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
|
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -S -mtriple=amdgcn-- -passes=amdgpu-lower-module-lds < %s | FileCheck %s
; Not reached by a non-kernel function and therefore not changed by this pass
@dynamic_kernel_only = external addrspace(3) global [0 x double]
; shared8 is used directly by a kernel so remains in the outbout
; the other three are only used by functions and will be replaced with
; table lookups and dropped from the IR
@dynamic_shared1 = external addrspace(3) global [0 x i8], align 1
@dynamic_shared2 = external addrspace(3) global [0 x i16], align 2
@dynamic_shared4 = external addrspace(3) global [0 x i32], align 4
@dynamic_shared8 = external addrspace(3) global [0 x i64], align 8
; CHECK: %llvm.amdgcn.module.lds.t = type { i32 }
; CHECK: @dynamic_kernel_only = external addrspace(3) global [0 x double]
; CHECK: @dynamic_shared8 = external addrspace(3) global [0 x i64], align 8
; CHECK: @llvm.amdgcn.module.lds = internal addrspace(3) global %llvm.amdgcn.module.lds.t undef, align 4, !absolute_symbol !0
; CHECK: @llvm.compiler.used = appending global [1 x ptr] [ptr addrspacecast (ptr addrspace(3) @llvm.amdgcn.module.lds to ptr)], section "llvm.metadata"
; Alignment of these must be the maximum of the alignment of the reachable symbols
; CHECK: @llvm.amdgcn.expect_align1.dynlds = external addrspace(3) global [0 x i8], align 1, !absolute_symbol !0
; CHECK: @llvm.amdgcn.expect_align2.dynlds = external addrspace(3) global [0 x i8], align 2, !absolute_symbol !0
; CHECK: @llvm.amdgcn.expect_align4.dynlds = external addrspace(3) global [0 x i8], align 4, !absolute_symbol !1
; CHECK: @llvm.amdgcn.expect_align8.dynlds = external addrspace(3) global [0 x i8], align 8, !absolute_symbol !0
; Align 4 and symbol at address [4 5) as module.lds is reachable
; CHECK: @llvm.amdgcn.expect_max_of_2_and_4.dynlds = external addrspace(3) global [0 x i8], align 4, !absolute_symbol !1
; Builds a lookup table out of the newly created (suffixed .dynlds) variables in kernel.id order
; CHECK: @llvm.amdgcn.dynlds.offset.table = internal addrspace(4) constant [5 x i32] [i32 ptrtoint (ptr addrspace(3) @llvm.amdgcn.expect_align1.dynlds to i32), i32 ptrtoint (ptr addrspace(3) @llvm.amdgcn.expect_align2.dynlds to i32), i32 ptrtoint (ptr addrspace(3) @llvm.amdgcn.expect_align4.dynlds to i32), i32 ptrtoint (ptr addrspace(3) @llvm.amdgcn.expect_align8.dynlds to i32), i32 ptrtoint (ptr addrspace(3) @llvm.amdgcn.expect_max_of_2_and_4.dynlds to i32)]
define amdgpu_kernel void @kernel_only() {
; CHECK-LABEL: @kernel_only() {
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [0 x double], ptr addrspace(3) @dynamic_kernel_only, i32 0, i32 0
; CHECK-NEXT: store double 3.140000e+00, ptr addrspace(3) [[ARRAYIDX]], align 8
; CHECK-NEXT: ret void
;
%arrayidx = getelementptr inbounds [0 x double], ptr addrspace(3) @dynamic_kernel_only, i32 0, i32 0
store double 3.140000e+00, ptr addrspace(3) %arrayidx
ret void
}
; The accesses from functions are rewritten to go through the llvm.amdgcn.dynlds.offset.table
define void @use_shared1() {
; CHECK-LABEL: @use_shared1() {
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.amdgcn.lds.kernel.id()
