clang/test/Driver/linker-wrapper-image.c

   1 // REQUIRES: x86-registered-target
   2 // REQUIRES: nvptx-registered-target
   3 // REQUIRES: amdgpu-registered-target
   4
   5 // RUN: %clang -cc1 %s -triple x86_64-unknown-linux-gnu -emit-obj -o %t.elf.o
   6
   7 // RUN: clang-offload-packager -o %t.out --image=file=%t.elf.o,kind=openmp,triple=nvptx64-nvidia-cuda,arch=sm_70
   8 // RUN: %clang -cc1 %s -triple x86_64-unknown-linux-gnu -emit-obj -o %t.o \
   9 // RUN:   -fembed-offload-object=%t.out
  10 // RUN: clang-linker-wrapper --print-wrapped-module --dry-run --host-triple=x86_64-unknown-linux-gnu \
  11 // RUN:   --linker-path=/usr/bin/ld -- %t.o -o a.out 2>&1 | FileCheck %s --check-prefix=OPENMP
  12
  13 //      OPENMP: @__start_omp_offloading_entries = external hidden constant %__tgt_offload_entry
  14 // OPENMP-NEXT: @__stop_omp_offloading_entries = external hidden constant %__tgt_offload_entry
  15 // OPENMP-NEXT: @__dummy.omp_offloading.entry = hidden constant [0 x %__tgt_offload_entry] zeroinitializer, section "omp_offloading_entries"
  16 // OPENMP-NEXT: @.omp_offloading.device_image = internal unnamed_addr constant [[[SIZE:[0-9]+]] x i8] c"\10\FF\10\AD{{.*}}"
  17 // OPENMP-NEXT: @.omp_offloading.device_images = internal unnamed_addr constant [1 x %__tgt_device_image] [%__tgt_device_image { ptr @.omp_offloading.device_image, ptr getelementptr inbounds ([[[SIZE]] x i8], ptr @.omp_offloading.device_image, i64 1, i64 0), ptr @__start_omp_offloading_entries, ptr @__stop_omp_offloading_entries }]
  18 // OPENMP-NEXT: @.omp_offloading.descriptor = internal constant %__tgt_bin_desc { i32 1, ptr @.omp_offloading.device_images, ptr @__start_omp_offloading_entries, ptr @__stop_omp_offloading_entries }
  19 // OPENMP-NEXT: @llvm.global_ctors = appending global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 1, ptr @.omp_offloading.descriptor_reg, ptr null }]
  20 // OPENMP-NEXT: @llvm.global_dtors = appending global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 1, ptr @.omp_offloading.descriptor_unreg, ptr null }]
  21
  22 //      OPENMP: define internal void @.omp_offloading.descriptor_reg() section ".text.startup" {
  23 // OPENMP-NEXT: entry:
  24 // OPENMP-NEXT:   call void @__tgt_register_lib(ptr @.omp_offloading.descriptor)
  25 // OPENMP-NEXT:   ret void
  26 // OPENMP-NEXT: }
  27
  28 //      OPENMP: define internal void @.omp_offloading.descriptor_unreg() section ".text.startup" {
  29 // OPENMP-NEXT: entry:
  30 // OPENMP-NEXT:   call void @__tgt_unregister_lib(ptr @.omp_offloading.descriptor)
  31 // OPENMP-NEXT:   ret void
  32 // OPENMP-NEXT: }
  33
  34 // RUN: clang-offload-packager -o %t.out --image=file=%t.elf.o,kind=cuda,triple=nvptx64-nvidia-cuda,arch=sm_70
  35 // RUN: %clang -cc1 %s -triple x86_64-unknown-linux-gnu -emit-obj -o %t.o \
  36 // RUN:   -fembed-offload-object=%t.out
