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
|
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
# RUN: llc -opaque-pointers=0 -mtriple=thumbv8.1m.main -mattr=+mve -run-pass=arm-low-overhead-loops %s -o - | FileCheck %s
--- |
@arr = external dso_local local_unnamed_addr global [0 x i32], align 4
define dso_local arm_aapcs_vfpcc void @foo(i32 %i) {
entry:
%tobool.not11 = icmp eq i32 %i, 0
br i1 %tobool.not11, label %for.end5, label %vector.ph.preheader
vector.ph.preheader: ; preds = %entry
%active.lane.mask = call <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32 0, i32 3)
br label %vector.ph
vector.ph: ; preds = %vector.ph.preheader, %vector.ph
%i.addr.012 = phi i32 [ %math, %vector.ph ], [ %i, %vector.ph.preheader ]
call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> <i32 2, i32 2, i32 2, i32 2>, <4 x i32>* bitcast ([0 x i32]* @arr to <4 x i32>*), i32 4, <4 x i1> %active.lane.mask)
%0 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 %i.addr.012, i32 1)
%math = extractvalue { i32, i1 } %0, 0
%ov = extractvalue { i32, i1 } %0, 1
br i1 %ov, label %for.end5, label %vector.ph
for.end5: ; preds = %vector.ph, %entry
ret void
}
declare <4 x i1> @llvm.get.active.lane.mask.v4i1.i32(i32, i32)
declare void @llvm.masked.store.v4i32.p0v4i32(<4 x i32>, <4 x i32>*, i32 immarg, <4 x i1>)
declare { i32, i1 } @llvm.uadd.with.overflow.i32(i32, i32)
...
---
name: foo
alignment: 8
tracksRegLiveness: true
registers: []
liveins:
- { reg: '$r0', virtual-reg: '' }
frameInfo:
maxAlignment: 1
stackProtector: ''
maxCallFrameSize: 0
cvBytesOfCalleeSavedRegisters: 0
localFrameSize: 0
savePoint: ''
restorePoint: ''
fixedStack: []
stack: []
callSites: []
constants:
- id: 0
value: '<4 x i32> <i32 0, i32 1, i32 2, i32 3>'
alignment: 8
isTargetSpecific: false
machineFunctionInfo: {}
body: |
; CHECK-LABEL: name: foo
; CHECK: bb.0.entry:
; CHECK: successors: %bb.3(0x30000000), %bb.1(0x50000000)
; CHECK: liveins: $r0
; CHECK: tCBZ $r0, %bb.3
; CHECK: bb.1.vector.ph.preheader:
; CHECK: successors: %bb.2(0x80000000)
; CHECK: liveins: $r0
; CHECK: renamable $r1 = tLEApcrel %const.0, 14 /* CC::al */, $noreg
; CHECK: renamable $q0 = MVE_VMOVimmi32 3, 0, $noreg, $noreg, undef renamable $q0
; CHECK: renamable $q1 = MVE_VLDRWU32 killed renamable $r1, 0, 0, $noreg, $noreg :: (load (s128) from constant-pool, align 8)
; CHECK: $r1 = t2MOVi16 target-flags(arm-lo16) @arr, 14 /* CC::al */, $noreg
; CHECK: $r1 = t2MOVTi16 killed $r1, target-flags(arm-hi16) @arr, 14 /* CC::al */, $noreg
; CHECK: renamable $vpr = MVE_VCMPu32 killed renamable $q0, killed renamable $q1, 8, 0, $noreg, $noreg
; CHECK: renamable $q0 = MVE_VMOVimmi32 2, 0, $noreg, $noreg, undef renamable $q0
; CHECK: bb.2.vector.ph:
; CHECK: successors: %bb.3(0x04000000), %bb.2(0x7c000000)
; CHECK: liveins: $vpr, $q0, $r0, $r1
; CHECK: renamable $r0, $cpsr = tADDi8 killed renamable $r0, 1, 14 /* CC::al */, $noreg
; CHECK: MVE_VPST 8, implicit $vpr
; CHECK: MVE_VSTRWU32 renamable $q0, renamable $r1, 0, 1, renamable $vpr, $noreg :: (store (s128) into `<4 x i32>* bitcast ([0 x i32]* @arr to <4 x i32>*)`, align 4)
; CHECK: tBcc %bb.2, 3 /* CC::lo */, killed $cpsr
; CHECK: bb.3.for.end5:
; CHECK: tBX_RET 14 /* CC::al */, $noreg
; CHECK: bb.4 (align 8):
; CHECK: CONSTPOOL_ENTRY 0, %const.0, 16
bb.0.entry:
successors: %bb.3(0x30000000), %bb.1(0x50000000)
liveins: $r0
tCBZ $r0, %bb.3
bb.1.vector.ph.preheader:
successors: %bb.2(0x80000000)
liveins: $r0
renamable $r1 = tLEApcrel %const.0, 14 /* CC::al */, $noreg
renamable $q0 = MVE_VMOVimmi32 3, 0, $noreg, $noreg, undef renamable $q0
renamable $q1 = MVE_VLDRWU32 killed renamable $r1, 0, 0, $noreg, $noreg :: (load (s128) from constant-pool, align 8)
$r1 = t2MOVi16 target-flags(arm-lo16) @arr, 14 /* CC::al */, $noreg
$r1 = t2MOVTi16 killed $r1, target-flags(arm-hi16) @arr, 14 /* CC::al */, $noreg
renamable $vpr = MVE_VCMPu32 killed renamable $q0, killed renamable $q1, 8, 0, $noreg, $noreg
renamable $q0 = MVE_VMOVimmi32 2, 0, $noreg, $noreg, undef renamable $q0
bb.2.vector.ph:
successors: %bb.3(0x04000000), %bb.2(0x7c000000)
liveins: $vpr, $q0, $r0, $r1
renamable $r0, $cpsr = tADDi8 killed renamable $r0, 1, 14 /* CC::al */, $noreg
MVE_VPST 8, implicit $vpr
MVE_VSTRWU32 renamable $q0, renamable $r1, 0, 1, renamable $vpr, $noreg :: (store (s128) into `<4 x i32>* bitcast ([0 x i32]* @arr to <4 x i32>*)`, align 4)
tBcc %bb.2, 3 /* CC::lo */, killed $cpsr
bb.3.for.end5:
tBX_RET 14 /* CC::al */, $noreg
bb.4 (align 8):
CONSTPOOL_ENTRY 0, %const.0, 16
...
|
