diff --git a/hipamd/src/hip_clang.cpp b/hipamd/src/hip_clang.cpp
index 635aa8391f..e8f3e86881 100644
--- a/hipamd/src/hip_clang.cpp
+++ b/hipamd/src/hip_clang.cpp
@@ -28,6 +28,7 @@ THE SOFTWARE.
 #include "hip_hcc_internal.h"
 #include "hip_fatbin.h"
 #include "trace_helper.h"
+#include "program_state.inl"
 
 #ifdef __GNUC__
 #pragma GCC visibility push (default)
@@ -94,8 +95,10 @@ __hipRegisterFatBinary(const void* data)
                                                                          agent);
 
       if (module->executable.handle) {
-        modules->at(deviceId) = module;
-        tprintf(DB_FB, "Loaded code object for %s\n", name);
+         hip_impl::program_state_impl::read_kernarg_metadata(image, module->kernargs);
+         modules->at(deviceId) = module;
+
+         tprintf(DB_FB, "Loaded code object for %s, args size=%ld\n", name, module->kernargs.size());
       } else {
         fprintf(stderr, "Failed to load code object for %s\n", name);
         abort();
@@ -232,7 +235,7 @@ static DeviceVar* findVar(std::string hostVar, int deviceId, hipModule_t hmod) {
 }
 
 hipError_t ihipGetGlobalVar(hipDeviceptr_t* dev_ptr, size_t* size_ptr,
-		              const char* hostVar, hipModule_t hmod) {
+                             const char* hostVar, hipModule_t hmod) {
   GET_TLS();
   auto ctx = ihipGetTlsDefaultCtx();
 
@@ -425,6 +428,41 @@ extern "C" hipError_t __hipPopCallConfiguration(
   return hipSuccess;
 }
 
+int getCurrentDeviceId()
+{
+  GET_TLS();
+
+  int deviceId = 0;
+  auto ctx = ihipGetTlsDefaultCtx();
+
+  if(!ctx) return deviceId;
+
+  LockedAccessor_CtxCrit_t crit(ctx->criticalData());
+
+  if(crit->_execStack.size() != 0)
+  {
+    auto &exec = crit->_execStack.top();
+
+    if (exec._hStream) {
+      deviceId = exec._hStream->getDevice()->_deviceId;
+    } else if (ctx->getDevice()) {
+      deviceId = ctx->getDevice()->_deviceId;
+    }
+  } else if (ctx->getDevice()) {
+    deviceId = ctx->getDevice()->_deviceId;
+  }
+  return deviceId;
+}
+
+hipFunction_t ihipGetDeviceFunction(const void *hostFunction)
+{
+  int deviceId = getCurrentDeviceId();
+  auto it = g_functions.find(hostFunction);
+  if (it == g_functions.end() || !it->second[deviceId]) {
+    return nullptr;
+  }
+  return it->second[deviceId];
+}
 
 hipError_t hipSetupArgument(
   const void *arg,
diff --git a/hipamd/src/hip_hcc_internal.h b/hipamd/src/hip_hcc_internal.h
index 8eec7292c2..104fd910a8 100644
--- a/hipamd/src/hip_hcc_internal.h
+++ b/hipamd/src/hip_hcc_internal.h
@@ -33,7 +33,7 @@ THE SOFTWARE.
 #include "hip_prof_api.h"
 #include "hip_util.h"
 #include "env.h"
-
+#include <unordered_map>
 
 #if (__hcc_workweek__ < 16354)
 #error("This version of HIP requires a newer version of HCC.");
@@ -1079,4 +1079,14 @@ static inline ihipCtx_t* iihipGetTlsDefaultCtx(TlsData* tls) {
     return tls->defaultCtx;
 }
 