; CHECK-NEXT: [[DYNAMIC_SHARED1:%.*]] = getelementptr inbounds [5 x i32], ptr addrspace(4) @llvm.amdgcn.dynlds.offset.table, i32 0, i32 [[TMP1]]
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(4) [[DYNAMIC_SHARED1]], align 4
; CHECK-NEXT: [[DYNAMIC_SHARED11:%.*]] = inttoptr i32 [[TMP2]] to ptr addrspace(3)
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [0 x i8], ptr addrspace(3) [[DYNAMIC_SHARED11]], i32 0, i32 1
; CHECK-NEXT: store i8 0, ptr addrspace(3) [[ARRAYIDX]], align 1
; CHECK-NEXT: ret void
;
%arrayidx = getelementptr inbounds [0 x i8], ptr addrspace(3) @dynamic_shared1, i32 0, i32 1
store i8 0, ptr addrspace(3) %arrayidx
ret void
}
define void @use_shared2() #0 {
; CHECK-LABEL: @use_shared2() #0 {
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.amdgcn.lds.kernel.id()
; CHECK-NEXT: [[DYNAMIC_SHARED2:%.*]] = getelementptr inbounds [5 x i32], ptr addrspace(4) @llvm.amdgcn.dynlds.offset.table, i32 0, i32 [[TMP1]]
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(4) [[DYNAMIC_SHARED2]], align 4
; CHECK-NEXT: [[DYNAMIC_SHARED21:%.*]] = inttoptr i32 [[TMP2]] to ptr addrspace(3)
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [0 x i16], ptr addrspace(3) [[DYNAMIC_SHARED21]], i32 0, i32 3
; CHECK-NEXT: store i16 1, ptr addrspace(3) [[ARRAYIDX]], align 2
; CHECK-NEXT: ret void
;
%arrayidx = getelementptr inbounds [0 x i16], ptr addrspace(3) @dynamic_shared2, i32 0, i32 3
store i16 1, ptr addrspace(3) %arrayidx
ret void
}
; Include a normal variable so that the new variables aren't all at the same absolute_symbol
@static_shared = addrspace(3) global i32 undef
define void @use_shared4() #0 {
; CHECK-LABEL: @use_shared4() #0 {
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.amdgcn.lds.kernel.id()
; CHECK-NEXT: store i32 4, ptr addrspace(3) @llvm.amdgcn.module.lds, align 4
; CHECK-NEXT: [[DYNAMIC_SHARED4:%.*]] = getelementptr inbounds [5 x i32], ptr addrspace(4) @llvm.amdgcn.dynlds.offset.table, i32 0, i32 [[TMP1]]
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(4) [[DYNAMIC_SHARED4]], align 4
; CHECK-NEXT: [[DYNAMIC_SHARED41:%.*]] = inttoptr i32 [[TMP2]] to ptr addrspace(3)
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [0 x i32], ptr addrspace(3) [[DYNAMIC_SHARED41]], i32 0, i32 5
; CHECK-NEXT: store i32 2, ptr addrspace(3) [[ARRAYIDX]], align 4
; CHECK-NEXT: ret void
;
store i32 4, ptr addrspace(3) @static_shared
%arrayidx = getelementptr inbounds [0 x i32], ptr addrspace(3) @dynamic_shared4, i32 0, i32 5
store i32 2, ptr addrspace(3) %arrayidx
ret void
}
define void @use_shared8() #0 {
; CHECK-LABEL: @use_shared8() #0 {
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @llvm.amdgcn.lds.kernel.id()
; CHECK-NEXT: [[DYNAMIC_SHARED8:%.*]] = getelementptr inbounds [5 x i32], ptr addrspace(4) @llvm.amdgcn.dynlds.offset.table, i32 0, i32 [[TMP1]]
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr addrspace(4) [[DYNAMIC_SHARED8]], align 4
; CHECK-NEXT: [[DYNAMIC_SHARED81:%.*]] = inttoptr i32 [[TMP2]] to ptr addrspace(3)
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [0 x i64], ptr addrspace(3) [[DYNAMIC_SHARED81]], i32 0, i32 7
; CHECK-NEXT: store i64 3, ptr addrspace(3) [[ARRAYIDX]], align 4
; CHECK-NEXT: ret void
;
%arrayidx = getelementptr inbounds [0 x i64], ptr addrspace(3) @dynamic_shared8, i32 0, i32 7
