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
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
| ; Test 64-bit floating-point subtraction.
;
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z10 \
; RUN: | FileCheck -check-prefix=CHECK -check-prefix=CHECK-SCALAR %s
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 | FileCheck %s
declare double @foo()
; Check register subtraction.
define double @f1(double %f1, double %f2) {
; CHECK-LABEL: f1:
; CHECK: sdbr %f0, %f2
; CHECK: br %r14
%res = fsub double %f1, %f2
ret double %res
}
; Check the low end of the SDB range.
define double @f2(double %f1, double *%ptr) {
; CHECK-LABEL: f2:
; CHECK: sdb %f0, 0(%r2)
; CHECK: br %r14
%f2 = load double, double *%ptr
%res = fsub double %f1, %f2
ret double %res
}
; Check the high end of the aligned SDB range.
define double @f3(double %f1, double *%base) {
; CHECK-LABEL: f3:
; CHECK: sdb %f0, 4088(%r2)
; CHECK: br %r14
%ptr = getelementptr double, double *%base, i64 511
%f2 = load double, double *%ptr
%res = fsub double %f1, %f2
ret double %res
}
; Check the next doubleword up, which needs separate address logic.
; Other sequences besides this one would be OK.
define double @f4(double %f1, double *%base) {
; CHECK-LABEL: f4:
; CHECK: aghi %r2, 4096
; CHECK: sdb %f0, 0(%r2)
; CHECK: br %r14
%ptr = getelementptr double, double *%base, i64 512
%f2 = load double, double *%ptr
%res = fsub double %f1, %f2
ret double %res
}
; Check negative displacements, which also need separate address logic.
define double @f5(double %f1, double *%base) {
; CHECK-LABEL: f5:
; CHECK: aghi %r2, -8
; CHECK: sdb %f0, 0(%r2)
; CHECK: br %r14
%ptr = getelementptr double, double *%base, i64 -1
%f2 = load double, double *%ptr
%res = fsub double %f1, %f2
ret double %res
}
; Check that SDB allows indices.
define double @f6(double %f1, double *%base, i64 %index) {
; CHECK-LABEL: f6:
; CHECK: sllg %r1, %r3, 3
; CHECK: sdb %f0, 800(%r1,%r2)
; CHECK: br %r14
%ptr1 = getelementptr double, double *%base, i64 %index
%ptr2 = getelementptr double, double *%ptr1, i64 100
%f2 = load double, double *%ptr2
%res = fsub double %f1, %f2
ret double %res
}
; Check that subtractions of spilled values can use SDB rather than SDBR.
define double @f7(double *%ptr0) {
; CHECK-LABEL: f7:
; CHECK: brasl %r14, foo@PLT
; CHECK-SCALAR: sdb %f0, 16{{[04]}}(%r15)
; CHECK: br %r14
%ptr1 = getelementptr double, double *%ptr0, i64 2
%ptr2 = getelementptr double, double *%ptr0, i64 4
%ptr3 = getelementptr double, double *%ptr0, i64 6
%ptr4 = getelementptr double, double *%ptr0, i64 8
%ptr5 = getelementptr double, double *%ptr0, i64 10
%ptr6 = getelementptr double, double *%ptr0, i64 12
%ptr7 = getelementptr double, double *%ptr0, i64 14
%ptr8 = getelementptr double, double *%ptr0, i64 16
%ptr9 = getelementptr double, double *%ptr0, i64 18
%ptr10 = getelementptr double, double *%ptr0, i64 20
%val0 = load double, double *%ptr0
%val1 = load double, double *%ptr1
%val2 = load double, double *%ptr2
%val3 = load double, double *%ptr3
%val4 = load double, double *%ptr4
%val5 = load double, double *%ptr5
%val6 = load double, double *%ptr6
%val7 = load double, double *%ptr7
%val8 = load double, double *%ptr8
%val9 = load double, double *%ptr9
%val10 = load double, double *%ptr10
%ret = call double @foo()
%sub0 = fsub double %ret, %val0
%sub1 = fsub double %sub0, %val1
%sub2 = fsub double %sub1, %val2
%sub3 = fsub double %sub2, %val3
%sub4 = fsub double %sub3, %val4
%sub5 = fsub double %sub4, %val5
%sub6 = fsub double %sub5, %val6
%sub7 = fsub double %sub6, %val7
%sub8 = fsub double %sub7, %val8
%sub9 = fsub double %sub8, %val9
%sub10 = fsub double %sub9, %val10
ret double %sub10
}
|