Fix merge conflict

16 files changed, 260 insertions, 238 deletions

diff --git a/indra/llcommon/lldate.cpp b/indra/llcommon/lldate.cpp
index 7a2a0869f4..2ddcf40895 100644
--- a/indra/llcommon/lldate.cpp
+++ b/indra/llcommon/lldate.cpp
@@ -86,11 +86,9 @@ std::string LLDate::asRFC1123() const
 	return toHTTPDateString (std::string ("%A, %d %b %Y %H:%M:%S GMT"));
 }
 
-LLTrace::BlockTimerStatHandle FT_DATE_FORMAT("Date Format");
-
 std::string LLDate::toHTTPDateString (std::string fmt) const
 {
-	LL_RECORD_BLOCK_TIME(FT_DATE_FORMAT);
+    LL_PROFILE_ZONE_SCOPED;
 	
 	time_t locSeconds = (time_t) mSecondsSinceEpoch;
 	struct tm * gmt = gmtime (&locSeconds);
@@ -99,7 +97,7 @@ std::string LLDate::toHTTPDateString (std::string fmt) const
 
 std::string LLDate::toHTTPDateString (tm * gmt, std::string fmt)
 {
-	LL_RECORD_BLOCK_TIME(FT_DATE_FORMAT);
+    LL_PROFILE_ZONE_SCOPED;
 
 	// avoid calling setlocale() unnecessarily - it's expensive.
 	static std::string prev_locale = "";
diff --git a/indra/llcommon/llfasttimer.cpp b/indra/llcommon/llfasttimer.cpp
index 5b6a7b82f8..d38946004f 100644
--- a/indra/llcommon/llfasttimer.cpp
+++ b/indra/llcommon/llfasttimer.cpp
@@ -191,29 +191,30 @@ TimeBlockTreeNode& BlockTimerStatHandle::getTreeNode() const
 }
 
 
+
 void BlockTimer::bootstrapTimerTree()
 {
-	for (auto& base : BlockTimerStatHandle::instance_snapshot())
-	{
-		// because of indirect derivation from LLInstanceTracker, have to downcast
-		BlockTimerStatHandle& timer = static_cast<BlockTimerStatHandle&>(base);
-		if (&timer == &BlockTimer::getRootTimeBlock()) continue;
-
-		// bootstrap tree construction by attaching to last timer to be on stack
-		// when this timer was called
-		if (timer.getParent() == &BlockTimer::getRootTimeBlock())
-		{
-			TimeBlockAccumulator& accumulator = timer.getCurrentAccumulator();
-
-			if (accumulator.mLastCaller)
-			{
-				timer.setParent(accumulator.mLastCaller);
-				accumulator.mParent = accumulator.mLastCaller;
-			}
-			// no need to push up tree on first use, flag can be set spuriously
-			accumulator.mMoveUpTree = false;
-		}
-	}
+    for (auto& base : BlockTimerStatHandle::instance_snapshot())
+    {
+        // because of indirect derivation from LLInstanceTracker, have to downcast
+        BlockTimerStatHandle& timer = static_cast<BlockTimerStatHandle&>(base);
+        if (&timer == &BlockTimer::getRootTimeBlock()) continue;
+
+        // bootstrap tree construction by attaching to last timer to be on stack
+        // when this timer was called
+        if (timer.getParent() == &BlockTimer::getRootTimeBlock())
+        {
+            TimeBlockAccumulator& accumulator = timer.getCurrentAccumulator();
+
+            if (accumulator.mLastCaller)
+            {
+                timer.setParent(accumulator.mLastCaller);
+                accumulator.mParent = accumulator.mLastCaller;
+            }
+            // no need to push up tree on first use, flag can be set spuriously
+            accumulator.mMoveUpTree = false;
+        }
+    }
 }
 
 // bump timers up tree if they have been flagged as being in the wrong place
@@ -221,6 +222,7 @@ void BlockTimer::bootstrapTimerTree()
 // this preserves partial order derived from current frame's observations
 void BlockTimer::incrementalUpdateTimerTree()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	for(block_timer_tree_df_post_iterator_t it = begin_block_timer_tree_df_post(BlockTimer::getRootTimeBlock());
 		it != end_block_timer_tree_df_post();
 		++it)
@@ -260,7 +262,8 @@ void BlockTimer::incrementalUpdateTimerTree()
 
 
 void BlockTimer::updateTimes()
-	{
+{
+    LL_PROFILE_ZONE_SCOPED;
 	// walk up stack of active timers and accumulate current time while leaving timing structures active
 	BlockTimerStackRecord* stack_record	= LLThreadLocalSingletonPointer<BlockTimerStackRecord>::getInstance();
 	if (!stack_record) return;
@@ -271,7 +274,7 @@ void BlockTimer::updateTimes()
 
 	while(cur_timer 
 		&& cur_timer->mParentTimerData.mActiveTimer != cur_timer) // root defined by parent pointing to self
-		{
+	{
 		U64 cumulative_time_delta = cur_time - cur_timer->mStartTime;
 		cur_timer->mStartTime = cur_time;
 
diff --git a/indra/llcommon/llmemory.h b/indra/llcommon/llmemory.h
index 2704a495e0..41023b4ba4 100644
--- a/indra/llcommon/llmemory.h
+++ b/indra/llcommon/llmemory.h
@@ -111,6 +111,16 @@ public:                                     \
     void operator delete(void* ptr)         \
     {                                       \
         ll_aligned_free_16(ptr);            \
+    }                                       \
+                                            \
+    void* operator new[](size_t size)       \
+    {                                       \
+        return ll_aligned_malloc_16(size);  \
+    }                                       \
+                                            \
+    void operator delete[](void* ptr)       \
+    {                                       \
+        ll_aligned_free_16(ptr);            \
     }
 
 
@@ -126,8 +136,9 @@ public:                                     \
 #else
 	inline void* ll_aligned_malloc_fallback( size_t size, int align )
 	{
+        LL_PROFILE_ZONE_SCOPED;
 	#if defined(LL_WINDOWS)
-		return _aligned_malloc(size, align);
+        void* ret = _aligned_malloc(size, align);
 	#else
         char* aligned = NULL;
 		void* mem = malloc( size + (align - 1) + sizeof(void*) );
@@ -138,12 +149,16 @@ public:                                     \
 
             ((void**)aligned)[-1] = mem;
         }
-		return aligned;
+		void* ret = aligned;
 	#endif
+        LL_PROFILE_ALLOC(ret, size);
+        return ret;
 	}
 
 	inline void ll_aligned_free_fallback( void* ptr )
 	{
+        LL_PROFILE_ZONE_SCOPED;
+        LL_PROFILE_FREE(ptr);
 	#if defined(LL_WINDOWS)
 		_aligned_free(ptr);
 	#else
@@ -159,21 +174,24 @@ public:                                     \
 
 inline void* ll_aligned_malloc_16(size_t size) // returned hunk MUST be freed with ll_aligned_free_16().
 {
+    LL_PROFILE_ZONE_SCOPED;
 #if defined(LL_WINDOWS)
-	return _aligned_malloc(size, 16);
+	void* ret = _aligned_malloc(size, 16);
 #elif defined(LL_DARWIN)
-	return malloc(size); // default osx malloc is 16 byte aligned.
+	void* ret = malloc(size); // default osx malloc is 16 byte aligned.
 #else
-	void *rtn;
-	if (LL_LIKELY(0 == posix_memalign(&rtn, 16, size)))
-		return rtn;
-	else // bad alignment requested, or out of memory
-		return NULL;
+	void *ret;
+    if (0 != posix_memalign(&ret, 16, size))
+        return nullptr;
 #endif
+    LL_PROFILE_ALLOC(ret, size);
+    return ret;
 }
 
 inline void ll_aligned_free_16(void *p)
 {
+    LL_PROFILE_ZONE_SCOPED;
+    LL_PROFILE_FREE(p);
 #if defined(LL_WINDOWS)
 	_aligned_free(p);
 #elif defined(LL_DARWIN)
@@ -185,10 +203,12 @@ inline void ll_aligned_free_16(void *p)
 
 inline void* ll_aligned_realloc_16(void* ptr, size_t size, size_t old_size) // returned hunk MUST be freed with ll_aligned_free_16().
 {
+    LL_PROFILE_ZONE_SCOPED;
+    LL_PROFILE_FREE(ptr);
 #if defined(LL_WINDOWS)
-	return _aligned_realloc(ptr, size, 16);
+	void* ret = _aligned_realloc(ptr, size, 16);
 #elif defined(LL_DARWIN)
-	return realloc(ptr,size); // default osx malloc is 16 byte aligned.
+	void* ret = realloc(ptr,size); // default osx malloc is 16 byte aligned.
 #else
 	//FIXME: memcpy is SLOW
 	void* ret = ll_aligned_malloc_16(size);
@@ -201,27 +221,31 @@ inline void* ll_aligned_realloc_16(void* ptr, size_t size, size_t old_size) // r
 		}
 		ll_aligned_free_16(ptr);
 	}
-	return ret;
 #endif
+    LL_PROFILE_ALLOC(ptr, size);
+    return ret;
 }
 
