diff --git a/llvm/include/llvm/Frontend/OpenMP/OMPKinds.def b/llvm/include/llvm/Frontend/OpenMP/OMPKinds.def
--- a/llvm/include/llvm/Frontend/OpenMP/OMPKinds.def
+++ b/llvm/include/llvm/Frontend/OpenMP/OMPKinds.def
@@ -87,6 +87,7 @@
 
 __OMP_STRUCT_TYPE(Ident, ident_t, Int32, Int32, Int32, Int32, Int8Ptr)
 __OMP_STRUCT_TYPE(AsyncInfo, __tgt_async_info, Int8Ptr)
+__OMP_STRUCT_TYPE(KernelEnvironmentTy, KernelEnvironmentTy, Ident)
 
 #undef __OMP_STRUCT_TYPE
 #undef OMP_STRUCT_TYPE
@@ -423,8 +424,8 @@
           /* Int */ Int32, /* kmp_task_t */ VoidPtr)
 
 /// OpenMP Device runtime functions
-__OMP_RTL(__kmpc_target_init, false, Int32, IdentPtr, Int8, Int1, Int1)
-__OMP_RTL(__kmpc_target_deinit, false, Void, IdentPtr, Int8, Int1)
+__OMP_RTL(__kmpc_target_init, false, Int32, KernelEnvironmentTyPtr, Int8, Int1, Int1)
+__OMP_RTL(__kmpc_target_deinit, false, Void, Int8, Int1)
 __OMP_RTL(__kmpc_kernel_prepare_parallel, false, Void, VoidPtr)
 __OMP_RTL(__kmpc_parallel_51, false, Void, IdentPtr, Int32, Int32, Int32, Int32,
           VoidPtr, VoidPtr, VoidPtrPtr, SizeTy)
diff --git a/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp b/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
--- a/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
+++ b/llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp
@@ -2811,8 +2811,25 @@
   Function *Fn = getOrCreateRuntimeFunctionPtr(
       omp::RuntimeFunction::OMPRTL___kmpc_target_init);
 
-  CallInst *ThreadKind = Builder.CreateCall(
-      Fn, {Ident, IsSPMDVal, UseGenericStateMachine, RequiresFullRuntimeVal});
+  Function *Kernel = Builder.GetInsertBlock()->getParent();
+  const DataLayout &DL = Fn->getParent()->getDataLayout();
+  Constant *KernelEnvironmentInitializer = ConstantStruct::get(
+      KernelEnvironmentTy, {cast<GlobalVariable>(Ident)->getInitializer()});
+  std::string KernelEnvironmentName =
+      (Kernel->getName() + "_kernel_info").str();
+  GlobalVariable *KernelEnvironment = new GlobalVariable(
+      M, KernelEnvironmentTy, /*IsConstant*/ false,
+      llvm::GlobalValue::ExternalLinkage, KernelEnvironmentInitializer,
+      KernelEnvironmentName,
+      /*InsertBefore=*/nullptr, llvm::GlobalValue::NotThreadLocal,
+      DL.getDefaultGlobalsAddressSpace());
+  auto *KernelEnvironmentCasted =
+      ConstantExpr::getPointerBitCastOrAddrSpaceCast(KernelEnvironment,
+                                                     KernelEnvironmentTyPtr);
+
+  CallInst *ThreadKind =
+      Builder.CreateCall(Fn, {KernelEnvironmentCasted, IsSPMDVal,
+                              UseGenericStateMachine, RequiresFullRuntimeVal});
 
   Value *ExecUserCode = Builder.CreateICmpEQ(
       ThreadKind, ConstantInt::get(ThreadKind->getType(), -1),
@@ -2851,9 +2868,6 @@
   if (!updateToLocation(Loc))
     return;
 
-  uint32_t SrcLocStrSize;
-  Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
-  Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
   ConstantInt *IsSPMDVal = ConstantInt::getSigned(
       IntegerType::getInt8Ty(Int8->getContext()),
       IsSPMD ? OMP_TGT_EXEC_MODE_SPMD : OMP_TGT_EXEC_MODE_GENERIC);
@@ -2863,7 +2877,7 @@
   Function *Fn = getOrCreateRuntimeFunctionPtr(
       omp::RuntimeFunction::OMPRTL___kmpc_target_deinit);
 
