Decouple VTune and ETW Trace, enable VTune JIT profiling on ChakraCore.

1. Fix prefast warning on FUNC_NotifyEvent for release build
2. Remove unnecessary comments and CDECL, ORIGINAL_CDECL defines since the latest VTune SDK file jitprofiling.h uses JITAPI_CDECL instead of CDECL, also remove #pragma optimize off line
3. Fix jitprofiling.h name to be lower case to pass linux build
4. Add  #ifdef VTUNE_PROFILING at beginning of the file

bin/ChakraCore/ChakraCoreDllFunc.cpp

@@ -12,6 +12,9 @@
 #endif
 #include "JsrtContext.h"
 #include "TestHooks.h"
+#ifdef VTUNE_PROFILING
+#include "Base/VTuneChakraProfile.h"
+#endif
 
 extern HANDLE g_hInstance;
 #ifdef _WIN32
@@ -56,6 +59,9 @@ static BOOL AttachProcess(HANDLE hmod)
 #ifdef ENABLE_JS_ETW
     EtwTrace::Register();
 #endif
+#ifdef VTUNE_PROFILING
+    VTuneChakraProfile::Register();
+#endif 
     ValueType::Initialize();
     ThreadContext::GlobalInitialize();
 
@@ -148,6 +154,9 @@ EXTERN_C BOOL WINAPI DllMain(HINSTANCE hmod, DWORD dwReason, PVOID pvReserved)
         // Do this before DetachProcess() so that we won't have ETW rundown callbacks while destroying threadContexts.
         EtwTrace::UnRegister();
 #endif
+#ifdef VTUNE_PROFILING
+        VTuneChakraProfile::UnRegister();
+#endif 
 
         // don't do anything if we are in forceful shutdown
         // try to clean up handles in graceful shutdown

lib/Backend/Func.cpp

@@ -5,6 +5,9 @@
 #include "Backend.h"
 #include "Base/EtwTrace.h"
 #include "Base/ScriptContextProfiler.h"
+#ifdef VTUNE_PROFILING
+#include "Base/VTuneChakraProfile.h"
+#endif
 
 Func::Func(JitArenaAllocator *alloc, CodeGenWorkItem* workItem, const Js::FunctionCodeGenRuntimeData *const runtimeData,
     Js::PolymorphicInlineCacheInfo * const polymorphicInlineCacheInfo, CodeGenAllocators *const codeGenAllocators,
@@ -1892,7 +1895,7 @@ Func::GetVtableName(INT_PTR address)
 bool Func::DoRecordNativeMap() const
 {
 #if defined(VTUNE_PROFILING)
-    if (EtwTrace::isJitProfilingActive)
+    if (VTuneChakraProfile::isJitProfilingActive)
     {
         return true;
     }

lib/Common/CommonDefines.h

@@ -301,12 +301,13 @@
 
 #ifdef ENABLE_JS_ETW
 #define TEST_ETW_EVENTS
+#endif
 
 // VTUNE profiling requires ETW trace
 #if defined(_M_IX86) || defined(_M_X64)
 #define VTUNE_PROFILING
 #endif
-#endif
+
 
 #ifdef NTBUILD
 #define PERF_COUNTERS

lib/Runtime/Base/Chakra.Runtime.Base.vcxproj

@@ -67,9 +67,14 @@
     <ClCompile Include="$(MSBuildThisFileDirectory)RuntimeBasePch.cpp">
       <PrecompiledHeader>Create</PrecompiledHeader>
     </ClCompile>
+    <ClCompile Include="jitprofiling.cpp" />
+    <ClCompile Include="VTuneChakraProfile.cpp" />
   </ItemGroup>
   <ItemGroup>
     <ClInclude Include="CompactCounters.h" />
+    <ClInclude Include="ittnotify_config.h" />
+    <ClInclude Include="ittnotify_types.h" />
+    <ClInclude Include="jitprofiling.h" />
     <ClInclude Include="RuntimeBasePch.h" />
     <ClInclude Include="AuxPtrs.h" />
     <ClInclude Include="CallInfo.h" />
@@ -109,6 +114,7 @@
     <ClInclude Include="ThreadServiceWrapper.h" />
     <ClInclude Include="ThreadServiceWrapperBase.h" />
     <ClInclude Include="Utf8SourceInfo.h" />
+    <ClInclude Include="VTuneChakraProfile.h" />
     <ClInclude Include="WindowsFoundationAdapter.h" />
     <ClInclude Include="WindowsGlobalizationAdapter.h" />
   </ItemGroup>

lib/Runtime/Base/EtwTrace.cpp

@@ -8,19 +8,6 @@
 #include "Core/EtwTraceCore.h"
 #include "Base/EtwTrace.h"
 
-#ifdef VTUNE_PROFILING
-#ifdef CDECL
-#define ORIGINAL_CDECL CDECL
-#undef CDECL
-#endif
-// Not enabled in ChakraCore
-#include "jitProfiling.h"
-#ifdef ORIGINAL_CDECL
-#undef CDECL
-#endif
-#define CDECL ORIGINAL_CDECL
-#endif
-
 using namespace Js;
 
 //
@@ -28,9 +15,6 @@ using namespace Js;
 // by an ETW controller for the Jscript and MSHTML providers.
 //
 
-static const char LoopStr[] = "Loop";
-static const char16 LoopWStr[] = _u("Loop");
-
 void EtwCallbackApi::OnSessionChange(ULONG controlCode, PVOID callbackContext)
 {
     PMCGEN_TRACE_CONTEXT context = (PMCGEN_TRACE_CONTEXT)callbackContext;
@@ -64,11 +48,6 @@ void EtwCallbackApi::OnSessionChange(ULONG controlCode, PVOID callbackContext)
     }
 }
 
-#ifdef VTUNE_PROFILING
-const utf8char_t EtwTrace::DynamicCode[] = "Dynamic code";
-bool EtwTrace::isJitProfilingActive = false;
-#endif
-
 //
 // Registers the ETW provider - this is usually done on Jscript DLL load
 // After registration, we will receive callbacks when ETW tracing is enabled/disabled.
@@ -80,10 +59,6 @@ void EtwTrace::Register()
 #ifdef TEST_ETW_EVENTS
     TestEtwEventSink::Load();
 #endif
-
-#ifdef VTUNE_PROFILING
-    isJitProfilingActive = (iJIT_IsProfilingActive() == iJIT_SAMPLING_ON);
-#endif
 }
 
 //
@@ -96,13 +71,6 @@ void EtwTrace::UnRegister()
 #ifdef TEST_ETW_EVENTS
     TestEtwEventSink::Unload();
 #endif
-
-#ifdef VTUNE_PROFILING
-    if(isJitProfilingActive)
-    {
-        iJIT_NotifyEvent(iJVM_EVENT_TYPE_SHUTDOWN, NULL);
-    }
-#endif
 }
 
 //
@@ -296,119 +264,12 @@ void EtwTrace::LogMethodInterpreterThunkLoadEvent(FunctionBody* body)
 void EtwTrace::LogMethodNativeLoadEvent(FunctionBody* body, FunctionEntryPointInfo* entryPoint)
 {
     LogMethodNativeEvent(EventWriteMethodLoad, body, entryPoint);
-
-#ifdef VTUNE_PROFILING
-    if(isJitProfilingActive)
-    {
-        iJIT_Method_Load methodInfo;
-        memset(&methodInfo, 0, sizeof(iJIT_Method_Load));
-        const char16* methodName = body->GetExternalDisplayName();
-        // Append function line number info to method name so that VTune can distinguish between polymorphic methods
-        char16 methodNameBuffer[_MAX_PATH];
-        ULONG lineNumber = body->GetLineNumber();
-        char16 numberBuffer[20];
-        _ltow_s(lineNumber, numberBuffer, 10);
-        wcscpy_s(methodNameBuffer, methodName);
-        if(entryPoint->GetJitMode() == ExecutionMode::SimpleJit)
-        {
-            wcscat_s(methodNameBuffer, _u(" Simple"));
-        }
-        wcscat_s(methodNameBuffer, _u(" {line:"));
-        wcscat_s(methodNameBuffer, numberBuffer);
-        wcscat_s(methodNameBuffer, _u("}"));
-
-        size_t methodLength = wcslen(methodNameBuffer);
-        Assert(methodLength < _MAX_PATH);
-        size_t length = methodLength * 3 + 1;
-        utf8char_t* utf8MethodName = HeapNewNoThrowArray(utf8char_t, length);
-        if(utf8MethodName)
-        {
-            methodInfo.method_id = iJIT_GetNewMethodID();
-            utf8::EncodeIntoAndNullTerminate(utf8MethodName, methodNameBuffer, (charcount_t)methodLength);
-            methodInfo.method_name = (char*)utf8MethodName;
-            methodInfo.method_load_address = (void*)entryPoint->GetNativeAddress();
-            methodInfo.method_size = (uint)entryPoint->GetCodeSize();        // Size in memory - Must be exact
-
-            LineNumberInfo numberInfo[1];
-
-            uint lineCount = (entryPoint->GetNativeOffsetMapCount()) * 2 + 1; // may need to record both .begin and .end for all elements
-            LineNumberInfo* pLineInfo = HeapNewNoThrowArray(LineNumberInfo, lineCount);
-
-            if (pLineInfo == NULL || Js::Configuration::Global.flags.DisableVTuneSourceLineInfo)
-            {
-                // resort to original implementation, attribute all samples to first line
-                numberInfo[0].LineNumber = lineNumber;
-                numberInfo[0].Offset = 0;
-                methodInfo.line_number_size = 1;
-                methodInfo.line_number_table = numberInfo;
-            }
-            else
-            {
-                int size = entryPoint->PopulateLineInfo(pLineInfo, body);
-                methodInfo.line_number_size = size;
-                methodInfo.line_number_table = pLineInfo;
-            }
-
-            size_t urlLength  = 0;
-            utf8char_t* utf8Url = GetUrl(body, &urlLength);
-            methodInfo.source_file_name = (char*)utf8Url;
-            methodInfo.env = iJDE_JittingAPI;
-            OUTPUT_TRACE(Js::ProfilerPhase, _u("Method load event: %s\n"), methodNameBuffer);
-            iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED, &methodInfo);
-
-            HeapDeleteArray(lineCount, pLineInfo);
-
-            if(urlLength > 0)
-            {
-                HeapDeleteArray(urlLength, utf8Url);
-            }
-
-            HeapDeleteArray(length, utf8MethodName);
-        }
-    }
-#endif
 }
 
