nuw(i32) {
88; CHECK-LABEL: @no_fold_nuw(
89; CHECK-NEXT:    [[TMP2:%.*]] = add nuw i32 [[TMP0:%.*]], 12
90; CHECK-NEXT:    [[TMP3:%.*]] = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 [[TMP2]], i32 30)
91; CHECK-NEXT:    ret { i32, i1 } [[TMP3]]
92;
93  %2 = add nuw i32 %0, 12
94  %3 = tail call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 %2, i32 30)
95  ret { i32, i1 } %3
96}
97