1 // RUN: mlir-opt %s --gpu-to-llvm="use-bare-pointers-for-kernels=1" -split-input-file | FileCheck %s
3 module attributes {gpu.container_module} {
4 gpu.module @kernels [#nvvm.target] {
5 llvm.func @kernel_1(%arg0: f32, %arg1: !llvm.ptr<1>) attributes {gpu.kernel, nvvm.kernel} {
6 %0 = llvm.mlir.undef : !llvm.struct<(ptr<1>, ptr<1>, i64, array<1 x i64>, array<1 x i64>)>
7 %1 = llvm.insertvalue %arg1, %0[0] : !llvm.struct<(ptr<1>, ptr<1>, i64, array<1 x i64>, array<1 x i64>)>
8 %2 = llvm.insertvalue %arg1, %1[1] : !llvm.struct<(ptr<1>, ptr<1>, i64, array<1 x i64>, array<1 x i64>)>
9 %3 = llvm.mlir.constant(0 : index) : i64
10 %4 = llvm.insertvalue %3, %2[2] : !llvm.struct<(ptr<1>, ptr<1>, i64, array<1 x i64>, array<1 x i64>)>
11 %5 = llvm.mlir.constant(10 : index) : i64
12 %6 = llvm.insertvalue %5, %4[3, 0] : !llvm.struct<(ptr<1>, ptr<1>, i64, array<1 x i64>, array<1 x i64>)>
13 %7 = llvm.mlir.constant(1 : index) : i64
14 %8 = llvm.insertvalue %7, %6[4, 0] : !llvm.struct<(ptr<1>, ptr<1>, i64, array<1 x i64>, array<1 x i64>)>
19 // CHECK: [[MEMREF:%.*]] = gpu.alloc () : memref<10xf32, 1>
20 // CHECK: [[DESCRIPTOR:%.*]] = builtin.unrealized_conversion_cast [[MEMREF]] : memref<10xf32, 1> to !llvm.struct<(ptr<1>, ptr<1>, i64, array<1 x i64>, array<1 x i64>)>
21 // CHECK: [[PTR:%.*]] = llvm.extractvalue [[DESCRIPTOR]][1] : !llvm.struct<(ptr<1>, ptr<1>, i64, array<1 x i64>, array<1 x i64>)>
22 // CHECK: gpu.launch_func @kernels::@kernel_1 blocks in ({{.*}}) threads in ({{.*}}) : i64
23 // CHECK: args(%{{.*}} : f32, [[PTR]] : !llvm.ptr<1>)
24 %0 = arith.constant 0. : f32
25 %1 = gpu.alloc () : memref<10xf32, 1>
26 %c8 = arith.constant 8 : index
27 gpu.launch_func @kernels::@kernel_1 blocks in (%c8, %c8, %c8) threads in (%c8, %c8, %c8) args(%0 : f32, %1 : memref<10xf32, 1>)