clang/test/Driver/openmp-offload-gpu.c

   1 ///
   2 /// Perform several driver tests for OpenMP offloading
   3 ///
   4
   5 // REQUIRES: x86-registered-target
   6 // REQUIRES: powerpc-registered-target
   7 // REQUIRES: nvptx-registered-target
   8 // REQUIRES: amdgpu-registered-target
   9
  10 /// ###########################################################################
  11
  12 /// Check -Xopenmp-target uses one of the archs provided when several archs are used.
  13 // RUN:   %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -nogpulib -nogpuinc \
  14 // RUN:          -Xopenmp-target -march=sm_52 -Xopenmp-target -march=sm_60 %s 2>&1 \
  15 // RUN:   | FileCheck -check-prefix=CHK-FOPENMP-TARGET-ARCHS %s
  16
  17 // CHK-FOPENMP-TARGET-ARCHS: ptxas{{.*}}" "--gpu-name" "sm_60"
  18
  19 /// ###########################################################################
  20
  21 /// Check -Xopenmp-target -march=sm_52 works as expected when two triples are present.
  22 // RUN:   %clang -### -fopenmp=libomp \
  23 // RUN:          -fopenmp-targets=powerpc64le-ibm-linux-gnu,nvptx64-nvidia-cuda \
  24 // RUN:          -nogpulib -nogpuinc -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 %s 2>&1 \
  25 // RUN:   | FileCheck -check-prefix=CHK-FOPENMP-TARGET-COMPILATION %s
  26
  27 // CHK-FOPENMP-TARGET-COMPILATION: ptxas{{.*}}" "--gpu-name" "sm_52"
  28
  29 /// Check PTXAS is passed -c flag when offloading to an NVIDIA device using OpenMP.
  30 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 \
  31 // RUN:          -fopenmp-targets=nvptx64-nvidia-cuda %s 2>&1 | FileCheck -check-prefix=CHK-PTXAS-DEFAULT %s
  32
  33 // CHK-PTXAS-DEFAULT: ptxas{{.*}}" "-c"
  34
  35 /// ###########################################################################
  36
  37 /// PTXAS is passed -c flag by default when offloading to an NVIDIA device using OpenMP - disable it.
  38 // RUN:   %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -fnoopenmp-relocatable-target \
  39 // RUN:         -nogpulib -nogpuinc --offload-arch=sm_52  -save-temps %s 2>&1 \
  40 // RUN:   | FileCheck -check-prefix=CHK-PTXAS-NORELO %s
  41
  42 // CHK-PTXAS-NORELO-NOT: ptxas{{.*}}" "-c"
  43
  44 /// ###########################################################################
  45
  46 /// PTXAS is passed -c flag by default when offloading to an NVIDIA device using OpenMP
  47 /// Check that the flag is passed when -fopenmp-relocatable-target is used.
  48 // RUN:   %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -fopenmp-relocatable-target \
  49 // RUN:          -nogpulib -nogpuinc --offload-arch=sm_52 -save-temps %s 2>&1 \
  50 // RUN:   | FileCheck -check-prefix=CHK-PTXAS-RELO %s
  51
  52 // CHK-PTXAS-RELO: ptxas{{.*}}" "-c"
  53
  54 /// ###########################################################################
  55
  56 /// Check that error is not thrown by toolchain when no cuda lib flag is used.
  57 /// Check that the flag is passed when -fopenmp-relocatable-target is used.
  58 // RUN:   %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 \
  59 // RUN:   -nogpulib -nogpuinc -fopenmp-relocatable-target -save-temps %s 2>&1 \
  60 // RUN:   | FileCheck -check-prefix=CHK-FLAG-NOLIBDEVICE %s
  61
  62 // CHK-FLAG-NOLIBDEVICE-NOT: error:{{.*}}sm_60
  63
  64 /// ###########################################################################
  65
  66 /// Check that error is not thrown by toolchain when no cuda lib device is found when using -S.
  67 /// Check that the flag is passed when -fopenmp-relocatable-target is used.
  68 // RUN:   %clang -### -S -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 \
  69 // RUN:   -fopenmp-relocatable-target -save-temps %s 2>&1 \
  70 // RUN:   | FileCheck -check-prefix=CHK-NOLIBDEVICE %s
  71
  72 // CHK-NOLIBDEVICE-NOT: error:{{.*}}sm_60
  73
  74 /// ###########################################################################
  75
  76 /// Check that the runtime bitcode library is part of the compile line.
