xref: /llvm-project/mlir/test/Target/LLVMIR/omptarget-constant-indexing-device-region.mlir (revision d2d4c3bd598453a1d871096466953ec25483034f)
1// RUN: mlir-translate -mlir-to-llvmir %s | FileCheck %s
2
3module attributes {llvm.target_triple = "amdgcn-amd-amdhsa", omp.is_target_device = true} {
4  llvm.func @_QQmain() attributes {bindc_name = "main"} {
5    %0 = llvm.mlir.addressof @_QFEsp : !llvm.ptr
6    %1 = llvm.mlir.constant(10 : index) : i64
7    %2 = llvm.mlir.constant(1 : index) : i64
8    %3 = llvm.mlir.constant(0 : index) : i64
9    %4 = llvm.mlir.constant(9 : index) : i64
10    %5 = omp.map.bounds lower_bound(%3 : i64) upper_bound(%4 : i64) extent(%1 : i64) stride(%2 : i64) start_idx(%2 : i64)
11    %6 = omp.map.info var_ptr(%0 : !llvm.ptr, !llvm.array<10 x i32>) map_clauses(tofrom) capture(ByRef) bounds(%5) -> !llvm.ptr {name = "sp"}
12    omp.target map_entries(%6 -> %arg0 : !llvm.ptr) {
13      %7 = llvm.mlir.constant(20 : i32) : i32
14      %8 = llvm.mlir.constant(0 : i64) : i64
15      %9 = llvm.getelementptr %arg0[0, %8] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i32>
16      llvm.store %7, %9 : i32, !llvm.ptr
17      %10 = llvm.mlir.constant(10 : i32) : i32
18      %11 = llvm.mlir.constant(4 : i64) : i64
19      %12 = llvm.getelementptr %arg0[0, %11] : (!llvm.ptr, i64) -> !llvm.ptr, !llvm.array<10 x i32>
20      llvm.store %10, %12 : i32, !llvm.ptr
21      omp.terminator
22    }
23    llvm.return
24  }
25  llvm.mlir.global internal @_QFEsp(dense<0> : tensor<10xi32>) {addr_space = 0 : i32} : !llvm.array<10 x i32>
26  llvm.mlir.global external constant @_QQEnvironmentDefaults() {addr_space = 0 : i32} : !llvm.ptr {
27    %0 = llvm.mlir.zero : !llvm.ptr
28    llvm.return %0 : !llvm.ptr
29  }
30}
31
32
33// CHECK: define {{.*}} void @__omp_offloading_{{.*}}_{{.*}}__QQmain_{{.*}}(ptr %{{.*}}, ptr %[[ARG1:.*]]) #{{[0-9]+}} {
34
35// CHECK: %[[ARG1_ALLOCA:.*]] = alloca ptr, align 8
36// CHECK: store ptr %[[ARG1]], ptr %[[ARG1_ALLOCA]], align 8
37// CHECK: %[[LOAD_ARG1_ALLOCA:.*]] = load ptr, ptr %[[ARG1_ALLOCA]], align 8
38// CHECK: store i32 20, ptr %[[LOAD_ARG1_ALLOCA]], align 4
39// CHECK: %[[GEP_ARG1_ALLOCA:.*]] = getelementptr [10 x i32], ptr %[[LOAD_ARG1_ALLOCA]], i32 0, i64 4
40// CHECK: store i32 10, ptr %[[GEP_ARG1_ALLOCA]], align 4
41
42