  37 // RUN: clang-linker-wrapper --print-wrapped-module --dry-run --host-triple=x86_64-unknown-linux-gnu \
  38 // RUN:   --linker-path=/usr/bin/ld -- %t.o -o a.out 2>&1 | FileCheck %s --check-prefix=CUDA
  39
  40 //      CUDA: @.fatbin_image = internal constant [0 x i8] zeroinitializer, section ".nv_fatbin"
  41 // CUDA-NEXT: @.fatbin_wrapper = internal constant %fatbin_wrapper { i32 1180844977, i32 1, ptr @.fatbin_image, ptr null }, section ".nvFatBinSegment", align 8
  42 // CUDA-NEXT: @__dummy.cuda_offloading.entry = hidden constant [0 x %__tgt_offload_entry] zeroinitializer, section "cuda_offloading_entries"
  43 // CUDA-NEXT: @.cuda.binary_handle = internal global ptr null
  44 // CUDA-NEXT: @__start_cuda_offloading_entries = external hidden constant [0 x %__tgt_offload_entry]
  45 // CUDA-NEXT: @__stop_cuda_offloading_entries = external hidden constant [0 x %__tgt_offload_entry]
  46 // CUDA-NEXT: @llvm.global_ctors = appending global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 1, ptr @.cuda.fatbin_reg, ptr null }]
  47
  48 //      CUDA: define internal void @.cuda.fatbin_reg() section ".text.startup" {
  49 // CUDA-NEXT: entry:
  50 // CUDA-NEXT:   %0 = call ptr @__cudaRegisterFatBinary(ptr @.fatbin_wrapper)
  51 // CUDA-NEXT:   store ptr %0, ptr @.cuda.binary_handle, align 8
  52 // CUDA-NEXT:   call void @.cuda.globals_reg(ptr %0)
  53 // CUDA-NEXT:   call void @__cudaRegisterFatBinaryEnd(ptr %0)
  54 // CUDA-NEXT:   %1 = call i32 @atexit(ptr @.cuda.fatbin_unreg)
  55 // CUDA-NEXT:   ret void
  56 // CUDA-NEXT: }
  57
  58 //      CUDA: define internal void @.cuda.fatbin_unreg() section ".text.startup" {
  59 // CUDA-NEXT: entry:
  60 // CUDA-NEXT:   %0 = load ptr, ptr @.cuda.binary_handle, align 8
  61 // CUDA-NEXT:   call void @__cudaUnregisterFatBinary(ptr %0)
  62 // CUDA-NEXT:   ret void
  63 // CUDA-NEXT: }
  64
  65 //      CUDA: define internal void @.cuda.globals_reg(ptr %0) section ".text.startup" {
  66 // CUDA-NEXT: entry:
  67 // CUDA-NEXT:   br i1 icmp ne (ptr @__start_cuda_offloading_entries, ptr @__stop_cuda_offloading_entries), label %while.entry, label %while.end
  68
  69 //      CUDA: while.entry:
  70 // CUDA-NEXT:  %entry1 = phi ptr [ @__start_cuda_offloading_entries, %entry ], [ %7, %if.end ]
  71 // CUDA-NEXT:  %1 = getelementptr inbounds %__tgt_offload_entry, ptr %entry1, i64 0, i32 0
  72 // CUDA-NEXT:  %addr = load ptr, ptr %1, align 8
  73 // CUDA-NEXT:  %2 = getelementptr inbounds %__tgt_offload_entry, ptr %entry1, i64 0, i32 1
  74 // CUDA-NEXT:  %name = load ptr, ptr %2, align 8
  75 // CUDA-NEXT:  %3 = getelementptr inbounds %__tgt_offload_entry, ptr %entry1, i64 0, i32 2
  76 // CUDA-NEXT:  %size = load i64, ptr %3, align 4
  77 // CUDA-NEXT:  %4 = getelementptr inbounds %__tgt_offload_entry, ptr %entry1, i64 0, i32 3
  78 // CUDA-NEXT:  %flag = load i32, ptr %4, align 4
  79 // CUDA-NEXT:  %5 = icmp eq i64 %size, 0
