1// RUN: %clang_cc1 %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir-unknown-unknown" | FileCheck %s --check-prefix=COMMON --check-prefix=B32
2// RUN: %clang_cc1 %s -cl-std=CL2.0 -ffake-address-space-map -O0 -emit-llvm -o - -triple "spir64-unknown-unknown" | FileCheck %s --check-prefix=COMMON --check-prefix=B64
3
4#pragma OPENCL EXTENSION cl_khr_subgroups : enable
5
6typedef void (^bl_t)(local void *);
7typedef struct {int a;} ndrange_t;
8
9// N.B. The check here only exists to set BL_GLOBAL
10// COMMON: @block_G =  addrspace(1) constant void (i8 addrspace(3)*) addrspace(4)* addrspacecast (void (i8 addrspace(3)*) addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* [[BL_GLOBAL:@__block_literal_global(\.[0-9]+)?]] to void (i8 addrspace(3)*) addrspace(1)*) to void (i8 addrspace(3)*) addrspace(4)*)
11const bl_t block_G = (bl_t) ^ (local void *a) {};
12
13kernel void device_side_enqueue(global int *a, global int *b, int i) {
14  // COMMON: %default_queue = alloca %opencl.queue_t*
15  queue_t default_queue;
16  // COMMON: %flags = alloca i32
17  unsigned flags = 0;
18  // COMMON: %ndrange = alloca %struct.ndrange_t
19  ndrange_t ndrange;
20  // COMMON: %clk_event = alloca %opencl.clk_event_t*
21  clk_event_t clk_event;
22  // COMMON: %event_wait_list = alloca %opencl.clk_event_t*
23  clk_event_t event_wait_list;
24  // COMMON: %event_wait_list2 = alloca [1 x %opencl.clk_event_t*]
25  clk_event_t event_wait_list2[] = {clk_event};
26
27  // COMMON: [[NDR:%[a-z0-9]+]] = alloca %struct.ndrange_t, align 4
28  // COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue
29  // COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags
30  // COMMON: [[BL:%[0-9]+]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor addrspace(2)*, i32{{.*}}, i32{{.*}}, i32{{.*}} }>* %block to void ()*
31  // COMMON: [[BL_I8:%[0-9]+]] = addrspacecast void ()* [[BL]] to i8 addrspace(4)*
32  // COMMON: call i32 @__enqueue_kernel_basic(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* byval [[NDR]]{{(.[0-9]+)?}}, i8 addrspace(4)* [[BL_I8]])
33  enqueue_kernel(default_queue, flags, ndrange,
34                 ^(void) {
35                   a[i] = b[i];
36                 });
37
38  // COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue
39  // COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags
40  // COMMON: [[WAIT_EVNT:%[0-9]+]] = addrspacecast %opencl.clk_event_t{{.*}}** %event_wait_list to %opencl.clk_event_t{{.*}}* addrspace(4)*
41  // COMMON: [[EVNT:%[0-9]+]] = addrspacecast %opencl.clk_event_t{{.*}}** %clk_event to %opencl.clk_event_t{{.*}}* addrspace(4)*
42  // COMMON: [[BL:%[0-9]+]] = bitcast <{ i8*, i32, i32, i8*, %struct.__block_descriptor addrspace(2)*, i32{{.*}}, i32{{.*}}, i32{{.*}} }>* %block3 to void ()*
43  // COMMON: [[BL_I8:%[0-9]+]] = addrspacecast void ()* [[BL]] to i8 addrspace(4)*
44  // COMMON: call i32 @__enqueue_kernel_basic_events(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]],  %struct.ndrange_t* {{.*}}, i32 2, %opencl.clk_event_t{{.*}}* addrspace(4)* [[WAIT_EVNT]], %opencl.clk_event_t{{.*}}* addrspace(4)* [[EVNT]], i8 addrspace(4)* [[BL_I8]])
45  enqueue_kernel(default_queue, flags, ndrange, 2, &event_wait_list, &clk_event,
46                 ^(void) {
47                   a[i] = b[i];
48                 });
49
50  // COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue
51  // COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags
52  // B32: %[[TMP:.*]] = alloca [1 x i32]
53  // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0
54  // B32: store i32 256, i32* %[[TMP1]], align 4
55  // B32: call i32 @__enqueue_kernel_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{(.[0-9]+)?}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i32* %[[TMP1]])
