Update NEWS.

https://gist.github.com/ikrima/1184c75b979cbfa655c0883c0d4ab068

NEWS

@@ -13,6 +13,9 @@ v0.x.x (xxxx-xx-xx)
 - Manually disconnecting from the server will no longer display erroneous
   warning message.
 - Added ability to display sample time spent in child function calls.
+- Fixed issue which may have prevented sampling on ARM64.
+- Added TRACY_NO_FRAME_IMAGE macro to disable frame image compression
+  thread.
 
 
 v0.7.7 (2021-04-01)

client/TracyProfiler.cpp

@@ -940,7 +940,9 @@ enum { QueuePrealloc = 256 * 1024 };
 
 static Profiler* s_instance = nullptr;
 static Thread* s_thread;
+#ifndef TRACY_NO_FRAME_IMAGE
 static Thread* s_compressThread;
+#endif
 
 #ifdef TRACY_HAS_SYSTEM_TRACING
 static Thread* s_sysTraceThread = nullptr;
@@ -1203,8 +1205,10 @@ Profiler::Profiler()
     , m_lz4Buf( (char*)tracy_malloc( LZ4Size + sizeof( lz4sz_t ) ) )
     , m_serialQueue( 1024*1024 )
     , m_serialDequeue( 1024*1024 )
+#ifndef TRACY_NO_FRAME_IMAGE
     , m_fiQueue( 16 )
     , m_fiDequeue( 16 )
+#endif
     , m_frameCount( 0 )
     , m_isConnected( false )
 #ifdef TRACY_ON_DEMAND
@@ -1254,8 +1258,10 @@ void Profiler::SpawnWorkerThreads()
     s_thread = (Thread*)tracy_malloc( sizeof( Thread ) );
     new(s_thread) Thread( LaunchWorker, this );
 
+#ifndef TRACY_NO_FRAME_IMAGE
     s_compressThread = (Thread*)tracy_malloc( sizeof( Thread ) );
     new(s_compressThread) Thread( LaunchCompressWorker, this );
+#endif
 
 #ifdef TRACY_HAS_SYSTEM_TRACING
     if( SysTraceStart( m_samplingPeriod ) )
@@ -1307,8 +1313,11 @@ Profiler::~Profiler()
     }
 #endif
 
+#ifndef TRACY_NO_FRAME_IMAGE
     s_compressThread->~Thread();
     tracy_free( s_compressThread );
+#endif
+
     s_thread->~Thread();
     tracy_free( s_thread );
 
@@ -1808,6 +1817,7 @@ void Profiler::Worker()
     }
 }
 
+#ifndef TRACY_NO_FRAME_IMAGE
 void Profiler::CompressWorker()
 {
     ThreadExitHandler threadExitHandler;
@@ -1874,6 +1884,7 @@ void Profiler::CompressWorker()
         }
     }
 }
+#endif
 
 static void FreeAssociatedMemory( const QueueItem& item )
 {

client/TracyProfiler.hpp

@@ -213,11 +213,12 @@ public:
 
     static tracy_force_inline void SendFrameImage( const void* image, uint16_t w, uint16_t h, uint8_t offset, bool flip )
     {
+#ifndef TRACY_NO_FRAME_IMAGE
         auto& profiler = GetProfiler();
         assert( profiler.m_frameCount.load( std::memory_order_relaxed ) < std::numeric_limits<uint32_t>::max() );
-#ifdef TRACY_ON_DEMAND
+#  ifdef TRACY_ON_DEMAND
         if( !profiler.IsConnected() ) return;
-#endif
+#  endif
         const auto sz = size_t( w ) * size_t( h ) * 4;
         auto ptr = (char*)tracy_malloc( sz );
         memcpy( ptr, image, sz );
@@ -231,6 +232,7 @@ public:
         fi->flip = flip;
         profiler.m_fiQueue.commit_next();
         profiler.m_fiLock.unlock();
+#endif
     }
 
     static tracy_force_inline void PlotData( const char* name, int64_t val )
@@ -642,8 +644,10 @@ private:
     static void LaunchWorker( void* ptr ) { ((Profiler*)ptr)->Worker(); }
     void Worker();
 