 void EtwTrace::LogLoopBodyLoadEvent(FunctionBody* body, LoopHeader* loopHeader, LoopEntryPointInfo* entryPoint, uint16 loopNumber)
 {
     Assert(loopNumber == body->GetLoopNumberWithLock(loopHeader));
     LogLoopBodyEventBG(EventWriteMethodLoad, body, loopHeader, entryPoint, loopNumber);
-
-#ifdef VTUNE_PROFILING
-    if(isJitProfilingActive)
-    {
-        iJIT_Method_Load methodInfo;
-        memset(&methodInfo, 0, sizeof(iJIT_Method_Load));
-        const char16* methodName = body->GetExternalDisplayName();
-        size_t methodLength = wcslen(methodName);
-        methodLength = min(methodLength, (size_t)UINT_MAX); // Just truncate if it is too big
-        size_t length = methodLength * 3 + /* spaces */ 2 + _countof(LoopStr) + /*size of loop number*/ 10 + /*NULL*/ 1;
-        utf8char_t* utf8MethodName = HeapNewNoThrowArray(utf8char_t, length);
-        if(utf8MethodName)
-        {
-            methodInfo.method_id = iJIT_GetNewMethodID();
-            size_t len = utf8::EncodeInto(utf8MethodName, methodName, (charcount_t)methodLength);
-            sprintf_s((char*)(utf8MethodName + len), length - len," %s %d", LoopStr, loopNumber + 1);
-            methodInfo.method_name = (char*)utf8MethodName;
-            methodInfo.method_load_address = (void*)entryPoint->GetNativeAddress();
-            methodInfo.method_size = (uint)entryPoint->GetCodeSize();        // Size in memory - Must be exact
-
-            size_t urlLength  = 0;
-            utf8char_t* utf8Url = GetUrl(body, &urlLength);
-            methodInfo.source_file_name = (char*)utf8Url;
-            methodInfo.env = iJDE_JittingAPI;
-
-            iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED, &methodInfo);
-            OUTPUT_TRACE(Js::ProfilerPhase, _u("Loop body load event: %s Loop %d\n"), methodName, loopNumber + 1);
-
-            if(urlLength > 0)
-            {
-                HeapDeleteArray(urlLength, utf8Url);
-            }
-            HeapDeleteArray(length, utf8MethodName);
-        }
-    }
-#endif
 }
 
 void EtwTrace::LogMethodInterpreterThunkUnloadEvent(FunctionBody* body)
@@ -490,32 +351,4 @@ size_t EtwTrace::GetSimpleJitFunctionName(
     return 0;
 }
 
-#ifdef VTUNE_PROFILING
-utf8char_t* EtwTrace::GetUrl( FunctionBody* body, size_t* urlBufferLength )
-{
-    utf8char_t* utf8Url = NULL;
-    if(!body->GetSourceContextInfo()->IsDynamic())
-    {
-        const wchar* url = body->GetSourceContextInfo()->url;
-        if(url)
-        {
-            size_t urlCharLength = wcslen(url);
-            urlCharLength = min(urlCharLength, (size_t)UINT_MAX);       // Just truncate if it is too big
-
-            *urlBufferLength = urlCharLength * 3 + 1;
-            utf8Url = HeapNewNoThrowArray(utf8char_t, *urlBufferLength);
-            if (utf8Url)
-            {
-                utf8::EncodeIntoAndNullTerminate(utf8Url, url, (charcount_t)urlCharLength);
-            }
-        }
-    }
-    else
-    {
-        utf8Url = (utf8char_t*)EtwTrace::DynamicCode;
-    }
-    return utf8Url;
-}
-#endif
-
 #endif

lib/Runtime/Base/EtwTrace.h

@@ -291,16 +291,6 @@ public:
     static size_t GetSimpleJitFunctionName(Js::FunctionBody *const body, _Out_writes_opt_z_(nameCharCapacity) char16 *const name, const size_t nameCharCapacity);
     static DWORD_PTR GetSourceId(Js::FunctionBody* body);
     static uint GetFunctionId(Js::FunctionProxy* body);
-
-#ifdef VTUNE_PROFILING
-    static const utf8char_t DynamicCode[];
-    static bool isJitProfilingActive;
-#endif
-
-private:
-#ifdef VTUNE_PROFILING
-    static utf8char_t* GetUrl(Js::FunctionBody* body, size_t* urlLength);
-#endif
 };
 
 #endif

lib/Runtime/Base/FunctionBody.cpp

@@ -12,6 +12,9 @@
 
 #include "ByteCode/ScopeInfo.h"
 #include "Base/EtwTrace.h"
+#ifdef VTUNE_PROFILING
+#include "Base/VTuneChakraProfile.h"
+#endif
 
 #ifdef DYNAMIC_PROFILE_MUTATOR
 #include "Language/DynamicProfileMutator.h"
@@ -3237,6 +3240,9 @@ namespace Js
         TraceExecutionMode();
 
         JS_ETW(EtwTrace::LogMethodNativeLoadEvent(this, entryPointInfo));
+#ifdef VTUNE_PROFILING
+        VTuneChakraProfile::LogMethodNativeLoadEvent(this, entryPointInfo);
+#endif
 
 #ifdef _M_ARM
         // For ARM we need to make sure that pipeline is synchronized with memory/cache for newly jitted code.
@@ -3295,6 +3301,9 @@ namespace Js
             loopHeader->interpretCount = entryPointInfo->GetFunctionBody()->GetLoopInterpretCount(loopHeader) - 1;
         }
         JS_ETW(EtwTrace::LogLoopBodyLoadEvent(this, loopHeader, ((LoopEntryPointInfo*) entryPointInfo), ((uint16)this->GetLoopNumberWithLock(loopHeader))));
+#ifdef VTUNE_PROFILING
+        VTuneChakraProfile::LogLoopBodyLoadEvent(this, loopHeader, ((LoopEntryPointInfo*)entryPointInfo), ((uint16)this->GetLoopNumberWithLock(loopHeader)));
+#endif
     }
 #endif
 
@@ -4431,16 +4440,7 @@ namespace Js
 #endif /* IR_VIEWER */
 
 #ifdef VTUNE_PROFILING
-#ifdef CDECL
-#define ORIGINAL_CDECL CDECL
-#undef CDECL
-#endif
-    // Not enabled in ChakraCore
-#include "jitProfiling.h"
-#ifdef ORIGINAL_CDECL
-#undef CDECL
-#endif
-#define CDECL ORIGINAL_CDECL
+#include "jitprofiling.h"
 
     int EntryPointInfo::GetNativeOffsetMapCount() const
     {

lib/Runtime/Base/VTuneChakraProfile.cpp

@@ -0,0 +1,178 @@
+//-------------------------------------------------------------------------------------------------------
+// Copyright (C) Microsoft. All rights reserved.
+// Licensed under the MIT license. See LICENSE.txt file in the project root for full license information.
+//-------------------------------------------------------------------------------------------------------
+#include "RuntimeBasePch.h"
+
+#ifdef VTUNE_PROFILING
+
+#include "VTuneChakraProfile.h"
+#include "jitprofiling.h"
+
+static const char LoopStr[] = "Loop";
+const utf8char_t VTuneChakraProfile::DynamicCode[] = "Dynamic code";
+bool VTuneChakraProfile::isJitProfilingActive = false;
+
+//
+// Registers the VTune profiler, if VTune sampling is running, we will get  
+// true value for isJitProfilingActive variable.
+//
+void VTuneChakraProfile::Register()
+{
+    isJitProfilingActive = (iJIT_IsProfilingActive() == iJIT_SAMPLING_ON);
+}
+
+//
+// Unregister and notify VTune that even sampling is done.
+//
+void VTuneChakraProfile::UnRegister()
+{
+    if(isJitProfilingActive)
+    {
+        iJIT_NotifyEvent(iJVM_EVENT_TYPE_SHUTDOWN, NULL);
+    }
+}
+
+//
+// Log JIT method native load event to VTune 
+//
+void VTuneChakraProfile::LogMethodNativeLoadEvent(Js::FunctionBody* body, Js::FunctionEntryPointInfo* entryPoint)
+{
+    if (isJitProfilingActive)
+    {
+        iJIT_Method_Load methodInfo;
+        memset(&methodInfo, 0, sizeof(iJIT_Method_Load));
+        const char16* methodName = body->GetExternalDisplayName();
+        // Append function line number info to method name so that VTune can distinguish between polymorphic methods
+        char16 methodNameBuffer[_MAX_PATH];
+        ULONG lineNumber = body->GetLineNumber();
+        char16 numberBuffer[20];
+        _ltow_s(lineNumber, numberBuffer, 10);
+        wcscpy_s(methodNameBuffer, methodName);
+        if (entryPoint->GetJitMode() == ExecutionMode::SimpleJit)
+        {
+            wcscat_s(methodNameBuffer, _u(" Simple"));
+        }
+        wcscat_s(methodNameBuffer, _u(" {line:"));
+        wcscat_s(methodNameBuffer, numberBuffer);
+        wcscat_s(methodNameBuffer, _u("}"));
+
+        size_t methodLength = wcslen(methodNameBuffer);
+        Assert(methodLength < _MAX_PATH);
+        size_t length = methodLength * 3 + 1;
+        utf8char_t* utf8MethodName = HeapNewNoThrowArray(utf8char_t, length);
+        if (utf8MethodName)
+        {
+            methodInfo.method_id = iJIT_GetNewMethodID();
+            utf8::EncodeIntoAndNullTerminate(utf8MethodName, methodNameBuffer, (charcount_t)methodLength);
+            methodInfo.method_name = (char*)utf8MethodName;
+            methodInfo.method_load_address = (void*)entryPoint->GetNativeAddress();
+            methodInfo.method_size = (uint)entryPoint->GetCodeSize();        // Size in memory - Must be exact
+
+            LineNumberInfo numberInfo[1];
+
+            uint lineCount = (entryPoint->GetNativeOffsetMapCount()) * 2 + 1; // may need to record both .begin and .end for all elements
+            LineNumberInfo* pLineInfo = HeapNewNoThrowArray(LineNumberInfo, lineCount);
+
+            if (pLineInfo == NULL || Js::Configuration::Global.flags.DisableVTuneSourceLineInfo)
+            {
+                // resort to original implementation, attribute all samples to first line
+                numberInfo[0].LineNumber = lineNumber;
+                numberInfo[0].Offset = 0;
+                methodInfo.line_number_size = 1;
+                methodInfo.line_number_table = numberInfo;
+            }
+            else
+            {
+                int size = entryPoint->PopulateLineInfo(pLineInfo, body);
+                methodInfo.line_number_size = size;
+                methodInfo.line_number_table = pLineInfo;
+            }
+
+            size_t urlLength = 0;
+            utf8char_t* utf8Url = GetUrl(body, &urlLength);
+            methodInfo.source_file_name = (char*)utf8Url;
+            OUTPUT_TRACE(Js::ProfilerPhase, _u("Method load event: %s\n"), methodNameBuffer);
+            iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED, &methodInfo);
+
+            HeapDeleteArray(lineCount, pLineInfo);
+
+            if (urlLength > 0)
+            {
+                HeapDeleteArray(urlLength, utf8Url);
+            }
+
+            HeapDeleteArray(length, utf8MethodName);
+        }
+    }
+}
+
+//
+// Log loop body load event to VTune 
+//
+void VTuneChakraProfile::LogLoopBodyLoadEvent(Js::FunctionBody* body, Js::LoopHeader* loopHeader, Js::LoopEntryPointInfo* entryPoint, uint16 loopNumber)
+{
+    if (isJitProfilingActive)
+    {
+        iJIT_Method_Load methodInfo;
+        memset(&methodInfo, 0, sizeof(iJIT_Method_Load));
+        const char16* methodName = body->GetExternalDisplayName();
+        size_t methodLength = wcslen(methodName);
+        methodLength = min(methodLength, (size_t)UINT_MAX); // Just truncate if it is too big
+        size_t length = methodLength * 3 + /* spaces */ 2 + _countof(LoopStr) + /*size of loop number*/ 10 + /*NULL*/ 1;
+        utf8char_t* utf8MethodName = HeapNewNoThrowArray(utf8char_t, length);
+        if(utf8MethodName)
+        {
+            methodInfo.method_id = iJIT_GetNewMethodID();
+            size_t len = utf8::EncodeInto(utf8MethodName, methodName, (charcount_t)methodLength);
+            sprintf_s((char*)(utf8MethodName + len), length - len," %s %d", LoopStr, loopNumber + 1);
+            methodInfo.method_name = (char*)utf8MethodName;
+            methodInfo.method_load_address = (void*)entryPoint->GetNativeAddress();
+            methodInfo.method_size = (uint)entryPoint->GetCodeSize();        // Size in memory - Must be exact
+
+            size_t urlLength  = 0;
+            utf8char_t* utf8Url = GetUrl(body, &urlLength);
+            methodInfo.source_file_name = (char*)utf8Url;
+
+            iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED, &methodInfo);
+            OUTPUT_TRACE(Js::ProfilerPhase, _u("Loop body load event: %s Loop %d\n"), methodName, loopNumber + 1);
+
+            if(urlLength > 0)
+            {
+                HeapDeleteArray(urlLength, utf8Url);
+            }
+            HeapDeleteArray(length, utf8MethodName);
+        }
+    }
+}
+
+//
+// Get URL from source context, called by LogMethodNativeLoadEvent and LogLoopBodyLoadEvent
+//
+utf8char_t* VTuneChakraProfile::GetUrl(Js::FunctionBody* body, size_t* urlBufferLength )
+{
+    utf8char_t* utf8Url = NULL;
+    if (!body->GetSourceContextInfo()->IsDynamic())
+    {
+        const wchar* url = body->GetSourceContextInfo()->url;
+        if (url)
+        {
+            size_t urlCharLength = wcslen(url);
+            urlCharLength = min(urlCharLength, (size_t)UINT_MAX);       // Just truncate if it is too big
+
+            *urlBufferLength = urlCharLength * 3 + 1;
+            utf8Url = HeapNewNoThrowArray(utf8char_t, *urlBufferLength);
+            if (utf8Url)
+            {
+                utf8::EncodeIntoAndNullTerminate(utf8Url, url, (charcount_t)urlCharLength);
+            }
+        }
+    }
+    else
+    {
+        utf8Url = (utf8char_t*)VTuneChakraProfile::DynamicCode;
+    }
+    return utf8Url;
+}
+
+#endif /* VTUNE_PROFILING */

