1// RUN: %clang_cc1 %s -cl-std=CL2.0 -emit-llvm -o - -O0 -triple spir-unknown-unknown | FileCheck -check-prefixes=COMMON,SPIR %s
2// RUN: %clang_cc1 %s -cl-std=CL2.0 -emit-llvm -o - -O0 -triple amdgcn-amd-amdhsa | FileCheck -check-prefixes=COMMON,AMDGCN %s
3
4// SPIR: %struct.__opencl_block_literal_generic = type { i32, i32, i8 addrspace(4)* }
5// AMDGCN: %struct.__opencl_block_literal_generic = type { i32, i32, i8* }
6// SPIR: @__block_literal_global = internal addrspace(1) constant { i32, i32, i8 addrspace(4)* } { i32 12, i32 4, i8 addrspace(4)* addrspacecast (i8* bitcast (void (i8 addrspace(4)*, i8 addrspace(3)*)* @block_A_block_invoke to i8*) to i8 addrspace(4)*) }
7// AMDGCN: @__block_literal_global = internal addrspace(1) constant { i32, i32, i8* } { i32 16, i32 8, i8* bitcast (void (i8*, i8 addrspace(3)*)* @block_A_block_invoke to i8*) }
8// COMMON-NOT: .str
9
10// SPIR-LABEL: define internal {{.*}}void @block_A_block_invoke(i8 addrspace(4)* %.block_descriptor, i8 addrspace(3)* %a)
11// AMDGCN-LABEL: define internal {{.*}}void @block_A_block_invoke(i8* %.block_descriptor, i8 addrspace(3)* %a)
12void (^block_A)(local void *) = ^(local void *a) {
13  return;
14};
15
16// COMMON-LABEL: define {{.*}}void @foo()
17void foo(){
18  int i;
19  // COMMON-NOT: %block.isa
20  // COMMON-NOT: %block.flags
21  // COMMON-NOT: %block.reserved
22  // COMMON-NOT: %block.descriptor
23  // SPIR: %[[block_size:.*]] = getelementptr inbounds <{ i32, i32, i8 addrspace(4)*, i32 }>, <{ i32, i32, i8 addrspace(4)*, i32 }>* %block, i32 0, i32 0
24  // AMDGCN: %[[block_size:.*]] = getelementptr inbounds <{ i32, i32, i8*, i32 }>, <{ i32, i32, i8*, i32 }> addrspace(5)* %block, i32 0, i32 0
25  // SPIR: store i32 16, i32* %[[block_size]]
26  // AMDGCN: store i32 20, i32 addrspace(5)* %[[block_size]]
27  // SPIR: %[[block_align:.*]] = getelementptr inbounds <{ i32, i32, i8 addrspace(4)*, i32 }>, <{ i32, i32, i8 addrspace(4)*, i32 }>* %block, i32 0, i32 1
28  // AMDGCN: %[[block_align:.*]] = getelementptr inbounds <{ i32, i32, i8*, i32 }>, <{ i32, i32, i8*, i32 }> addrspace(5)* %block, i32 0, i32 1
29  // SPIR: store i32 4, i32* %[[block_align]]
30  // AMDGCN: store i32 8, i32 addrspace(5)* %[[block_align]]
31  // SPIR: %[[block_invoke:.*]] = getelementptr inbounds <{ i32, i32, i8 addrspace(4)*, i32 }>, <{ i32, i32, i8 addrspace(4)*, i32 }>* %[[block:.*]], i32 0, i32 2
32  // SPIR: store i8 addrspace(4)* addrspacecast (i8* bitcast (i32 (i8 addrspace(4)*)* @__foo_block_invoke to i8*) to i8 addrspace(4)*), i8 addrspace(4)** %[[block_invoke]]
33  // SPIR: %[[block_captured:.*]] = getelementptr inbounds <{ i32, i32, i8 addrspace(4)*, i32 }>, <{ i32, i32, i8 addrspace(4)*, i32 }>* %[[block]], i32 0, i32 3
34  // SPIR: %[[i_value:.*]] = load i32, i32* %i
35  // SPIR: store i32 %[[i_value]], i32* %[[block_captured]],
36  // SPIR: %[[blk_ptr:.*]] = bitcast <{ i32, i32, i8 addrspace(4)*, i32 }>* %[[block]] to i32 ()*
37  // SPIR: %[[blk_gen_ptr:.*]] = addrspacecast i32 ()* %[[blk_ptr]] to i32 () addrspace(4)*
38  // SPIR: store i32 () addrspace(4)* %[[blk_gen_ptr]], i32 () addrspace(4)** %[[block_B:.*]],
39  // SPIR: %[[blk_gen_ptr:.*]] = load i32 () addrspace(4)*, i32 () addrspace(4)** %[[block_B]]
40  // SPIR: %[[block_literal:.*]] = bitcast i32 () addrspace(4)* %[[blk_gen_ptr]] to %struct.__opencl_block_literal_generic addrspace(4)*