-  Builder.CreateCall(Fn, {Ident, IsSPMDVal, RequiresFullRuntimeVal});
+  Builder.CreateCall(Fn, {IsSPMDVal, RequiresFullRuntimeVal});
 }
 
 std::string OpenMPIRBuilder::getNameWithSeparators(ArrayRef<StringRef> Parts,
diff --git a/openmp/libomptarget/DeviceRTL/CMakeLists.txt b/openmp/libomptarget/DeviceRTL/CMakeLists.txt
--- a/openmp/libomptarget/DeviceRTL/CMakeLists.txt
+++ b/openmp/libomptarget/DeviceRTL/CMakeLists.txt
@@ -117,7 +117,7 @@
   ${source_directory}/Workshare.cpp
 )
 
-set(clang_opt_flags -O1 -mllvm -openmp-opt-disable -DSHARED_SCRATCHPAD_SIZE=2048)
+set(clang_opt_flags -O1 -mllvm -openmp-opt-disable -DSHARED_SCRATCHPAD_SIZE=2048 -DOMPTARGET_DEVICE_RUNTIME)
 set(link_opt_flags  -O1        -openmp-opt-disable)
 
 # Prepend -I to each list element
diff --git a/openmp/libomptarget/DeviceRTL/include/Interface.h b/openmp/libomptarget/DeviceRTL/include/Interface.h
--- a/openmp/libomptarget/DeviceRTL/include/Interface.h
+++ b/openmp/libomptarget/DeviceRTL/include/Interface.h
@@ -12,6 +12,7 @@
 #ifndef OMPTARGET_DEVICERTL_INTERFACE_H
 #define OMPTARGET_DEVICERTL_INTERFACE_H
 
+#include "KernelEnvironment.h"
 #include "Types.h"
 
 /// External API
@@ -211,10 +212,10 @@
 ///{
 int8_t __kmpc_is_spmd_exec_mode();
 
-int32_t __kmpc_target_init(IdentTy *Ident, int8_t Mode,
+int32_t __kmpc_target_init(_OMP::KernelEnvironmentTy KernelEnv, int8_t Mode,
                            bool UseGenericStateMachine, bool);
 
-void __kmpc_target_deinit(IdentTy *Ident, int8_t Mode, bool);
+void __kmpc_target_deinit(int8_t Mode, bool);
 
 ///}
 
diff --git a/openmp/libomptarget/DeviceRTL/include/State.h b/openmp/libomptarget/DeviceRTL/include/State.h
--- a/openmp/libomptarget/DeviceRTL/include/State.h
+++ b/openmp/libomptarget/DeviceRTL/include/State.h
@@ -13,6 +13,7 @@
 #define OMPTARGET_STATE_H
 
 #include "Debug.h"
+#include "KernelEnvironment.h"
 #include "Types.h"
 
 #pragma omp declare target
@@ -24,7 +25,10 @@
 inline constexpr uint32_t SharedScratchpadSize = SHARED_SCRATCHPAD_SIZE;
 
 /// Initialize the state machinery. Must be called by all threads.
-void init(bool IsSPMD);
+void init(bool IsSPMD, KernelEnvironmentTy &KernelEnv);
+
+/// Return the kernel environment associated with the current kernel.
+KernelEnvironmentTy &getKernelEnvironment();
 
 /// TODO
 enum ValueKind {
diff --git a/openmp/libomptarget/DeviceRTL/src/Kernel.cpp b/openmp/libomptarget/DeviceRTL/src/Kernel.cpp
--- a/openmp/libomptarget/DeviceRTL/src/Kernel.cpp
+++ b/openmp/libomptarget/DeviceRTL/src/Kernel.cpp
@@ -12,6 +12,7 @@
 
 #include "Debug.h"
 #include "Interface.h"
+#include "KernelEnvironment.h"
 #include "Mapping.h"
 #include "State.h"
 #include "Synchronization.h"
@@ -21,15 +22,15 @@
 
 #pragma omp declare target
 
-static void inititializeRuntime(bool IsSPMD) {
+static void inititializeRuntime(bool IsSPMD, KernelEnvironmentTy &KernelEnv) {
   // Order is important here.
   synchronize::init(IsSPMD);
   mapping::init(IsSPMD);
-  state::init(IsSPMD);
+  state::init(IsSPMD, KernelEnv);
 }
 