 inline void* ll_aligned_malloc_32(size_t size) // returned hunk MUST be freed with ll_aligned_free_32().
 {
+    LL_PROFILE_ZONE_SCOPED;
 #if defined(LL_WINDOWS)
-	return _aligned_malloc(size, 32);
+	void* ret = _aligned_malloc(size, 32);
 #elif defined(LL_DARWIN)
-	return ll_aligned_malloc_fallback( size, 32 );
+	void* ret = ll_aligned_malloc_fallback( size, 32 );
 #else
-	void *rtn;
-	if (LL_LIKELY(0 == posix_memalign(&rtn, 32, size)))
-		return rtn;
-	else // bad alignment requested, or out of memory
-		return NULL;
+	void *ret;
+    if (0 != posix_memalign(&ret, 32, size))
+        return nullptr;
 #endif
+    LL_PROFILE_ALLOC(ret, size);
+    return ret;
 }
 
 inline void ll_aligned_free_32(void *p)
 {
+    LL_PROFILE_ZONE_SCOPED;
+    LL_PROFILE_FREE(p);
 #if defined(LL_WINDOWS)
 	_aligned_free(p);
 #elif defined(LL_DARWIN)
@@ -235,29 +259,35 @@ inline void ll_aligned_free_32(void *p)
 template<size_t ALIGNMENT>
 LL_FORCE_INLINE void* ll_aligned_malloc(size_t size)
 {
+    LL_PROFILE_ZONE_SCOPED;
+    void* ret;
 	if (LL_DEFAULT_HEAP_ALIGN % ALIGNMENT == 0)
 	{
-		return malloc(size);
+		ret = malloc(size);
+        LL_PROFILE_ALLOC(ret, size);
 	}
 	else if (ALIGNMENT == 16)
 	{
-		return ll_aligned_malloc_16(size);
+		ret = ll_aligned_malloc_16(size);
 	}
 	else if (ALIGNMENT == 32)
 	{
-		return ll_aligned_malloc_32(size);
+		ret = ll_aligned_malloc_32(size);
 	}
 	else
 	{
-		return ll_aligned_malloc_fallback(size, ALIGNMENT);
+		ret = ll_aligned_malloc_fallback(size, ALIGNMENT);
 	}
+    return ret;
 }
 
 template<size_t ALIGNMENT>
 LL_FORCE_INLINE void ll_aligned_free(void* ptr)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	if (ALIGNMENT == LL_DEFAULT_HEAP_ALIGN)
 	{
+        LL_PROFILE_FREE(ptr);
 		free(ptr);
 	}
 	else if (ALIGNMENT == 16)
@@ -279,6 +309,7 @@ LL_FORCE_INLINE void ll_aligned_free(void* ptr)
 //
 inline void ll_memcpy_nonaliased_aligned_16(char* __restrict dst, const char* __restrict src, size_t bytes)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	assert(src != NULL);
 	assert(dst != NULL);
 	assert(bytes > 0);
diff --git a/indra/llcommon/llprofiler.h b/indra/llcommon/llprofiler.h
index e36f693dd3..ca60d23248 100644
--- a/indra/llcommon/llprofiler.h
+++ b/indra/llcommon/llprofiler.h
@@ -66,6 +66,8 @@ extern thread_local bool gProfilerEnabled;
         #define LL_PROFILE_ZONE_ERR(name)               LL_PROFILE_ZONE_NAMED_COLOR( name, 0XFF0000  )  // RGB yellow
         #define LL_PROFILE_ZONE_INFO(name)              LL_PROFILE_ZONE_NAMED_COLOR( name, 0X00FFFF  )  // RGB cyan
         #define LL_PROFILE_ZONE_WARN(name)              LL_PROFILE_ZONE_NAMED_COLOR( name, 0x0FFFF00 )  // RGB red
+        #define LL_PROFILE_ALLOC(ptr, size)             TracyAlloc(ptr, size)
+        #define LL_PROFILE_FREE(ptr)                    TracyFree(ptr)
     #endif
     #if LL_PROFILER_CONFIGURATION == LL_PROFILER_CONFIG_FAST_TIMER
         #define LL_PROFILER_FRAME_END
@@ -81,11 +83,13 @@ extern thread_local bool gProfilerEnabled;
         #define LL_PROFILE_ZONE_ERR(name)               (void)(name); // Not supported
         #define LL_PROFILE_ZONE_INFO(name)              (void)(name); // Not supported
         #define LL_PROFILE_ZONE_WARN(name)              (void)(name); // Not supported
+        #define LL_PROFILE_ALLOC(ptr, size)             (void)(ptr); (void)(size);
+        #define LL_PROFILE_FREE(ptr)                    (void)(ptr);
     #endif
     #if LL_PROFILER_CONFIGURATION == LL_PROFILER_CONFIG_TRACY_FAST_TIMER
         #define LL_PROFILER_FRAME_END                   FrameMark
         #define LL_PROFILER_SET_THREAD_NAME( name )     tracy::SetThreadName( name );    gProfilerEnabled = true;
-        #define LL_RECORD_BLOCK_TIME(name)              ZoneScoped                                          const LLTrace::BlockTimer& LL_GLUE_TOKENS(block_time_recorder, __LINE__)(LLTrace::timeThisBlock(name)); (void)LL_GLUE_TOKENS(block_time_recorder, __LINE__);
+        #define LL_RECORD_BLOCK_TIME(name)              ZoneNamedN(___tracy_scoped_zone, #name, true);   const LLTrace::BlockTimer& LL_GLUE_TOKENS(block_time_recorder, __LINE__)(LLTrace::timeThisBlock(name)); (void)LL_GLUE_TOKENS(block_time_recorder, __LINE__);
         #define LL_PROFILE_ZONE_NAMED(name)             ZoneNamedN( ___tracy_scoped_zone, #name, true );
         #define LL_PROFILE_ZONE_NAMED_COLOR(name,color) ZoneNamedNC( ___tracy_scopped_zone, name, color, true ) // RGB
         #define LL_PROFILE_ZONE_SCOPED                  ZoneScoped
@@ -96,6 +100,8 @@ extern thread_local bool gProfilerEnabled;
         #define LL_PROFILE_ZONE_ERR(name)               LL_PROFILE_ZONE_NAMED_COLOR( name, 0XFF0000  )  // RGB yellow
         #define LL_PROFILE_ZONE_INFO(name)              LL_PROFILE_ZONE_NAMED_COLOR( name, 0X00FFFF  )  // RGB cyan
         #define LL_PROFILE_ZONE_WARN(name)              LL_PROFILE_ZONE_NAMED_COLOR( name, 0x0FFFF00 )  // RGB red
+        #define LL_PROFILE_ALLOC(ptr, size)             TracyAlloc(ptr, size)
+        #define LL_PROFILE_FREE(ptr)                    TracyFree(ptr)
     #endif
 #else
     #define LL_PROFILER_FRAME_END
diff --git a/indra/llcommon/llsdparam.cpp b/indra/llcommon/llsdparam.cpp
index 2e7b46f885..af4ccf25fd 100644
--- a/indra/llcommon/llsdparam.cpp
+++ b/indra/llcommon/llsdparam.cpp
@@ -37,8 +37,6 @@ static 	LLInitParam::Parser::parser_write_func_map_t sWriteFuncs;
 static 	LLInitParam::Parser::parser_inspect_func_map_t sInspectFuncs;
 static const LLSD NO_VALUE_MARKER;
 
-LLTrace::BlockTimerStatHandle FTM_SD_PARAM_ADAPTOR("LLSD to LLInitParam conversion");
-
 //
 // LLParamSDParser
 //
diff --git a/indra/llcommon/llsdparam.h b/indra/llcommon/llsdparam.h
index 93910b70ae..82a623a8a0 100644
--- a/indra/llcommon/llsdparam.h
+++ b/indra/llcommon/llsdparam.h
@@ -110,7 +110,6 @@ private:
 };
 
 
-extern LL_COMMON_API LLTrace::BlockTimerStatHandle FTM_SD_PARAM_ADAPTOR;
 template<typename T>
 class LLSDParamAdapter : public T
 {
@@ -118,7 +117,7 @@ public:
 	LLSDParamAdapter() {}
 	LLSDParamAdapter(const LLSD& sd)
 	{
-		LL_RECORD_BLOCK_TIME(FTM_SD_PARAM_ADAPTOR);
+        LL_PROFILE_ZONE_SCOPED;
 		LLParamSDParser parser;
 		// don't spam for implicit parsing of LLSD, as we want to allow arbitrary freeform data and ignore most of it
 		bool parse_silently = true;
diff --git a/indra/llcommon/llsingleton.h b/indra/llcommon/llsingleton.h
index 2e43a3cbed..10a8ecfedb 100644
--- a/indra/llcommon/llsingleton.h
+++ b/indra/llcommon/llsingleton.h
@@ -839,4 +839,30 @@ private:                                                                \
     /* LLSINGLETON() is carefully implemented to permit exactly this */ \
     LLSINGLETON_C11(DERIVED_CLASS) {}
 
+// Relatively unsafe singleton implementation that is much faster
+// and simpler than LLSingleton, but has no dependency tracking
+// or inherent thread safety and requires manual invocation of 
+// createInstance before first use.
+template<class T>
+class LLSimpleton
+{
+public:
+    static T* sInstance;
+    
+    static void createInstance() 
+    { 
+        llassert(sInstance == nullptr);
+        sInstance = new T(); 
+    }
+    
+    static inline T* getInstance() { return sInstance; }
+    static inline T& instance() { return *getInstance(); }
+    static inline bool instanceExists() { return sInstance != nullptr; }
+
+    static void deleteSingleton() { 
+        delete sInstance; 
+        sInstance = nullptr; 
+    }
+};
+
 #endif
diff --git a/indra/llcommon/llstring.cpp b/indra/llcommon/llstring.cpp
index 0290eea143..f6f9f97809 100644
--- a/indra/llcommon/llstring.cpp
+++ b/indra/llcommon/llstring.cpp
@@ -37,9 +37,6 @@
 #include <winnls.h> // for WideCharToMultiByte
 #endif
 