  80 // CUDA-NEXT:  br i1 %5, label %if.then, label %if.else
  81
  82 //      CUDA: if.then:
  83 // CUDA-NEXT:   %6 = call i32 @__cudaRegisterFunction(ptr %0, ptr %addr, ptr %name, ptr %name, i32 -1, ptr null, ptr null, ptr null, ptr null, ptr null)
  84 // CUDA-NEXT:   br label %if.end
  85
  86 //      CUDA: if.else:
  87 // CUDA-NEXT:   switch i32 %flag, label %if.end [
  88 // CUDA-NEXT:     i32 0, label %sw.global
  89 // CUDA-NEXT:     i32 1, label %sw.managed
  90 // CUDA-NEXT:     i32 2, label %sw.surface
  91 // CUDA-NEXT:     i32 3, label %sw.texture
  92 // CUDA-NEXT:   ]
  93
  94 //      CUDA: sw.global:
  95 // CUDA-NEXT:   call void @__cudaRegisterVar(ptr %0, ptr %addr, ptr %name, ptr %name, i32 0, i64 %size, i32 0, i32 0)
  96 // CUDA-NEXT:   br label %if.end
  97
  98 //      CUDA: sw.managed:
  99 // CUDA-NEXT:   br label %if.end
 100
 101 //      CUDA: sw.surface:
 102 // CUDA-NEXT:   br label %if.end
 103
 104 //      CUDA: sw.texture:
 105 // CUDA-NEXT:   br label %if.end
 106
 107 //      CUDA: if.end:
 108 // CUDA-NEXT:   %7 = getelementptr inbounds %__tgt_offload_entry, ptr %entry1, i64 1
 109 // CUDA-NEXT:   %8 = icmp eq ptr %7, @__stop_cuda_offloading_entries
 110 // CUDA-NEXT:   br i1 %8, label %while.end, label %while.entry
 111
 112 //      CUDA: while.end:
 113 // CUDA-NEXT:   ret void
 114 // CUDA-NEXT: }
 115
 116 // RUN: clang-offload-packager -o %t.out --image=file=%t.elf.o,kind=hip,triple=amdgcn-amd-amdhsa,arch=gfx908
 117 // RUN: %clang -cc1 %s -triple x86_64-unknown-linux-gnu -emit-obj -o %t.o \
 118 // RUN:   -fembed-offload-object=%t.out
 119 // RUN: clang-linker-wrapper --print-wrapped-module --dry-run --host-triple=x86_64-unknown-linux-gnu \
 120 // RUN:   --linker-path=/usr/bin/ld -- %t.o -o a.out 2>&1 | FileCheck %s --check-prefix=HIP
 121
 122 //      HIP: @.fatbin_image = internal constant [0 x i8] zeroinitializer, section ".hip_fatbin"
 123 // HIP-NEXT: @.fatbin_wrapper = internal constant %fatbin_wrapper { i32 1212764230, i32 1, ptr @.fatbin_image, ptr null }, section ".hipFatBinSegment", align 8
 124 // HIP-NEXT: @__dummy.hip_offloading.entry = hidden constant [0 x %__tgt_offload_entry] zeroinitializer, section "hip_offloading_entries"
 125 // HIP-NEXT: @.hip.binary_handle = internal global ptr null
 126 // HIP-NEXT: @__start_hip_offloading_entries = external hidden constant [0 x %__tgt_offload_entry]
 127 // HIP-NEXT: @__stop_hip_offloading_entries = external hidden constant [0 x %__tgt_offload_entry]
 128 // HIP-NEXT: @llvm.global_ctors = appending global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 1, ptr @.hip.fatbin_reg, ptr null }]
 129
 130 //      HIP: define internal void @.hip.fatbin_reg() section ".text.startup" {
 131 // HIP-NEXT: entry:
 132 // HIP-NEXT:   %0 = call ptr @__hipRegisterFatBinary(ptr @.fatbin_wrapper)