lib/Runtime/Base/VTuneChakraProfile.h

@@ -0,0 +1,28 @@
+//-------------------------------------------------------------------------------------------------------
+// Copyright (C) Microsoft. All rights reserved.
+// Licensed under the MIT license. See LICENSE.txt file in the project root for full license information.
+//-------------------------------------------------------------------------------------------------------
+#pragma once
+
+#ifdef VTUNE_PROFILING
+
+//
+// Encapsulates all VTune Chakra profiling event logging and registration etc..
+//
+class VTuneChakraProfile
+{
+public:
+    static void Register();
+    static void UnRegister();
+
+	static void LogMethodNativeLoadEvent(Js::FunctionBody* body, Js::FunctionEntryPointInfo* entryPoint);
+    static void LogLoopBodyLoadEvent(Js::FunctionBody* body, Js::LoopHeader* loopHeader, Js::LoopEntryPointInfo* entryPoint, uint16 loopNumber);
+
+    static const utf8char_t DynamicCode[];
+    static bool isJitProfilingActive;
+
+private:
+    static utf8char_t* GetUrl(Js::FunctionBody* body, size_t* urlLength);
+};
+
+#endif

lib/Runtime/Base/ittnotify_config.h

@@ -0,0 +1,545 @@
+//-------------------------------------------------------------------------------------------------------
+// Copyright (C) Microsoft Corporation and contributors. All rights reserved.
+// Licensed under the MIT license. See LICENSE.txt file in the project root for full license information.
+//-------------------------------------------------------------------------------------------------------
+#ifndef _ITTNOTIFY_CONFIG_H_
+#define _ITTNOTIFY_CONFIG_H_
+
+/** @cond exclude_from_documentation */
+#ifndef ITT_OS_WIN
+#  define ITT_OS_WIN   1
+#endif /* ITT_OS_WIN */
+
+#ifndef ITT_OS_LINUX
+#  define ITT_OS_LINUX 2
+#endif /* ITT_OS_LINUX */
+
+#ifndef ITT_OS_MAC
+#  define ITT_OS_MAC   3
+#endif /* ITT_OS_MAC */
+
+#ifndef ITT_OS_FREEBSD
+#  define ITT_OS_FREEBSD   4
+#endif /* ITT_OS_FREEBSD */
+
+#ifndef ITT_OS
+#  if defined WIN32 || defined _WIN32
+#    define ITT_OS ITT_OS_WIN
+#  elif defined( __APPLE__ ) && defined( __MACH__ )
+#    define ITT_OS ITT_OS_MAC
+#  elif defined( __FreeBSD__ )
+#    define ITT_OS ITT_OS_FREEBSD
+#  else
+#    define ITT_OS ITT_OS_LINUX
+#  endif
+#endif /* ITT_OS */
+
+#ifndef ITT_PLATFORM_WIN
+#  define ITT_PLATFORM_WIN 1
+#endif /* ITT_PLATFORM_WIN */
+
+#ifndef ITT_PLATFORM_POSIX
+#  define ITT_PLATFORM_POSIX 2
+#endif /* ITT_PLATFORM_POSIX */
+
+#ifndef ITT_PLATFORM_MAC
+#  define ITT_PLATFORM_MAC 3
+#endif /* ITT_PLATFORM_MAC */
+
+#ifndef ITT_PLATFORM_FREEBSD
+#  define ITT_PLATFORM_FREEBSD 4
+#endif /* ITT_PLATFORM_FREEBSD */
+
+#ifndef ITT_PLATFORM
+#  if ITT_OS==ITT_OS_WIN
+#    define ITT_PLATFORM ITT_PLATFORM_WIN
+#  elif ITT_OS==ITT_OS_MAC
+#    define ITT_PLATFORM ITT_PLATFORM_MAC
+#  elif ITT_OS==ITT_OS_FREEBSD
+#    define ITT_PLATFORM ITT_PLATFORM_FREEBSD
+#  else
+#    define ITT_PLATFORM ITT_PLATFORM_POSIX
+#  endif
+#endif /* ITT_PLATFORM */
+
+#if defined(_UNICODE) && !defined(UNICODE)
+#define UNICODE
+#endif
+
+#include <stddef.h>
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+// 4/29/2016 disable warning 4995 on deprecated _tccpy, wcsncpy, wcsncat, wcstok funcs in tchar.h
+#pragma warning( disable : 4995 )   
+#include <tchar.h>
+#else  /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+#include <stdint.h>
+#if defined(UNICODE) || defined(_UNICODE)
+#include <wchar.h>
+#endif /* UNICODE || _UNICODE */
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+
+#ifndef ITTAPI_CDECL
+#  if ITT_PLATFORM==ITT_PLATFORM_WIN
+#    define ITTAPI_CDECL __cdecl
+#  else /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+#    if defined _M_IX86 || defined __i386__
+#      define ITTAPI_CDECL __attribute__ ((cdecl))
+#    else  /* _M_IX86 || __i386__ */
+#      define ITTAPI_CDECL /* actual only on x86 platform */
+#    endif /* _M_IX86 || __i386__ */
+#  endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+#endif /* ITTAPI_CDECL */
+
+#ifndef STDCALL
+#  if ITT_PLATFORM==ITT_PLATFORM_WIN
+#    define STDCALL __stdcall
+#  else /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+#    if defined _M_IX86 || defined __i386__
+#      define STDCALL __attribute__ ((stdcall))
+#    else  /* _M_IX86 || __i386__ */
+#      define STDCALL /* supported only on x86 platform */
+#    endif /* _M_IX86 || __i386__ */
+#  endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+#endif /* STDCALL */
+
+#define ITTAPI    ITTAPI_CDECL
+#define LIBITTAPI ITTAPI_CDECL
+
+/* TODO: Temporary for compatibility! */
+#define ITTAPI_CALL    ITTAPI_CDECL
+#define LIBITTAPI_CALL ITTAPI_CDECL
+
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+/* use __forceinline (VC++ specific) */
+#define ITT_INLINE           __forceinline
+#define ITT_INLINE_ATTRIBUTE /* nothing */
+#else  /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+/*
+ * Generally, functions are not inlined unless optimization is specified.
+ * For functions declared inline, this attribute inlines the function even
+ * if no optimization level was specified.
+ */
+#ifdef __STRICT_ANSI__
+#define ITT_INLINE           static
+#define ITT_INLINE_ATTRIBUTE __attribute__((unused))
+#else  /* __STRICT_ANSI__ */
+#define ITT_INLINE           static inline
+#define ITT_INLINE_ATTRIBUTE __attribute__((always_inline, unused))
+#endif /* __STRICT_ANSI__ */
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+/** @endcond */
+
+#ifndef ITT_ARCH_IA32
+#  define ITT_ARCH_IA32  1
+#endif /* ITT_ARCH_IA32 */
+
+#ifndef ITT_ARCH_IA32E
+#  define ITT_ARCH_IA32E 2
+#endif /* ITT_ARCH_IA32E */
+
+#ifndef ITT_ARCH_ARM
+#  define ITT_ARCH_ARM  4
+#endif /* ITT_ARCH_ARM */
+
+#ifndef ITT_ARCH_PPC64
+#  define ITT_ARCH_PPC64  5
+#endif /* ITT_ARCH_PPC64 */
+
+#ifndef ITT_ARCH
+#  if defined _M_IX86 || defined __i386__
+#    define ITT_ARCH ITT_ARCH_IA32
+#  elif defined _M_X64 || defined _M_AMD64 || defined __x86_64__
+#    define ITT_ARCH ITT_ARCH_IA32E
+#  elif defined _M_IA64 || defined __ia64__
+#    define ITT_ARCH ITT_ARCH_IA64
+#  elif defined _M_ARM || defined __arm__
+#    define ITT_ARCH ITT_ARCH_ARM
+#  elif defined __powerpc64__
+#    define ITT_ARCH ITT_ARCH_PPC64
+#  endif
+#endif
+
+#ifdef __cplusplus
+#  define ITT_EXTERN_C extern "C"
+#  define ITT_EXTERN_C_BEGIN extern "C" {
+#  define ITT_EXTERN_C_END }
+#else
+#  define ITT_EXTERN_C /* nothing */
+#  define ITT_EXTERN_C_BEGIN /* nothing */
+#  define ITT_EXTERN_C_END /* nothing */
+#endif /* __cplusplus */
+
+#define ITT_TO_STR_AUX(x) #x
+#define ITT_TO_STR(x)     ITT_TO_STR_AUX(x)
+
+#define __ITT_BUILD_ASSERT(expr, suffix) do { \
+    static char __itt_build_check_##suffix[(expr) ? 1 : -1]; \
+    __itt_build_check_##suffix[0] = 0; \
+} while(0)
+#define _ITT_BUILD_ASSERT(expr, suffix)  __ITT_BUILD_ASSERT((expr), suffix)
+#define ITT_BUILD_ASSERT(expr)           _ITT_BUILD_ASSERT((expr), __LINE__)
+
+#define ITT_MAGIC { 0xED, 0xAB, 0xAB, 0xEC, 0x0D, 0xEE, 0xDA, 0x30 }
+
+/* Replace with snapshot date YYYYMMDD for promotion build. */
+#define API_VERSION_BUILD    20151119
+
+#ifndef API_VERSION_NUM
+#define API_VERSION_NUM 0.0.0
+#endif /* API_VERSION_NUM */
+
+#define API_VERSION "ITT-API-Version " ITT_TO_STR(API_VERSION_NUM) \
+                                " (" ITT_TO_STR(API_VERSION_BUILD) ")"
+
+/* OS communication functions */
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+#include <windows.h>
+typedef HMODULE           lib_t;
+typedef DWORD             TIDT;
+typedef CRITICAL_SECTION  mutex_t;
+#define MUTEX_INITIALIZER { 0 }
+#define strong_alias(name, aliasname) /* empty for Windows */
+#else  /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+#include <dlfcn.h>
+#if defined(UNICODE) || defined(_UNICODE)
+#include <wchar.h>
+#endif /* UNICODE */
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE 1 /* need for PTHREAD_MUTEX_RECURSIVE */
+#endif /* _GNU_SOURCE */
+#ifndef __USE_UNIX98
+#define __USE_UNIX98 1 /* need for PTHREAD_MUTEX_RECURSIVE, on SLES11.1 with gcc 4.3.4 wherein pthread.h missing dependency on __USE_XOPEN2K8 */
+#endif /*__USE_UNIX98*/
+#include <pthread.h>
+typedef void*             lib_t;
+typedef pthread_t         TIDT;
+typedef pthread_mutex_t   mutex_t;
+#define MUTEX_INITIALIZER PTHREAD_MUTEX_INITIALIZER
+#define _strong_alias(name, aliasname) \
+            extern __typeof (name) aliasname __attribute__ ((alias (#name)));
+#define strong_alias(name, aliasname) _strong_alias(name, aliasname)
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+#define __itt_get_proc(lib, name) GetProcAddress(lib, name)
+#define __itt_mutex_init(mutex)   InitializeCriticalSection(mutex)
+#define __itt_mutex_lock(mutex)   EnterCriticalSection(mutex)
+#define __itt_mutex_unlock(mutex) LeaveCriticalSection(mutex)
+#define __itt_load_lib(name)      LoadLibraryA(name)
+#define __itt_unload_lib(handle)  FreeLibrary(handle)
+#define __itt_system_error()      (int)GetLastError()
+#define __itt_fstrcmp(s1, s2)     lstrcmpA(s1, s2)
+#define __itt_fstrnlen(s, l)      strnlen_s(s, l)
+#define __itt_fstrcpyn(s1, b, s2, l) strncpy_s(s1, b, s2, l)
+#define __itt_fstrdup(s)          _strdup(s)
+#define __itt_thread_id()         GetCurrentThreadId()
+#define __itt_thread_yield()      SwitchToThread()
+#ifndef ITT_SIMPLE_INIT
+ITT_INLINE long
+__itt_interlocked_increment(volatile long* ptr) ITT_INLINE_ATTRIBUTE;
+ITT_INLINE long __itt_interlocked_increment(volatile long* ptr)
+{
+    return InterlockedIncrement(ptr);
+}
+#endif /* ITT_SIMPLE_INIT */
+#else /* ITT_PLATFORM!=ITT_PLATFORM_WIN */
+#define __itt_get_proc(lib, name) dlsym(lib, name)
+#define __itt_mutex_init(mutex)   {\
+    pthread_mutexattr_t mutex_attr;                                         \
+    int error_code = pthread_mutexattr_init(&mutex_attr);                   \
+    if (error_code)                                                         \
+        __itt_report_error(__itt_error_system, "pthread_mutexattr_init",    \
+                           error_code);                                     \
+    error_code = pthread_mutexattr_settype(&mutex_attr,                     \
+                                           PTHREAD_MUTEX_RECURSIVE);        \