-LLTrace::BlockTimerStatHandle FT_STRING_FORMAT("String Format");
-
-
 std::string ll_safe_string(const char* in)
 {
 	if(in) return std::string(in);
@@ -1356,7 +1353,7 @@ bool LLStringUtil::formatDatetime(std::string& replacement, std::string token,
 template<> 
 S32 LLStringUtil::format(std::string& s, const format_map_t& substitutions)
 {
-	LL_RECORD_BLOCK_TIME(FT_STRING_FORMAT);
+    LL_PROFILE_ZONE_SCOPED;
 	S32 res = 0;
 
 	std::string output;
@@ -1429,7 +1426,7 @@ S32 LLStringUtil::format(std::string& s, const format_map_t& substitutions)
 template<> 
 S32 LLStringUtil::format(std::string& s, const LLSD& substitutions)
 {
-	LL_RECORD_BLOCK_TIME(FT_STRING_FORMAT);
+    LL_PROFILE_ZONE_SCOPED;
 	S32 res = 0;
 
 	if (!substitutions.isMap()) 
diff --git a/indra/llcommon/llsys.cpp b/indra/llcommon/llsys.cpp
index 6d5d043e8d..306ef05b6d 100644
--- a/indra/llcommon/llsys.cpp
+++ b/indra/llcommon/llsys.cpp
@@ -871,11 +871,9 @@ LLMemoryInfo& LLMemoryInfo::refresh()
 	return *this;
 }
 
-static LLTrace::BlockTimerStatHandle FTM_MEMINFO_LOAD_STATS("MemInfo Load Stats");
-
 LLSD LLMemoryInfo::loadStatsMap()
 {
-	LL_RECORD_BLOCK_TIME(FTM_MEMINFO_LOAD_STATS);
+    LL_PROFILE_ZONE_SCOPED;
 
 	// This implementation is derived from stream() code (as of 2011-06-29).
 	Stats stats;
diff --git a/indra/llcommon/lltrace.cpp b/indra/llcommon/lltrace.cpp
index 54079a4689..f59b207ded 100644
--- a/indra/llcommon/lltrace.cpp
+++ b/indra/llcommon/lltrace.cpp
@@ -61,6 +61,7 @@ TimeBlockTreeNode::TimeBlockTreeNode()
 
 void TimeBlockTreeNode::setParent( BlockTimerStatHandle* parent )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	llassert_always(parent != mBlock);
 	llassert_always(parent != NULL);
 
diff --git a/indra/llcommon/lltrace.h b/indra/llcommon/lltrace.h
index 0d0cd6f581..4051c558a4 100644
--- a/indra/llcommon/lltrace.h
+++ b/indra/llcommon/lltrace.h
@@ -227,6 +227,7 @@ public:
 
 	void setName(const char* name)
 	{
+        LL_PROFILE_ZONE_SCOPED;
 		mName = name;
 		setKey(name);
 	}
@@ -234,12 +235,14 @@ public:
 	/*virtual*/ const char* getUnitLabel() const { return "KB"; }
 
 	StatType<MemAccumulator::AllocationFacet>& allocations() 
-	{ 
+	{
+        LL_PROFILE_ZONE_SCOPED;
 		return static_cast<StatType<MemAccumulator::AllocationFacet>&>(*(StatType<MemAccumulator>*)this);
 	}
 
 	StatType<MemAccumulator::DeallocationFacet>& deallocations() 
-	{ 
+	{
+        LL_PROFILE_ZONE_SCOPED;
 		return static_cast<StatType<MemAccumulator::DeallocationFacet>&>(*(StatType<MemAccumulator>*)this);
 	}
 };
@@ -261,6 +264,7 @@ struct MeasureMem<T, typename T::mem_trackable_tag_t, IS_BYTES>
 {
 	static size_t measureFootprint(const T& value)
 	{
+        LL_PROFILE_ZONE_SCOPED;
 		return sizeof(T) + value.getMemFootprint();
 	}
 };
@@ -270,6 +274,7 @@ struct MeasureMem<T, IS_MEM_TRACKABLE, typename T::is_unit_t>
 {
 	static size_t measureFootprint(const T& value)
 	{
+        LL_PROFILE_ZONE_SCOPED;
 		return U32Bytes(value).value();
 	}
 };
@@ -279,6 +284,7 @@ struct MeasureMem<T*, IS_MEM_TRACKABLE, IS_BYTES>
 {
 	static size_t measureFootprint(const T* value)
 	{
+        LL_PROFILE_ZONE_SCOPED;
 		if (!value)
 		{
 			return 0;
@@ -323,6 +329,7 @@ struct MeasureMem<std::basic_string<T>, IS_MEM_TRACKABLE, IS_BYTES>
 {
 	static size_t measureFootprint(const std::basic_string<T>& value)
 	{
+        LL_PROFILE_ZONE_SCOPED;
 		return value.capacity() * sizeof(T);
 	}
 };
@@ -331,6 +338,7 @@ struct MeasureMem<std::basic_string<T>, IS_MEM_TRACKABLE, IS_BYTES>
 template<typename T>
 inline void claim_alloc(MemStatHandle& measurement, const T& value)
 {
+    LL_PROFILE_ZONE_SCOPED;
 #if LL_TRACE_ENABLED
 	S32 size = MeasureMem<T>::measureFootprint(value);
 	if(size == 0) return;
@@ -343,6 +351,7 @@ inline void claim_alloc(MemStatHandle& measurement, const T& value)
 template<typename T>
 inline void disclaim_alloc(MemStatHandle& measurement, const T& value)
 {
+    LL_PROFILE_ZONE_SCOPED;
 #if LL_TRACE_ENABLED
 	S32 size = MeasureMem<T>::measureFootprint(value);
 	if(size == 0) return;
@@ -352,141 +361,6 @@ inline void disclaim_alloc(MemStatHandle& measurement, const T& value)
 #endif
 }
 
-template<typename DERIVED, size_t ALIGNMENT = LL_DEFAULT_HEAP_ALIGN>
-class MemTrackableNonVirtual
-{
-public:
-	typedef void mem_trackable_tag_t;
-
-	MemTrackableNonVirtual(const char* name)
-#if LL_TRACE_ENABLED
-	:	mMemFootprint(0)
-#endif
-	{
-#if LL_TRACE_ENABLED
-		static bool name_initialized = false;
-		if (!name_initialized)
-		{
-			name_initialized = true;
-			sMemStat.setName(name);
-		}
-#endif
-	}
-
-#if LL_TRACE_ENABLED
-	~MemTrackableNonVirtual()
-	{
-		disclaimMem(mMemFootprint);
-	}
-
-	static MemStatHandle& getMemStatHandle()
-	{
-		return sMemStat;
-	}
-
-	S32 getMemFootprint() const	{ return mMemFootprint; }
-#endif
-
-	void* operator new(size_t size) 
-	{
-#if LL_TRACE_ENABLED
-		claim_alloc(sMemStat, size);
-#endif
-		return ll_aligned_malloc<ALIGNMENT>(size);
-	}
-
-	template<int CUSTOM_ALIGNMENT>
-	static void* aligned_new(size_t size)
-	{
-#if LL_TRACE_ENABLED
-		claim_alloc(sMemStat, size);
-#endif
-		return ll_aligned_malloc<CUSTOM_ALIGNMENT>(size);
-	}
-
-	void operator delete(void* ptr, size_t size)
-	{
-#if LL_TRACE_ENABLED
-		disclaim_alloc(sMemStat, size);
-#endif
-		ll_aligned_free<ALIGNMENT>(ptr);
-	}
-
-	template<int CUSTOM_ALIGNMENT>
-	static void aligned_delete(void* ptr, size_t size)
-	{
-#if LL_TRACE_ENABLED
-		disclaim_alloc(sMemStat, size);
-#endif
-		ll_aligned_free<CUSTOM_ALIGNMENT>(ptr);
-	}
-
-	void* operator new [](size_t size)
-	{
-#if LL_TRACE_ENABLED
-		claim_alloc(sMemStat, size);
-#endif
-		return ll_aligned_malloc<ALIGNMENT>(size);
-	}
-
-	void operator delete[](void* ptr, size_t size)
-	{
-#if LL_TRACE_ENABLED
-		disclaim_alloc(sMemStat, size);
-#endif
-		ll_aligned_free<ALIGNMENT>(ptr);
-	}
-
-	// claim memory associated with other objects/data as our own, adding to our calculated footprint
-	template<typename CLAIM_T>
-	void claimMem(const CLAIM_T& value) const
-	{
-#if LL_TRACE_ENABLED
-		S32 size = MeasureMem<CLAIM_T>::measureFootprint(value);
-		claim_alloc(sMemStat, size);
-		mMemFootprint += size;
-#endif
-	}
-
-	// remove memory we had claimed from our calculated footprint
-	template<typename CLAIM_T>
-	void disclaimMem(const CLAIM_T& value) const
-	{
-#if LL_TRACE_ENABLED
-		S32 size = MeasureMem<CLAIM_T>::measureFootprint(value);
-		disclaim_alloc(sMemStat, size);
-		mMemFootprint -= size;
-#endif
-	}
-
-private:
-#if LL_TRACE_ENABLED
-	// use signed values so that we can temporarily go negative
-	// and reconcile in destructor
-	// NB: this assumes that no single class is responsible for > 2GB of allocations
-	mutable S32 mMemFootprint;
-	
-	static	MemStatHandle	sMemStat;
-#endif
-
-};
-
-#if LL_TRACE_ENABLED
-template<typename DERIVED, size_t ALIGNMENT>
-MemStatHandle MemTrackableNonVirtual<DERIVED, ALIGNMENT>::sMemStat(typeid(MemTrackableNonVirtual<DERIVED, ALIGNMENT>).name());
-#endif
-
-template<typename DERIVED, size_t ALIGNMENT = LL_DEFAULT_HEAP_ALIGN>
-class MemTrackable : public MemTrackableNonVirtual<DERIVED, ALIGNMENT>
-{
-public:
-	MemTrackable(const char* name)
-	:	MemTrackableNonVirtual<DERIVED, ALIGNMENT>(name)
-	{}
-
-	virtual ~MemTrackable()
-	{}
-};
 }
 
