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
87
88
89
90
91
92
93
94
95
96
97
|
; Test stores of byte-swapped vector elements.
;
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z15 | FileCheck %s
declare <8 x i16> @llvm.bswap.v8i16(<8 x i16>)
declare <4 x i32> @llvm.bswap.v4i32(<4 x i32>)
declare <2 x i64> @llvm.bswap.v2i64(<2 x i64>)
; Test v8i16 stores.
define void @f1(<8 x i16> %val, <8 x i16> *%ptr) {
; CHECK-LABEL: f1:
; CHECK: vstbrh %v24, 0(%r2)
; CHECK: br %r14
%swap = call <8 x i16> @llvm.bswap.v8i16(<8 x i16> %val)
store <8 x i16> %swap, <8 x i16> *%ptr
ret void
}
; Test v4i32 stores.
define void @f2(<4 x i32> %val, <4 x i32> *%ptr) {
; CHECK-LABEL: f2:
; CHECK: vstbrf %v24, 0(%r2)
; CHECK: br %r14
%swap = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %val)
store <4 x i32> %swap, <4 x i32> *%ptr
ret void
}
; Test v2i64 stores.
define void @f3(<2 x i64> %val, <2 x i64> *%ptr) {
; CHECK-LABEL: f3:
; CHECK: vstbrg %v24, 0(%r2)
; CHECK: br %r14
%swap = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> %val)
store <2 x i64> %swap, <2 x i64> *%ptr
ret void
}
; Test the highest aligned in-range offset.
define void @f4(<4 x i32> %val, <4 x i32> *%base) {
; CHECK-LABEL: f4:
; CHECK: vstbrf %v24, 4080(%r2)
; CHECK: br %r14
%ptr = getelementptr <4 x i32>, <4 x i32> *%base, i64 255
%swap = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %val)
store <4 x i32> %swap, <4 x i32> *%ptr
ret void
}
; Test the highest unaligned in-range offset.
define void @f5(<4 x i32> %val, i8 *%base) {
; CHECK-LABEL: f5:
; CHECK: vstbrf %v24, 4095(%r2)
; CHECK: br %r14
%addr = getelementptr i8, i8 *%base, i64 4095
%ptr = bitcast i8 *%addr to <4 x i32> *
%swap = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %val)
store <4 x i32> %swap, <4 x i32> *%ptr, align 1
ret void
}
; Test the next offset up, which requires separate address logic,
define void @f6(<4 x i32> %val, <4 x i32> *%base) {
; CHECK-LABEL: f6:
; CHECK: aghi %r2, 4096
; CHECK: vstbrf %v24, 0(%r2)
; CHECK: br %r14
%ptr = getelementptr <4 x i32>, <4 x i32> *%base, i64 256
%swap = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %val)
store <4 x i32> %swap, <4 x i32> *%ptr
ret void
}
; Test negative offsets, which also require separate address logic,
define void @f7(<4 x i32> %val, <4 x i32> *%base) {
; CHECK-LABEL: f7:
; CHECK: aghi %r2, -16
; CHECK: vstbrf %v24, 0(%r2)
; CHECK: br %r14
%ptr = getelementptr <4 x i32>, <4 x i32> *%base, i64 -1
%swap = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %val)
store <4 x i32> %swap, <4 x i32> *%ptr
ret void
}
; Check that indexes are allowed.
define void @f8(<4 x i32> %val, i8 *%base, i64 %index) {
; CHECK-LABEL: f8:
; CHECK: vstbrf %v24, 0(%r3,%r2)
; CHECK: br %r14
%addr = getelementptr i8, i8 *%base, i64 %index
%ptr = bitcast i8 *%addr to <4 x i32> *
%swap = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %val)
store <4 x i32> %swap, <4 x i32> *%ptr, align 1
ret void
}
|