store i64 3, ptr addrspace(3) %arrayidx
ret void
}
; The kernels are annotated with kernel.id and llvm.donothing use of the corresponding variable
define amdgpu_kernel void @expect_align1() {
; CHECK-LABEL: @expect_align1() !llvm.amdgcn.lds.kernel.id !2
; CHECK-NEXT: call void @llvm.donothing() [ "ExplicitUse"(ptr addrspace(3) @llvm.amdgcn.expect_align1.dynlds) ]
; CHECK-NEXT: call void @use_shared1()
; CHECK-NEXT: ret void
;
call void @use_shared1()
ret void
}
define amdgpu_kernel void @expect_align2() {
; CHECK-LABEL: @expect_align2() !llvm.amdgcn.lds.kernel.id !3
; CHECK-NEXT: call void @llvm.donothing() [ "ExplicitUse"(ptr addrspace(3) @llvm.amdgcn.expect_align2.dynlds) ]
; CHECK-NEXT: call void @use_shared2()
; CHECK-NEXT: ret void
;
call void @use_shared2()
ret void
}
define amdgpu_kernel void @expect_align4() {
; CHECK-LABEL: @expect_align4() #1 !llvm.amdgcn.lds.kernel.id !4 {
; CHECK-NEXT: call void @llvm.donothing() [ "ExplicitUse"(ptr addrspace(3) @llvm.amdgcn.expect_align4.dynlds) ]
; CHECK-NEXT: call void @llvm.donothing() [ "ExplicitUse"(ptr addrspace(3) @llvm.amdgcn.module.lds) ]
; CHECK-NEXT: call void @use_shared4()
; CHECK-NEXT: ret void
;
call void @use_shared4()
ret void
}
; Use dynamic_shared directly too.
define amdgpu_kernel void @expect_align8() {
; CHECK-LABEL: @expect_align8() !llvm.amdgcn.lds.kernel.id !5 {
; CHECK-NEXT: call void @llvm.donothing() [ "ExplicitUse"(ptr addrspace(3) @llvm.amdgcn.expect_align8.dynlds) ]
; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [0 x i64], ptr addrspace(3) @dynamic_shared8, i32 0, i32 9
; CHECK-NEXT: store i64 3, ptr addrspace(3) [[ARRAYIDX]], align 4
; CHECK-NEXT: call void @use_shared8()
; CHECK-NEXT: ret void
;
%arrayidx = getelementptr inbounds [0 x i64], ptr addrspace(3) @dynamic_shared8, i32 0, i32 9
store i64 3, ptr addrspace(3) %arrayidx
call void @use_shared8()
ret void
}
; Note: use_shared4 uses module.lds so this will allocate at offset 4
define amdgpu_kernel void @expect_max_of_2_and_4() {
; CHECK-LABEL: @expect_max_of_2_and_4() #1 !llvm.amdgcn.lds.kernel.id !6 {
; CHECK-NEXT: call void @llvm.donothing() [ "ExplicitUse"(ptr addrspace(3) @llvm.amdgcn.expect_max_of_2_and_4.dynlds) ]
; CHECK-NEXT: call void @llvm.donothing() [ "ExplicitUse"(ptr addrspace(3) @llvm.amdgcn.module.lds) ]
; CHECK-NEXT: call void @use_shared2()
; CHECK-NEXT: call void @use_shared4()
; CHECK-NEXT: ret void
;
call void @use_shared2()
call void @use_shared4()
ret void
}
attributes #0 = { noinline }
; Function Attrs: nocallback nofree nosync nounwind willreturn memory(none)
; CHECK: declare void @llvm.donothing() #2
; Function Attrs: nocallback nofree nosync nounwind speculatable willreturn memory(none)
; CHECK: declare i32 @llvm.amdgcn.lds.kernel.id() #3
; CHECK: attributes #0 = { noinline }
; CHECK: attributes #1 = { "amdgpu-lds-size"="4,4" }
; CHECK: attributes #2 = { nocallback nofree nosync nounwind willreturn memory(none) }
; CHECK: attributes #3 = { nocallback nofree nosync nounwind speculatable willreturn memory(none) }
; CHECK: !0 = !{i64 0, i64 1}
; CHECK: !1 = !{i64 4, i64 5}
; CHECK: !2 = !{i32 0}
; CHECK: !3 = !{i32 1}
; CHECK: !4 = !{i32 2}
; CHECK: !5 = !{i32 3}
; CHECK: !6 = !{i32 4}
|