 #endif // LL_LLTRACE_H
diff --git a/indra/llcommon/lltraceaccumulators.cpp b/indra/llcommon/lltraceaccumulators.cpp
index b1c23c6fb7..8e9aaee0e6 100644
--- a/indra/llcommon/lltraceaccumulators.cpp
+++ b/indra/llcommon/lltraceaccumulators.cpp
@@ -41,6 +41,7 @@ extern MemStatHandle gTraceMemStat;
 
 AccumulatorBufferGroup::AccumulatorBufferGroup() 
 {
+    LL_PROFILE_ZONE_SCOPED;
 	claim_alloc(gTraceMemStat, mCounts.capacity() * sizeof(CountAccumulator));
 	claim_alloc(gTraceMemStat, mSamples.capacity() * sizeof(SampleAccumulator));
 	claim_alloc(gTraceMemStat, mEvents.capacity() * sizeof(EventAccumulator));
@@ -55,6 +56,7 @@ AccumulatorBufferGroup::AccumulatorBufferGroup(const AccumulatorBufferGroup& oth
 	mStackTimers(other.mStackTimers),
 	mMemStats(other.mMemStats)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	claim_alloc(gTraceMemStat, mCounts.capacity() * sizeof(CountAccumulator));
 	claim_alloc(gTraceMemStat, mSamples.capacity() * sizeof(SampleAccumulator));
 	claim_alloc(gTraceMemStat, mEvents.capacity() * sizeof(EventAccumulator));
@@ -64,6 +66,7 @@ AccumulatorBufferGroup::AccumulatorBufferGroup(const AccumulatorBufferGroup& oth
 
 AccumulatorBufferGroup::~AccumulatorBufferGroup()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	disclaim_alloc(gTraceMemStat, mCounts.capacity() * sizeof(CountAccumulator));
 	disclaim_alloc(gTraceMemStat, mSamples.capacity() * sizeof(SampleAccumulator));
 	disclaim_alloc(gTraceMemStat, mEvents.capacity() * sizeof(EventAccumulator));
@@ -73,6 +76,7 @@ AccumulatorBufferGroup::~AccumulatorBufferGroup()
 
 void AccumulatorBufferGroup::handOffTo(AccumulatorBufferGroup& other)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	other.mCounts.reset(&mCounts);
 	other.mSamples.reset(&mSamples);
 	other.mEvents.reset(&mEvents);
@@ -82,6 +86,7 @@ void AccumulatorBufferGroup::handOffTo(AccumulatorBufferGroup& other)
 
 void AccumulatorBufferGroup::makeCurrent()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mCounts.makeCurrent();
 	mSamples.makeCurrent();
 	mEvents.makeCurrent();
@@ -104,6 +109,7 @@ void AccumulatorBufferGroup::makeCurrent()
 //static
 void AccumulatorBufferGroup::clearCurrent()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	AccumulatorBuffer<CountAccumulator>::clearCurrent();	
 	AccumulatorBuffer<SampleAccumulator>::clearCurrent();
 	AccumulatorBuffer<EventAccumulator>::clearCurrent();
@@ -118,6 +124,7 @@ bool AccumulatorBufferGroup::isCurrent() const
 
 void AccumulatorBufferGroup::append( const AccumulatorBufferGroup& other )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mCounts.addSamples(other.mCounts, SEQUENTIAL);
 	mSamples.addSamples(other.mSamples, SEQUENTIAL);
 	mEvents.addSamples(other.mEvents, SEQUENTIAL);
@@ -127,6 +134,7 @@ void AccumulatorBufferGroup::append( const AccumulatorBufferGroup& other )
 
 void AccumulatorBufferGroup::merge( const AccumulatorBufferGroup& other)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mCounts.addSamples(other.mCounts, NON_SEQUENTIAL);
 	mSamples.addSamples(other.mSamples, NON_SEQUENTIAL);
 	mEvents.addSamples(other.mEvents, NON_SEQUENTIAL);
@@ -137,6 +145,7 @@ void AccumulatorBufferGroup::merge( const AccumulatorBufferGroup& other)
 
 void AccumulatorBufferGroup::reset(AccumulatorBufferGroup* other)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mCounts.reset(other ? &other->mCounts : NULL);
 	mSamples.reset(other ? &other->mSamples : NULL);
 	mEvents.reset(other ? &other->mEvents : NULL);
@@ -146,6 +155,7 @@ void AccumulatorBufferGroup::reset(AccumulatorBufferGroup* other)
 
 void AccumulatorBufferGroup::sync()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	if (isCurrent())
 	{
 		F64SecondsImplicit time_stamp = LLTimer::getTotalSeconds();
@@ -190,7 +200,7 @@ F64 SampleAccumulator::mergeSumsOfSquares(const SampleAccumulator& a, const Samp
 
 void SampleAccumulator::addSamples( const SampleAccumulator& other, EBufferAppendType append_type )
 {
-	if (append_type == NON_SEQUENTIAL)
+    if (append_type == NON_SEQUENTIAL)
 	{
 		return;
 	}
@@ -289,7 +299,7 @@ void EventAccumulator::addSamples( const EventAccumulator& other, EBufferAppendT
 
 void EventAccumulator::reset( const EventAccumulator* other )
 {
-	mNumSamples = 0;
+    mNumSamples = 0;
 	mSum = 0;
 	mMin = F32(NaN);
 	mMax = F32(NaN);
diff --git a/indra/llcommon/lltraceaccumulators.h b/indra/llcommon/lltraceaccumulators.h
index 8eb5338a2a..b183fcd14a 100644
--- a/indra/llcommon/lltraceaccumulators.h
+++ b/indra/llcommon/lltraceaccumulators.h
@@ -66,6 +66,7 @@ namespace LLTrace
 			: mStorageSize(0),
 			mStorage(NULL)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			const AccumulatorBuffer& other = *getDefaultBuffer();
 			resize(sNextStorageSlot);
 			for (S32 i = 0; i < sNextStorageSlot; i++)
@@ -76,6 +77,7 @@ namespace LLTrace
 
 		~AccumulatorBuffer()
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			if (isCurrent())
 			{
 				LLThreadLocalSingletonPointer<ACCUMULATOR>::setInstance(NULL);
@@ -98,6 +100,7 @@ namespace LLTrace
 			: mStorageSize(0),
 			mStorage(NULL)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			resize(sNextStorageSlot);
 			for (S32 i = 0; i < sNextStorageSlot; i++)
 			{
@@ -107,6 +110,7 @@ namespace LLTrace
 
 		void addSamples(const AccumulatorBuffer<ACCUMULATOR>& other, EBufferAppendType append_type)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			llassert(mStorageSize >= sNextStorageSlot && other.mStorageSize >= sNextStorageSlot);
 			for (size_t i = 0; i < sNextStorageSlot; i++)
 			{
@@ -116,6 +120,7 @@ namespace LLTrace
 
 		void copyFrom(const AccumulatorBuffer<ACCUMULATOR>& other)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			llassert(mStorageSize >= sNextStorageSlot && other.mStorageSize >= sNextStorageSlot);
 			for (size_t i = 0; i < sNextStorageSlot; i++)
 			{
@@ -125,6 +130,7 @@ namespace LLTrace
 
 		void reset(const AccumulatorBuffer<ACCUMULATOR>* other = NULL)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			llassert(mStorageSize >= sNextStorageSlot);
 			for (size_t i = 0; i < sNextStorageSlot; i++)
 			{
@@ -134,6 +140,7 @@ namespace LLTrace
 
 		void sync(F64SecondsImplicit time_stamp)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			llassert(mStorageSize >= sNextStorageSlot);
 			for (size_t i = 0; i < sNextStorageSlot; i++)
 			{
@@ -153,12 +160,13 @@ namespace LLTrace
 
 		static void clearCurrent()
 		{
-			LLThreadLocalSingletonPointer<ACCUMULATOR>::setInstance(NULL);
+            LLThreadLocalSingletonPointer<ACCUMULATOR>::setInstance(NULL);
 		}
 