+    if (error_code)                                                         \
+        __itt_report_error(__itt_error_system, "pthread_mutexattr_settype", \
+                           error_code);                                     \
+    error_code = pthread_mutex_init(mutex, &mutex_attr);                    \
+    if (error_code)                                                         \
+        __itt_report_error(__itt_error_system, "pthread_mutex_init",        \
+                           error_code);                                     \
+    error_code = pthread_mutexattr_destroy(&mutex_attr);                    \
+    if (error_code)                                                         \
+        __itt_report_error(__itt_error_system, "pthread_mutexattr_destroy", \
+                           error_code);                                     \
+}
+#define __itt_mutex_lock(mutex)   pthread_mutex_lock(mutex)
+#define __itt_mutex_unlock(mutex) pthread_mutex_unlock(mutex)
+#define __itt_load_lib(name)      dlopen(name, RTLD_LAZY)
+#define __itt_unload_lib(handle)  dlclose(handle)
+#define __itt_system_error()      errno
+#define __itt_fstrcmp(s1, s2)     strcmp(s1, s2)
+
+/* makes customer code define safe APIs for SDL_STRNLEN_S and SDL_STRNCPY_S */
+#ifdef SDL_STRNLEN_S
+#define __itt_fstrnlen(s, l)      SDL_STRNLEN_S(s, l)
+#else
+#define __itt_fstrnlen(s, l)      strlen(s)
+#endif /* SDL_STRNLEN_S */
+#ifdef SDL_STRNCPY_S
+#define __itt_fstrcpyn(s1, b, s2, l) SDL_STRNCPY_S(s1, b, s2, l)
+#else
+#define __itt_fstrcpyn(s1, b, s2, l) strncpy(s1, s2, l)
+#endif /* SDL_STRNCPY_S */
+
+#define __itt_fstrdup(s)          strdup(s)
+#define __itt_thread_id()         pthread_self()
+#define __itt_thread_yield()      sched_yield()
+#if ITT_ARCH==ITT_ARCH_IA64
+#ifdef __INTEL_COMPILER
+#define __TBB_machine_fetchadd4(addr, val) __fetchadd4_acq((void *)addr, val)
+#else  /* __INTEL_COMPILER */
+/* TODO: Add Support for not Intel compilers for IA-64 architecture */
+#endif /* __INTEL_COMPILER */
+#elif ITT_ARCH==ITT_ARCH_IA32 || ITT_ARCH==ITT_ARCH_IA32E /* ITT_ARCH!=ITT_ARCH_IA64 */
+ITT_INLINE long
+__TBB_machine_fetchadd4(volatile void* ptr, long addend) ITT_INLINE_ATTRIBUTE;
+ITT_INLINE long __TBB_machine_fetchadd4(volatile void* ptr, long addend)
+{
+    long result;
+    __asm__ __volatile__("lock\nxadd %0,%1"
+                          : "=r"(result),"=m"(*(int*)ptr)
+                          : "0"(addend), "m"(*(int*)ptr)
+                          : "memory");
+    return result;
+}
+#elif ITT_ARCH==ITT_ARCH_ARM || ITT_ARCH==ITT_ARCH_PPC64
+#define __TBB_machine_fetchadd4(addr, val) __sync_fetch_and_add(addr, val)
+#endif /* ITT_ARCH==ITT_ARCH_IA64 */
+#ifndef ITT_SIMPLE_INIT
+ITT_INLINE long
+__itt_interlocked_increment(volatile long* ptr) ITT_INLINE_ATTRIBUTE;
+ITT_INLINE long __itt_interlocked_increment(volatile long* ptr)
+{
+    return __TBB_machine_fetchadd4(ptr, 1) + 1L;
+}
+#endif /* ITT_SIMPLE_INIT */
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+
+typedef enum {
+    __itt_collection_normal = 0,
+    __itt_collection_paused = 1
+} __itt_collection_state;
+
+typedef enum {
+    __itt_thread_normal  = 0,
+    __itt_thread_ignored = 1
+} __itt_thread_state;
+
+#pragma pack(push, 8)
+
+typedef struct ___itt_thread_info
+{
+    const char* nameA; /*!< Copy of original name in ASCII. */
+#if defined(UNICODE) || defined(_UNICODE)
+    const wchar_t* nameW; /*!< Copy of original name in UNICODE. */
+#else  /* UNICODE || _UNICODE */
+    void* nameW;
+#endif /* UNICODE || _UNICODE */
+    TIDT               tid;
+    __itt_thread_state state;   /*!< Thread state (paused or normal) */
+    int                extra1;  /*!< Reserved to the runtime */
+    void*              extra2;  /*!< Reserved to the runtime */
+    struct ___itt_thread_info* next;
+} __itt_thread_info;
+
+#include "ittnotify_types.h" /* For __itt_group_id definition */
+
+typedef struct ___itt_api_info_20101001
+{
+    const char*    name;
+    void**         func_ptr;
+    void*          init_func;
+    __itt_group_id group;
+}  __itt_api_info_20101001;
+
+typedef struct ___itt_api_info
+{
+    const char*    name;
+    void**         func_ptr;
+    void*          init_func;
+    void*          null_func;
+    __itt_group_id group;
+}  __itt_api_info;
+
+typedef struct __itt_counter_info
+{
+    const char* nameA;  /*!< Copy of original name in ASCII. */
+#if defined(UNICODE) || defined(_UNICODE)
+    const wchar_t* nameW; /*!< Copy of original name in UNICODE. */
+#else  /* UNICODE || _UNICODE */
+    void* nameW;
+#endif /* UNICODE || _UNICODE */
+    const char* domainA;  /*!< Copy of original name in ASCII. */
+#if defined(UNICODE) || defined(_UNICODE)
+    const wchar_t* domainW; /*!< Copy of original name in UNICODE. */
+#else  /* UNICODE || _UNICODE */
+    void* domainW;
+#endif /* UNICODE || _UNICODE */
+    int type;
+    long index;
+    int   extra1; /*!< Reserved to the runtime */
+    void* extra2; /*!< Reserved to the runtime */
+    struct __itt_counter_info* next;
+}  __itt_counter_info_t;
+
+struct ___itt_domain;
+struct ___itt_string_handle;
+
+typedef struct ___itt_global
+{
+    unsigned char          magic[8];
+    unsigned long          version_major;
+    unsigned long          version_minor;
+    unsigned long          version_build;
+    volatile long          api_initialized;
+    volatile long          mutex_initialized;
+    volatile long          atomic_counter;
+    mutex_t                mutex;
+    lib_t                  lib;
+    void*                  error_handler;
+    const char**           dll_path_ptr;
+    __itt_api_info*        api_list_ptr;
+    struct ___itt_global*  next;
+    /* Joinable structures below */
+    __itt_thread_info*     thread_list;
+    struct ___itt_domain*  domain_list;
+    struct ___itt_string_handle* string_list;
+    __itt_collection_state state;
+    __itt_counter_info_t* counter_list;
+} __itt_global;
+
+#pragma pack(pop)
+
+#define NEW_THREAD_INFO_W(gptr,h,h_tail,t,s,n) { \
+    h = (__itt_thread_info*)malloc(sizeof(__itt_thread_info)); \
+    if (h != NULL) { \
+        h->tid    = t; \
+        h->nameA  = NULL; \
+        h->nameW  = n ? _wcsdup(n) : NULL; \
+        h->state  = s; \
+        h->extra1 = 0;    /* reserved */ \
+        h->extra2 = NULL; /* reserved */ \
+        h->next   = NULL; \
+        if (h_tail == NULL) \
+            (gptr)->thread_list = h; \
+        else \
+            h_tail->next = h; \
+    } \
+}
+
+#define NEW_THREAD_INFO_A(gptr,h,h_tail,t,s,n) { \
+    h = (__itt_thread_info*)malloc(sizeof(__itt_thread_info)); \
+    if (h != NULL) { \
+        h->tid    = t; \
+        h->nameA  = n ? __itt_fstrdup(n) : NULL; \
+        h->nameW  = NULL; \
+        h->state  = s; \
+        h->extra1 = 0;    /* reserved */ \
+        h->extra2 = NULL; /* reserved */ \
+        h->next   = NULL; \
+        if (h_tail == NULL) \
+            (gptr)->thread_list = h; \
+        else \
+            h_tail->next = h; \
+    } \
+}
+
+#define NEW_DOMAIN_W(gptr,h,h_tail,name) { \
+    h = (__itt_domain*)malloc(sizeof(__itt_domain)); \
+    if (h != NULL) { \
+        h->flags  = 1;    /* domain is enabled by default */ \
+        h->nameA  = NULL; \
+        h->nameW  = name ? _wcsdup(name) : NULL; \
+        h->extra1 = 0;    /* reserved */ \
+        h->extra2 = NULL; /* reserved */ \
+        h->next   = NULL; \
+        if (h_tail == NULL) \
+            (gptr)->domain_list = h; \
+        else \
+            h_tail->next = h; \
+    } \
+}
+
+#define NEW_DOMAIN_A(gptr,h,h_tail,name) { \
+    h = (__itt_domain*)malloc(sizeof(__itt_domain)); \
+    if (h != NULL) { \
+        h->flags  = 1;    /* domain is enabled by default */ \
+        h->nameA  = name ? __itt_fstrdup(name) : NULL; \
+        h->nameW  = NULL; \
+        h->extra1 = 0;    /* reserved */ \
+        h->extra2 = NULL; /* reserved */ \
+        h->next   = NULL; \
+        if (h_tail == NULL) \
+            (gptr)->domain_list = h; \
+        else \
+            h_tail->next = h; \
+    } \
+}
+
+#define NEW_STRING_HANDLE_W(gptr,h,h_tail,name) { \
+    h = (__itt_string_handle*)malloc(sizeof(__itt_string_handle)); \
+    if (h != NULL) { \
+        h->strA   = NULL; \
+        h->strW   = name ? _wcsdup(name) : NULL; \
+        h->extra1 = 0;    /* reserved */ \
+        h->extra2 = NULL; /* reserved */ \
+        h->next   = NULL; \
+        if (h_tail == NULL) \
+            (gptr)->string_list = h; \
+        else \
+            h_tail->next = h; \
+    } \
+}
+
+#define NEW_STRING_HANDLE_A(gptr,h,h_tail,name) { \
+    h = (__itt_string_handle*)malloc(sizeof(__itt_string_handle)); \
+    if (h != NULL) { \
+        h->strA   = name ? __itt_fstrdup(name) : NULL; \
+        h->strW   = NULL; \
+        h->extra1 = 0;    /* reserved */ \
+        h->extra2 = NULL; /* reserved */ \
+        h->next   = NULL; \
+        if (h_tail == NULL) \
+            (gptr)->string_list = h; \
+        else \
+            h_tail->next = h; \
+    } \
+}
+
+#define NEW_COUNTER_W(gptr,h,h_tail,name,domain,type) { \
+    h = (__itt_counter_info_t*)malloc(sizeof(__itt_counter_info_t)); \
+    if (h != NULL) { \
+        h->nameA   = NULL; \
+        h->nameW   = name ? _wcsdup(name) : NULL; \
+        h->domainA   = NULL; \
+        h->domainW   = name ? _wcsdup(domain) : NULL; \
+        h->type = type; \
+        h->index = 0; \
+        h->next   = NULL; \
+        if (h_tail == NULL) \
+            (gptr)->counter_list = h; \
+        else \
+            h_tail->next = h; \
+    } \
+}
+
+#define NEW_COUNTER_A(gptr,h,h_tail,name,domain,type) { \
+    h = (__itt_counter_info_t*)malloc(sizeof(__itt_counter_info_t)); \
+    if (h != NULL) { \
+        h->nameA   = name ? __itt_fstrdup(name) : NULL; \
+        h->nameW   = NULL; \
+        h->domainA   = domain ? __itt_fstrdup(domain) : NULL; \
+        h->domainW   = NULL; \
+        h->type = type; \
+        h->index = 0; \
+        h->next   = NULL; \
+        if (h_tail == NULL) \
+            (gptr)->counter_list = h; \
+        else \
+            h_tail->next = h; \
+    } \
+}
+
+#endif /* _ITTNOTIFY_CONFIG_H_ */