  77 /// Create a bogus bitcode library and specify it with libomptarget-nvptx-bc-path
  78 // RUN:   %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda \
  79 // RUN:   --libomptarget-nvptx-bc-path=%S/Inputs/libomptarget/libomptarget-nvptx-test.bc \
  80 // RUN:   -Xopenmp-target -march=sm_52 --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
  81 // RUN:   -fopenmp-relocatable-target -save-temps %s 2>&1 \
  82 // RUN:   | FileCheck -check-prefix=CHK-BCLIB %s
  83
  84 /// Specify the directory containing the bitcode lib, check clang picks the right one
  85 // RUN:   %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda \
  86 // RUN:   --libomptarget-nvptx-bc-path=%S/Inputs/libomptarget \
  87 // RUN:   -Xopenmp-target -march=sm_52 --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
  88 // RUN:   -fopenmp-relocatable-target -save-temps \
  89 // RUN:   %s 2>&1 | FileCheck -check-prefix=CHK-BCLIB-DIR %s
  90
  91 /// Create a bogus bitcode library and find it with LIBRARY_PATH
  92 // RUN:   env LIBRARY_PATH=%S/Inputs/libomptarget/subdir %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda \
  93 // RUN:   -Xopenmp-target -march=sm_52 --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
  94 // RUN:   -fopenmp-relocatable-target -save-temps \
  95 // RUN:   %s 2>&1 | FileCheck -check-prefix=CHK-ENV-BCLIB %s
  96
  97 // CHK-BCLIB: clang{{.*}}-triple{{.*}}nvptx64-nvidia-cuda{{.*}}-mlink-builtin-bitcode{{.*}}libomptarget-nvptx-test.bc
  98 // CHK-BCLIB-DIR: clang{{.*}}-triple{{.*}}nvptx64-nvidia-cuda{{.*}}-mlink-builtin-bitcode{{.*}}libomptarget{{/|\\\\}}libomptarget-nvptx-sm_52.bc
  99 // CHK-ENV-BCLIB: clang{{.*}}-triple{{.*}}nvptx64-nvidia-cuda{{.*}}-mlink-builtin-bitcode{{.*}}subdir{{/|\\\\}}libomptarget-nvptx-sm_52.bc
 100 // CHK-BCLIB-NOT: {{error:|warning:}}
 101
 102 /// ###########################################################################
 103
 104 /// Check that the warning is thrown when the libomptarget bitcode library is not found.
 105 /// Libomptarget requires sm_52 or newer so an sm_52 bitcode library should never exist.
 106 // RUN:   not %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda \
 107 // RUN:   -Xopenmp-target -march=sm_52 --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
 108 // RUN:   -fopenmp-relocatable-target -save-temps %s 2>&1 \
 109 // RUN:   | FileCheck -check-prefix=CHK-BCLIB-WARN %s
 110
 111 // CHK-BCLIB-WARN: no library 'libomptarget-nvptx-sm_52.bc' found in the default clang lib directory or in LIBRARY_PATH; use '--libomptarget-nvptx-bc-path' to specify nvptx bitcode library
 112
 113 /// ###########################################################################
 114
 115 /// Check that the error is thrown when the libomptarget bitcode library does not exist.
 116 // RUN:   not %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda \
 117 // RUN:   -Xopenmp-target -march=sm_52 --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
 118 // RUN:   --libomptarget-nvptx-bc-path=not-exist.bc \
 119 // RUN:   -fopenmp-relocatable-target -save-temps %s 2>&1 \
 120 // RUN:   | FileCheck -check-prefix=CHK-BCLIB-ERROR %s
 121
 122 // CHK-BCLIB-ERROR: bitcode library 'not-exist.bc' does not exist
 123
 124 /// ###########################################################################
 125
 126 /// Check that the error is thrown when CUDA 9.1 or lower version is used.