 		// NOTE: this is not thread-safe.  We assume that slots are reserved in the main thread before any child threads are spawned
 		size_t reserveSlot()
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			size_t next_slot = sNextStorageSlot++;
 			if (next_slot >= mStorageSize)
 			{
@@ -172,6 +180,7 @@ namespace LLTrace
 
 		void resize(size_t new_size)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			if (new_size <= mStorageSize) return;
 
 			ACCUMULATOR* old_storage = mStorage;
@@ -212,6 +221,7 @@ namespace LLTrace
 
 		static self_t* getDefaultBuffer()
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			static bool sInitialized = false;
 			if (!sInitialized)
 			{
@@ -326,6 +336,7 @@ namespace LLTrace
 
 		void sample(F64 value)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			F64SecondsImplicit time_stamp = LLTimer::getTotalSeconds();
 
 			// store effect of last value
@@ -444,9 +455,9 @@ namespace LLTrace
 		S32	mNumSamples;
 	};
 
-	class TimeBlockAccumulator
+	class alignas(32) TimeBlockAccumulator
 	{
-	public:
+    public:
 		typedef F64Seconds value_t;
 		static F64Seconds getDefaultValue() { return F64Seconds(0); }
 
@@ -539,6 +550,7 @@ namespace LLTrace
 
 		void addSamples(const MemAccumulator& other, EBufferAppendType append_type)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			mAllocations.addSamples(other.mAllocations, append_type);
 			mDeallocations.addSamples(other.mDeallocations, append_type);
 
@@ -557,6 +569,7 @@ namespace LLTrace
 
 		void reset(const MemAccumulator* other)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			mSize.reset(other ? &other->mSize : NULL);
 			mAllocations.reset(other ? &other->mAllocations : NULL);
 			mDeallocations.reset(other ? &other->mDeallocations : NULL);
diff --git a/indra/llcommon/lltracerecording.cpp b/indra/llcommon/lltracerecording.cpp
index 3094b627a2..c72a64d086 100644
--- a/indra/llcommon/lltracerecording.cpp
+++ b/indra/llcommon/lltracerecording.cpp
@@ -50,6 +50,7 @@ Recording::Recording(EPlayState state)
 :	mElapsedSeconds(0),
 	mActiveBuffers(NULL)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	claim_alloc(gTraceMemStat, this);
 	mBuffers = new AccumulatorBufferGroup();
 	claim_alloc(gTraceMemStat, mBuffers);
@@ -59,12 +60,14 @@ Recording::Recording(EPlayState state)
 Recording::Recording( const Recording& other )
 :	mActiveBuffers(NULL)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	claim_alloc(gTraceMemStat, this);
 	*this = other;
 }
 
 Recording& Recording::operator = (const Recording& other)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	// this will allow us to seamlessly start without affecting any data we've acquired from other
 	setPlayState(PAUSED);
 
@@ -85,6 +88,7 @@ Recording& Recording::operator = (const Recording& other)
 
 Recording::~Recording()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	disclaim_alloc(gTraceMemStat, this);
 	disclaim_alloc(gTraceMemStat, mBuffers);
 
@@ -103,6 +107,7 @@ void Recording::update()
 #if LL_TRACE_ENABLED
 	if (isStarted())
 	{
+        LL_PROFILE_ZONE_SCOPED;
 		mElapsedSeconds += mSamplingTimer.getElapsedTimeF64();
 
 		// must have 
@@ -123,6 +128,7 @@ void Recording::update()
 
 void Recording::handleReset()
 {
+    LL_PROFILE_ZONE_SCOPED;
 #if LL_TRACE_ENABLED
 	mBuffers.write()->reset();
 
@@ -133,6 +139,7 @@ void Recording::handleReset()
 
 void Recording::handleStart()
 {
+    LL_PROFILE_ZONE_SCOPED;
 #if LL_TRACE_ENABLED
 	mSamplingTimer.reset();
 	mBuffers.setStayUnique(true);
@@ -144,6 +151,7 @@ void Recording::handleStart()
 
 void Recording::handleStop()
 {
+    LL_PROFILE_ZONE_SCOPED;
 #if LL_TRACE_ENABLED
 	mElapsedSeconds += mSamplingTimer.getElapsedTimeF64();
 	// must have thread recorder running on this thread
@@ -273,7 +281,7 @@ F64Kilobytes Recording::getMean(const StatType<MemAccumulator>& stat)
 
 F64Kilobytes Recording::getMax(const StatType<MemAccumulator>& stat)
 {
-	update();
+    update();
 	const MemAccumulator& accumulator = mBuffers->mMemStats[stat.getIndex()];
 	const MemAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mMemStats[stat.getIndex()] : NULL;
 	return F64Bytes(llmax(accumulator.mSize.getMax(), active_accumulator && active_accumulator->mSize.hasValue() ? active_accumulator->mSize.getMax() : F32_MIN));
@@ -281,7 +289,7 @@ F64Kilobytes Recording::getMax(const StatType<MemAccumulator>& stat)
 
 F64Kilobytes Recording::getStandardDeviation(const StatType<MemAccumulator>& stat)
 {
-	update();
+    update();
 	const MemAccumulator& accumulator = mBuffers->mMemStats[stat.getIndex()];
 	const MemAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mMemStats[stat.getIndex()] : NULL;
 	if (active_accumulator && active_accumulator->hasValue())
@@ -297,7 +305,7 @@ F64Kilobytes Recording::getStandardDeviation(const StatType<MemAccumulator>& sta
 
 F64Kilobytes Recording::getLastValue(const StatType<MemAccumulator>& stat)
 {
-	update();
+    update();
 	const MemAccumulator& accumulator = mBuffers->mMemStats[stat.getIndex()];
 	const MemAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mMemStats[stat.getIndex()] : NULL;
 	return F64Bytes(active_accumulator ? active_accumulator->mSize.getLastValue() : accumulator.mSize.getLastValue());
@@ -305,7 +313,7 @@ F64Kilobytes Recording::getLastValue(const StatType<MemAccumulator>& stat)
 
 bool Recording::hasValue(const StatType<MemAccumulator::AllocationFacet>& stat)
 {
-	update();
+    update();
 	const MemAccumulator& accumulator = mBuffers->mMemStats[stat.getIndex()];
 	const MemAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mMemStats[stat.getIndex()] : NULL;
 	return accumulator.mAllocations.hasValue() || (active_accumulator ? active_accumulator->mAllocations.hasValue() : false);
@@ -313,7 +321,7 @@ bool Recording::hasValue(const StatType<MemAccumulator::AllocationFacet>& stat)
 
 F64Kilobytes Recording::getSum(const StatType<MemAccumulator::AllocationFacet>& stat)
 {
-	update();
+    update();
 	const MemAccumulator& accumulator = mBuffers->mMemStats[stat.getIndex()];
 	const MemAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mMemStats[stat.getIndex()] : NULL;
 	return F64Bytes(accumulator.mAllocations.getSum() + (active_accumulator ? active_accumulator->mAllocations.getSum() : 0));
@@ -321,7 +329,7 @@ F64Kilobytes Recording::getSum(const StatType<MemAccumulator::AllocationFacet>&
 
 F64Kilobytes Recording::getPerSec(const StatType<MemAccumulator::AllocationFacet>& stat)
 {
-	update();
+    update();
 	const MemAccumulator& accumulator = mBuffers->mMemStats[stat.getIndex()];
 	const MemAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mMemStats[stat.getIndex()] : NULL;
 	return F64Bytes((accumulator.mAllocations.getSum() + (active_accumulator ? active_accumulator->mAllocations.getSum() : 0)) / mElapsedSeconds.value());
@@ -329,7 +337,7 @@ F64Kilobytes Recording::getPerSec(const StatType<MemAccumulator::AllocationFacet
 
 S32 Recording::getSampleCount(const StatType<MemAccumulator::AllocationFacet>& stat)
 {
-	update();
+    update();
 	const MemAccumulator& accumulator = mBuffers->mMemStats[stat.getIndex()];
 	const MemAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mMemStats[stat.getIndex()] : NULL;
 	return accumulator.mAllocations.getSampleCount() + (active_accumulator ? active_accumulator->mAllocations.getSampleCount() : 0);
@@ -337,7 +345,7 @@ S32 Recording::getSampleCount(const StatType<MemAccumulator::AllocationFacet>& s
 
 bool Recording::hasValue(const StatType<MemAccumulator::DeallocationFacet>& stat)
 {
-	update();
+    update();
 	const MemAccumulator& accumulator = mBuffers->mMemStats[stat.getIndex()];
 	const MemAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mMemStats[stat.getIndex()] : NULL;
 	return accumulator.mDeallocations.hasValue() || (active_accumulator ? active_accumulator->mDeallocations.hasValue() : false);
@@ -346,7 +354,7 @@ bool Recording::hasValue(const StatType<MemAccumulator::DeallocationFacet>& stat
 
 F64Kilobytes Recording::getSum(const StatType<MemAccumulator::DeallocationFacet>& stat)
 {
-	update();
+    update();
 	const MemAccumulator& accumulator = mBuffers->mMemStats[stat.getIndex()];
 	const MemAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mMemStats[stat.getIndex()] : NULL;
 	return F64Bytes(accumulator.mDeallocations.getSum() + (active_accumulator ? active_accumulator->mDeallocations.getSum() : 0));
@@ -354,7 +362,7 @@ F64Kilobytes Recording::getSum(const StatType<MemAccumulator::DeallocationFacet>
 