lib/Runtime/Base/ittnotify_types.h

@@ -0,0 +1,62 @@
+//-------------------------------------------------------------------------------------------------------
+// Copyright (C) Microsoft Corporation and contributors. All rights reserved.
+// Licensed under the MIT license. See LICENSE.txt file in the project root for full license information.
+//-------------------------------------------------------------------------------------------------------
+
+#ifndef _ITTNOTIFY_TYPES_H_
+#define _ITTNOTIFY_TYPES_H_
+
+typedef enum ___itt_group_id
+{
+    __itt_group_none      = 0,
+    __itt_group_legacy    = 1<<0,
+    __itt_group_control   = 1<<1,
+    __itt_group_thread    = 1<<2,
+    __itt_group_mark      = 1<<3,
+    __itt_group_sync      = 1<<4,
+    __itt_group_fsync     = 1<<5,
+    __itt_group_jit       = 1<<6,
+    __itt_group_model     = 1<<7,
+    __itt_group_splitter_min = 1<<7,
+    __itt_group_counter   = 1<<8,
+    __itt_group_frame     = 1<<9,
+    __itt_group_stitch    = 1<<10,
+    __itt_group_heap      = 1<<11,
+    __itt_group_splitter_max = 1<<12,
+    __itt_group_structure = 1<<12,
+    __itt_group_suppress = 1<<13,
+    __itt_group_arrays    = 1<<14,
+    __itt_group_all       = -1
+} __itt_group_id;
+
+#pragma pack(push, 8)
+
+typedef struct ___itt_group_list
+{
+    __itt_group_id id;
+    const char*    name;
+} __itt_group_list;
+
+#pragma pack(pop)
+
+#define ITT_GROUP_LIST(varname) \
+    static __itt_group_list varname[] = {       \
+        { __itt_group_all,       "all"       }, \
+        { __itt_group_control,   "control"   }, \
+        { __itt_group_thread,    "thread"    }, \
+        { __itt_group_mark,      "mark"      }, \
+        { __itt_group_sync,      "sync"      }, \
+        { __itt_group_fsync,     "fsync"     }, \
+        { __itt_group_jit,       "jit"       }, \
+        { __itt_group_model,     "model"     }, \
+        { __itt_group_counter,   "counter"   }, \
+        { __itt_group_frame,     "frame"     }, \
+        { __itt_group_stitch,    "stitch"    }, \
+        { __itt_group_heap,      "heap"      }, \
+        { __itt_group_structure, "structure" }, \
+        { __itt_group_suppress,  "suppress"  }, \
+        { __itt_group_arrays,    "arrays"    }, \
+        { __itt_group_none,      NULL        }  \
+    }
+
+#endif /* _ITTNOTIFY_TYPES_H_ */