 127 // RUN:   not %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda \
 128 // RUN:   -Xopenmp-target -march=sm_52 --cuda-path=%S/Inputs/CUDA_90/usr/local/cuda \
 129 // RUN:   -fopenmp-relocatable-target -save-temps %s 2>&1 \
 130 // RUN:   | FileCheck -check-prefix=CHK-CUDA-VERSION-ERROR %s
 131
 132 // CHK-CUDA-VERSION-ERROR: NVPTX target requires CUDA 9.2 or above; CUDA 9.0 detected
 133
 134 /// Check that debug info is emitted in dwarf-2
 135 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O1 --no-cuda-noopt-device-debug 2>&1 \
 136 // RUN:   | FileCheck -check-prefix=DEBUG_DIRECTIVES %s
 137 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O3 2>&1 \
 138 // RUN:   | FileCheck -check-prefix=DEBUG_DIRECTIVES %s
 139 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O3 --no-cuda-noopt-device-debug 2>&1 \
 140 // RUN:   | FileCheck -check-prefix=DEBUG_DIRECTIVES %s
 141 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g0 2>&1 \
 142 // RUN:   | FileCheck -check-prefix=NO_DEBUG %s
 143 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -ggdb0 -O3 --cuda-noopt-device-debug 2>&1 \
 144 // RUN:   | FileCheck -check-prefix=NO_DEBUG %s
 145 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -gline-directives-only 2>&1 \
 146 // RUN:   | FileCheck -check-prefix=DEBUG_DIRECTIVES %s
 147
 148 // DEBUG_DIRECTIVES-NOT: warning: debug
 149 // NO_DEBUG-NOT: warning: debug
 150 // NO_DEBUG: "-fopenmp-is-target-device"
 151 // NO_DEBUG-NOT: "-debug-info-kind=
 152 // NO_DEBUG: ptxas
 153 // DEBUG_DIRECTIVES: "-triple" "nvptx64-nvidia-cuda"
 154 // DEBUG_DIRECTIVES-SAME: "-debug-info-kind=line-directives-only"
 155 // DEBUG_DIRECTIVES-SAME: "-fopenmp-is-target-device"
 156 // DEBUG_DIRECTIVES: ptxas
 157 // DEBUG_DIRECTIVES: "-lineinfo"
 158
 159 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O0 --no-cuda-noopt-device-debug 2>&1 \
 160 // RUN:   | FileCheck -check-prefix=HAS_DEBUG %s
 161 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g 2>&1 \
 162 // RUN:   | FileCheck -check-prefix=HAS_DEBUG %s
 163 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O0 --cuda-noopt-device-debug 2>&1 \
 164 // RUN:   | FileCheck -check-prefix=HAS_DEBUG %s
 165 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O3 --cuda-noopt-device-debug 2>&1 \
 166 // RUN:   | FileCheck -check-prefix=HAS_DEBUG %s
 167 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g2 2>&1 \
 168 // RUN:   | FileCheck -check-prefix=HAS_DEBUG %s
 169 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -ggdb2 -O0 --cuda-noopt-device-debug 2>&1 \
 170 // RUN:   | FileCheck -check-prefix=HAS_DEBUG %s
 171 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g3 -O3 --cuda-noopt-device-debug 2>&1 \
 172 // RUN:   | FileCheck -check-prefix=HAS_DEBUG %s
 173 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -ggdb3 -O2 --cuda-noopt-device-debug 2>&1 \
 174 // RUN:   | FileCheck -check-prefix=HAS_DEBUG %s
 175 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -gline-tables-only 2>&1 \
 176 // RUN:   | FileCheck -check-prefix=HAS_DEBUG %s
 177 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -ggdb1 -O2 --cuda-noopt-device-debug 2>&1 \
 178 // RUN:   | FileCheck -check-prefix=HAS_DEBUG %s
 179
 180 // HAS_DEBUG-NOT: warning: debug
 181 // HAS_DEBUG: "-triple" "nvptx64-nvidia-cuda"
