|
| 1 | +; RUN: llvm-offload-wrapper --triple=x86_64-unknown-linux-gnu -kind=openmp %s -o %t.bc |
| 2 | +; RUN: llvm-dis %t.bc -o - | FileCheck %s --check-prefix=OMP |
| 3 | + |
| 4 | +; OMP: @__start_llvm_offload_entries = external hidden constant [0 x %struct.__tgt_offload_entry] |
| 5 | +; OMP-NEXT: @__stop_llvm_offload_entries = external hidden constant [0 x %struct.__tgt_offload_entry] |
| 6 | +; OMP-NEXT: @__dummy.llvm_offload_entries = internal constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "llvm_offload_entries", align 8 |
| 7 | +; OMP-NEXT: @llvm.compiler.used = appending global [1 x ptr] [ptr @__dummy.llvm_offload_entries], section "llvm.metadata" |
| 8 | +; OMP-NEXT: @.omp_offloading.device_image = internal unnamed_addr constant [[[SIZE:[0-9]+]] x i8] c"{{.*}}", section ".llvm.offloading", align 8 |
| 9 | +; OMP-NEXT: @.omp_offloading.device_images = internal unnamed_addr constant [1 x %__tgt_device_image] [%__tgt_device_image { ptr @.omp_offloading.device_image, ptr getelementptr ([[[SIZE]] x i8], ptr @.omp_offloading.device_image, i64 0, i64 [[SIZE]]), ptr @__start_llvm_offload_entries, ptr @__stop_llvm_offload_entries }] |
| 10 | +; OMP-NEXT: @.omp_offloading.descriptor = internal constant %__tgt_bin_desc { i32 1, ptr @.omp_offloading.device_images, ptr @__start_llvm_offload_entries, ptr @__stop_llvm_offload_entries } |
| 11 | +; OMP-NEXT: @llvm.global_ctors = appending global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 101, ptr @.omp_offloading.descriptor_reg, ptr null }] |
| 12 | + |
| 13 | +; OMP: define internal void @.omp_offloading.descriptor_reg() section ".text.startup" { |
| 14 | +; OMP-NEXT: entry: |
| 15 | +; OMP-NEXT: call void @__tgt_register_lib(ptr @.omp_offloading.descriptor) |
| 16 | +; OMP-NEXT: %0 = call i32 @atexit(ptr @.omp_offloading.descriptor_unreg) |
| 17 | +; OMP-NEXT: ret void |
| 18 | +; OMP-NEXT: } |
| 19 | + |
| 20 | +; OMP: define internal void @.omp_offloading.descriptor_unreg() section ".text.startup" { |
| 21 | +; OMP-NEXT: entry: |
| 22 | +; OMP-NEXT: call void @__tgt_unregister_lib(ptr @.omp_offloading.descriptor) |
| 23 | +; OMP-NEXT: ret void |
| 24 | +; OMP-NEXT: } |
| 25 | + |
| 26 | +; RUN: llvm-offload-wrapper --triple=x86_64-unknown-linux-gnu -kind=hip %s -o %t.bc |
| 27 | +; RUN: llvm-dis %t.bc -o - | FileCheck %s --check-prefix=HIP |
| 28 | + |
| 29 | +; HIP: @__start_llvm_offload_entries = external hidden constant [0 x %struct.__tgt_offload_entry] |
| 30 | +; HIP-NEXT: @__stop_llvm_offload_entries = external hidden constant [0 x %struct.__tgt_offload_entry] |
| 31 | +; HIP-NEXT: @__dummy.llvm_offload_entries = internal constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "llvm_offload_entries", align 8 |
| 32 | +; HIP-NEXT: @llvm.compiler.used = appending global [1 x ptr] [ptr @__dummy.llvm_offload_entries], section "llvm.metadata" |
| 33 | +; HIP-NEXT: @.fatbin_image = internal constant {{.*}}, section ".hip_fatbin" |
| 34 | +; HIP-NEXT: @.fatbin_wrapper = internal constant %fatbin_wrapper { i32 1212764230, i32 1, ptr @.fatbin_image, ptr null }, section ".hipFatBinSegment", align 8 |