41  // SPIR: %[[invoke_addr:.*]] = getelementptr inbounds %struct.__opencl_block_literal_generic, %struct.__opencl_block_literal_generic addrspace(4)* %[[block_literal]], i32 0, i32 2
42  // SPIR: %[[blk_gen_ptr:.*]] = bitcast %struct.__opencl_block_literal_generic addrspace(4)* %[[block_literal]] to i8 addrspace(4)*
43  // SPIR: %[[invoke_func_ptr:.*]] = load i8 addrspace(4)*, i8 addrspace(4)* addrspace(4)* %[[invoke_addr]]
44  // SPIR: %[[invoke_func:.*]] = addrspacecast i8 addrspace(4)* %[[invoke_func_ptr]] to i32 (i8 addrspace(4)*)*
45  // SPIR: call {{.*}}i32 %[[invoke_func]](i8 addrspace(4)* %[[blk_gen_ptr]])
46  // AMDGCN: %[[block_invoke:.*]] = getelementptr inbounds <{ i32, i32, i8*, i32 }>, <{ i32, i32, i8*, i32 }> addrspace(5)* %[[block:.*]], i32 0, i32 2
47  // AMDGCN: store i8* bitcast (i32 (i8*)* @__foo_block_invoke to i8*), i8* addrspace(5)* %[[block_invoke]]
48  // AMDGCN: %[[block_captured:.*]] = getelementptr inbounds <{ i32, i32, i8*, i32 }>, <{ i32, i32, i8*, i32 }> addrspace(5)* %[[block]], i32 0, i32 3
49  // AMDGCN: %[[i_value:.*]] = load i32, i32 addrspace(5)* %i
50  // AMDGCN: store i32 %[[i_value]], i32 addrspace(5)* %[[block_captured]],
51  // AMDGCN: %[[blk_ptr:.*]] = bitcast <{ i32, i32, i8*, i32 }> addrspace(5)* %[[block]] to i32 () addrspace(5)*
52  // AMDGCN: %[[blk_gen_ptr:.*]] = addrspacecast i32 () addrspace(5)* %[[blk_ptr]] to i32 ()*
53  // AMDGCN: store i32 ()* %[[blk_gen_ptr]], i32 ()* addrspace(5)* %[[block_B:.*]],
54  // AMDGCN: %[[blk_gen_ptr:.*]] = load i32 ()*, i32 ()* addrspace(5)* %[[block_B]]
55  // AMDGCN: %[[block_literal:.*]] = bitcast i32 ()* %[[blk_gen_ptr]] to %struct.__opencl_block_literal_generic*
56  // AMDGCN: %[[invoke_addr:.*]] = getelementptr inbounds %struct.__opencl_block_literal_generic, %struct.__opencl_block_literal_generic* %[[block_literal]], i32 0, i32 2
57  // AMDGCN: %[[blk_gen_ptr:.*]] = bitcast %struct.__opencl_block_literal_generic* %[[block_literal]] to i8*
58  // AMDGCN: %[[invoke_func_ptr:.*]] = load i8*, i8** %[[invoke_addr]]
59  // AMDGCN: %[[invoke_func:.*]] = bitcast i8* %[[invoke_func_ptr]] to i32 (i8*)*
60  // AMDGCN: call {{.*}}i32 %[[invoke_func]](i8* %[[blk_gen_ptr]])
61
62  int (^ block_B)(void) = ^{
63    return i;
64  };
65  block_B();
66}
67
68// SPIR-LABEL: define internal {{.*}}i32 @__foo_block_invoke(i8 addrspace(4)* %.block_descriptor)
69// SPIR:  %[[block:.*]] = bitcast i8 addrspace(4)* %.block_descriptor to <{ i32, i32, i8 addrspace(4)*, i32 }> addrspace(4)*
70// SPIR:  %[[block_capture_addr:.*]] = getelementptr inbounds <{ i32, i32, i8 addrspace(4)*, i32 }>, <{ i32, i32, i8 addrspace(4)*, i32 }> addrspace(4)* %[[block]], i32 0, i32 3
71// SPIR:  %[[block_capture:.*]] = load i32, i32 addrspace(4)* %[[block_capture_addr]]
72// AMDGCN-LABEL: define internal {{.*}}i32 @__foo_block_invoke(i8* %.block_descriptor)
73// AMDGCN:  %[[block:.*]] = bitcast i8* %.block_descriptor to <{ i32, i32, i8*, i32 }>*
74// AMDGCN:  %[[block_capture_addr:.*]] = getelementptr inbounds <{ i32, i32, i8*, i32 }>, <{ i32, i32, i8*, i32 }>* %[[block]], i32 0, i32 3
75// AMDGCN:  %[[block_capture:.*]] = load i32, i32* %[[block_capture_addr]]
76
77// COMMON-NOT: define{{.*}}@__foo_block_invoke_kernel
78