A few fixes in the threadpool semaphore. Unify Windows/Unix implementation of LIFO policy.

docs/coding-guidelines/interop-guidelines.md

@@ -34,7 +34,7 @@ internal static partial class Interop
 ...
 internal static partial class Interop
 {
-    internal static partial class mincore { ... }
+    internal static partial class Mincore { ... }
 }
 ```
 - With few exceptions, the only methods that should be defined in these interop types are DllImports.
@@ -49,7 +49,9 @@ internal static partial class Interop
 ```
 \Common\src\Interop
     \Windows
-        \mincore
+        \Kernel32
+            ... interop files
+        \Mincore
             ... interop files
     \Unix
         \libc
@@ -71,8 +73,8 @@ As shown above, platforms may be additive, in that an assembly may use functiona
         \libc
             \Interop.strerror.cs
     \Windows
-        \mincore
-            \Interop.OutputDebugString.cs
+        \Mincore
+            \Interop.WaitOnAddress.cs    <-- Also contains WakeByAddressSingle
 ```
 
 - If structs/constants will be used on their own without an associated DllImport, or if they may be used with multiple DllImports not in the same file, they should be declared in a separate file.
@@ -81,9 +83,9 @@ As shown above, platforms may be additive, in that an assembly may use functiona
 ```
 \Common\src\Interop
     \Windows
-        \mincore
-            \Interop.DuplicateHandle_SafeTokenHandle.cs
-            \Interop.DuplicateHandle_IntPtr.cs
+        \Kernel32
+            \Interop.DuplicateHandle_SafeFileHandle.cs
+            \Interop.DuplicateHandle_SafePipeHandle.cs
 ```
 
 - The library names used per-platform are stored in internal constants in the Interop class in a private Libraries class in a per-platform file named Interop.Libraries.cs. These constants are then used for all DllImports to that library, rather than having the string duplicated each time, e.g.
@@ -94,12 +96,9 @@ internal static partial class Interop // contents of Common\src\Interop\Windows\
     private static class Libraries
     {
         internal const string Kernel32 = "kernel32.dll";
+        internal const string OleAut32 = "oleaut32.dll";
         internal const string Localization = "api-ms-win-core-localization-l1-2-0.dll";
-        internal const string Handle = "api-ms-win-core-handle-l1-1-0.dll";
-        internal const string ProcessThreads = "api-ms-win-core-processthreads-l1-1-0.dll";
-        internal const string File = "api-ms-win-core-file-l1-1-0.dll";
-        internal const string NamedPipe = "api-ms-win-core-namedpipe-l1-1-0.dll";
-        internal const string IO = "api-ms-win-core-io-l1-1-0.dll";
+        internal const string Synch = "api-ms-win-core-synch-l1-2-0.dll";
         ...
     }
 }

src/coreclr/nativeaot/BuildIntegration/Microsoft.NETCore.Native.Windows.targets

@@ -82,6 +82,7 @@ The .NET Foundation licenses this file to you under the MIT license.
       <SdkNativeLibrary Include="user32.lib" />
       <SdkNativeLibrary Include="version.lib" />
       <SdkNativeLibrary Include="ws2_32.lib" />
+      <SdkNativeLibrary Include="Synchronization.lib" />
     </ItemGroup>
 
     <ItemGroup>

src/coreclr/nativeaot/BuildIntegration/WindowsAPIs.txt

@@ -2340,6 +2340,12 @@ ws2_32!WSCUpdateProvider
 ws2_32!WSCWriteNameSpaceOrder
 ws2_32!WSCWriteProviderOrder
 