 /// Simple generic state machine for worker threads.
-static void genericStateMachine(IdentTy *Ident) {
+static void genericStateMachine(IdentTy &Ident) {
   FunctionTracingRAII();
 
   uint32_t TId = mapping::getThreadIdInBlock();
@@ -61,19 +62,15 @@
 
 extern "C" {
 
-/// Initialization
-///
-/// \param Ident               Source location identification, can be NULL.
-///
-int32_t __kmpc_target_init(IdentTy *Ident, int8_t Mode,
+int32_t __kmpc_target_init(KernelEnvironmentTy KernelEnv, int8_t Mode,
                            bool UseGenericStateMachine, bool) {
   FunctionTracingRAII();
   const bool IsSPMD = Mode & OMP_TGT_EXEC_MODE_SPMD;
   if (IsSPMD) {
-    inititializeRuntime(/* IsSPMD */ true);
+    inititializeRuntime(/* IsSPMD */ true, KernelEnv);
     synchronize::threadsAligned();
   } else {
-    inititializeRuntime(/* IsSPMD */ false);
+    inititializeRuntime(/* IsSPMD */ false, KernelEnv);
     // No need to wait since only the main threads will execute user
     // code and workers will run into a barrier right away.
   }
@@ -87,7 +84,7 @@
     return -1;
 
   if (UseGenericStateMachine)
-    genericStateMachine(Ident);
+    genericStateMachine(KernelEnv.Ident);
 
   return mapping::getThreadIdInBlock();
 }
@@ -96,10 +93,7 @@
 ///
 /// In non-SPMD, this function releases the workers trapped in a state machine
 /// and also any memory dynamically allocated by the runtime.
-///
-/// \param Ident Source location identification, can be NULL.
-///
-void __kmpc_target_deinit(IdentTy *Ident, int8_t Mode, bool) {
+void __kmpc_target_deinit(int8_t Mode, bool) {
   FunctionTracingRAII();
   const bool IsSPMD = Mode & OMP_TGT_EXEC_MODE_SPMD;
   state::assumeInitialState(IsSPMD);
diff --git a/openmp/libomptarget/DeviceRTL/src/State.cpp b/openmp/libomptarget/DeviceRTL/src/State.cpp
--- a/openmp/libomptarget/DeviceRTL/src/State.cpp
+++ b/openmp/libomptarget/DeviceRTL/src/State.cpp
@@ -32,6 +32,9 @@
 extern unsigned char DynamicSharedBuffer[] __attribute__((aligned(Alignment)));
 #pragma omp allocate(DynamicSharedBuffer) allocator(omp_pteam_mem_alloc)
 
+/// The kernel environment passed to the init method by the compiler.
+static KernelEnvironmentTy *SHARED(KernelEnvironmentPtr);
+
 namespace {
 
 /// Fallback implementations are missing to trigger a link time error.
@@ -364,14 +367,20 @@
   __builtin_unreachable();
 }
 
-void state::init(bool IsSPMD) {
+void state::init(bool IsSPMD, KernelEnvironmentTy &KernelEnv) {
   SharedMemorySmartStack.init(IsSPMD);
-  if (mapping::isInitialThreadInLevel0(IsSPMD))
+  if (mapping::isInitialThreadInLevel0(IsSPMD)) {
     TeamState.init(IsSPMD);
+    KernelEnvironmentPtr = &KernelEnv;
+  }
 
   ThreadStates[mapping::getThreadIdInBlock()] = nullptr;
 }
 
+KernelEnvironmentTy &state::getKernelEnvironment() {
+  return *KernelEnvironmentPtr;
+}
+
 void state::enterDataEnvironment() {
   unsigned TId = mapping::getThreadIdInBlock();
   ThreadStateTy *NewThreadState =
diff --git a/openmp/libomptarget/deviceRTLs/amdgcn/CMakeLists.txt b/openmp/libomptarget/deviceRTLs/amdgcn/CMakeLists.txt
--- a/openmp/libomptarget/deviceRTLs/amdgcn/CMakeLists.txt
+++ b/openmp/libomptarget/deviceRTLs/amdgcn/CMakeLists.txt
@@ -131,6 +131,7 @@
     -fvisibility=default
     -Wno-unused-value
     -nogpulib
+    -DOMPTARGET_OLD_DEVICE_RUNTIME
     -O${optimization_level}
     ${CUDA_DEBUG}
     -I${CMAKE_CURRENT_SOURCE_DIR}/src
diff --git a/openmp/libomptarget/deviceRTLs/common/src/omptarget.cu b/openmp/libomptarget/deviceRTLs/common/src/omptarget.cu
--- a/openmp/libomptarget/deviceRTLs/common/src/omptarget.cu
+++ b/openmp/libomptarget/deviceRTLs/common/src/omptarget.cu
@@ -207,7 +207,7 @@
 }
 