 182 // HAS_DEBUG-SAME: "-debug-info-kind={{constructor|line-tables-only}}"
 183 // HAS_DEBUG-SAME: "-dwarf-version=2"
 184 // HAS_DEBUG-SAME: "-fopenmp-is-target-device"
 185 // HAS_DEBUG: ptxas
 186 // HAS_DEBUG-SAME: "-g"
 187 // HAS_DEBUG-SAME: "--dont-merge-basicblocks"
 188 // HAS_DEBUG-SAME: "--return-at-end"
 189
 190 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -fopenmp-cuda-mode 2>&1 \
 191 // RUN:   | FileCheck -check-prefix=CUDA_MODE %s
 192 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -fno-openmp-cuda-mode -fopenmp-cuda-mode 2>&1 \
 193 // RUN:   | FileCheck -check-prefix=CUDA_MODE %s
 194 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=amdgcn-amd-amdhsa -Xopenmp-target -march=gfx906 %s -fopenmp-cuda-mode 2>&1 \
 195 // RUN:   | FileCheck -check-prefix=CUDA_MODE %s
 196 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=amdgcn-amd-amdhsa -Xopenmp-target -march=gfx906 %s -fno-openmp-cuda-mode -fopenmp-cuda-mode 2>&1 \
 197 // RUN:   | FileCheck -check-prefix=CUDA_MODE %s
 198 // CUDA_MODE: "-cc1"{{.*}}"-triple" "{{nvptx64-nvidia-cuda|amdgcn-amd-amdhsa}}"
 199 // CUDA_MODE-SAME: "-fopenmp-cuda-mode"
 200 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -fno-openmp-cuda-mode 2>&1 \
 201 // RUN:   | FileCheck -check-prefix=NO_CUDA_MODE %s
 202 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -fopenmp-cuda-mode -fno-openmp-cuda-mode 2>&1 \
 203 // RUN:   | FileCheck -check-prefix=NO_CUDA_MODE %s
 204 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=amdgcn-amd-amdhsa -Xopenmp-target -march=gfx906 %s -fno-openmp-cuda-mode 2>&1 \
 205 // RUN:   | FileCheck -check-prefix=NO_CUDA_MODE %s
 206 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=amdgcn-amd-amdhsa -Xopenmp-target -march=gfx906 %s -fopenmp-cuda-mode -fno-openmp-cuda-mode 2>&1 \
 207 // RUN:   | FileCheck -check-prefix=NO_CUDA_MODE %s
 208 // NO_CUDA_MODE-NOT: "-{{fno-|f}}openmp-cuda-mode"
 209
 210 // RUN:   %clang -### -nogpulib -nogpuinc -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -fopenmp-cuda-teams-reduction-recs-num=2048 2>&1 \
 211 // RUN:   | FileCheck -check-prefix=CUDA_RED_RECS %s
 212 // CUDA_RED_RECS: "-cc1"{{.*}}"-triple" "nvptx64-nvidia-cuda"
 213 // CUDA_RED_RECS-SAME: "-fopenmp-cuda-teams-reduction-recs-num=2048"
 214
 215 // RUN:   %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
 216 // RUN:          --offload-arch=sm_52 --libomptarget-nvptx-bc-path=%S/Inputs/libomptarget/libomptarget-nvptx-test.bc %s 2>&1 \
 217 // RUN:   | FileCheck -check-prefix=OPENMP_NVPTX_WRAPPERS %s
 218 // OPENMP_NVPTX_WRAPPERS: "-cc1"{{.*}}"-triple" "nvptx64-nvidia-cuda"
 219 // OPENMP_NVPTX_WRAPPERS-SAME: "-internal-isystem" "{{.*}}openmp_wrappers"
 220
 221 // RUN:   %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -nogpulib -nogpuinc \
 222 // RUN:          --offload-arch=sm_52 -save-temps -ccc-print-bindings %s -o openmp-offload-gpu 2>&1 \
 223 // RUN:   | FileCheck -check-prefix=SAVE_TEMPS_NAMES %s
 224
 225 // SAVE_TEMPS_NAMES-NOT: "GNU::Linker"{{.*}}["[[SAVE_TEMPS_INPUT1:.*\.o]]", "[[SAVE_TEMPS_INPUT1]]"]
 226
 227 // RUN:   %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64 -Xopenmp-target=nvptx64 -march=sm_52 \
 228 // RUN:          -nogpulib -nogpuinc -save-temps %s -o openmp-offload-gpu 2>&1 \