56  // B64: %[[TMP:.*]] = alloca [1 x i64]
57  // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0
58  // B64: store i64 256, i64* %[[TMP1]], align 8
59  // B64: call i32 @__enqueue_kernel_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{(.[0-9]+)?}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i64* %[[TMP1]])
60  enqueue_kernel(default_queue, flags, ndrange,
61                 ^(local void *p) {
62                   return;
63                 },
64                 256);
65  char c;
66  // COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue
67  // COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags
68  // B32: %[[TMP:.*]] = alloca [1 x i32]
69  // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0
70  // B32: store i32 %{{.*}}, i32* %[[TMP1]], align 4
71  // B32: call i32 @__enqueue_kernel_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{(.[0-9]+)?}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i32* %[[TMP1]])
72  // B64: %[[TMP:.*]] = alloca [1 x i64]
73  // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0
74  // B64: store i64 %{{.*}}, i64* %[[TMP1]], align 8
75  // B64: call i32 @__enqueue_kernel_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{(.[0-9]+)?}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i64* %[[TMP1]])
76  enqueue_kernel(default_queue, flags, ndrange,
77                 ^(local void *p) {
78                   return;
79                 },
80                 c);
81
82  // COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue
83  // COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags
84  // COMMON: [[AD:%arraydecay[0-9]*]] = getelementptr inbounds [1 x %opencl.clk_event_t*], [1 x %opencl.clk_event_t*]* %event_wait_list2, i32 0, i32 0
85  // COMMON: [[WAIT_EVNT:%[0-9]+]] = addrspacecast %opencl.clk_event_t{{.*}}** [[AD]] to %opencl.clk_event_t{{.*}}* addrspace(4)*
86  // COMMON: [[EVNT:%[0-9]+]]  = addrspacecast %opencl.clk_event_t{{.*}}** %clk_event to %opencl.clk_event_t{{.*}}* addrspace(4)*
87  // B32: %[[TMP:.*]] = alloca [1 x i32]
88  // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0
89  // B32: store i32 256, i32* %[[TMP1]], align 4
90  // B32: call i32 @__enqueue_kernel_events_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]],  %struct.ndrange_t* {{.*}}, i32 2, %opencl.clk_event_t{{.*}} [[WAIT_EVNT]], %opencl.clk_event_t{{.*}} [[EVNT]], i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i32* %[[TMP1]])
91  // B64: %[[TMP:.*]] = alloca [1 x i64]
92  // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0
93  // B64: store i64 256, i64* %[[TMP1]], align 8
94  // B64: call i32 @__enqueue_kernel_events_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]],  %struct.ndrange_t* {{.*}}, i32 2, %opencl.clk_event_t{{.*}} [[WAIT_EVNT]], %opencl.clk_event_t{{.*}} [[EVNT]], i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i64* %[[TMP1]])
95  enqueue_kernel(default_queue, flags, ndrange, 2, event_wait_list2, &clk_event,
96                 ^(local void *p) {
97                   return;
98                 },
99                 256);
100
101  // COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue
102  // COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags
103  // COMMON: [[AD:%arraydecay[0-9]*]] = getelementptr inbounds [1 x %opencl.clk_event_t*], [1 x %opencl.clk_event_t*]* %event_wait_list2, i32 0, i32 0
104  // COMMON: [[WAIT_EVNT:%[0-9]+]] = addrspacecast %opencl.clk_event_t{{.*}}** [[AD]] to %opencl.clk_event_t{{.*}}* addrspace(4)*
105  // COMMON: [[EVNT:%[0-9]+]]  = addrspacecast %opencl.clk_event_t{{.*}}** %clk_event to %opencl.clk_event_t{{.*}}* addrspace(4)*
106  // B32: %[[TMP:.*]] = alloca [1 x i32]