+#ifndef TRACY_NO_FRAME_IMAGE
     static void LaunchCompressWorker( void* ptr ) { ((Profiler*)ptr)->CompressWorker(); }
     void CompressWorker();
+#endif
 
     void ClearQueues( tracy::moodycamel::ConsumerToken& token );
     void ClearSerial();
@@ -790,8 +794,10 @@ private:
     FastVector<QueueItem> m_serialQueue, m_serialDequeue;
     TracyMutex m_serialLock;
 
+#ifndef TRACY_NO_FRAME_IMAGE
     FastVector<FrameImageQueueItem> m_fiQueue, m_fiDequeue;
     TracyMutex m_fiLock;
+#endif
 
     std::atomic<uint64_t> m_frameCount;
     std::atomic<bool> m_isConnected;

client/TracySysTrace.cpp

@@ -733,54 +733,59 @@ static void SetupSampling( int64_t& samplingPeriod )
                         s_ring[i].Read( &cnt, offset, sizeof( uint64_t ) );
                         offset += sizeof( uint64_t );
 
-                        auto trace = (uint64_t*)tracy_malloc( ( 1 + cnt ) * sizeof( uint64_t ) );
-                        s_ring[i].Read( trace+1, offset, sizeof( uint64_t ) * cnt );
+                        if( cnt > 0 )
+                        {
+                            auto trace = (uint64_t*)tracy_malloc( ( 1 + cnt ) * sizeof( uint64_t ) );
+                            s_ring[i].Read( trace+1, offset, sizeof( uint64_t ) * cnt );
 
-                        // remove non-canonical pointers
-                        do
-                        {
-                            const auto test = (int64_t)trace[cnt];
-                            const auto m1 = test >> 63;
-                            const auto m2 = test >> 47;
-                            if( m1 == m2 ) break;
-                        }
-                        while( --cnt > 0 );
-                        for( uint64_t j=1; j<cnt; j++ )
-                        {
-                            const auto test = (int64_t)trace[j];
-                            const auto m1 = test >> 63;
-                            const auto m2 = test >> 47;
-                            if( m1 != m2 ) trace[j] = 0;
-                        }
-
-                        // skip kernel frames
-                        uint64_t j;
-                        for( j=0; j<cnt; j++ )
-                        {
-                            if( (int64_t)trace[j+1] >= 0 ) break;
-                        }
-                        if( j == cnt )
-                        {
-                            tracy_free( trace );
-                        }
-                        else
-                        {
-                            if( j > 0 )
+#if defined __x86_64__ || defined _M_X64
+                            // remove non-canonical pointers
+                            do
                             {
-                                cnt -= j;
-                                memmove( trace+1, trace+1+j, sizeof( uint64_t ) * cnt );
+                                const auto test = (int64_t)trace[cnt];
+                                const auto m1 = test >> 63;
+                                const auto m2 = test >> 47;
+                                if( m1 == m2 ) break;
+                            }
+                            while( --cnt > 0 );
+                            for( uint64_t j=1; j<cnt; j++ )
+                            {
+                                const auto test = (int64_t)trace[j];
+                                const auto m1 = test >> 63;
+                                const auto m2 = test >> 47;
+                                if( m1 != m2 ) trace[j] = 0;
                             }
-                            memcpy( trace, &cnt, sizeof( uint64_t ) );
-
-#if defined TRACY_HW_TIMER && ( defined __i386 || defined _M_IX86 || defined __x86_64__ || defined _M_X64 )
-                            t0 = s_ring[i].ConvertTimeToTsc( t0 );
 #endif
 