 229 // RUN:   | FileCheck -check-prefix=TRIPLE %s
 230
 231 // TRIPLE: "-triple" "nvptx64-nvidia-cuda"
 232 // TRIPLE: "-target-cpu" "sm_52"
 233
 234 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda \
 235 // RUN:          -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
 236 // RUN:          --libomptarget-nvptx-bc-path=%S/Inputs/libomptarget/libomptarget-nvptx-test.bc %s 2>&1 \
 237 // RUN:   | FileCheck %s
 238 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda \
 239 // RUN:          --offload-arch=sm_52 --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
 240 // RUN:          --libomptarget-nvptx-bc-path=%S/Inputs/libomptarget/libomptarget-nvptx-test.bc %s 2>&1 \
 241 // RUN:   | FileCheck %s
 242
 243 // verify the tools invocations
 244 // CHECK: "-cc1" "-triple" "x86_64-unknown-linux-gnu"{{.*}}"-emit-llvm-bc"{{.*}}"-x" "c"
 245 // CHECK: "-cc1" "-triple" "nvptx64-nvidia-cuda" "-aux-triple" "x86_64-unknown-linux-gnu"{{.*}}"-target-cpu" "sm_52"
 246 // CHECK: "-cc1" "-triple" "x86_64-unknown-linux-gnu"{{.*}}"-emit-obj"
 247 // CHECK: clang-linker-wrapper{{.*}}"--"{{.*}} "-o" "a.out"
 248
 249 // RUN:   %clang -ccc-print-phases --target=x86_64-unknown-linux-gnu -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 %s 2>&1 \
 250 // RUN:   | FileCheck --check-prefix=CHECK-PHASES %s
 251 // CHECK-PHASES: 0: input, "[[INPUT:.+]]", c, (host-openmp)
 252 // CHECK-PHASES: 1: preprocessor, {0}, cpp-output, (host-openmp)
 253 // CHECK-PHASES: 2: compiler, {1}, ir, (host-openmp)
 254 // CHECK-PHASES: 3: input, "[[INPUT]]", c, (device-openmp)
 255 // CHECK-PHASES: 4: preprocessor, {3}, cpp-output, (device-openmp)
 256 // CHECK-PHASES: 5: compiler, {4}, ir, (device-openmp)
 257 // CHECK-PHASES: 6: offload, "host-openmp (x86_64-unknown-linux-gnu)" {2}, "device-openmp (nvptx64-nvidia-cuda)" {5}, ir
 258 // CHECK-PHASES: 7: backend, {6}, assembler, (device-openmp)
 259 // CHECK-PHASES: 8: assembler, {7}, object, (device-openmp)
 260 // CHECK-PHASES: 9: offload, "device-openmp (nvptx64-nvidia-cuda)" {8}, object
 261 // CHECK-PHASES: 10: clang-offload-packager, {9}, image
 262 // CHECK-PHASES: 11: offload, "host-openmp (x86_64-unknown-linux-gnu)" {2}, "device-openmp (x86_64-unknown-linux-gnu)" {10}, ir
 263 // CHECK-PHASES: 12: backend, {11}, assembler, (host-openmp)
 264 // CHECK-PHASES: 13: assembler, {12}, object, (host-openmp)
 265 // CHECK-PHASES: 14: clang-linker-wrapper, {13}, image, (host-openmp)
 266
 267 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-BINDINGS
 268 // CHECK-BINDINGS: "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[INPUT:.+]]"], output: "[[HOST_BC:.+]]"
 269 // CHECK-BINDINGS: "nvptx64-nvidia-cuda" - "clang", inputs: ["[[INPUT]]", "[[HOST_BC]]"], output: "[[DEVICE_BC:.+]]"
 270 // CHECK-BINDINGS: "nvptx64-nvidia-cuda" - "NVPTX::Assembler", inputs: ["[[DEVICE_BC]]"], output: "[[DEVICE_OBJ:.+]]"
 271 // CHECK-BINDINGS: "x86_64-unknown-linux-gnu" - "Offload::Packager", inputs: ["[[DEVICE_OBJ]]"], output: "[[BINARY:.+.out]]"
 272 // CHECK-BINDINGS: "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[HOST_BC]]", "[[BINARY]]"], output: "[[HOST_OBJ:.+]]"
 273 // CHECK-BINDINGS: "x86_64-unknown-linux-gnu" - "Offload::Linker", inputs: ["[[HOST_OBJ]]"], output: "a.out"
 274
