[Libomptarget] Change visibility to hidden for device RTL

[jhuber6]

Summary

This patch changes the visibility for all construct in the new device
RTL to be hidden by default. This is done after the changes introduced
in D117806 changed the visibility from being hidden by default for all
device compilations. This asserts that the visibility for the device
runtime library will be hidden except for the internal environment
variable. This is done to aid optimization and linking of the device
library.

[JonChesterfield]

LG. @ronlieb rocm may need similar annotation for the old runtime, there's a macro for making variables externally accessible that would apply visibility protected to the environment and the array used by the memory allocator.

openmp/libomptarget/DeviceRTL/src/Configuration.cpp

29

Protected? Don't want to preempt this one either

[jhuber6]

openmp/libomptarget/DeviceRTL/src/Configuration.cpp

29

Right, I'll change it.

[jhuber6]

Updating visibility and for old runtime.

openmp/libomptarget/DeviceRTL/CMakeLists.txt

 set(clang_opt_flags -O1 -mllvm -openmp-opt-disable -DSHARED_SCRATCHPAD_SIZE=2048)
 set(link_opt_flags  -O1        -openmp-opt-disable)
 
 # Prepend -I to each list element
 set (LIBOMPTARGET_LLVM_INCLUDE_DIRS_DEVICERTL "${LIBOMPTARGET_LLVM_INCLUDE_DIRS}")
 list(TRANSFORM LIBOMPTARGET_LLVM_INCLUDE_DIRS_DEVICERTL PREPEND "-I")
 
 # Set flags for LLVM Bitcode compilation.
-set(bc_flags -S -x c++ -std=c++17
+set(bc_flags -S -x c++ -std=c++17 -fvisibility=hidden
               ${clang_opt_flags}
              -Xclang -emit-llvm-bc
              -Xclang -aux-triple -Xclang ${aux_triple}
              -fopenmp -fopenmp-cuda-mode -Xclang -fopenmp-is-device
              -I${include_directory}
              -I${devicertl_base_directory}/../include
              ${LIBOMPTARGET_LLVM_INCLUDE_DIRS_DEVICERTL}
 )

 endfunction()
 
 # Generate a Bitcode library for all the compute capabilities the user requested
 foreach(sm ${nvptx_sm_list})
   compileDeviceRTLLibrary(sm_${sm} nvptx -target nvptx64-nvidia-cuda -Xclang -target-feature -Xclang +ptx61 "-D__CUDA_ARCH__=${sm}0")
 endforeach()
 
 foreach(mcpu ${amdgpu_mcpus})
-  compileDeviceRTLLibrary(${mcpu} amdgpu -target amdgcn-amd-amdhsa -D__AMDGCN__ -fvisibility=default -nogpulib)
+  compileDeviceRTLLibrary(${mcpu} amdgpu -target amdgcn-amd-amdhsa -D__AMDGCN__ -nogpulib)
 endforeach()

openmp/libomptarget/DeviceRTL/include/Types.h

 
 #define SHARED(NAME)                                                           \
   NAME [[clang::loader_uninitialized]];                                        \
   OMP_PRAGMA(allocate(NAME) allocator(omp_pteam_mem_alloc))
 
 // TODO: clang should use address space 5 for omp_thread_mem_alloc, but right
 //       now that's not the case.
 #define THREAD_LOCAL(NAME)                                                     \
-  NAME [[clang::loader_uninitialized, clang::address_space(5),                 \
-         gnu::visibility("hidden")]]
+  NAME [[clang::loader_uninitialized, clang::address_space(5)]]
 
 // TODO: clang should use address space 4 for omp_const_mem_alloc, maybe it
 //       does?
 #define CONSTANT(NAME)                                                         \
   NAME [[clang::loader_uninitialized, clang::address_space(4)]]
 
 ///}
 
 #endif

openmp/libomptarget/DeviceRTL/src/Configuration.cpp

 
 using namespace _OMP;
 
 #pragma omp declare target
 
 extern uint32_t __omp_rtl_debug_kind; // defined by CGOpenMPRuntimeGPU
 
 // TODO: We want to change the name as soon as the old runtime is gone.
+// This variable should be visibile to the plugin so we override the default
+// hidden visibility.
 DeviceEnvironmentTy CONSTANT(omptarget_device_environment)
-    __attribute__((used, retain, weak));
+    __attribute__((used, retain, weak, visibility("protected")));
 
 uint32_t config::getDebugKind() {
   return __omp_rtl_debug_kind & omptarget_device_environment.DebugKind;
 }
 
 uint32_t config::getNumDevices() {
   return omptarget_device_environment.NumDevices;
 }

openmp/libomptarget/deviceRTLs/amdgcn/CMakeLists.txt

set(cu_cmd ${CLANG_TOOL}
     -std=c++14
     -ffreestanding
     -target amdgcn-amd-amdhsa
     -emit-llvm
     -Xclang -aux-triple -Xclang ${aux_triple}
     -fopenmp -fopenmp-cuda-mode -Xclang -fopenmp-is-device
     -D__AMDGCN__
     -Xclang -target-cpu -Xclang ${mcpu}
-    -fvisibility=default
+    -fvisibility=hidden
     -Wno-unused-value
     -nogpulib
     -O${optimization_level}
     ${CUDA_DEBUG}
     -I${CMAKE_CURRENT_SOURCE_DIR}/src
     -I${devicertl_base_directory}/common/include
     -I${devicertl_base_directory}
     -I${devicertl_base_directory}/../include

openmp/libomptarget/deviceRTLs/amdgcn/src/target_impl.h

 
 #define INLINE inline
 #define NOINLINE __attribute__((noinline))
 #define ALIGN(N) __attribute__((aligned(N)))
 #define PLUGIN_ACCESSIBLE                                                      \
   __attribute__((used))   /* Don't discard values the plugin reads */          \
   __attribute__((weak))   /* We may have multiple definitions */               \
   __attribute__((retain)) /* Also needed to keep values alive */               \
-  __attribute__((visibility("default"))) /* Access via SHT_HASH */             \
-  __attribute__((section(".data")))      /* Not .bss, can write before load */
+  __attribute__((visibility("protected"))) /* Access via SHT_HASH */           \
+  __attribute__((section(".data")))        /* Not .bss, can write before load */
 
 #include "llvm/Frontend/OpenMP/OMPGridValues.h"
 
 INLINE constexpr const llvm::omp::GV &getGridValue() {
   return llvm::omp::getAMDGPUGridValues<__AMDGCN_WAVEFRONT_SIZE>();
 }
 
 ////////////////////////////////////////////////////////////////////////////////

openmp/libomptarget/deviceRTLs/nvptx/CMakeLists.txt

 set (LIBOMPTARGET_LLVM_INCLUDE_DIRS_NVPTX "${LIBOMPTARGET_LLVM_INCLUDE_DIRS}")
 list(TRANSFORM LIBOMPTARGET_LLVM_INCLUDE_DIRS_NVPTX PREPEND "-I")
 
 # Set flags for LLVM Bitcode compilation.
 set(bc_flags -S -x c++ -O1 -std=c++14
              -mllvm -openmp-opt-disable
              -ffreestanding
              -target nvptx64
+             -fvisibility=hidden
              -Xclang -emit-llvm-bc
              -Xclang -aux-triple -Xclang ${aux_triple}
              -fopenmp -fopenmp-cuda-mode -Xclang -fopenmp-is-device
              -Xclang -target-feature -Xclang +ptx61
              -D__CUDACC__
              -I${devicertl_base_directory}
              -I${devicertl_common_directory}/include
              -I${devicertl_nvptx_directory}/src