 F64Kilobytes Recording::getPerSec(const StatType<MemAccumulator::DeallocationFacet>& stat)
 {
-	update();
+    update();
 	const MemAccumulator& accumulator = mBuffers->mMemStats[stat.getIndex()];
 	const MemAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mMemStats[stat.getIndex()] : NULL;
 	return F64Bytes((accumulator.mDeallocations.getSum() + (active_accumulator ? active_accumulator->mDeallocations.getSum() : 0)) / mElapsedSeconds.value());
@@ -362,7 +370,7 @@ F64Kilobytes Recording::getPerSec(const StatType<MemAccumulator::DeallocationFac
 
 S32 Recording::getSampleCount(const StatType<MemAccumulator::DeallocationFacet>& stat)
 {
-	update();
+    update();
 	const MemAccumulator& accumulator = mBuffers->mMemStats[stat.getIndex()];
 	const MemAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mMemStats[stat.getIndex()] : NULL;
 	return accumulator.mDeallocations.getSampleCount() + (active_accumulator ? active_accumulator->mDeallocations.getSampleCount() : 0);
@@ -370,7 +378,7 @@ S32 Recording::getSampleCount(const StatType<MemAccumulator::DeallocationFacet>&
 
 bool Recording::hasValue(const StatType<CountAccumulator>& stat)
 {
-	update();
+    update();
 	const CountAccumulator& accumulator = mBuffers->mCounts[stat.getIndex()];
 	const CountAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mCounts[stat.getIndex()] : NULL;
 	return accumulator.hasValue() || (active_accumulator ? active_accumulator->hasValue() : false);
@@ -378,7 +386,7 @@ bool Recording::hasValue(const StatType<CountAccumulator>& stat)
 
 F64 Recording::getSum(const StatType<CountAccumulator>& stat)
 {
-	update();
+    update();
 	const CountAccumulator& accumulator = mBuffers->mCounts[stat.getIndex()];
 	const CountAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mCounts[stat.getIndex()] : NULL;
 	return accumulator.getSum() + (active_accumulator ? active_accumulator->getSum() : 0);
@@ -386,7 +394,7 @@ F64 Recording::getSum(const StatType<CountAccumulator>& stat)
 
 F64 Recording::getPerSec( const StatType<CountAccumulator>& stat )
 {
-	update();
+    update();
 	const CountAccumulator& accumulator = mBuffers->mCounts[stat.getIndex()];
 	const CountAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mCounts[stat.getIndex()] : NULL;
 	F64 sum = accumulator.getSum() + (active_accumulator ? active_accumulator->getSum() : 0);
@@ -395,7 +403,7 @@ F64 Recording::getPerSec( const StatType<CountAccumulator>& stat )
 
 S32 Recording::getSampleCount( const StatType<CountAccumulator>& stat )
 {
-	update();
+    update();
 	const CountAccumulator& accumulator = mBuffers->mCounts[stat.getIndex()];
 	const CountAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mCounts[stat.getIndex()] : NULL;
 	return accumulator.getSampleCount() + (active_accumulator ? active_accumulator->getSampleCount() : 0);
@@ -403,7 +411,7 @@ S32 Recording::getSampleCount( const StatType<CountAccumulator>& stat )
 
 bool Recording::hasValue(const StatType<SampleAccumulator>& stat)
 {
-	update();
+    update();
 	const SampleAccumulator& accumulator = mBuffers->mSamples[stat.getIndex()];
 	const SampleAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mSamples[stat.getIndex()] : NULL;
 	return accumulator.hasValue() || (active_accumulator && active_accumulator->hasValue());
@@ -411,7 +419,7 @@ bool Recording::hasValue(const StatType<SampleAccumulator>& stat)
 
 F64 Recording::getMin( const StatType<SampleAccumulator>& stat )
 {
-	update();
+    update();
 	const SampleAccumulator& accumulator = mBuffers->mSamples[stat.getIndex()];
 	const SampleAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mSamples[stat.getIndex()] : NULL;
 	return llmin(accumulator.getMin(), active_accumulator && active_accumulator->hasValue() ? active_accumulator->getMin() : F32_MAX);
@@ -419,7 +427,7 @@ F64 Recording::getMin( const StatType<SampleAccumulator>& stat )
 
 F64 Recording::getMax( const StatType<SampleAccumulator>& stat )
 {
-	update();
+    update();
 	const SampleAccumulator& accumulator = mBuffers->mSamples[stat.getIndex()];
 	const SampleAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mSamples[stat.getIndex()] : NULL;
 	return llmax(accumulator.getMax(), active_accumulator && active_accumulator->hasValue() ? active_accumulator->getMax() : F32_MIN);
@@ -427,7 +435,7 @@ F64 Recording::getMax( const StatType<SampleAccumulator>& stat )
 
 F64 Recording::getMean( const StatType<SampleAccumulator>& stat )
 {
-	update();
+    update();
 	const SampleAccumulator& accumulator = mBuffers->mSamples[stat.getIndex()];
 	const SampleAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mSamples[stat.getIndex()] : NULL;
 	if (active_accumulator && active_accumulator->hasValue())
@@ -448,7 +456,7 @@ F64 Recording::getMean( const StatType<SampleAccumulator>& stat )
 
 F64 Recording::getStandardDeviation( const StatType<SampleAccumulator>& stat )
 {
-	update();
+    update();
 	const SampleAccumulator& accumulator = mBuffers->mSamples[stat.getIndex()];
 	const SampleAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mSamples[stat.getIndex()] : NULL;
 
@@ -465,7 +473,7 @@ F64 Recording::getStandardDeviation( const StatType<SampleAccumulator>& stat )
 
 F64 Recording::getLastValue( const StatType<SampleAccumulator>& stat )
 {
-	update();
+    update();
 	const SampleAccumulator& accumulator = mBuffers->mSamples[stat.getIndex()];
 	const SampleAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mSamples[stat.getIndex()] : NULL;
 	return (active_accumulator && active_accumulator->hasValue() ? active_accumulator->getLastValue() : accumulator.getLastValue());
@@ -473,7 +481,7 @@ F64 Recording::getLastValue( const StatType<SampleAccumulator>& stat )
 
 S32 Recording::getSampleCount( const StatType<SampleAccumulator>& stat )
 {
-	update();
+    update();
 	const SampleAccumulator& accumulator = mBuffers->mSamples[stat.getIndex()];
 	const SampleAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mSamples[stat.getIndex()] : NULL;
 	return accumulator.getSampleCount() + (active_accumulator && active_accumulator->hasValue() ? active_accumulator->getSampleCount() : 0);
@@ -481,7 +489,7 @@ S32 Recording::getSampleCount( const StatType<SampleAccumulator>& stat )
 
 bool Recording::hasValue(const StatType<EventAccumulator>& stat)
 {
-	update();
+    update();
 	const EventAccumulator& accumulator = mBuffers->mEvents[stat.getIndex()];
 	const EventAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mEvents[stat.getIndex()] : NULL;
 	return accumulator.hasValue() || (active_accumulator && active_accumulator->hasValue());
@@ -489,7 +497,7 @@ bool Recording::hasValue(const StatType<EventAccumulator>& stat)
 
 F64 Recording::getSum( const StatType<EventAccumulator>& stat)
 {
-	update();
+    update();
 	const EventAccumulator& accumulator = mBuffers->mEvents[stat.getIndex()];
 	const EventAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mEvents[stat.getIndex()] : NULL;
 	return (F64)(accumulator.getSum() + (active_accumulator && active_accumulator->hasValue() ? active_accumulator->getSum() : 0));
@@ -497,7 +505,7 @@ F64 Recording::getSum( const StatType<EventAccumulator>& stat)
 
 F64 Recording::getMin( const StatType<EventAccumulator>& stat )
 {
-	update();
+    update();
 	const EventAccumulator& accumulator = mBuffers->mEvents[stat.getIndex()];
 	const EventAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mEvents[stat.getIndex()] : NULL;
 	return llmin(accumulator.getMin(), active_accumulator && active_accumulator->hasValue() ? active_accumulator->getMin() : F32_MAX);
@@ -505,7 +513,7 @@ F64 Recording::getMin( const StatType<EventAccumulator>& stat )
 
 F64 Recording::getMax( const StatType<EventAccumulator>& stat )
 {
-	update();
+    update();
 	const EventAccumulator& accumulator = mBuffers->mEvents[stat.getIndex()];
 	const EventAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mEvents[stat.getIndex()] : NULL;
 	return llmax(accumulator.getMax(), active_accumulator && active_accumulator->hasValue() ? active_accumulator->getMax() : F32_MIN);
@@ -513,7 +521,7 @@ F64 Recording::getMax( const StatType<EventAccumulator>& stat )
 
 F64 Recording::getMean( const StatType<EventAccumulator>& stat )
 {
-	update();
+    update();
 	const EventAccumulator& accumulator = mBuffers->mEvents[stat.getIndex()];
 	const EventAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mEvents[stat.getIndex()] : NULL;
 	if (active_accumulator && active_accumulator->hasValue())
@@ -534,7 +542,7 @@ F64 Recording::getMean( const StatType<EventAccumulator>& stat )
 
 F64 Recording::getStandardDeviation( const StatType<EventAccumulator>& stat )
 {
-	update();
+    update();
 	const EventAccumulator& accumulator = mBuffers->mEvents[stat.getIndex()];
 	const EventAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mEvents[stat.getIndex()] : NULL;
 
