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
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
| --- |
; pr37472
; These test verify that shrink-wrap does not set the restore point
; to a position where the stack might still be accessed by a load or store
; RUN: llc -x=mir -simplify-mir -run-pass=shrink-wrap -o - %s | FileCheck %s
; CHECK: name: compiler_pop_stack
; CHECK: frameInfo:
; CHECK-NOT: savePoint:
; CHECK-NOT: restorePoint:
; CHECK: stack:
; CHECK: name: f
; CHECK: frameInfo:
; CHECK: savePoint: '%bb.2'
; CHECK-NEXT: restorePoint: '%bb.4'
; CHECK-NEXT: stack:
target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
target triple = "aarch64-arm-none-eabi"
%struct.S = type { i32, i32 }
@__const.f.arr = private unnamed_addr constant [4 x i8] c"\01\02\03\04", align 1
declare void @llvm.lifetime.start.p0i8(i64 immarg, i8* nocapture)
declare void @llvm.lifetime.end.p0i8(i64 immarg, i8* nocapture)
declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture writeonly, i8* nocapture readonly, i64, i1 immarg)
; Test from: https://bugs.llvm.org/show_bug.cgi?id=42136
define void @compiler_pop_stack(i32 %num) {
entry:
%rstack = alloca [16 x i32], align 4
%tmp = bitcast [16 x i32]* %rstack to i8*
call void @llvm.lifetime.start.p0i8(i64 64, i8* nonnull %tmp)
%cmp = icmp ult i32 %num, 2
br i1 %cmp, label %cleanup, label %if.end
if.end:
%arrayidx1 = bitcast [16 x i32]* %rstack to i32*
store volatile i32 %num, i32* %arrayidx1, align 4
br label %while.body
while.body:
%ptr.017 = phi i32 [ 1, %if.end ], [ %ptr.1, %if.end7 ]
%dec = add i32 %ptr.017, -1
%idxprom = zext i32 %dec to i64
%arrayidx2 = getelementptr inbounds [16 x i32], [16 x i32]* %rstack, i64 0, i64 %idxprom
%tmp1 = load volatile i32, i32* %arrayidx2, align 4
%cmp3 = icmp eq i32 %tmp1, 0
br i1 %cmp3, label %if.end7, label %if.then4
if.then4:
%sunkaddr = mul i64 %idxprom, 4
%0 = bitcast [16 x i32]* %rstack to i8*
%sunkaddr2 = getelementptr inbounds i8, i8* %0, i64 %sunkaddr
%1 = bitcast i8* %sunkaddr2 to i32*
store volatile i32 %tmp1, i32* %1, align 4
br label %if.end7
if.end7:
%ptr.1 = phi i32 [ %ptr.017, %if.then4 ], [ %dec, %while.body ]
%cmp1 = icmp eq i32 %ptr.1, 0
br i1 %cmp1, label %cleanup, label %while.body
cleanup:
%2 = bitcast [16 x i32]* %rstack to i8*
call void @llvm.lifetime.end.p0i8(i64 64, i8* nonnull %2)
ret void
}
; Test from: https://bugs.llvm.org/show_bug.cgi?id=37472
define i32 @f(%struct.S* nocapture %arg, i32 %arg1) {
bb:
%tmp = alloca [4 x i8], align 1
%tmp2 = icmp ugt i32 %arg1, 4
br i1 %tmp2, label %bb16, label %bb3
bb3:
%tmp41 = bitcast [4 x i8]* %tmp to i8*
call void @llvm.lifetime.start.p0i8(i64 4, i8* nonnull %tmp41)
call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 1 %tmp41, i8* align 1 getelementptr inbounds ([4 x i8], [4 x i8]* @__const.f.arr, i64 0, i64 0), i64 4, i1 true)
%tmp5 = zext i32 %arg1 to i64
%tmp6 = getelementptr inbounds [4 x i8], [4 x i8]* %tmp, i64 0, i64 %tmp5
%tmp7 = load volatile i8, i8* %tmp6, align 1
%tmp8 = zext i8 %tmp7 to i32
%tmp92 = bitcast %struct.S* %arg to i32*
store i32 %tmp8, i32* %tmp92, align 4
%tmp10 = icmp ult i32 %arg1, 3
br i1 %tmp10, label %bb11, label %bb15
bb11:
%0 = bitcast [4 x i8]* %tmp to i8*
%sunkaddr = getelementptr inbounds i8, i8* %0, i64 %tmp5
%tmp12 = load volatile i8, i8* %sunkaddr, align 1
%tmp13 = zext i8 %tmp12 to i32
%tmp14 = getelementptr inbounds %struct.S, %struct.S* %arg, i64 0, i32 1
store i32 %tmp13, i32* %tmp14, align 4
br label %bb15
bb15:
%1 = bitcast [4 x i8]* %tmp to i8*
call void @llvm.lifetime.end.p0i8(i64 4, i8* nonnull %1)
br label %bb16
bb16:
%tmp17 = phi i32 [ 0, %bb15 ], [ 1, %bb ]
ret i32 %tmp17
}
declare void @llvm.stackprotector(i8*, i8**)
...