+#
+# Synchronization.lib
+#
+api-ms-win-core-synch-l1-2-0!WaitOnAddress
+api-ms-win-core-synch-l1-2-0!WakeByAddressSingle
+
 #
 # Include all memory allocation APIs from ucrt to ensure that all of them use allocator
 # from the same ucrt copy. Also include the frequently used basic memory manipulation APIs.

src/coreclr/tools/aot/ILCompiler/reproNative/reproNative.vcxproj

@@ -46,7 +46,7 @@
     <NativeAotSourceRoot>$(CoreClrSourceRoot)nativeaot\</NativeAotSourceRoot>
     <ArtifactsRoot>$(CoreClrSourceRoot)..\..\artifacts\</ArtifactsRoot>
     <NativeRoot>$(CoreClrSourceRoot)..\native\</NativeRoot>
-    <Win32SDKLibs>advapi32.lib;bcrypt.lib;crypt32.lib;iphlpapi.lib;kernel32.lib;mswsock.lib;ncrypt.lib;normaliz.lib;ntdll.lib;ole32.lib;oleaut32.lib;secur32.lib;user32.lib;version.lib;ws2_32.lib</Win32SDKLibs>
+    <Win32SDKLibs>advapi32.lib;bcrypt.lib;crypt32.lib;iphlpapi.lib;kernel32.lib;mswsock.lib;ncrypt.lib;normaliz.lib;ntdll.lib;ole32.lib;oleaut32.lib;secur32.lib;user32.lib;version.lib;ws2_32.lib;Synchronization.lib</Win32SDKLibs>
   </PropertyGroup>
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">

src/libraries/Common/src/Interop/Unix/System.Native/Interop.Futex.cs

@@ -0,0 +1,21 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+
+
+using System.Runtime.InteropServices;
+
+internal static partial class Interop
+{
+    internal static partial class Sys
+    {
+        [LibraryImport(Libraries.SystemNative, EntryPoint = "SystemNative_LowLevelFutex_WaitOnAddress")]
+        internal static unsafe partial void LowLevelFutex_WaitOnAddress(int* address, int comparand);
+
+        [LibraryImport(Libraries.SystemNative, EntryPoint = "SystemNative_LowLevelFutex_WaitOnAddressTimeout")]
+        [return: MarshalAs(UnmanagedType.Bool)]
+        internal static unsafe partial bool LowLevelFutex_WaitOnAddressTimeout(int* address, int comparand, int timeoutMilliseconds);
+
+        [LibraryImport(Libraries.SystemNative, EntryPoint = "SystemNative_LowLevelFutex_WakeByAddressSingle")]
+        internal static unsafe partial void LowLevelFutex_WakeByAddressSingle(int* address);
+    }
+}

src/libraries/Common/src/Interop/Unix/System.Native/Interop.LowLevelMonitor.cs

@@ -17,6 +17,7 @@ internal static partial class Sys
         [LibraryImport(Libraries.SystemNative, EntryPoint = "SystemNative_LowLevelMonitor_Acquire")]
         internal static partial void LowLevelMonitor_Acquire(IntPtr monitor);
 
+        [SuppressGCTransition]
         [LibraryImport(Libraries.SystemNative, EntryPoint = "SystemNative_LowLevelMonitor_Release")]
         internal static partial void LowLevelMonitor_Release(IntPtr monitor);
 

src/libraries/Common/src/Interop/Windows/Interop.Libraries.cs

@@ -48,7 +48,7 @@ internal static partial class Libraries
         internal const string Xolehlp = "xolehlp.dll";
         internal const string Comdlg32 = "comdlg32.dll";
         internal const string Gdiplus = "gdiplus.dll";
-        internal const string Oleaut32 = "oleaut32.dll";
         internal const string Winspool = "winspool.drv";
+        internal const string Synch = "api-ms-win-core-synch-l1-2-0.dll";
     }
 }

src/libraries/Common/src/Interop/Windows/Kernel32/Interop.CriticalSection.cs

@@ -25,6 +25,7 @@ internal struct CRITICAL_SECTION
         [LibraryImport(Libraries.Kernel32)]
         internal static unsafe partial void EnterCriticalSection(CRITICAL_SECTION* lpCriticalSection);
 
+        [SuppressGCTransition]
         [LibraryImport(Libraries.Kernel32)]
         internal static unsafe partial void LeaveCriticalSection(CRITICAL_SECTION* lpCriticalSection);
 

src/libraries/Common/src/Interop/Windows/Mincore/Interop.WaitOnAddress.cs

@@ -0,0 +1,17 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+
+
+using System.Runtime.InteropServices;
+
+internal static partial class Interop
+{
+    internal static partial class Mincore
+    {
+        [LibraryImport(Libraries.Synch, SetLastError = true)]
+        internal static unsafe partial BOOL WaitOnAddress(void* Address, void* CompareAddress, nint AddressSize, int dwMilliseconds);
+