 275 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 -nogpulib -save-temps %s 2>&1 | FileCheck %s --check-prefix=CHECK-TEMP-BINDINGS
 276 // CHECK-TEMP-BINDINGS: "x86_64-unknown-linux-gnu" - "Offload::Packager", inputs: ["[[DEVICE_OBJ:.+]]"], output: "[[BINARY:.+.out]]"
 277
 278 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda --offload-arch=sm_52 --offload-arch=sm_70 -nogpuinc -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-ARCH-BINDINGS
 279 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda --offload-arch=sm_52,sm_70 -nogpuinc -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-ARCH-BINDINGS
 280 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda --offload-arch=sm_52,sm_60,sm_70,sm_80 --no-offload-arch=sm_60,sm_80 -nogpuinc -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-ARCH-BINDINGS
 281 // CHECK-ARCH-BINDINGS: "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[INPUT:.*]]"], output: "[[HOST_BC:.*]]"
 282 // CHECK-ARCH-BINDINGS: "nvptx64-nvidia-cuda" - "clang", inputs: ["[[INPUT]]", "[[HOST_BC]]"], output: "[[DEVICE_BC_SM_52:.*]]"
 283 // CHECK-ARCH-BINDINGS: "nvptx64-nvidia-cuda" - "NVPTX::Assembler", inputs: ["[[DEVICE_BC_SM_52]]"], output: "[[DEVICE_OBJ_SM_52:.*]]"
 284 // CHECK-ARCH-BINDINGS: "nvptx64-nvidia-cuda" - "clang", inputs: ["[[INPUT]]", "[[HOST_BC]]"], output: "[[DEVICE_BC_SM_70:.*]]"
 285 // CHECK-ARCH-BINDINGS: "nvptx64-nvidia-cuda" - "NVPTX::Assembler", inputs: ["[[DEVICE_BC_SM_70]]"], output: "[[DEVICE_OBJ_SM_70:.*]]"
 286 // CHECK-ARCH-BINDINGS: "x86_64-unknown-linux-gnu" - "Offload::Packager", inputs: ["[[DEVICE_OBJ_SM_52]]", "[[DEVICE_OBJ_SM_70]]"], output: "[[BINARY:.*]]"
 287 // CHECK-ARCH-BINDINGS: "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[HOST_BC]]", "[[BINARY]]"], output: "[[HOST_OBJ:.*]]"
 288 // CHECK-ARCH-BINDINGS: "x86_64-unknown-linux-gnu" - "Offload::Linker", inputs: ["[[HOST_OBJ]]"], output: "a.out"
 289
 290 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp \
 291 // RUN:     -fopenmp-targets=nvptx64-nvidia-cuda,amdgcn-amd-amdhsa -Xopenmp-target=nvptx64-nvidia-cuda --offload-arch=sm_70 \
 292 // RUN:     -fopenmp-targets=nvptx64-nvidia-cuda,amdgcn-amd-amdhsa -Xopenmp-target=amdgcn-amd-amdhsa --offload-arch=gfx908  \
 293 // RUN:     -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-NVIDIA-AMDGPU
 294
 295 // CHECK-NVIDIA-AMDGPU: "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[INPUT:.+]]"], output: "[[HOST_BC:.+]]"
 296 // CHECK-NVIDIA-AMDGPU: "amdgcn-amd-amdhsa" - "clang", inputs: ["[[INPUT]]", "[[HOST_BC]]"], output: "[[AMD_BC:.+]]"
 297 // CHECK-NVIDIA-AMDGPU: "nvptx64-nvidia-cuda" - "clang", inputs: ["[[INPUT]]", "[[HOST_BC]]"], output: "[[NVIDIA_PTX:.+]]"
 298 // CHECK-NVIDIA-AMDGPU: "nvptx64-nvidia-cuda" - "NVPTX::Assembler", inputs: ["[[NVIDIA_PTX]]"], output: "[[NVIDIA_CUBIN:.+]]"
 299 // CHECK-NVIDIA-AMDGPU: "x86_64-unknown-linux-gnu" - "Offload::Packager", inputs: ["[[AMD_BC]]", "[[NVIDIA_CUBIN]]"], output: "[[BINARY:.*]]"
 300 // CHECK-NVIDIA-AMDGPU: "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[HOST_BC]]", "[[BINARY]]"], output: "[[HOST_OBJ:.+]]"
 301 // CHECK-NVIDIA-AMDGPU: "x86_64-unknown-linux-gnu" - "Offload::Linker", inputs: ["[[HOST_OBJ]]"], output: "a.out"
 302