-                            TracyLfqPrepare( QueueType::CallstackSample );
-                            MemWrite( &item->callstackSampleFat.time, t0 );
-                            MemWrite( &item->callstackSampleFat.thread, (uint64_t)tid );
-                            MemWrite( &item->callstackSampleFat.ptr, (uint64_t)trace );
-                            TracyLfqCommit;
+                            // skip kernel frames
+                            uint64_t j;
+                            for( j=0; j<cnt; j++ )
+                            {
+                                if( (int64_t)trace[j+1] >= 0 ) break;
+                            }
+                            if( j == cnt )
+                            {
+                                tracy_free( trace );
+                            }
+                            else
+                            {
+                                if( j > 0 )
+                                {
+                                    cnt -= j;
+                                    memmove( trace+1, trace+1+j, sizeof( uint64_t ) * cnt );
+                                }
+                                memcpy( trace, &cnt, sizeof( uint64_t ) );
+
+#if defined TRACY_HW_TIMER && ( defined __i386 || defined _M_IX86 || defined __x86_64__ || defined _M_X64 )
+                                t0 = s_ring[i].ConvertTimeToTsc( t0 );
+#endif
+
+                                TracyLfqPrepare( QueueType::CallstackSample );
+                                MemWrite( &item->callstackSampleFat.time, t0 );
+                                MemWrite( &item->callstackSampleFat.thread, (uint64_t)tid );
+                                MemWrite( &item->callstackSampleFat.ptr, (uint64_t)trace );
+                                TracyLfqCommit;
+                            }
                         }
                     }
                 }

common/TracySystem.cpp

@@ -99,6 +99,29 @@ std::atomic<ThreadNameData*>& GetThreadNameData();
 TRACY_API void InitRPMallocThread();
 #endif
 
+#ifdef _MSC_VER
+#  pragma pack( push, 8 )
+struct THREADNAME_INFO
+{
+    DWORD dwType;
+    LPCSTR szName;
+    DWORD dwThreadID;
+    DWORD dwFlags;
+};
+#  pragma pack(pop)
+
+void ThreadNameMsvcMagic( const THREADNAME_INFO& info )
+{
+    __try
+    {
+        RaiseException( 0x406D1388, 0, sizeof(info)/sizeof(ULONG_PTR), (ULONG_PTR*)&info );
+    }
+    __except(EXCEPTION_EXECUTE_HANDLER)
+    {
+    }
+}
+#endif
+
 TRACY_API void SetThreadName( const char* name )
 {
 #if defined _WIN32 || defined __CYGWIN__
@@ -112,31 +135,12 @@ TRACY_API void SetThreadName( const char* name )
     else
     {
 #  if defined _MSC_VER
-        const DWORD MS_VC_EXCEPTION=0x406D1388;
-#    pragma pack( push, 8 )
-        struct THREADNAME_INFO
-        {
-            DWORD dwType;
-            LPCSTR szName;
-            DWORD dwThreadID;
-            DWORD dwFlags;
-        };
-#    pragma pack(pop)
-
-        DWORD ThreadId = GetCurrentThreadId();
         THREADNAME_INFO info;
         info.dwType = 0x1000;
         info.szName = name;
-        info.dwThreadID = ThreadId;
+        info.dwThreadID = GetCurrentThreadId();
         info.dwFlags = 0;
-
-        __try
-        {
-            RaiseException( MS_VC_EXCEPTION, 0, sizeof(info)/sizeof(ULONG_PTR), (ULONG_PTR*)&info );
-        }
-        __except(EXCEPTION_EXECUTE_HANDLER)
-        {
-        }
+        ThreadNameMsvcMagic( info );
 #  endif
     }
 #elif defined _GNU_SOURCE && !defined __EMSCRIPTEN__ && !defined __CYGWIN__

manual/tracy.tex

@@ -914,6 +914,7 @@ logo=\bcattention
 ]{Caveats}
 \begin{itemize}
 \item Frame images are compressed on a second client profiler thread\footnote{Small part of compression task is performed on the server.}, to reduce memory usage of queued images. This might have impact on the performance of the profiled application.
+\item This second thread will be periodically woken up, even if there are no frame images to compress\footnote{This way of doing things is required to prevent a deadlock in specific circumstances.}. If you are not using the frame image capture functionality and you don't wish this thread to be running, you can define the \texttt{TRACY\_NO\_FRAME\_IMAGE} macro.
 \item Due to implementation details of the network buffer, single frame image cannot be greater than 256 KB after compression. Note that a $960\times540$ image fits in this limit.
 \end{itemize}
 \end{bclogo}