lib/Runtime/Base/jitprofiling.cpp

@@ -0,0 +1,269 @@
+//-------------------------------------------------------------------------------------------------------
+// Copyright (C) Microsoft Corporation and contributors. All rights reserved.
+// Licensed under the MIT license. See LICENSE.txt file in the project root for full license information.
+//-------------------------------------------------------------------------------------------------------
+#include "RuntimeBasePch.h"
+
+#ifdef VTUNE_PROFILING
+
+#include "ittnotify_config.h"
+
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+#include <windows.h>
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+#if ITT_PLATFORM != ITT_PLATFORM_MAC && ITT_PLATFORM != ITT_PLATFORM_FREEBSD
+#include <malloc.h>
+#endif
+#include <stdlib.h>
+
+#include "jitprofiling.h"
+
+static const char rcsid[] = "\n@(#) $Revision: 463960 $\n";
+
+#define DLL_ENVIRONMENT_VAR             "VS_PROFILER"
+
+#ifndef NEW_DLL_ENVIRONMENT_VAR
+#if ITT_ARCH==ITT_ARCH_IA32
+#define NEW_DLL_ENVIRONMENT_VAR	        "INTEL_JIT_PROFILER32"
+#else
+#define NEW_DLL_ENVIRONMENT_VAR	        "INTEL_JIT_PROFILER64"
+#endif
+#endif /* NEW_DLL_ENVIRONMENT_VAR */
+
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+#define DEFAULT_DLLNAME                 "JitPI.dll"
+HINSTANCE m_libHandle = NULL;
+#elif ITT_PLATFORM==ITT_PLATFORM_MAC
+#define DEFAULT_DLLNAME                 "libJitPI.dylib"
+void* m_libHandle = NULL;
+#else
+#define DEFAULT_DLLNAME                 "libJitPI.so"
+void* m_libHandle = NULL;
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+
+/* default location of JIT profiling agent on Android */
+#define ANDROID_JIT_AGENT_PATH  "/data/intel/libittnotify.so"
+
+/* the function pointers */
+typedef unsigned int(JITAPI *TPInitialize)(void);
+static TPInitialize FUNC_Initialize=NULL;
+
+typedef unsigned int(JITAPI *TPNotify)(unsigned int, void*);
+static TPNotify FUNC_NotifyEvent=NULL;
+
+static iJIT_IsProfilingActiveFlags executionMode = iJIT_NOTHING_RUNNING;
+
+/* end collector dll part. */
+
+/* loadiJIT_Funcs() : this function is called just in the beginning
+ * and is responsible to load the functions from BistroJavaCollector.dll
+ * result:
+ *  on success: the functions loads, iJIT_DLL_is_missing=0, return value = 1
+ *  on failure: the functions are NULL, iJIT_DLL_is_missing=1, return value = 0
+ */
+static int loadiJIT_Funcs(void);
+
+/* global representing whether the collector can't be loaded */
+static int iJIT_DLL_is_missing = 0;
+
+ITT_EXTERN_C int JITAPI
+iJIT_NotifyEvent(iJIT_JVM_EVENT event_type, void *EventSpecificData)
+{
+    int ReturnValue = 0;
+
+    /* initialization part - the collector has not been loaded yet. */
+    if (!FUNC_NotifyEvent)
+    {
+        if (iJIT_DLL_is_missing)
+            return 0;
+
+        if (!loadiJIT_Funcs())
+            return 0;
+    }
+
+    if (event_type == iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED ||
+        event_type == iJVM_EVENT_TYPE_METHOD_UPDATE)
+    {
+        if (((piJIT_Method_Load)EventSpecificData)->method_id == 0)
+            return 0;
+    }
+    else if (event_type == iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED_V2)
+    {
+        if (((piJIT_Method_Load_V2)EventSpecificData)->method_id == 0)
+            return 0;
+    }
+    else if (event_type == iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED_V3)
+    {
+        if (((piJIT_Method_Load_V3)EventSpecificData)->method_id == 0)
+            return 0;
+    }
+    else if (event_type == iJVM_EVENT_TYPE_METHOD_INLINE_LOAD_FINISHED)
+    {
+        if (((piJIT_Method_Inline_Load)EventSpecificData)->method_id == 0 ||
+            ((piJIT_Method_Inline_Load)EventSpecificData)->parent_method_id == 0)
+            return 0;
+    }
+
+    if (FUNC_NotifyEvent)
+    {
+        ReturnValue = (int)FUNC_NotifyEvent(event_type, EventSpecificData);
+    }
+
+    return ReturnValue;
+}
+
+ITT_EXTERN_C iJIT_IsProfilingActiveFlags JITAPI iJIT_IsProfilingActive()
+{
+    if (!iJIT_DLL_is_missing)
+    {
+        loadiJIT_Funcs();
+    }
+
+    return executionMode;
+}
+
+/* This function loads the collector dll and the relevant functions.
+ * on success: all functions load,     iJIT_DLL_is_missing = 0, return value = 1
+ * on failure: all functions are NULL, iJIT_DLL_is_missing = 1, return value = 0
+ */
+static int loadiJIT_Funcs()
+{
+    static int bDllWasLoaded = 0;
+    char *dllName = (char*)rcsid; /* !! Just to avoid unused code elimination */
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+    DWORD dNameLength = 0;
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+
+    if(bDllWasLoaded)
+    {
+        /* dll was already loaded, no need to do it for the second time */
+        return 1;
+    }
+
+    /* Assumes that the DLL will not be found */
+    iJIT_DLL_is_missing = 1;
+    FUNC_NotifyEvent = NULL;
+
+    if (m_libHandle)
+    {
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+        FreeLibrary(m_libHandle);
+#else  /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+        dlclose(m_libHandle);
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+        m_libHandle = NULL;
+    }
+
+    /* Try to get the dll name from the environment */
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+    dNameLength = GetEnvironmentVariableA(NEW_DLL_ENVIRONMENT_VAR, NULL, 0);
+    if (dNameLength)
+    {
+        DWORD envret = 0;
+        dllName = (char*)malloc(sizeof(char) * (dNameLength + 1));
+        if(dllName != NULL)
+        {
+            envret = GetEnvironmentVariableA(NEW_DLL_ENVIRONMENT_VAR, 
+                                             dllName, dNameLength);
+            if (envret)
+            {
+                /* Try to load the dll from the PATH... */
+                m_libHandle = LoadLibraryExA(dllName, 
+                                             NULL, LOAD_WITH_ALTERED_SEARCH_PATH);
+            }
+            free(dllName);
+        }
+    } else {
+        /* Try to use old VS_PROFILER variable */
+        dNameLength = GetEnvironmentVariableA(DLL_ENVIRONMENT_VAR, NULL, 0);
+        if (dNameLength)
+        {
+            DWORD envret = 0;
+            dllName = (char*)malloc(sizeof(char) * (dNameLength + 1));
+            if(dllName != NULL)
+            {
+                envret = GetEnvironmentVariableA(DLL_ENVIRONMENT_VAR, 
+                                                 dllName, dNameLength);
+                if (envret)
+                {
+                    /* Try to load the dll from the PATH... */
+                    m_libHandle = LoadLibraryA(dllName);
+                }
+                free(dllName);
+            }
+        }
+    }
+#else  /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+    dllName = getenv(NEW_DLL_ENVIRONMENT_VAR);
+    if (!dllName)
+        dllName = getenv(DLL_ENVIRONMENT_VAR);
+#if defined(__ANDROID__) || defined(ANDROID)
+    if (!dllName)
+        dllName = ANDROID_JIT_AGENT_PATH;
+#endif
+    if (dllName)
+    {
+        /* Try to load the dll from the PATH... */
+        m_libHandle = dlopen(dllName, RTLD_LAZY);
+    }
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+
+    if (!m_libHandle)
+    {
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+        m_libHandle = LoadLibraryA(DEFAULT_DLLNAME);
+#else  /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+        m_libHandle = dlopen(DEFAULT_DLLNAME, RTLD_LAZY);
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+    }
+
+    /* if the dll wasn't loaded - exit. */
+    if (!m_libHandle)
+    {
+        iJIT_DLL_is_missing = 1; /* don't try to initialize
+                                  * JIT agent the second time
+                                  */
+        return 0;
+    }
+
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+    FUNC_NotifyEvent = (TPNotify)GetProcAddress(m_libHandle, "NotifyEvent");
+#else  /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+    FUNC_NotifyEvent = (TPNotify)dlsym(m_libHandle, "NotifyEvent");
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+    if (!FUNC_NotifyEvent) 
+    {
+        FUNC_Initialize = NULL;
+        return 0;
+    }
+
+#if ITT_PLATFORM==ITT_PLATFORM_WIN
+    FUNC_Initialize = (TPInitialize)GetProcAddress(m_libHandle, "Initialize");
+#else  /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+    FUNC_Initialize = (TPInitialize)dlsym(m_libHandle, "Initialize");
+#endif /* ITT_PLATFORM==ITT_PLATFORM_WIN */
+    if (!FUNC_Initialize) 
+    {
+        FUNC_NotifyEvent = NULL;
+        return 0;
+    }
+
+    executionMode = (iJIT_IsProfilingActiveFlags)FUNC_Initialize();
+
+    bDllWasLoaded = 1;
+    iJIT_DLL_is_missing = 0; /* DLL is ok. */
+
+    return 1;
+}
+
+ITT_EXTERN_C unsigned int JITAPI iJIT_GetNewMethodID()
+{
+    static unsigned int methodID = 1;
+
+    if (methodID == 0)
+        return 0;  /* ERROR : this is not a valid value */
+
+    return methodID++;
+}
+
+#endif /* VTUNE_PROFILING */