 EXTERN
-int32_t __kmpc_target_init(ident_t *Ident, int8_t Mode,
+int32_t __kmpc_target_init(KernelEnvironmentTy KernelEnv, int8_t Mode,
                            bool UseGenericStateMachine,
                            bool RequiresFullRuntime) {
   const bool IsSPMD = Mode & OMP_TGT_EXEC_MODE_SPMD;
@@ -218,7 +218,7 @@
     __kmpc_generic_kernel_init();
 
    if (IsSPMD) {
-    __kmpc_barrier_simple_spmd(Ident, TId);
+    __kmpc_barrier_simple_spmd(&KernelEnv.Ident, TId);
      return -1;
    }
 
@@ -226,13 +226,13 @@
      return -1;
 
   if (UseGenericStateMachine)
-    __kmpc_target_region_state_machine(Ident);
+    __kmpc_target_region_state_machine(&KernelEnv.Ident);
 
   return TId;
 }
 
 EXTERN
-void __kmpc_target_deinit(ident_t *Ident, int8_t Mode,
+void __kmpc_target_deinit(int8_t Mode,
                           bool RequiresFullRuntime) {
   const bool IsSPMD = Mode & OMP_TGT_EXEC_MODE_SPMD;
   if (IsSPMD)
diff --git a/openmp/libomptarget/deviceRTLs/interface.h b/openmp/libomptarget/deviceRTLs/interface.h
--- a/openmp/libomptarget/deviceRTLs/interface.h
+++ b/openmp/libomptarget/deviceRTLs/interface.h
@@ -439,11 +439,12 @@
 EXTERN int32_t __kmpc_cancel(kmp_Ident *loc, int32_t global_tid,
                              int32_t cancelVal);
 
+#include "KernelEnvironment.h"
 // non standard
-EXTERN int32_t __kmpc_target_init(ident_t *Ident, int8_t Mode,
+EXTERN int32_t __kmpc_target_init(KernelEnvironmentTy, int8_t Mode,
                                   bool UseGenericStateMachine,
                                   bool RequiresFullRuntime);
-EXTERN void __kmpc_target_deinit(ident_t *Ident, int8_t Mode,
+EXTERN void __kmpc_target_deinit(int8_t Mode,
                                  bool RequiresFullRuntime);
 EXTERN void __kmpc_kernel_prepare_parallel(void *WorkFn);
 EXTERN bool __kmpc_kernel_parallel(void **WorkFn);
diff --git a/openmp/libomptarget/deviceRTLs/nvptx/CMakeLists.txt b/openmp/libomptarget/deviceRTLs/nvptx/CMakeLists.txt
--- a/openmp/libomptarget/deviceRTLs/nvptx/CMakeLists.txt
+++ b/openmp/libomptarget/deviceRTLs/nvptx/CMakeLists.txt
@@ -187,7 +187,7 @@
 
 # Generate a Bitcode library for all the compute capabilities the user requested
 foreach(sm ${nvptx_sm_list})
-  set(cuda_flags -Xclang -target-cpu -Xclang sm_${sm} "-D__CUDA_ARCH__=${sm}0")
+  set(cuda_flags -Xclang -target-cpu -Xclang sm_${sm} "-D__CUDA_ARCH__=${sm}0" -DOMPTARGET_OLD_DEVICE_RUNTIME)
   set(bc_files "")
   foreach(src ${cuda_src_files})
     get_filename_component(infile ${src} ABSOLUTE)
diff --git a/openmp/libomptarget/include/KernelEnvironment.h b/openmp/libomptarget/include/KernelEnvironment.h
new file mode 100644
--- /dev/null
+++ b/openmp/libomptarget/include/KernelEnvironment.h
@@ -0,0 +1,40 @@
+//===---- KernelEnvironment.h - OpenMP GPU kernel environment ----- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef OMPTARGET_KERNEL_ENVIRONMENT_H
+#define OMPTARGET_KERNEL_ENVIRONMENT_H
+
+// deviceRTL uses <stdint> and DeviceRTL uses explicit definitions
+
+#ifdef OMPTARGET_DEVICE_RUNTIME
+#include "Types.h"
+#else
+#ifdef OMPTARGET_OLD_DEVICE_RUNTIME
+#include "interface.h"
+using IdentTy = ident_t;
+#else
+#include <SourceInfo.h>
+using IdentTy = ident_t;
+#endif
+#endif
+
+#ifdef OMPTARGET_DEVICE_RUNTIME
+namespace _OMP {
+#endif
+
+struct KernelEnvironmentTy {
+  IdentTy Ident;
+};
+
+#ifdef OMPTARGET_DEVICE_RUNTIME
+} // namespace _OMP
+#endif
+
+#endif