+/**
+ *  @brief Get device function from host kernel function pointer
+ *  Needed only for clang + HIP-HCC RT
+ *
+ *  @param [in] hostFunction host kernel function pointer
+ *
+ *  @returns hipFuntion_t, nullptr
+ */
+hipFunction_t ihipGetDeviceFunction(const void *hostFunction);
+
 #endif
diff --git a/hipamd/src/hip_module.cpp b/hipamd/src/hip_module.cpp
index b6fbc45866..0a7348a3a2 100644
--- a/hipamd/src/hip_module.cpp
+++ b/hipamd/src/hip_module.cpp
@@ -344,6 +344,8 @@ hipError_t ihipExtLaunchMultiKernelMultiDevice(hipLaunchParams* launchParamsList
             free(kds);
             return hipErrorInvalidValue;
         }
+        if (!kds[i]->_kernarg_layout.empty()) continue;
+
         hip_impl::kernargs_size_align kargs = ps.get_kernargs_size_align(
                 reinterpret_cast<std::uintptr_t>(lp.func));
         kds[i]->_kernarg_layout = *reinterpret_cast<const std::vector<std::pair<std::size_t, std::size_t>>*>(
diff --git a/hipamd/src/program_state.inl b/hipamd/src/program_state.inl
index d77abd9674..548a56795f 100644
--- a/hipamd/src/program_state.inl
+++ b/hipamd/src/program_state.inl
@@ -19,6 +19,8 @@
 #include <hsa/hsa_ven_amd_loader.h>
 #include <amd_comgr.h>
 #include "hc.hpp"
+#include "hip_hcc_internal.h"
+#include "trace_helper.h"
 
 #include <link.h>
 
@@ -944,14 +946,16 @@ public:
 
         auto it0 = get_functions(agent).find(function_address);
 
-        if (it0 == get_functions(agent).cend()) {
-            hip_throw(std::runtime_error{
+        if (it0 != get_functions(agent).cend()) return it0->second;
+
+        // For hip-clang compiler + Hcc RT
+        hipFunction_t f = ihipGetDeviceFunction((const void*)function_address);
+        if (f) return reinterpret_cast<Kernel_descriptor&>(*f);
+
+        hip_throw(std::runtime_error{
                     "No device code available for function: " +
                     std::string(name(function_address)) +
                     ", for agent: " + name(agent)});
-        }
-
-        return it0->second;
     }
 
     const std::vector<std::pair<std::size_t, std::size_t>>& 
diff --git a/hipamd/tests/src/gcc/LaunchKernel.c b/hipamd/tests/src/gcc/LaunchKernel.c
index ef910a423e..d2fc854510 100644
--- a/hipamd/tests/src/gcc/LaunchKernel.c
+++ b/hipamd/tests/src/gcc/LaunchKernel.c
@@ -19,10 +19,10 @@
 
 
 /* HIT_START
- * BUILD_CMD: gpu.o %hc -I%hip-path/include -g -c %S/gpu.cpp -o %T/gpu.o EXCLUDE_HIP_PLATFORM nvcc vdi EXCLUDE_HIP_COMPILER clang
- * BUILD_CMD: launchkernel.o %hc -D__HIP_PLATFORM_HCC__ -g -I%hip-path/include -c %S/LaunchKernel.c -o %T/launchkernel.o EXCLUDE_HIP_PLATFORM nvcc vdi EXCLUDE_HIP_COMPILER clang
- * BUILD_CMD: LaunchKernel %hc %T/launchkernel.o %T/gpu.o -g -Wl,--rpath=%hip-path/lib %hip-path/lib/libhip_hcc.so -o %T/%t DEPENDS gpu.o launchkernel.o EXCLUDE_HIP_PLATFORM nvcc vdi EXCLUDE_HIP_COMPILER clang
- * TEST: %t EXCLUDE_HIP_PLATFORM nvcc vdi EXCLUDE_HIP_COMPILER clang
+ * BUILD_CMD: gpu.o %hc -I%hip-path/include -g -c %S/gpu.cpp -o %T/gpu.o EXCLUDE_HIP_PLATFORM nvcc vdi
+ * BUILD_CMD: launchkernel.o %hc -D__HIP_PLATFORM_HCC__ -g -I%hip-path/include -c %S/LaunchKernel.c -o %T/launchkernel.o EXCLUDE_HIP_PLATFORM nvcc vdi
+ * BUILD_CMD: LaunchKernel %hc %T/launchkernel.o %T/gpu.o -g -Wl,--rpath=%hip-path/lib %hip-path/lib/libhip_hcc.so -o %T/%t DEPENDS gpu.o launchkernel.o EXCLUDE_HIP_PLATFORM nvcc vdi
+ * TEST: %t EXCLUDE_HIP_PLATFORM nvcc vdi
  * HIT_END
  */