lib/Runtime/Base/jitprofiling.h

@@ -0,0 +1,641 @@
+//-------------------------------------------------------------------------------------------------------
+// Copyright (C) Microsoft Corporation and contributors. All rights reserved.
+// Licensed under the MIT license. See LICENSE.txt file in the project root for full license information.
+//-------------------------------------------------------------------------------------------------------
+
+#ifndef __JITPROFILING_H__
+#define __JITPROFILING_H__
+
+/**
+ * @brief JIT Profiling APIs
+ *
+ * The JIT Profiling API is used to report information about just-in-time
+ * generated code that can be used by performance tools. The user inserts
+ * calls in the code generator to report information before JIT-compiled
+ * code goes to execution. This information is collected at runtime and used
+ * by tools like Intel(R) VTune(TM) Amplifier to display performance metrics
+ * associated with JIT-compiled code.
+ *
+ * These APIs can be used to\n
+ * - **Profile trace-based and method-based JIT-compiled
+ * code**. Some examples of environments that you can profile with these APIs:
+ * dynamic JIT compilation of JavaScript code traces, JIT execution in OpenCL(TM)
+ * software technology, Java/.NET managed execution environments, and custom
+ * ISV JIT engines.
+ * @code
+ * #include <jitprofiling.h>
+ *
+ * if (iJIT_IsProfilingActive != iJIT_SAMPLING_ON) {
+ *     return;
+ * }
+ *
+ * iJIT_Method_Load jmethod = {0};
+ * jmethod.method_id = iJIT_GetNewMethodID();
+ * jmethod.method_name = "method_name";
+ * jmethod.class_file_name = "class_name";
+ * jmethod.source_file_name = "source_file_name";
+ * jmethod.method_load_address = code_addr;
+ * jmethod.method_size = code_size;
+ *
+ * iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED, (void*)&jmethod);
+ * iJIT_NotifyEvent(iJVM_EVENT_TYPE_SHUTDOWN, NULL);
+ * @endcode
+ *
+ *  * Expected behavior:
+ *    * If any iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED event overwrites an
+ *      already reported method, then such a method becomes invalid and its
+ *      memory region is treated as unloaded. VTune Amplifier displays the metrics
+ *      collected by the method until it is overwritten.
+ *    * If supplied line number information contains multiple source lines for
+ *      the same assembly instruction (code location), then VTune Amplifier picks up
+ *      the first line number.
+ *    * Dynamically generated code can be associated with a module name.
+ *      Use the iJIT_Method_Load_V2 structure.\n
+ *      Clarification of some cases:
+ *        * If you register a function with the same method ID multiple times,
+ *          specifying different module names, then the VTune Amplifier picks up
+ *          the module name registered first. If you want to distinguish the same
+ *          function between different JIT engines, supply different method IDs for
+ *          each function. Other symbolic information (for example, source file)
+ *          can be identical.
+ *
+ * - **Analyze split functions** (multiple joint or disjoint code regions
+ * belonging to the same function) **including re-JIT**
+ * with potential overlapping of code regions in time, which is common in
+ * resource-limited environments.
+ * @code
+ * #include <jitprofiling.h>
+ *
+ * unsigned int method_id = iJIT_GetNewMethodID();
+ *
+ * iJIT_Method_Load a = {0};
+ * a.method_id = method_id;
+ * a.method_load_address = 0x100;
+ * a.method_size = 0x20;
+ *
+ * iJIT_Method_Load b = {0};
+ * b.method_id = method_id;
+ * b.method_load_address = 0x200;
+ * b.method_size = 0x30;
+ *
+ * iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED, (void*)&a);
+ * iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED, (void*)&b);
+ * @endcode
+ *
+ *  * Expected behaviour:
+ *      * If a iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED event overwrites an
+ *        already reported method, then such a method becomes invalid and
+ *        its memory region is treated as unloaded.
+ *      * All code regions reported with the same method ID are considered as
+ *        belonging to the same method. Symbolic information (method name,
+ *        source file name) will be taken from the first notification, and all
+ *        subsequent notifications with the same method ID will be processed
+ *        only for line number table information. So, the VTune Amplifier will map
+ *        samples to a source line using the line number table from the current
+ *        notification while taking the source file name from the very first one.\n
+ *        Clarification of some cases:\n
+ *          * If you register a second code region with a different source file
+ *          name and the same method ID, then this information will be saved and
+ *          will not be considered as an extension of the first code region, but
+ *          VTune Amplifier will use the source file of the first code region and map
+ *          performance metrics incorrectly.
+ *          * If you register a second code region with the same source file as
+ *          for the first region and the same method ID, then the source file will be
+ *          discarded but VTune Amplifier will map metrics to the source file correctly.
+ *          * If you register a second code region with a null source file and
+ *          the same method ID, then provided line number info will be associated
+ *          with the source file of the first code region.
+ *
+ * - **Explore inline functions** including multi-level hierarchy of
+ * nested inline methods which shows how performance metrics are distributed through them.
+ * @code
+ * #include <jitprofiling.h>
+ *
+ *  //                                    method_id   parent_id
+ *  //   [-- c --]                          3000        2000
+ *  //                  [---- d -----]      2001        1000
+ *  //  [---- b ----]                       2000        1000
+ *  // [------------ a ----------------]    1000         n/a
+ *
+ * iJIT_Method_Load a = {0};
+ * a.method_id = 1000;
+ *
+ * iJIT_Method_Inline_Load b = {0};
+ * b.method_id = 2000;
+ * b.parent_method_id = 1000;
+ *
+ * iJIT_Method_Inline_Load c = {0};
+ * c.method_id = 3000;
+ * c.parent_method_id = 2000;
+ *
+ * iJIT_Method_Inline_Load d = {0};
+ * d.method_id = 2001;
+ * d.parent_method_id = 1000;
+ *
+ * iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED, (void*)&a);
+ * iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_INLINE_LOAD_FINISHED, (void*)&b);
+ * iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_INLINE_LOAD_FINISHED, (void*)&c);
+ * iJIT_NotifyEvent(iJVM_EVENT_TYPE_METHOD_INLINE_LOAD_FINISHED, (void*)&d);
+ * @endcode
+ *
+ *  * Requirements:
+ *      * Each inline (iJIT_Method_Inline_Load) method should be associated
+ *        with two method IDs: one for itself; one for its immediate parent.
+ *      * Address regions of inline methods of the same parent method cannot
+ *        overlap each other.
+ *      * Execution of the parent method must not be started until it and all
+ *        its inline methods are reported.
+ *  * Expected behaviour:
+ *      * In case of nested inline methods an order of
+ *        iJVM_EVENT_TYPE_METHOD_INLINE_LOAD_FINISHED events is not important.
+ *      * If any event overwrites either inline method or top parent method,
+ *        then the parent, including inline methods, becomes invalid and its memory
+ *        region is treated as unloaded.
+ *
+ * **Life time of allocated data**\n
+ * The client sends an event notification to the agent with event-specific
+ * data, which is a structure. The pointers in the structure refer to memory
+ * allocated by the client, which responsible for releasing it. The pointers are
+ * used by the iJIT_NotifyEvent method to copy client's data in a trace file,
+ * and they are not used after the iJIT_NotifyEvent method returns.
+ */
+
+/**
+ * @defgroup jitapi JIT Profiling
+ * @ingroup internal
+ * @{
+ */
+
+/**
+ * @brief Enumerator for the types of notifications
+ */
+typedef enum iJIT_jvm_event
+{
+    iJVM_EVENT_TYPE_SHUTDOWN = 2,               /**<\brief Send this to shutdown the agent.
+                                                 * Use NULL for event data. */
+
+    iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED = 13,  /**<\brief Send when dynamic code is
+                                                 * JIT compiled and loaded into
+                                                 * memory by the JIT engine, but
+                                                 * before the code is executed.
+                                                 * Use iJIT_Method_Load as event
+                                                 * data. */
+/** @cond exclude_from_documentation */
+    iJVM_EVENT_TYPE_METHOD_UNLOAD_START,    /**<\brief Send when compiled dynamic
+                                             * code is being unloaded from memory.
+                                             * Use iJIT_Method_Load as event data.*/
+/** @endcond */
+
+    iJVM_EVENT_TYPE_METHOD_UPDATE,   /**<\brief Send to provide new content for
+                                      * a previously reported dynamic code.
+                                      * The previous content will be invalidated
+                                      * starting from the time of the notification.
+                                      * Use iJIT_Method_Load as event data but
+                                      * required fields are following:
+                                      * - method_id    identify the code to update.
+                                      * - method_load_address    specify start address
+                                      *                          within identified code range
+                                      *                          where update should be started.
+                                      * - method_size            specify length of updated code
+                                      *                          range. */
+
+
+    iJVM_EVENT_TYPE_METHOD_INLINE_LOAD_FINISHED, /**<\brief Send when an inline dynamic
+                                                  * code is JIT compiled and loaded
+                                                  * into memory by the JIT engine,
+                                                  * but before the parent code region
+                                                  * starts executing.
+                                                  * Use iJIT_Method_Inline_Load as event data.*/
+
+/** @cond exclude_from_documentation */
+    iJVM_EVENT_TYPE_METHOD_UPDATE_V2,
+/** @endcond */
+
+    iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED_V2 = 21, /**<\brief Send when a dynamic code is
+                                                   * JIT compiled and loaded into
+                                                   * memory by the JIT engine, but
+                                                   * before the code is executed.
+                                                   * Use iJIT_Method_Load_V2 as event data. */
+
+    iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED_V3       /**<\brief Send when a dynamic code is
+                                                   * JIT compiled and loaded into
+                                                   * memory by the JIT engine, but
+                                                   * before the code is executed.
+                                                   * Use iJIT_Method_Load_V3 as event data. */
+} iJIT_JVM_EVENT;
+
+/**
+ * @brief Enumerator for the agent's mode
+ */
+typedef enum _iJIT_IsProfilingActiveFlags
+{
+    iJIT_NOTHING_RUNNING           = 0x0000,    /**<\brief The agent is not running;
+                                                 * iJIT_NotifyEvent calls will
+                                                 * not be processed. */
+    iJIT_SAMPLING_ON               = 0x0001,    /**<\brief The agent is running and
+                                                 * ready to process notifications. */
+} iJIT_IsProfilingActiveFlags;
+
+/**
+ * @brief Description of a single entry in the line number information of a code region.
+ * @details A table of line number entries gives information about how the reported code region
+ * is mapped to source file.
+ * Intel(R) VTune(TM) Amplifier uses line number information to attribute
+ * the samples (virtual address) to a line number. \n
+ * It is acceptable to report different code addresses for the same source line:
+ * @code
+ *   Offset LineNumber
+ *      1       2
+ *      12      4
+ *      15      2
+ *      18      1
+ *      21      30
+ *
+ *  VTune Amplifier constructs the following table using the client data
+ *
+ *   Code subrange  Line number
+ *      0-1             2
+ *      1-12            4
+ *      12-15           2
+ *      15-18           1
+ *      18-21           30
+ * @endcode
+ */
+typedef struct _LineNumberInfo
+{
+    unsigned int Offset;     /**<\brief Offset from the begining of the code region. */
+    unsigned int LineNumber; /**<\brief Matching source line number offset (from beginning of source file). */
+
+} *pLineNumberInfo, LineNumberInfo;
+
+/**
+ * @brief Enumerator for the code architecture.
+ */
+typedef enum _iJIT_CodeArchitecture
+{
+    iJIT_CA_NATIVE = 0, /**<\brief Native to the process architecture that is calling it. */
+
+    iJIT_CA_32,         /**<\brief 32-bit machine code. */
+
+    iJIT_CA_64          /**<\brief 64-bit machine code. */
+
+} iJIT_CodeArchitecture;
+
+#pragma pack(push, 8)
+
+/**
+ * @brief Description of a JIT-compiled method
+ * @details When you use the iJIT_Method_Load structure to describe
+ *  the JIT compiled method, use iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED
+ *  as an event type to report it.
+ */
+typedef struct _iJIT_Method_Load
+{
+    unsigned int method_id; /**<\brief Unique method ID. Cannot be 0.
+                             *  You must either use the API function
+                             *  iJIT_GetNewMethodID to get a valid and unique
+                             *  method ID, or else manage ID uniqueness
+                             *  and correct range by yourself.\n
+                             *  You must use the same method ID for all code
+                             *  regions of the same method, otherwise different
+                             *  method IDs specify different methods. */
+
+    char* method_name; /**<\brief The name of the method. It can be optionally
+                        *  prefixed with its class name and appended with
+                        *  its complete signature. Can't be NULL. */
+
+    void* method_load_address; /**<\brief The start virtual address of the method code
+                                *  region. If NULL, data provided with
+                                *  event are not accepted. */
+
+    unsigned int method_size; /**<\brief The code size of the method in memory.
+                               *  If 0, then data provided with the event are not
+                               *  accepted. */
+
+    unsigned int line_number_size; /**<\brief The number of entries in the line number
+                                    *  table.0 if none. */