@@ -551,7 +559,7 @@ F64 Recording::getStandardDeviation( const StatType<EventAccumulator>& stat )
 
 F64 Recording::getLastValue( const StatType<EventAccumulator>& stat )
 {
-	update();
+    update();
 	const EventAccumulator& accumulator = mBuffers->mEvents[stat.getIndex()];
 	const EventAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mEvents[stat.getIndex()] : NULL;
 	return active_accumulator ? active_accumulator->getLastValue() : accumulator.getLastValue();
@@ -559,7 +567,7 @@ F64 Recording::getLastValue( const StatType<EventAccumulator>& stat )
 
 S32 Recording::getSampleCount( const StatType<EventAccumulator>& stat )
 {
-	update();
+    update();
 	const EventAccumulator& accumulator = mBuffers->mEvents[stat.getIndex()];
 	const EventAccumulator* active_accumulator = mActiveBuffers ? &mActiveBuffers->mEvents[stat.getIndex()] : NULL;
 	return accumulator.getSampleCount() + (active_accumulator ? active_accumulator->getSampleCount() : 0);
@@ -575,17 +583,20 @@ PeriodicRecording::PeriodicRecording( S32 num_periods, EPlayState state)
 	mNumRecordedPeriods(0),
 	mRecordingPeriods(num_periods ? num_periods : 1)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	setPlayState(state);
 	claim_alloc(gTraceMemStat, this);
 }
 
 PeriodicRecording::~PeriodicRecording()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	disclaim_alloc(gTraceMemStat, this);
 }
 
 void PeriodicRecording::nextPeriod()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	if (mAutoResize)
 	{
 		mRecordingPeriods.push_back(Recording());
@@ -600,6 +611,7 @@ void PeriodicRecording::nextPeriod()
 
 void PeriodicRecording::appendRecording(Recording& recording)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	getCurRecording().appendRecording(recording);
 	nextPeriod();
 }
@@ -607,6 +619,7 @@ void PeriodicRecording::appendRecording(Recording& recording)
 
 void PeriodicRecording::appendPeriodicRecording( PeriodicRecording& other )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	if (other.mRecordingPeriods.empty()) return;
 
 	getCurRecording().update();
@@ -680,6 +693,7 @@ void PeriodicRecording::appendPeriodicRecording( PeriodicRecording& other )
 
 F64Seconds PeriodicRecording::getDuration() const
 {
+    LL_PROFILE_ZONE_SCOPED;
 	F64Seconds duration;
 	S32 num_periods = mRecordingPeriods.size();
 	for (S32 i = 1; i <= num_periods; i++)
@@ -693,6 +707,7 @@ F64Seconds PeriodicRecording::getDuration() const
 
 LLTrace::Recording PeriodicRecording::snapshotCurRecording() const
 {
+    LL_PROFILE_ZONE_SCOPED;
 	Recording recording_copy(getCurRecording());
 	recording_copy.stop();
 	return recording_copy;
@@ -735,16 +750,19 @@ const Recording& PeriodicRecording::getPrevRecording( S32 offset ) const
 
 void PeriodicRecording::handleStart()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	getCurRecording().start();
 }
 
 void PeriodicRecording::handleStop()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	getCurRecording().pause();
 }
 
 void PeriodicRecording::handleReset()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	getCurRecording().stop();
 
 	if (mAutoResize)
@@ -768,11 +786,13 @@ void PeriodicRecording::handleReset()
 
 void PeriodicRecording::handleSplitTo(PeriodicRecording& other)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	getCurRecording().splitTo(other.getCurRecording());
 }
 
 F64 PeriodicRecording::getPeriodMin( const StatType<EventAccumulator>& stat, S32 num_periods /*= S32_MAX*/ )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	bool has_value = false;
@@ -794,6 +814,7 @@ F64 PeriodicRecording::getPeriodMin( const StatType<EventAccumulator>& stat, S32
 
 F64 PeriodicRecording::getPeriodMax( const StatType<EventAccumulator>& stat, S32 num_periods /*= S32_MAX*/ )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	bool has_value = false;
@@ -816,6 +837,7 @@ F64 PeriodicRecording::getPeriodMax( const StatType<EventAccumulator>& stat, S32
 // calculates means using aggregates per period
 F64 PeriodicRecording::getPeriodMean( const StatType<EventAccumulator>& stat, S32 num_periods /*= S32_MAX*/ )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	F64 mean = 0;
@@ -839,6 +861,7 @@ F64 PeriodicRecording::getPeriodMean( const StatType<EventAccumulator>& stat, S3
 
 F64 PeriodicRecording::getPeriodStandardDeviation( const StatType<EventAccumulator>& stat, S32 num_periods /*= S32_MAX*/ )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	F64 period_mean = getPeriodMean(stat, num_periods);
@@ -863,6 +886,7 @@ F64 PeriodicRecording::getPeriodStandardDeviation( const StatType<EventAccumulat
 
 F64 PeriodicRecording::getPeriodMin( const StatType<SampleAccumulator>& stat, S32 num_periods /*= S32_MAX*/ )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	bool has_value = false;
@@ -884,6 +908,7 @@ F64 PeriodicRecording::getPeriodMin( const StatType<SampleAccumulator>& stat, S3
 
 F64 PeriodicRecording::getPeriodMax(const StatType<SampleAccumulator>& stat, S32 num_periods /*= S32_MAX*/)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	bool has_value = false;
@@ -906,6 +931,7 @@ F64 PeriodicRecording::getPeriodMax(const StatType<SampleAccumulator>& stat, S32
 
 F64 PeriodicRecording::getPeriodMean( const StatType<SampleAccumulator>& stat, S32 num_periods /*= S32_MAX*/ )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	S32 valid_period_count = 0;
@@ -928,6 +954,7 @@ F64 PeriodicRecording::getPeriodMean( const StatType<SampleAccumulator>& stat, S
 
 F64 PeriodicRecording::getPeriodStandardDeviation( const StatType<SampleAccumulator>& stat, S32 num_periods /*= S32_MAX*/ )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	F64 period_mean = getPeriodMean(stat, num_periods);
@@ -953,6 +980,7 @@ F64 PeriodicRecording::getPeriodStandardDeviation( const StatType<SampleAccumula
 
 F64Kilobytes PeriodicRecording::getPeriodMin( const StatType<MemAccumulator>& stat, S32 num_periods /*= S32_MAX*/ )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	F64Kilobytes min_val(std::numeric_limits<F64>::max());
@@ -972,6 +1000,7 @@ F64Kilobytes PeriodicRecording::getPeriodMin(const MemStatHandle& stat, S32 num_
 
 F64Kilobytes PeriodicRecording::getPeriodMax(const StatType<MemAccumulator>& stat, S32 num_periods /*= S32_MAX*/)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	F64Kilobytes max_val(0.0);
@@ -991,6 +1020,7 @@ F64Kilobytes PeriodicRecording::getPeriodMax(const MemStatHandle& stat, S32 num_
 
 F64Kilobytes PeriodicRecording::getPeriodMean( const StatType<MemAccumulator>& stat, S32 num_periods /*= S32_MAX*/ )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	F64Kilobytes mean(0);
@@ -1011,6 +1041,7 @@ F64Kilobytes PeriodicRecording::getPeriodMean(const MemStatHandle& stat, S32 num
 
 F64Kilobytes PeriodicRecording::getPeriodStandardDeviation( const StatType<MemAccumulator>& stat, S32 num_periods /*= S32_MAX*/ )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 	F64Kilobytes period_mean = getPeriodMean(stat, num_periods);
@@ -1044,6 +1075,7 @@ F64Kilobytes PeriodicRecording::getPeriodStandardDeviation(const MemStatHandle&
 
 void ExtendableRecording::extend()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	// push the data back to accepted recording
 	mAcceptedRecording.appendRecording(mPotentialRecording);
 	// flush data, so we can start from scratch
@@ -1052,22 +1084,26 @@ void ExtendableRecording::extend()
 
 void ExtendableRecording::handleStart()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mPotentialRecording.start();
 }
 
 void ExtendableRecording::handleStop()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mPotentialRecording.pause();
 }
 
 void ExtendableRecording::handleReset()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mAcceptedRecording.reset();
 	mPotentialRecording.reset();
 }
 
 void ExtendableRecording::handleSplitTo(ExtendableRecording& other)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mPotentialRecording.splitTo(other.mPotentialRecording);
 }
 
@@ -1084,6 +1120,7 @@ ExtendablePeriodicRecording::ExtendablePeriodicRecording()
 
 void ExtendablePeriodicRecording::extend()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	// push the data back to accepted recording
 	mAcceptedRecording.appendPeriodicRecording(mPotentialRecording);
 	// flush data, so we can start from scratch
@@ -1093,22 +1130,26 @@ void ExtendablePeriodicRecording::extend()
 
 void ExtendablePeriodicRecording::handleStart()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mPotentialRecording.start();
 }
 
 void ExtendablePeriodicRecording::handleStop()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mPotentialRecording.pause();
 }
 
 void ExtendablePeriodicRecording::handleReset()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mAcceptedRecording.reset();
 	mPotentialRecording.reset();
 }
 
 void ExtendablePeriodicRecording::handleSplitTo(ExtendablePeriodicRecording& other)
 {
+    LL_PROFILE_ZONE_SCOPED;
 	mPotentialRecording.splitTo(other.mPotentialRecording);
 }
 