107  // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0
108  // B32: store i32 %{{.*}}, i32* %[[TMP1]], align 4
109  // B32: call i32 @__enqueue_kernel_events_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]],  %struct.ndrange_t* {{.*}}, i32 2, %opencl.clk_event_t{{.*}}* addrspace(4)* [[WAIT_EVNT]], %opencl.clk_event_t{{.*}}* addrspace(4)* [[EVNT]], i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i32* %[[TMP1]])
110  // B64: %[[TMP:.*]] = alloca [1 x i64]
111  // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0
112  // B64: store i64 %{{.*}}, i64* %[[TMP1]], align 8
113  // B64: call i32 @__enqueue_kernel_events_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]],  %struct.ndrange_t* {{.*}}, i32 2, %opencl.clk_event_t{{.*}}* addrspace(4)* [[WAIT_EVNT]], %opencl.clk_event_t{{.*}}* addrspace(4)* [[EVNT]], i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i64* %[[TMP1]])
114  enqueue_kernel(default_queue, flags, ndrange, 2, event_wait_list2, &clk_event,
115                 ^(local void *p) {
116                   return;
117                 },
118                 c);
119
120  long l;
121  // COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue
122  // COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags
123  // B32: %[[TMP:.*]] = alloca [1 x i32]
124  // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0
125  // B32: store i32 %{{.*}}, i32* %[[TMP1]], align 4
126  // B32: call i32 @__enqueue_kernel_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{(.[0-9]+)?}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i32* %[[TMP1]])
127  // B64: %[[TMP:.*]] = alloca [1 x i64]
128  // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0
129  // B64: store i64 %{{.*}}, i64* %[[TMP1]], align 8
130  // B64: call i32 @__enqueue_kernel_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{(.[0-9]+)?}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i64* %[[TMP1]])
131  enqueue_kernel(default_queue, flags, ndrange,
132                 ^(local void *p) {
133                   return;
134                 },
135                 l);
136
137  // COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t{{.*}}*, %opencl.queue_t{{.*}}** %default_queue
138  // COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags
139  // B32: %[[TMP:.*]] = alloca [3 x i32]
140  // B32: %[[TMP1:.*]] = getelementptr [3 x i32], [3 x i32]* %[[TMP]], i32 0, i32 0
141  // B32: store i32 1, i32* %[[TMP1]], align 4
142  // B32: %[[TMP2:.*]] = getelementptr [3 x i32], [3 x i32]* %[[TMP]], i32 0, i32 1
143  // B32: store i32 2, i32* %[[TMP2]], align 4
144  // B32: %[[TMP3:.*]] = getelementptr [3 x i32], [3 x i32]* %[[TMP]], i32 0, i32 2
145  // B32: store i32 4, i32* %[[TMP3]], align 4
146  // B32: call i32 @__enqueue_kernel_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{(.[0-9]+)?}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 3, i32* %[[TMP1]])
147  // B64: %[[TMP:.*]] = alloca [3 x i64]
148  // B64: %[[TMP1:.*]] = getelementptr [3 x i64], [3 x i64]* %[[TMP]], i32 0, i32 0
149  // B64: store i64 1, i64* %[[TMP1]], align 8
150  // B64: %[[TMP2:.*]] = getelementptr [3 x i64], [3 x i64]* %[[TMP]], i32 0, i32 1
151  // B64: store i64 2, i64* %[[TMP2]], align 8
152  // B64: %[[TMP3:.*]] = getelementptr [3 x i64], [3 x i64]* %[[TMP]], i32 0, i32 2
153  // B64: store i64 4, i64* %[[TMP3]], align 8
154  // B64: call i32 @__enqueue_kernel_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{(.[0-9]+)?}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 3, i64* %[[TMP1]])
155  enqueue_kernel(default_queue, flags, ndrange,
156                 ^(local void *p1, local void *p2, local void *p3) {
157                   return;