 303 // RUN:   %clang -x ir -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp --offload-arch=sm_52 -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-IR
 304
 305 // CHECK-IR: "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[INPUT_IR:.+]]"], output: "[[OBJECT:.+]]"
 306 // CHECK-IR: "x86_64-unknown-linux-gnu" - "Offload::Linker", inputs: ["[[OBJECT]]"], output: "a.out"
 307
 308 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -emit-llvm -S -fopenmp=libomp --offload-device-only \
 309 // RUN:     -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 -nogpulib %s 2>&1 \
 310 // RUN:   | FileCheck %s --check-prefix=CHECK-EMIT-LLVM-IR
 311 // CHECK-EMIT-LLVM-IR: "-cc1"{{.*}}"-triple" "nvptx64-nvidia-cuda"{{.*}}"-emit-llvm"
 312
 313 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -emit-llvm -S -fopenmp=libomp \
 314 // RUN:     -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 -nogpulib %s 2>&1 \
 315 // RUN:   | FileCheck %s --check-prefix=CHECK-EMIT-LLVM-IR-BC
 316 // CHECK-EMIT-LLVM-IR-BC: "-cc1"{{.*}}"-triple" "nvptx64-nvidia-cuda"{{.*}}"-emit-llvm-bc"
 317
 318 // RUN:   %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target=nvptx64-nvida-cuda -march=sm_70 \
 319 // RUN:          --libomptarget-nvptx-bc-path=%S/Inputs/libomptarget/libomptarget-new-nvptx-test.bc \
 320 // RUN:          -nogpulib %s -o openmp-offload-gpu 2>&1 \
 321 // RUN:   | FileCheck -check-prefix=DRIVER_EMBEDDING %s
 322
 323 // DRIVER_EMBEDDING: -fembed-offload-object={{.*}}.out
 324
 325 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda \
 326 // RUN:     --offload-host-only -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-HOST-ONLY
 327 // CHECK-HOST-ONLY: "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[INPUT:.*]]"], output: "[[OUTPUT:.*]]"
 328 // CHECK-HOST-ONLY: "x86_64-unknown-linux-gnu" - "Offload::Linker", inputs: ["[[OUTPUT]]"], output: "a.out"
 329
 330 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda \
 331 // RUN:     -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 --offload-device-only -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-DEVICE-ONLY
 332 // CHECK-DEVICE-ONLY: "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[INPUT:.*]]"], output: "[[HOST_BC:.*]]"
 333 // CHECK-DEVICE-ONLY: "nvptx64-nvidia-cuda" - "clang", inputs: ["[[INPUT]]", "[[HOST_BC]]"], output: "[[DEVICE_ASM:.*]]"
 334 // CHECK-DEVICE-ONLY: "nvptx64-nvidia-cuda" - "NVPTX::Assembler", inputs: ["[[DEVICE_ASM]]"], output: "{{.*}}-openmp-nvptx64-nvidia-cuda.o"
 335
 336 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda \
 337 // RUN:     -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 --offload-device-only -E -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-DEVICE-ONLY-PP
 338 // CHECK-DEVICE-ONLY-PP: "nvptx64-nvidia-cuda" - "clang", inputs: ["[[INPUT:.*]]"], output: "-"
 339
 340 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp --offload-arch=sm_52 --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
 341 // RUN:     -foffload-lto %s 2>&1 | FileCheck --check-prefix=CHECK-LTO-LIBRARY %s
 342
 343 // CHECK-LTO-LIBRARY: {{.*}}-lomptarget{{.*}}-lomptarget.devicertl
 344
 345 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp --offload-arch=sm_52 --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
