/
memcpy-to-load.ll
87 lines (75 loc) · 3.4 KB
/
memcpy-to-load.ll
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -passes=instcombine -S | FileCheck %s
; RUN: opt < %s -passes=instcombine -S -data-layout=n32 | FileCheck %s
; RUN: opt < %s -passes=instcombine -S -data-layout=n32:64 | FileCheck %s
; RUN: opt < %s -passes=instcombine -S -data-layout=n32:64:128 | FileCheck %s
declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture, i8* nocapture, i32, i1) nounwind
; memcpy can be expanded inline with load/store (based on the datalayout?)
define void @copy_1_byte(i8* %d, i8* %s) {
; CHECK-LABEL: @copy_1_byte(
; CHECK-NEXT: [[TMP1:%.*]] = load i8, i8* [[S:%.*]], align 1
; CHECK-NEXT: store i8 [[TMP1]], i8* [[D:%.*]], align 1
; CHECK-NEXT: ret void
;
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 1, i1 false)
ret void
}
define void @copy_2_bytes(i8* %d, i8* %s) {
; CHECK-LABEL: @copy_2_bytes(
; CHECK-NEXT: [[TMP1:%.*]] = bitcast i8* [[S:%.*]] to i16*
; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[D:%.*]] to i16*
; CHECK-NEXT: [[TMP3:%.*]] = load i16, i16* [[TMP1]], align 1
; CHECK-NEXT: store i16 [[TMP3]], i16* [[TMP2]], align 1
; CHECK-NEXT: ret void
;
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 2, i1 false)
ret void
}
; We don't expand small non-power-of-2. Should we? Might be a target-dependent choice.
define void @copy_3_bytes(i8* %d, i8* %s) {
; CHECK-LABEL: @copy_3_bytes(
; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* noundef nonnull align 1 dereferenceable(3) [[D:%.*]], i8* noundef nonnull align 1 dereferenceable(3) [[S:%.*]], i32 3, i1 false)
; CHECK-NEXT: ret void
;
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 3, i1 false)
ret void
}
define void @copy_4_bytes(i8* %d, i8* %s) {
; CHECK-LABEL: @copy_4_bytes(
; CHECK-NEXT: [[TMP1:%.*]] = bitcast i8* [[S:%.*]] to i32*
; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[D:%.*]] to i32*
; CHECK-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP1]], align 1
; CHECK-NEXT: store i32 [[TMP3]], i32* [[TMP2]], align 1
; CHECK-NEXT: ret void
;
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 4, i1 false)
ret void
}
; We don't expand small non-power-of-2. Should we? Might be a target-dependent choice.
define void @copy_5_bytes(i8* %d, i8* %s) {
; CHECK-LABEL: @copy_5_bytes(
; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* noundef nonnull align 1 dereferenceable(5) [[D:%.*]], i8* noundef nonnull align 1 dereferenceable(5) [[S:%.*]], i32 5, i1 false)
; CHECK-NEXT: ret void
;
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 5, i1 false)
ret void
}
define void @copy_8_bytes(i8* %d, i8* %s) {
; CHECK-LABEL: @copy_8_bytes(
; CHECK-NEXT: [[TMP1:%.*]] = bitcast i8* [[S:%.*]] to i64*
; CHECK-NEXT: [[TMP2:%.*]] = bitcast i8* [[D:%.*]] to i64*
; CHECK-NEXT: [[TMP3:%.*]] = load i64, i64* [[TMP1]], align 1
; CHECK-NEXT: store i64 [[TMP3]], i64* [[TMP2]], align 1
; CHECK-NEXT: ret void
;
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 8, i1 false)
ret void
}
define void @copy_16_bytes(i8* %d, i8* %s) {
; CHECK-LABEL: @copy_16_bytes(
; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* noundef nonnull align 1 dereferenceable(16) [[D:%.*]], i8* noundef nonnull align 1 dereferenceable(16) [[S:%.*]], i32 16, i1 false)
; CHECK-NEXT: ret void
;
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %d, i8* %s, i32 16, i1 false)
ret void
}