158                 },
159                 1, 2, 4);
160
161  // COMMON: [[DEF_Q:%[0-9]+]] = load %opencl.queue_t*, %opencl.queue_t** %default_queue
162  // COMMON: [[FLAGS:%[0-9]+]] = load i32, i32* %flags
163  // B32: %[[TMP:.*]] = alloca [1 x i32]
164  // B32: %[[TMP1:.*]] = getelementptr [1 x i32], [1 x i32]* %[[TMP]], i32 0, i32 0
165  // B32: store i32 0, i32* %[[TMP1]], align 4
166  // B32: call i32 @__enqueue_kernel_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{(.[0-9]+)?}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i32* %[[TMP1]])
167  // B64: %[[TMP:.*]] = alloca [1 x i64]
168  // B64: %[[TMP1:.*]] = getelementptr [1 x i64], [1 x i64]* %[[TMP]], i32 0, i32 0
169  // B64: store i64 4294967296, i64* %[[TMP1]], align 8
170  // B64: call i32 @__enqueue_kernel_vaargs(%opencl.queue_t{{.*}}* [[DEF_Q]], i32 [[FLAGS]], %struct.ndrange_t* [[NDR]]{{(.[0-9]+)?}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* @__block_literal_global{{(.[0-9]+)?}} to i8 addrspace(1)*) to i8 addrspace(4)*), i32 1, i64* %[[TMP1]])
171  enqueue_kernel(default_queue, flags, ndrange,
172                 ^(local void *p) {
173                   return;
174                 },
175                 4294967296L);
176
177  // The full type of these expressions are long (and repeated elsewhere), so we
178  // capture it as part of the regex for convenience and clarity.
179  // COMMON: store void () addrspace(4)* addrspacecast (void () addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* [[BL_A:@__block_literal_global(\.[0-9]+)?]] to void () addrspace(1)*) to void () addrspace(4)*), void () addrspace(4)** %block_A
180  void (^const block_A)(void) = ^{
181    return;
182  };
183
184  // COMMON: store void (i8 addrspace(3)*) addrspace(4)* addrspacecast (void (i8 addrspace(3)*) addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* [[BL_B:@__block_literal_global(\.[0-9]+)?]] to void (i8 addrspace(3)*) addrspace(1)*) to void (i8 addrspace(3)*) addrspace(4)*), void (i8 addrspace(3)*) addrspace(4)** %block_B
185  void (^const block_B)(local void *) = ^(local void *a) {
186    return;
187  };
188
189  // COMMON: call i32 @__get_kernel_work_group_size_impl(i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* [[BL_A]] to i8 addrspace(1)*) to i8 addrspace(4)*))
190  unsigned size = get_kernel_work_group_size(block_A);
191  // COMMON: call i32 @__get_kernel_work_group_size_impl(i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* [[BL_B]] to i8 addrspace(1)*) to i8 addrspace(4)*))
192  size = get_kernel_work_group_size(block_B);
193  // COMMON: call i32 @__get_kernel_preferred_work_group_multiple_impl(i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* [[BL_A]] to i8 addrspace(1)*) to i8 addrspace(4)*))
194  size = get_kernel_preferred_work_group_size_multiple(block_A);
195  // COMMON: call i32 @__get_kernel_preferred_work_group_multiple_impl(i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* [[BL_GLOBAL]] to i8 addrspace(1)*) to i8 addrspace(4)*))
196  size = get_kernel_preferred_work_group_size_multiple(block_G);
197
198  // COMMON: call i32 @__get_kernel_max_sub_group_size_for_ndrange_impl(%struct.ndrange_t* {{.*}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* {{.*}} to i8 addrspace(1)*) to i8 addrspace(4)*))
199  size = get_kernel_max_sub_group_size_for_ndrange(ndrange, ^(){});
200  // COMMON: call i32 @__get_kernel_sub_group_count_for_ndrange_impl(%struct.ndrange_t* {{.*}}, i8 addrspace(4)* addrspacecast (i8 addrspace(1)* bitcast ({ i8**, i32, i32, i8*, %struct.__block_descriptor addrspace(2)* } addrspace(1)* {{.*}} to i8 addrspace(1)*) to i8 addrspace(4)*))
201  size = get_kernel_sub_group_count_for_ndrange(ndrange, ^(){});
202}
203