profiler/build/win32/DebugVis.natvis

@@ -1,4 +1,4 @@
-<?xml version="1.0" encoding="utf-8"?> 
+<?xml version="1.0" encoding="utf-8"?>
 <AutoVisualizer xmlns="http://schemas.microsoft.com/vstudio/debugger/natvis/2010">
   <Type Name="tracy::Vector&lt;*&gt;">
     <DisplayString>{{ size={m_size} }}</DisplayString>
@@ -44,4 +44,67 @@
   <Type Name="tracy::Int48">
     <DisplayString>{{ value={int64_t( uint64_t(m_val[0]) | (uint64_t(m_val[1])&lt;&lt;8) | (uint64_t(m_val[2])&lt;&lt;16) | (uint64_t(m_val[3])&lt;&lt;24) | (uint64_t(m_val[4])&lt;&lt;32) | (uint64_t(m_val[5])&lt;&lt;40) )} }}</DisplayString>
   </Type>
+  <Type Name="tracy::detail::Table&lt;*,*,*,*,*,*&gt;">
+    <!--
+  $T1 = bool IsFlat
+  $T2 = size_t MaxLoadFactor100
+  $T3 = typename Key
+  $T4 = typename T
+  $T5 = typename Hash
+  $T6 = typename KeyEqual
+-->
+    <!-- <DisplayString>{map}</DisplayString> -->
+    <Expand>
+      <Synthetic Name="[elements]">
+        <DisplayString>{{size={mNumElements}}}</DisplayString>
+        <Expand>
+          <CustomListItems MaxItemsPerView="5000" >
+            <Variable Name="itKeyVals"    InitialValue="mKeyVals     " />
+            <Variable Name="itInfo"       InitialValue="mInfo        " />
+            <Variable Name="itEndKeyVals" InitialValue="(void *)mInfo " />
+            <Variable Name="n"            InitialValue="0ULL         " />
+            <Variable Name="inc"          InitialValue="(unsigned long)0" />
+
+            <Size>mNumElements</Size>
+            <Loop>
+              <!-- Fast forward -->
+              <Exec>n = *((size_t*)itInfo)</Exec>
+              <Loop>
+                <Break Condition="n != 0" />
+                <Exec>itInfo += sizeof(size_t)</Exec>
+                <Exec>itKeyVals += sizeof(size_t)</Exec>
+              </Loop>
+              <!-- Count Trailing Zeros -->
+              <Exec>
+                inc = n == 0
+                ? 64
+                : (
+                63
+                - (((n &amp; (~n + 1)) &amp; 0x00000000FFFFFFFF) ? 32 : 0)
+                - (((n &amp; (~n + 1)) &amp; 0x0000FFFF0000FFFF) ? 16 : 0)
+                - (((n &amp; (~n + 1)) &amp; 0x00FF00FF00FF00FF) ? 8  : 0)
+                - (((n &amp; (~n + 1)) &amp; 0x0F0F0F0F0F0F0F0F) ? 4  : 0)
+                - (((n &amp; (~n + 1)) &amp; 0x3333333333333333) ? 2  : 0)
+                - (((n &amp; (~n + 1)) &amp; 0x5555555555555555) ? 1  : 0)
+                )
+              </Exec>
+              <Exec>itInfo    += inc / 8</Exec>
+              <Exec>itKeyVals += inc / 8</Exec>
+              <!-- Fast forward -->
+              <Break Condition="(void*)itKeyVals == itEndKeyVals" />
+              <Item Name="[{itKeyVals-&gt;mData.first}]">itKeyVals-&gt;mData.second</Item>
+              <!-- <Item>itKeyVals-&gt;mData</Item> -->
+              <Exec>itInfo++   </Exec>
+              <Exec>itKeyVals++</Exec>
+            </Loop>
+          </CustomListItems>
+        </Expand>
+      </Synthetic>
+      <Item Name="[load_factor]"    >float(mNumElements) / float(mMask + 1)</Item>
+      <Item Name="[max_load_factor]">$T2</Item>
+      <Item Name="[IsFlat]"         >$T1</Item>
+      <Item Name="[hash_function]"  >*(WrapHash&lt;$T5&gt;*)this,nd</Item>
+      <Item Name="[key_eq]"         >*(WrapKeyEqual&lt;$T6&gt;*)this,nd</Item>
+    </Expand>
+  </Type>
 </AutoVisualizer>