+        [LibraryImport(Libraries.Synch)]
+        internal static unsafe partial void WakeByAddressSingle(void* Address);
+    }
+}

src/libraries/System.Private.CoreLib/src/System.Private.CoreLib.Shared.projitems

@@ -1840,6 +1840,9 @@
     <Compile Include="$(CommonPath)Interop\Windows\Kernel32\Interop.CreateSymbolicLink.cs">
       <Link>Common\Interop\Windows\Kernel32\Interop.CreateSymbolicLink.cs</Link>
     </Compile>
+    <Compile Include="$(CommonPath)Interop\Windows\Mincore\Interop.WaitOnAddress.cs">
+      <Link>Common\Interop\Windows\Mincore\Interop.WaitOnAddress.cs</Link>
+    </Compile>
     <Compile Include="$(CommonPath)Interop\Windows\Kernel32\Interop.CriticalSection.cs">
       <Link>Common\Interop\Windows\Kernel32\Interop.CriticalSection.cs</Link>
     </Compile>
@@ -2471,6 +2474,9 @@
     <Compile Include="$(CommonPath)Interop\Unix\System.Native\Interop.LowLevelMonitor.cs">
       <Link>Common\Interop\Unix\System.Native\Interop.LowLevelMonitor.cs</Link>
     </Compile>
+    <Compile Include="$(CommonPath)Interop\Unix\System.Native\Interop.Futex.cs">
+      <Link>Common\Interop\Unix\System.Native\Interop.Futex.cs</Link>
+    </Compile>
     <Compile Include="$(CommonPath)Interop\Unix\System.Native\Interop.LSeek.cs">
       <Link>Common\Interop\Unix\System.Native\Interop.LSeek.cs</Link>
     </Compile>
@@ -2825,8 +2831,9 @@
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PortableThreadPool.Unix.cs" Condition="'$(TargetsUnix)' == 'true' or ('$(TargetsBrowser)' == 'true' and '$(FeatureWasmManagedThreads)' != 'true') or '$(TargetsWasi)' == 'true'" />
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PortableThreadPool.Windows.cs" Condition="'$(TargetsWindows)' == 'true'" />
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\LowLevelLifoSemaphore.cs" />
-    <Compile Include="$(MSBuildThisFileDirectory)System\Threading\LowLevelLifoSemaphore.Windows.cs" Condition="'$(TargetsWindows)' == 'true'" />
-    <Compile Include="$(MSBuildThisFileDirectory)System\Threading\LowLevelLifoSemaphore.Unix.cs" Condition="'$(TargetsUnix)' == 'true' or '$(TargetsBrowser)' == 'true' or '$(TargetsWasi)' == 'true'"  />
+    <Compile Include="$(MSBuildThisFileDirectory)System\Threading\LowLevelThreadBlocker.cs" />
+    <Compile Include="$(MSBuildThisFileDirectory)System\Threading\LowLevelFutex.Windows.cs" Condition="'$(TargetsWindows)' == 'true'" />
+    <Compile Include="$(MSBuildThisFileDirectory)System\Threading\LowLevelFutex.Unix.cs" Condition="'$(TargetsUnix)' == 'true'" />
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PreAllocatedOverlapped.Windows.cs" Condition="'$(TargetsWindows)' == 'true'" />
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PreAllocatedOverlapped.Unix.cs" Condition="'$(TargetsUnix)' == 'true' or ('$(TargetsBrowser)' == 'true' and '$(FeatureWasmManagedThreads)' == 'true')" />
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PreAllocatedOverlapped.Portable.cs" Condition="('$(TargetsBrowser)' != 'true' and '$(TargetsWasi)' != 'true') or '$(FeatureWasmManagedThreads)' == 'true'" />
@@ -2930,4 +2937,4 @@
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\Wasi\WasiPollWorld.wit.imports.wasi.io.v0_2_0.IPoll.cs" />
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\Wasi\WasiPollWorld.wit.imports.wasi.io.v0_2_0.PollInterop.cs" />
   </ItemGroup>
-</Project>
+</Project>

src/libraries/System.Private.CoreLib/src/System/Threading/Backoff.cs

@@ -15,21 +15,27 @@ internal static class Backoff
         // the exponential backoff will generally be not more than 2X worse than the perfect guess and
         // will do a lot less attempts than a simple retry. On multiprocessor machine fruitless attempts
         // will cause unnecessary sharing of the contended state which may make modifying the state more expensive.
-        // To protect against degenerate cases we will cap the per-iteration wait to a few thousand spinwaits.
-        private const uint MaxExponentialBackoffBits = 14;
+        // To protect against degenerate cases we will cap the per-iteration wait to 1-2 thousand spinwaits.