| 35 | +; HIP-NEXT: @.hip.binary_handle = internal global ptr null |
| 36 | +; HIP-NEXT: @llvm.global_ctors = appending global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 101, ptr @.hip.fatbin_reg, ptr null }] |
| 37 | + |
| 38 | +; HIP: define internal void @.hip.fatbin_reg() section ".text.startup" { |
| 39 | +; HIP-NEXT: entry: |
| 40 | +; HIP-NEXT: %0 = call ptr @__hipRegisterFatBinary(ptr @.fatbin_wrapper) |
| 41 | +; HIP-NEXT: store ptr %0, ptr @.hip.binary_handle, align 8 |
| 42 | +; HIP-NEXT: call void @.hip.globals_reg(ptr %0) |
| 43 | +; HIP-NEXT: %1 = call i32 @atexit(ptr @.hip.fatbin_unreg) |
| 44 | +; HIP-NEXT: ret void |
| 45 | +; HIP-NEXT: } |
| 46 | + |
| 47 | +; HIP: define internal void @.hip.fatbin_unreg() section ".text.startup" { |
| 48 | +; HIP-NEXT: entry: |
| 49 | +; HIP-NEXT: %0 = load ptr, ptr @.hip.binary_handle, align 8 |
| 50 | +; HIP-NEXT: call void @__hipUnregisterFatBinary(ptr %0) |
| 51 | +; HIP-NEXT: ret void |
| 52 | +; HIP-NEXT: } |
| 53 | + |
| 54 | +; RUN: llvm-offload-wrapper --triple=x86_64-unknown-linux-gnu -kind=cuda %s -o %t.bc |
| 55 | +; RUN: llvm-dis %t.bc -o - | FileCheck %s --check-prefix=CUDA |
| 56 | + |
| 57 | +; CUDA: @__start_llvm_offload_entries = external hidden constant [0 x %struct.__tgt_offload_entry] |
| 58 | +; CUDA-NEXT: @__stop_llvm_offload_entries = external hidden constant [0 x %struct.__tgt_offload_entry] |
| 59 | +; CUDA-NEXT: @__dummy.llvm_offload_entries = internal constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "llvm_offload_entries", align 8 |
| 60 | +; CUDA-NEXT: @llvm.compiler.used = appending global [1 x ptr] [ptr @__dummy.llvm_offload_entries], section "llvm.metadata" |
| 61 | +; CUDA-NEXT: @.fatbin_image = internal constant {{.*}}, section ".nv_fatbin" |
| 62 | +; CUDA-NEXT: @.fatbin_wrapper = internal constant %fatbin_wrapper { i32 1180844977, i32 1, ptr @.fatbin_image, ptr null }, section ".nvFatBinSegment", align 8 |
| 63 | +; CUDA-NEXT: @.cuda.binary_handle = internal global ptr null |
| 64 | +; CUDA-NEXT: @llvm.global_ctors = appending global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 101, ptr @.cuda.fatbin_reg, ptr null }] |
| 65 | + |
| 66 | +; CUDA: define internal void @.cuda.fatbin_reg() section ".text.startup" { |
| 67 | +; CUDA-NEXT: entry: |
| 68 | +; CUDA-NEXT: %0 = call ptr @__cudaRegisterFatBinary(ptr @.fatbin_wrapper) |
| 69 | +; CUDA-NEXT: store ptr %0, ptr @.cuda.binary_handle, align 8 |
| 70 | +; CUDA-NEXT: call void @.cuda.globals_reg(ptr %0) |
| 71 | +; CUDA-NEXT: call void @__cudaRegisterFatBinaryEnd(ptr %0) |
| 72 | +; CUDA-NEXT: %1 = call i32 @atexit(ptr @.cuda.fatbin_unreg) |
| 73 | +; CUDA-NEXT: ret void |
| 74 | +; CUDA-NEXT: } |
| 75 | + |
| 76 | +; CUDA: define internal void @.cuda.fatbin_unreg() section ".text.startup" { |
| 77 | +; CUDA-NEXT: entry: |
| 78 | +; CUDA-NEXT: %0 = load ptr, ptr @.cuda.binary_handle, align 8 |
| 79 | +; CUDA-NEXT: call void @__cudaUnregisterFatBinary(ptr %0) |
| 80 | +; CUDA-NEXT: ret void |
| 81 | +; CUDA-NEXT: } |
0 commit comments