server/TracySourceView.cpp

@@ -2723,13 +2723,42 @@ void SourceView::RenderAsmLine( AsmLine& line, const AddrStat& ipcnt, const Addr
     {
         TextDisabledUnformatted( buf );
     }
-    if( ImGui::IsItemClicked( 0 ) )
+    if( ImGui::IsItemHovered() )
     {
-        m_asmCountBase = asmIdx;
-    }
-    else if( ImGui::IsItemClicked( 1 ) )
-    {
-        m_asmCountBase = -1;
+        if( m_font ) ImGui::PopFont();
+        ImGui::BeginTooltip();
+        if( m_asmCountBase >= 0 )
+        {
+            TextDisabledUnformatted( "Absolute address:" );
+            ImGui::SameLine();
+            ImGui::Text( "%" PRIx64, line.addr );
+            TextDisabledUnformatted( "Relative address:" );
+            ImGui::SameLine();
+            ImGui::Text( "+%" PRIx64, line.addr - m_baseAddr );
+        }
+        else if( m_asmRelative )
+        {
+            TextDisabledUnformatted( "Absolute address:" );
+            ImGui::SameLine();
+            ImGui::Text( "%" PRIx64, line.addr );
+        }
+        else
+        {
+            TextDisabledUnformatted( "Relative address:" );
+            ImGui::SameLine();
+            ImGui::Text( "+%" PRIx64, line.addr - m_baseAddr );
+        }
+        ImGui::EndTooltip();
+        if( m_font ) ImGui::PushFont( m_font );
+
+        if( ImGui::IsItemClicked( 0 ) )
+        {
+            m_asmCountBase = asmIdx;
+        }
+        else if( ImGui::IsItemClicked( 1 ) )
+        {
+            m_asmCountBase = -1;
+        }
     }
 
     const auto stw = ImGui::CalcTextSize( " " ).x;
@@ -3025,6 +3054,17 @@ void SourceView::RenderAsmLine( AsmLine& line, const AddrStat& ipcnt, const Addr
         ImGui::TextUnformatted( buf );
     }
 