@@ -1123,6 +1164,7 @@ PeriodicRecording& get_frame_recording()
 
 void LLStopWatchControlsMixinCommon::start()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	switch (mPlayState)
 	{
 	case STOPPED:
@@ -1144,6 +1186,7 @@ void LLStopWatchControlsMixinCommon::start()
 
 void LLStopWatchControlsMixinCommon::stop()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	switch (mPlayState)
 	{
 	case STOPPED:
@@ -1163,6 +1206,7 @@ void LLStopWatchControlsMixinCommon::stop()
 
 void LLStopWatchControlsMixinCommon::pause()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	switch (mPlayState)
 	{
 	case STOPPED:
@@ -1182,6 +1226,7 @@ void LLStopWatchControlsMixinCommon::pause()
 
 void LLStopWatchControlsMixinCommon::unpause()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	switch (mPlayState)
 	{
 	case STOPPED:
@@ -1201,6 +1246,7 @@ void LLStopWatchControlsMixinCommon::unpause()
 
 void LLStopWatchControlsMixinCommon::resume()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	switch (mPlayState)
 	{
 	case STOPPED:
@@ -1221,6 +1267,7 @@ void LLStopWatchControlsMixinCommon::resume()
 
 void LLStopWatchControlsMixinCommon::restart()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	switch (mPlayState)
 	{
 	case STOPPED:
@@ -1244,11 +1291,13 @@ void LLStopWatchControlsMixinCommon::restart()
 
 void LLStopWatchControlsMixinCommon::reset()
 {
+    LL_PROFILE_ZONE_SCOPED;
 	handleReset();
 }
 
 void LLStopWatchControlsMixinCommon::setPlayState( EPlayState state )
 {
+    LL_PROFILE_ZONE_SCOPED;
 	switch(state)
 	{
 	case STOPPED:
diff --git a/indra/llcommon/lltracerecording.h b/indra/llcommon/lltracerecording.h
index d0b4a842a6..6715104613 100644
--- a/indra/llcommon/lltracerecording.h
+++ b/indra/llcommon/lltracerecording.h
@@ -355,6 +355,7 @@ namespace LLTrace
 		template <typename T>
 		S32 getSampleCount(const StatType<T>& stat, S32 num_periods = S32_MAX)
         {
+            LL_PROFILE_ZONE_SCOPED;
 			num_periods = llmin(num_periods, getNumRecordedPeriods());
 
             S32 num_samples = 0;
@@ -374,6 +375,7 @@ namespace LLTrace
 		template <typename T>
 		typename T::value_t getPeriodMin(const StatType<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 			bool has_value = false;
@@ -396,6 +398,7 @@ namespace LLTrace
 		template<typename T>
 		T getPeriodMin(const CountStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return T(getPeriodMin(static_cast<const StatType<CountAccumulator>&>(stat), num_periods));
 		}
 
@@ -403,6 +406,7 @@ namespace LLTrace
 		template<typename T>
 		T getPeriodMin(const SampleStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return T(getPeriodMin(static_cast<const StatType<SampleAccumulator>&>(stat), num_periods));
 		}
 
@@ -410,6 +414,7 @@ namespace LLTrace
 		template<typename T>
 		T getPeriodMin(const EventStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return T(getPeriodMin(static_cast<const StatType<EventAccumulator>&>(stat), num_periods));
 		}
 
@@ -419,6 +424,7 @@ namespace LLTrace
 		template <typename T>
 		typename RelatedTypes<typename T::value_t>::fractional_t getPeriodMinPerSec(const StatType<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 			typename RelatedTypes<typename T::value_t>::fractional_t min_val(std::numeric_limits<F64>::max());
@@ -433,6 +439,7 @@ namespace LLTrace
 		template<typename T>
 		typename RelatedTypes<T>::fractional_t getPeriodMinPerSec(const CountStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return typename RelatedTypes<T>::fractional_t(getPeriodMinPerSec(static_cast<const StatType<CountAccumulator>&>(stat), num_periods));
 		}
 
@@ -444,6 +451,7 @@ namespace LLTrace
 		template <typename T>
 		typename T::value_t getPeriodMax(const StatType<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 			bool has_value = false;
@@ -466,6 +474,7 @@ namespace LLTrace
 		template<typename T>
 		T getPeriodMax(const CountStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return T(getPeriodMax(static_cast<const StatType<CountAccumulator>&>(stat), num_periods));
 		}
 
@@ -473,6 +482,7 @@ namespace LLTrace
 		template<typename T>
 		T getPeriodMax(const SampleStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return T(getPeriodMax(static_cast<const StatType<SampleAccumulator>&>(stat), num_periods));
 		}
 
@@ -480,6 +490,7 @@ namespace LLTrace
 		template<typename T>
 		T getPeriodMax(const EventStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return T(getPeriodMax(static_cast<const StatType<EventAccumulator>&>(stat), num_periods));
 		}
 
@@ -489,6 +500,7 @@ namespace LLTrace
 		template <typename T>
 		typename RelatedTypes<typename T::value_t>::fractional_t getPeriodMaxPerSec(const StatType<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 			F64 max_val = std::numeric_limits<F64>::min();
@@ -503,6 +515,7 @@ namespace LLTrace
 		template<typename T>
 		typename RelatedTypes<T>::fractional_t getPeriodMaxPerSec(const CountStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return typename RelatedTypes<T>::fractional_t(getPeriodMaxPerSec(static_cast<const StatType<CountAccumulator>&>(stat), num_periods));
 		}
 
@@ -514,6 +527,7 @@ namespace LLTrace
 		template <typename T>
 		typename RelatedTypes<typename T::value_t>::fractional_t getPeriodMean(const StatType<T >& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 			typename RelatedTypes<typename T::value_t>::fractional_t mean(0);
@@ -534,12 +548,14 @@ namespace LLTrace
 		template<typename T>
 		typename RelatedTypes<T>::fractional_t getPeriodMean(const CountStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return typename RelatedTypes<T>::fractional_t(getPeriodMean(static_cast<const StatType<CountAccumulator>&>(stat), num_periods));
 		}
 		F64 getPeriodMean(const StatType<SampleAccumulator>& stat, S32 num_periods = S32_MAX);
 		template<typename T> 
 		typename RelatedTypes<T>::fractional_t getPeriodMean(const SampleStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return typename RelatedTypes<T>::fractional_t(getPeriodMean(static_cast<const StatType<SampleAccumulator>&>(stat), num_periods));
 		}
 
@@ -547,6 +563,7 @@ namespace LLTrace
 		template<typename T>
 		typename RelatedTypes<T>::fractional_t getPeriodMean(const EventStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return typename RelatedTypes<T>::fractional_t(getPeriodMean(static_cast<const StatType<EventAccumulator>&>(stat), num_periods));
 		}
 
@@ -556,6 +573,7 @@ namespace LLTrace
 		template <typename T>
 		typename RelatedTypes<typename T::value_t>::fractional_t getPeriodMeanPerSec(const StatType<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			num_periods = llmin(num_periods, getNumRecordedPeriods());
 
 			typename RelatedTypes<typename T::value_t>::fractional_t mean = 0;
@@ -577,6 +595,7 @@ namespace LLTrace
 		template<typename T>
 		typename RelatedTypes<T>::fractional_t getPeriodMeanPerSec(const CountStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return typename RelatedTypes<T>::fractional_t(getPeriodMeanPerSec(static_cast<const StatType<CountAccumulator>&>(stat), num_periods));
 		}
 
@@ -589,6 +608,7 @@ namespace LLTrace
 		template<typename T> 
 		typename RelatedTypes<T>::fractional_t getPeriodStandardDeviation(const SampleStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return typename RelatedTypes<T>::fractional_t(getPeriodStandardDeviation(static_cast<const StatType<SampleAccumulator>&>(stat), num_periods));
 		}
 
@@ -596,6 +616,7 @@ namespace LLTrace
 		template<typename T>
 		typename RelatedTypes<T>::fractional_t getPeriodStandardDeviation(const EventStatHandle<T>& stat, S32 num_periods = S32_MAX)
 		{
+            LL_PROFILE_ZONE_SCOPED;
 			return typename RelatedTypes<T>::fractional_t(getPeriodStandardDeviation(static_cast<const StatType<EventAccumulator>&>(stat), num_periods));
 		}
 
diff --git a/indra/llcommon/lltracethreadrecorder.cpp b/indra/llcommon/lltracethreadrecorder.cpp
index 025dc57044..7ae1e72784 100644
--- a/indra/llcommon/lltracethreadrecorder.cpp
+++ b/indra/llcommon/lltracethreadrecorder.cpp
@@ -274,12 +274,10 @@ void ThreadRecorder::pushToParent()
 }
 
 
-static LLTrace::BlockTimerStatHandle FTM_PULL_TRACE_DATA_FROM_CHILDREN("Pull child thread trace data");
-
 void ThreadRecorder::pullFromChildren()
 {
 #if LL_TRACE_ENABLED
-	LL_RECORD_BLOCK_TIME(FTM_PULL_TRACE_DATA_FROM_CHILDREN);
+    LL_PROFILE_ZONE_SCOPED;
 	if (mActiveRecordings.empty()) return;
 
 	{ LLMutexLock lock(&mChildListMutex);