 346 // RUN:     --libomptarget-nvptx-bc-path=%S/Inputs/libomptarget/libomptarget-nvptx-test.bc %s 2>&1 \
 347 // RUN:   | FileCheck --check-prefix=CHECK-NO-LTO-LIBRARY %s
 348
 349 // CHECK-NO-LTO-LIBRARY: {{.*}}-lomptarget{{.*}}-lomptarget.devicertl
 350
 351 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp --offload-arch=sm_52 -nogpulib \
 352 // RUN:     -foffload-lto %s 2>&1 | FileCheck --check-prefix=CHECK-NO-LIBRARY %s
 353
 354 // CHECK-NO-LIBRARY-NOT: {{.*}}-lomptarget{{.*}}-lomptarget.devicertl
 355
 356 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp --offload-arch=sm_52 -nogpulib \
 357 // RUN:     -Xoffload-linker a -Xoffload-linker-nvptx64-nvidia-cuda b -Xoffload-linker-nvptx64 c \
 358 // RUN:     %s 2>&1 | FileCheck --check-prefix=CHECK-XLINKER %s
 359
 360 // CHECK-XLINKER: -device-linker=a{{.*}}-device-linker=nvptx64-nvidia-cuda=b{{.*}}-device-linker=nvptx64-nvidia-cuda=c{{.*}}--
 361
 362 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp --offload-arch=sm_52 -nogpulib \
 363 // RUN:     -foffload-lto %s 2>&1 | FileCheck --check-prefix=CHECK-LTO-FEATURES %s
 364
 365 // CHECK-LTO-FEATURES: clang-offload-packager{{.*}}--image={{.*}}feature=+ptx{{[0-9]+}}
 366
 367 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp --offload-arch=sm_52 -nogpulib \
 368 // RUN:     -Xopenmp-target=nvptx64-nvidia-cuda --cuda-feature=+ptx64 -foffload-lto %s 2>&1 \
 369 // RUN:    | FileCheck --check-prefix=CHECK-SET-FEATURES %s
 370
 371 // CHECK-SET-FEATURES: clang-offload-packager{{.*}}--image={{.*}}feature=+ptx64
 372
 373 //
 374 // Check that `-Xarch_host` works for OpenMP offloading.
 375 //
 376 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp \
 377 // RUN:     --offload-arch=sm_52 -nogpulib -nogpuinc -Xarch_host -O3 %s 2>&1 \
 378 // RUN:   | FileCheck --check-prefix=XARCH-HOST %s
 379 // XARCH-HOST: "-cc1" "-triple" "x86_64-unknown-linux-gnu"{{.*}}"-O3"
 380 // XARCH-HOST-NOT: "-cc1" "-triple" "nvptx64-nvidia-cuda"{{.*}}"-O3"
 381
 382 //
 383 // Check that `-Xarch_device` works for OpenMP offloading.
 384 //
 385 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp \
 386 // RUN:     --offload-arch=sm_52 -nogpulib -nogpuinc -Xarch_device -O3 %s 2>&1 \
 387 // RUN:   | FileCheck --check-prefix=XARCH-DEVICE %s
 388 // XARCH-DEVICE: "-cc1" "-triple" "nvptx64-nvidia-cuda"{{.*}}"-O3"
 389 // XARCH-DEVICE-NOT: "-cc1" "-triple" "x86_64-unknown-linux-gnu"{{.*}}"-O3"
 390
 391 //
 392 // Check that `-gpulibc` includes the LLVM C libraries for the GPU.
 393 //
 394 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp \
 395 // RUN:      --libomptarget-nvptx-bc-path=%S/Inputs/libomptarget/libomptarget-nvptx-test.bc \
 396 // RUN:      --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
 397 // RUN:      --offload-arch=sm_52 -gpulibc -nogpuinc %s 2>&1 \
 398 // RUN:   | FileCheck --check-prefix=LIBC-GPU %s
 399 // LIBC-GPU: "-lcgpu"{{.*}}"-lmgpu"
 400
 401 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -fopenmp=libomp \
 402 // RUN:      --libomptarget-nvptx-bc-path=%S/Inputs/libomptarget/libomptarget-nvptx-test.bc \
 403 // RUN:      --cuda-path=%S/Inputs/CUDA_102/usr/local/cuda \
 404 // RUN:      --offload-arch=sm_52 -nogpulibc -nogpuinc %s 2>&1 \
 405 // RUN:   | FileCheck --check-prefix=NO-LIBC-GPU %s
 406 // NO-LIBC-GPU-NOT: "-lcgpu"{{.*}}"-lmgpu"