+    uint32_t jumpOffset;
+    uint64_t jumpBase;
+    const char* jumpName = nullptr;
+    if( line.jumpAddr != 0 )
+    {
+        jumpOffset = 0;
+        jumpBase = worker.GetSymbolForAddress( line.jumpAddr, jumpOffset );
+        auto jumpSym = jumpBase == 0 ? worker.GetSymbolData( line.jumpAddr ) : worker.GetSymbolData( jumpBase );
+        if( jumpSym ) jumpName = worker.GetString( jumpSym->name );
+    }
+
     if( ImGui::IsItemHovered() )
     {
         if( asmVar )
@@ -3032,16 +3072,31 @@ void SourceView::RenderAsmLine( AsmLine& line, const AddrStat& ipcnt, const Addr
             const auto& var = *asmVar;
             if( m_font ) ImGui::PopFont();
             ImGui::BeginTooltip();
-            if( opdesc != 0 )
+            if( jumpName || opdesc != 0 )
             {
-                ImGui::TextUnformatted( OpDescList[opdesc] );
+                if( opdesc != 0 ) ImGui::TextUnformatted( OpDescList[opdesc] );
+                if( jumpName )
+                {
+                    if( jumpBase == m_baseAddr )
+                    {
+                        TextDisabledUnformatted( "Local target:" );
+                    }
+                    else
+                    {
+                        TextDisabledUnformatted( "External target:" );
+                    }
+                    ImGui::SameLine();
+                    ImGui::Text( "%s+%" PRIu32, jumpName, jumpOffset );
+                }
                 ImGui::Separator();
             }
+
             TextFocused( "Throughput:", RealToString( var.tp ) );
             ImGui::SameLine();
             TextDisabledUnformatted( "(cycles per instruction, lower is better)" );
             if( var.maxlat >= 0 )
             {
+                bool exact = false;
                 TextDisabledUnformatted( "Latency:" );
                 ImGui::SameLine();
                 if( var.minlat == var.maxlat && var.minbound == var.maxbound )
@@ -3053,6 +3108,7 @@ void SourceView::RenderAsmLine( AsmLine& line, const AddrStat& ipcnt, const Addr
                     else
                     {
                         ImGui::TextUnformatted( RealToString( var.minlat ) );
+                        exact = true;
                     }
                 }
                 else
@@ -3076,7 +3132,14 @@ void SourceView::RenderAsmLine( AsmLine& line, const AddrStat& ipcnt, const Addr
                     }
                 }
                 ImGui::SameLine();
-                TextDisabledUnformatted( "(cycles in execution, may vary by used output)" );
+                if( exact )
+                {
+                    TextDisabledUnformatted( "(cycles in execution)" );
+                }
+                else
+                {
+                    TextDisabledUnformatted( "(cycles in execution, may vary by used output)" );
+                }
             }
             TextFocused( "\xce\xbcops:", RealToString( var.uops ) );
             if( var.port != -1 ) TextFocused( "Ports:", PortList[var.port] );
@@ -3134,6 +3197,23 @@ void SourceView::RenderAsmLine( AsmLine& line, const AddrStat& ipcnt, const Addr
             ImGui::EndTooltip();
             if( m_font ) ImGui::PushFont( m_font );
         }
+        else if( jumpName )
+        {
+            if( m_font ) ImGui::PopFont();
+            ImGui::BeginTooltip();
+            if( jumpBase == m_baseAddr )
+            {
+                TextDisabledUnformatted( "Local target:" );
+            }
+            else
+            {
+                TextDisabledUnformatted( "External target:" );
+            }
+            ImGui::SameLine();
+            ImGui::Text( "%s+%" PRIu32, jumpName, jumpOffset );
+            ImGui::EndTooltip();
+            if( m_font ) ImGui::PushFont( m_font );
+        }
         if( m_cpuArch == CpuArchX86 || m_cpuArch == CpuArchX64 )
         {
             if( line.readX86[0] != RegsX86::invalid || line.writeX86[0] != RegsX86::invalid )
@@ -3265,35 +3345,29 @@ void SourceView::RenderAsmLine( AsmLine& line, const AddrStat& ipcnt, const Addr
         TextColoredUnformatted( ImVec4( 0.5f, 0.5, 1, 1 ), "}" );
     }
 
-    if( line.jumpAddr != 0 )
+    if( jumpName )
     {
-        uint32_t offset = 0;
-        const auto base = worker.GetSymbolForAddress( line.jumpAddr, offset );
-        auto sym = base == 0 ? worker.GetSymbolData( line.jumpAddr ) : worker.GetSymbolData( base );
-        if( sym )
+        ImGui::SameLine();
+        ImGui::Spacing();
+        ImGui::SameLine();
+        if( jumpBase == m_baseAddr )
         {
-            ImGui::SameLine();
-            ImGui::Spacing();
-            ImGui::SameLine();
-            if( base == m_baseAddr )
+            ImGui::TextDisabled( "-> [%s+%" PRIu32"]", jumpName, jumpOffset );
+            if( ImGui::IsItemHovered() )
             {
-                ImGui::TextDisabled( "-> [%s+%" PRIu32"]", worker.GetString( sym->name ), offset );
-                if( ImGui::IsItemHovered() )
+                m_highlightAddr = line.jumpAddr;
+                if( ImGui::IsItemClicked() )
                 {
-                    m_highlightAddr = line.jumpAddr;
-                    if( ImGui::IsItemClicked() )
-                    {
-                        m_targetAddr = line.jumpAddr;
-                        m_selectedAddresses.clear();
-                        m_selectedAddresses.emplace( line.jumpAddr );
-                    }
+                    m_targetAddr = line.jumpAddr;
+                    m_selectedAddresses.clear();
+                    m_selectedAddresses.emplace( line.jumpAddr );
                 }
             }
-            else
-            {
-                ImGui::TextDisabled( "[%s+%" PRIu32"]", worker.GetString( sym->name ), offset );
-                if( ImGui::IsItemClicked() ) jumpOut = line.jumpAddr;
-            }
+        }
+        else
+        {
+            ImGui::TextDisabled( "[%s+%" PRIu32"]", jumpName, jumpOffset );
+            if( ImGui::IsItemClicked() ) jumpOut = line.jumpAddr;
         }
     }