 133 // HIP-NEXT:   store ptr %0, ptr @.hip.binary_handle, align 8
 134 // HIP-NEXT:   call void @.hip.globals_reg(ptr %0)
 135 // HIP-NEXT:   %1 = call i32 @atexit(ptr @.hip.fatbin_unreg)
 136 // HIP-NEXT:   ret void
 137 // HIP-NEXT: }
 138
 139 //      HIP: define internal void @.hip.fatbin_unreg() section ".text.startup" {
 140 // HIP-NEXT: entry:
 141 // HIP-NEXT:   %0 = load ptr, ptr @.hip.binary_handle, align 8
 142 // HIP-NEXT:   call void @__hipUnregisterFatBinary(ptr %0)
 143 // HIP-NEXT:   ret void
 144 // HIP-NEXT: }
 145
 146 //      HIP: define internal void @.hip.globals_reg(ptr %0) section ".text.startup" {
 147 // HIP-NEXT: entry:
 148 // HIP-NEXT:   br i1 icmp ne (ptr @__start_hip_offloading_entries, ptr @__stop_hip_offloading_entries), label %while.entry, label %while.end
 149
 150 //      HIP: while.entry:
 151 // HIP-NEXT:   %entry1 = phi ptr [ @__start_hip_offloading_entries, %entry ], [ %7, %if.end ]
 152 // HIP-NEXT:   %1 = getelementptr inbounds %__tgt_offload_entry, ptr %entry1, i64 0, i32 0
 153 // HIP-NEXT:   %addr = load ptr, ptr %1, align 8
 154 // HIP-NEXT:   %2 = getelementptr inbounds %__tgt_offload_entry, ptr %entry1, i64 0, i32 1
 155 // HIP-NEXT:   %name = load ptr, ptr %2, align 8
 156 // HIP-NEXT:   %3 = getelementptr inbounds %__tgt_offload_entry, ptr %entry1, i64 0, i32 2
 157 // HIP-NEXT:   %size = load i64, ptr %3, align 4
 158 // HIP-NEXT:   %4 = getelementptr inbounds %__tgt_offload_entry, ptr %entry1, i64 0, i32 3
 159 // HIP-NEXT:   %flag = load i32, ptr %4, align 4
 160 // HIP-NEXT:   %5 = icmp eq i64 %size, 0
 161 // HIP-NEXT:   br i1 %5, label %if.then, label %if.else
 162
 163 //      HIP: if.then:
 164 // HIP-NEXT:   %6 = call i32 @__hipRegisterFunction(ptr %0, ptr %addr, ptr %name, ptr %name, i32 -1, ptr null, ptr null, ptr null, ptr null, ptr null)
 165 // HIP-NEXT:   br label %if.end
 166
 167 //      HIP: if.else:
 168 // HIP-NEXT:   switch i32 %flag, label %if.end [
 169 // HIP-NEXT:     i32 0, label %sw.global
 170 // HIP-NEXT:     i32 1, label %sw.managed
 171 // HIP-NEXT:     i32 2, label %sw.surface
 172 // HIP-NEXT:     i32 3, label %sw.texture
 173 // HIP-NEXT:   ]
 174
 175 //      HIP: sw.global:
 176 // HIP-NEXT:   call void @__hipRegisterVar(ptr %0, ptr %addr, ptr %name, ptr %name, i32 0, i64 %size, i32 0, i32 0)
 177 // HIP-NEXT:   br label %if.end
 178
 179 //      HIP: sw.managed:
 180 // HIP-NEXT:   br label %if.end
 181
 182 //      HIP: sw.surface:
 183 // HIP-NEXT:   br label %if.end
 184
 185 //      HIP: sw.texture:
 186 // HIP-NEXT:   br label %if.end
 187
 188 //      HIP: if.end:
 189 // HIP-NEXT:   %7 = getelementptr inbounds %__tgt_offload_entry, ptr %entry1, i64 1
 190 // HIP-NEXT:   %8 = icmp eq ptr %7, @__stop_hip_offloading_entries
 191 // HIP-NEXT:   br i1 %8, label %while.end, label %while.entry
 192
 193 //      HIP: while.end:
 194 // HIP-NEXT:   ret void
 195 // HIP-NEXT: }