+
+    pLineNumberInfo line_number_table; /**<\brief Pointer to the line numbers info
+                                        *  array. Can be NULL if
+                                        *  line_number_size is 0. See
+                                        *  LineNumberInfo Structure for a
+                                        *  description of a single entry in
+                                        *  the line number info array */
+
+    unsigned int class_id; /**<\brief This field is obsolete. */
+
+    char* class_file_name; /**<\brief Class name. Can be NULL.*/
+
+    char* source_file_name; /**<\brief Source file name. Can be NULL.*/
+
+} *piJIT_Method_Load, iJIT_Method_Load;
+
+/**
+ * @brief Description of a JIT-compiled method
+ * @details When you use the iJIT_Method_Load_V2 structure to describe
+ *  the JIT compiled method, use iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED_V2
+ *  as an event type to report it.
+ */
+typedef struct _iJIT_Method_Load_V2
+{
+    unsigned int method_id; /**<\brief Unique method ID. Cannot be 0.
+                             *  You must either use the API function
+                             *  iJIT_GetNewMethodID to get a valid and unique
+                             *  method ID, or else manage ID uniqueness
+                             *  and correct range by yourself.\n
+                             *  You must use the same method ID for all code
+                             *  regions of the same method, otherwise different
+                             *  method IDs specify different methods. */
+
+    char* method_name; /**<\brief The name of the method. It can be optionally
+                        *  prefixed with its class name and appended with
+                        *  its complete signature. Can't be  NULL. */
+
+    void* method_load_address; /**<\brief The start virtual address of the method code
+                                *  region. If NULL, then data provided with the
+                                *  event are not accepted. */
+
+    unsigned int method_size; /**<\brief The code size of the method in memory.
+                               *  If 0, then data provided with the event are not
+                               *  accepted. */
+
+    unsigned int line_number_size; /**<\brief The number of entries in the line number
+                                    *  table. 0 if none. */
+
+    pLineNumberInfo line_number_table; /**<\brief Pointer to the line numbers info
+                                        *  array. Can be NULL if
+                                        *  line_number_size is 0. See
+                                        *  LineNumberInfo Structure for a
+                                        *  description of a single entry in
+                                        *  the line number info array. */
+
+    char* class_file_name; /**<\brief Class name. Can be NULL. */
+
+    char* source_file_name; /**<\brief Source file name. Can be NULL. */
+
+    char* module_name; /**<\brief Module name. Can be NULL.
+                           The module name can be useful for distinguishing among
+                           different JIT engines. VTune Amplifier will display
+                           reported methods grouped by specific module. */
+
+} *piJIT_Method_Load_V2, iJIT_Method_Load_V2;
+
+/**
+ * @brief Description of a JIT-compiled method
+ * @details The iJIT_Method_Load_V3 structure is the same as iJIT_Method_Load_V2
+ *  with a newly introduced 'arch' field that specifies architecture of the code region.
+ *  When you use the iJIT_Method_Load_V3 structure to describe
+ *  the JIT compiled method, use iJVM_EVENT_TYPE_METHOD_LOAD_FINISHED_V3
+ *  as an event type to report it.
+ */
+typedef struct _iJIT_Method_Load_V3
+{
+    unsigned int method_id; /**<\brief Unique method ID. Cannot be 0.
+                             *  You must either use the API function
+                             *  iJIT_GetNewMethodID to get a valid and unique
+                             *  method ID, or manage ID uniqueness
+                             *  and correct range by yourself.\n
+                             *  You must use the same method ID for all code
+                             *  regions of the same method, otherwise they are
+                             *  treated as regions of different methods. */
+
+    char* method_name; /**<\brief The name of the method. It can be optionally
+                        *  prefixed with its class name and appended with
+                        *  its complete signature. Cannot be  NULL. */
+
+    void* method_load_address; /**<\brief The start virtual address of the method code
+                                *  region. If NULL, then data provided with the
+                                *  event are not accepted. */
+
+    unsigned int method_size; /**<\brief The code size of the method in memory.
+                               *  If 0, then data provided with the event are not
+                               *  accepted. */
+
+    unsigned int line_number_size; /**<\brief The number of entries in the line number
+                                    *  table. 0 if none. */
+
+    pLineNumberInfo line_number_table; /**<\brief Pointer to the line numbers info
+                                        *  array. Can be NULL if
+                                        *  line_number_size is 0. See
+                                        *  LineNumberInfo Structure for a
+                                        *  description of a single entry in
+                                        *  the line number info array. */
+
+    char* class_file_name; /**<\brief Class name. Can be NULL. */
+
+    char* source_file_name; /**<\brief Source file name. Can be NULL. */
+
+    char* module_name; /**<\brief Module name. Can be NULL.
+                        *  The module name can be useful for distinguishing among
+                        *  different JIT engines. VTune Amplifier will display
+                        *  reported methods grouped by specific module. */
+
+    iJIT_CodeArchitecture module_arch; /**<\brief Architecture of the method's code region.
+                                        *  By default, it is the same as the process
+                                        *  architecture that is calling it.
+                                        *  For example, you can use it if your 32-bit JIT
+                                        *  engine generates 64-bit code.
+                                        *
+                                        *  If JIT engine reports both 32-bit and 64-bit types
+                                        *  of methods then VTune Amplifier splits the methods
+                                        *  with the same module name but with different
+                                        *  architectures in two different modules. VTune Amplifier
+                                        *  modifies the original name provided with a 64-bit method
+                                        *  version by ending it with '(64)' */
+
+} *piJIT_Method_Load_V3, iJIT_Method_Load_V3;
+
+/**
+ * @brief Description of an inline JIT-compiled method
+ * @details When you use the_iJIT_Method_Inline_Load structure to describe
+ *  the JIT compiled method, use iJVM_EVENT_TYPE_METHOD_INLINE_LOAD_FINISHED
+ *  as an event type to report it.
+ */
+typedef struct _iJIT_Method_Inline_Load
+{
+    unsigned int method_id; /**<\brief Unique method ID. Cannot be 0.
+                             *  You must either use the API function
+                             *  iJIT_GetNewMethodID to get a valid and unique
+                             *  method ID, or else manage ID uniqueness
+                             *  and correct range by yourself. */
+
+    unsigned int parent_method_id; /**<\brief Unique immediate parent's method ID.
+                                    *  Cannot be 0.
+                                    *  You must either use the API function
+                                    *  iJIT_GetNewMethodID to get a valid and unique
+                                    *  method ID, or else manage ID uniqueness
+                                    *  and correct range by yourself. */
+
+    char* method_name; /**<\brief The name of the method. It can be optionally
+                        *  prefixed with its class name and appended with
+                        *  its complete signature. Can't be NULL. */
+
+    void* method_load_address;  /** <\brief The virtual address on which the method
+                                 *  is inlined. If NULL, then data provided with
+                                 *  the event are not accepted. */
+
+    unsigned int method_size; /**<\brief The code size of the method in memory.
+                               *  If 0, then data provided with the event are not
+                               *  accepted. */
+
+    unsigned int line_number_size; /**<\brief The number of entries in the line number
+                                    *  table. 0 if none. */
+
+    pLineNumberInfo line_number_table; /**<\brief Pointer to the line numbers info
+                                        *  array. Can be NULL if
+                                        *  line_number_size is 0. See
+                                        *  LineNumberInfo Structure for a
+                                        *  description of a single entry in
+                                        *  the line number info array */
+
+    char* class_file_name; /**<\brief Class name. Can be NULL.*/
+
+    char* source_file_name; /**<\brief Source file name. Can be NULL.*/
+
+} *piJIT_Method_Inline_Load, iJIT_Method_Inline_Load;
+
+/** @cond exclude_from_documentation */
+/**
+ * @brief Description of a segment type
+ * @details Use the segment type to specify a type of data supplied
+ * with the iJVM_EVENT_TYPE_METHOD_UPDATE_V2 event to be applied to
+ * a certain code trace.
+ */
+typedef enum _iJIT_SegmentType
+{
+    iJIT_CT_UNKNOWN = 0,
+
+    iJIT_CT_CODE,           /**<\brief Executable code. */
+
+    iJIT_CT_DATA,           /**<\brief Data (not executable code).
+                             * VTune Amplifier uses the format string
+                             * (see iJIT_Method_Update) to represent
+                             * this data in the VTune Amplifier GUI */
+
+    iJIT_CT_KEEP,           /**<\brief Use the previous markup for the trace.
+                             * Can be used for the following
+                             * iJVM_EVENT_TYPE_METHOD_UPDATE_V2 events,
+                             * if the type of the previously reported segment
+                             * type is the same. */
+    iJIT_CT_EOF
+} iJIT_SegmentType;
+
+/**
+ * @brief Description of a dynamic update of the content within JIT-compiled method
+ * @details The JIT engine may generate the methods that are updated at runtime
+ * partially by mixed (data + executable code) content. When you use the iJIT_Method_Update
+ * structure to describe the update of the content within a JIT-compiled method,
+ * use iJVM_EVENT_TYPE_METHOD_UPDATE_V2 as an event type to report it.
+ *
+ * On the first Update event, VTune Amplifier copies the original code range reported by
+ * the iJVM_EVENT_TYPE_METHOD_LOAD event, then modifies it with the supplied bytes and
+ * adds the modified range to the original method. For next update events, VTune Amplifier
+ * does the same but it uses the latest modified version of a code region for update.
+ * Eventually, VTune Amplifier GUI displays multiple code ranges for the method reported by
+ * the iJVM_EVENT_TYPE_METHOD_LOAD event.
+ * Notes:
+ * - Multiple update events with different types for the same trace are allowed
+ *   but they must be reported for the same code ranges.
+ *   Example,
+ * @code
+ *                      [-- data---]        Allowed
+ *          [-- code --]                    Allowed
+ *        [code]                            Ignored
+ *                      [-- data---]        Allowed
+ *          [-- code --]                    Allowed
+ *      [------------ trace ---------]
+ * @endcode
+ * - The types of previously reported events can be changed but they must be reported
+ *   for the same code ranges.
+ *   Example,
+ * @code
+ *          [-- data---]                    Allowed
+ *          [-- code --]                    Allowed
+ *          [-- data---]                    Allowed
+ *          [-- code --]                    Allowed
+ *      [------------ trace ---------]
+ * @endcode
+ */
+
+typedef struct _iJIT_Method_Update
+{
+    void* load_address;         /**<\brief Start address of the update within a method */
+
+    unsigned int size;          /**<\brief The update size */
+
+    iJIT_SegmentType type;      /**<\brief Type of the update */
+
+    const char* data_format;    /**<\brief C string that contains a format string
+                                 * that follows the same specifications as format in printf.
+                                 * The format string is used for iJIT_CT_CODE only
+                                 * and cannot be NULL.
+                                 * Format can be changed on the fly. */
+} *piJIT_Method_Update, iJIT_Method_Update;
+
+/** @endcond */
+
+#pragma pack(pop)
+
+/** @cond exclude_from_documentation */
+#ifdef __cplusplus
+extern "C" {
+#endif /* __cplusplus */
+
+#ifndef JITAPI_CDECL
+#  if defined WIN32 || defined _WIN32
+#    define JITAPI_CDECL __cdecl
+#  else /* defined WIN32 || defined _WIN32 */
+#    if defined _M_IX86 || defined __i386__
+#      define JITAPI_CDECL __attribute__ ((cdecl))
+#    else  /* _M_IX86 || __i386__ */
+#      define JITAPI_CDECL /* actual only on x86_64 platform */
+#    endif /* _M_IX86 || __i386__ */
+#  endif /* defined WIN32 || defined _WIN32 */
+#endif /* JITAPI_CDECL */
+
+#define JITAPI JITAPI_CDECL
+/** @endcond */
+
+/**
+ * @brief Generates a new unique method ID.
+ *
+ * You must use this API to obtain unique and valid method IDs for methods or
+ * traces reported to the agent if you don't have your own mechanism to generate
+ * unique method IDs.
+ *
+ * @return a new unique method ID. When out of unique method IDs, this API
+ * returns 0, which is not an accepted value.
+ */
+unsigned int JITAPI iJIT_GetNewMethodID(void);
+
+/**
+ * @brief Returns the current mode of the agent.
+ *
+ * @return iJIT_SAMPLING_ON, indicating that agent is running, or
+ * iJIT_NOTHING_RUNNING if no agent is running.
+ */
+iJIT_IsProfilingActiveFlags JITAPI iJIT_IsProfilingActive(void);
+
+/**
+ * @brief Reports infomation about JIT-compiled code to the agent.
+ *
+ * The reported information is used to attribute samples obtained from any
+ * Intel(R) VTune(TM) Amplifier collector. This API needs to be called
+ * after JIT compilation and before the first entry into the JIT-compiled
+ * code.
+ *
+ * @param[in] event_type - type of the data sent to the agent
+ * @param[in] EventSpecificData - pointer to event-specific data
+ * 
+ * @returns 1 on success, otherwise 0.
+ */
+int JITAPI iJIT_NotifyEvent(iJIT_JVM_EVENT event_type, void *EventSpecificData);
+
+#ifdef __cplusplus
+}
+#endif /* __cplusplus */
+/** @endcond */
+
+/** @} jitapi group */
+
+#endif /* __JITPROFILING_H__ */