+        private const uint MaxExponentialBackoffBits = 10;
 
-        internal static unsafe void Exponential(uint attempt)
+        internal static unsafe int Exponential(uint attempt)
         {
-            attempt = Math.Min(attempt, MaxExponentialBackoffBits);
-            // We will backoff for some random number of spins that roughly grows as attempt^2
-            // No need for much randomness here, randomness is "good to have", we could do without it,
-            // so we will just cheaply hash in the stack location.
-            uint rand = (uint)&attempt * 2654435769u;
-            // Set the highmost bit to ensure minimum number of spins is exponentially increasing.
-            // It basically guarantees that we spin at least 0, 1, 2, 4, 8, 16, times, and so on
-            rand |= (1u << 31);
-            uint spins = rand >> (byte)(32 - attempt);
-            Thread.SpinWait((int)spins);
+            if (attempt > 0)
+            {
+                attempt = Math.Min(attempt, MaxExponentialBackoffBits);
+                // We will backoff for some random number of spins that roughly grows as attempt^2
+                // No need for much randomness here, randomness is "good to have", we could do without it,
+                // so we will just cheaply hash in the stack location.
+                uint rand = (uint)&attempt * 2654435769u;
+                // Set the highmost bit to ensure minimum number of spins is exponentially increasing.
+                // It basically guarantees that we spin at least 0, 1, 2, 4, 8, 16, times, and so on
+                rand |= (1u << 31);
+                uint spins = rand >> (byte)(32 - attempt);
+                Thread.SpinWait((int)spins);
+                return (int)spins;
+            }
+
+            return 0;
         }
     }
 }

src/libraries/System.Private.CoreLib/src/System/Threading/LowLevelFutex.Unix.cs

@@ -0,0 +1,41 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+
+using System.Diagnostics;
+
+
+namespace System.Threading
+{
+    // NOTE: Only supported on Linux for now.
+    //       Most OS have support for futex-like APIs and should be added in the future.
+    //       (ex: OSX has `os_sync_wait_on_address`, but support may vary by OS version)
+
+    /// <summary>
+    /// A compare-and-wait synchronization primitive.
+    /// Provides simple functionality to block and wake threads.
+    /// </summary>
+    internal static unsafe class LowLevelFutex
+    {
+        internal static void WaitOnAddress(int* address, int comparand)
+        {
+            Interop.Sys.LowLevelFutex_WaitOnAddress(address, comparand);
+        }
+
+        internal static bool WaitOnAddressTimeout(int* address, int comparand, int milliseconds)
+        {
+            Debug.Assert(milliseconds >= -1);
+            if (milliseconds == -1)
+            {
+                WaitOnAddress(address, comparand);
+                return true;
+            }
+
+            return Interop.Sys.LowLevelFutex_WaitOnAddressTimeout(address, comparand, milliseconds);
+        }
+
+        internal static void WakeByAddressSingle(int* address)
+        {
+            Interop.Sys.LowLevelFutex_WakeByAddressSingle(address);
+        }
+    }
+}

src/libraries/System.Private.CoreLib/src/System/Threading/LowLevelFutex.Windows.cs

@@ -0,0 +1,48 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+
+using System.Diagnostics;
+using System.Runtime.InteropServices;
+
+namespace System.Threading
+{
+    /// <summary>
+    /// A compare-and-wait synchronization primitive.
+    /// Provides simple functionality to block and wake threads.
+    /// </summary>
+    internal static unsafe class LowLevelFutex
+    {
+        internal static void WaitOnAddress(int* address, int comparand)
+        {
+            Interop.BOOL result = Interop.Mincore.WaitOnAddress(address, &comparand, sizeof(int), -1);
+            // assert success, but in release treat unexpected results as spurious wakes
+            Debug.Assert(result == Interop.BOOL.TRUE);
+        }
+
+        internal static bool WaitOnAddressTimeout(int* address, int comparand, int milliseconds)
+        {
+            Interop.BOOL result = Interop.Mincore.WaitOnAddress(address, &comparand, sizeof(int), milliseconds);
+            if (result == Interop.BOOL.TRUE)
+            {
+                // normal or spurious wake
+                return true;
+            }
+
+            int lastError = Marshal.GetLastWin32Error();
+            Debug.Assert(lastError == Interop.Errors.ERROR_TIMEOUT);
+            if (lastError == Interop.Errors.ERROR_TIMEOUT)
+            {
+                // timeout
+                return false;
+            }
+
+            // in release treat unexpected results as spurious wakes
+            return true;
+        }
+
+        internal static void WakeByAddressSingle(int* address)
+        {
+            Interop.Mincore.WakeByAddressSingle(address);
+        }
+    }
+}