---
name: compiler_pop_stack
alignment: 4
tracksRegLiveness: true
liveins:
- { reg: '$w0' }
frameInfo:
maxAlignment: 4
maxCallFrameSize: 0
localFrameSize: 64
stack:
- { id: 0, name: rstack, size: 64, alignment: 4, stack-id: default, local-offset: -64 }
machineFunctionInfo: {}
body: |
bb.0.entry:
liveins: $w0
dead $wzr = SUBSWri renamable $w0, 2, 0, implicit-def $nzcv
Bcc 3, %bb.6, implicit killed $nzcv
B %bb.1
bb.1.if.end:
liveins: $w0
STRWui killed renamable $w0, %stack.0.rstack, 0 :: (volatile store 4 into %ir.arrayidx1)
renamable $w9 = MOVi32imm 1
renamable $x8 = ADDXri %stack.0.rstack, 0, 0
bb.2.while.body:
successors: %bb.3(0x30000000), %bb.4(0x50000000)
liveins: $w9, $x8
renamable $w10 = SUBWri renamable $w9, 1, 0, implicit-def $x10
renamable $w11 = LDRWroW renamable $x8, renamable $w10, 0, 1 :: (volatile load 4 from %ir.arrayidx2)
CBNZW renamable $w11, %bb.4
bb.3:
liveins: $x8, $x10
renamable $w9 = COPY renamable $w10, implicit killed $x10
B %bb.5
bb.4.if.then4:
liveins: $w9, $w11, $x8, $x10
STRWroX killed renamable $w11, renamable $x8, killed renamable $x10, 0, 1 :: (volatile store 4 into %ir.1)
bb.5.if.end7:
successors: %bb.6(0x04000000), %bb.2(0x7c000000)
liveins: $w9, $x8
CBNZW renamable $w9, %bb.2
B %bb.6
bb.6.cleanup:
RET_ReallyLR
...
---
name: f
alignment: 4
tracksRegLiveness: true
liveins:
- { reg: '$x0' }
- { reg: '$w1' }
frameInfo:
maxAlignment: 4
maxCallFrameSize: 0
localFrameSize: 4
stack:
- { id: 0, name: tmp, size: 4, alignment: 4, stack-id: default, local-offset: -4 }
machineFunctionInfo: {}
body: |
bb.0.bb:
successors: %bb.1, %bb.2
liveins: $w1, $x0
dead $wzr = SUBSWri renamable $w1, 4, 0, implicit-def $nzcv
Bcc 9, %bb.2, implicit killed $nzcv
bb.1:
renamable $w0 = MOVi32imm 1
B %bb.5
bb.2.bb3:
successors: %bb.3, %bb.4
liveins: $w1, $x0
renamable $w9 = MOVi32imm 67305985
renamable $w8 = ORRWrs $wzr, renamable $w1, 0, implicit-def $x8
STRWui killed renamable $w9, %stack.0.tmp, 0 :: (volatile store 4 into %ir.tmp41)
renamable $x9 = ADDXri %stack.0.tmp, 0, 0
renamable $w10 = LDRBBroX renamable $x9, renamable $x8, 0, 0 :: (volatile load 1 from %ir.tmp6)
dead $wzr = SUBSWri killed renamable $w1, 2, 0, implicit-def $nzcv
STRWui killed renamable $w10, renamable $x0, 0 :: (store 4 into %ir.tmp92)
Bcc 8, %bb.4, implicit killed $nzcv
B %bb.3
bb.3.bb11:
liveins: $x0, $x8, $x9
renamable $w8 = LDRBBroX killed renamable $x9, killed renamable $x8, 0, 0 :: (volatile load 1 from %ir.sunkaddr)
STRWui killed renamable $w8, killed renamable $x0, 1 :: (store 4 into %ir.tmp14)
bb.4.bb15:
renamable $w0 = COPY $wzr
bb.5.bb16:
liveins: $w0
RET_ReallyLR implicit $w0
...
|