[wasm-mt] Support async JS interop on threadpool threads (#84494)

This is the last part of #84489 - initial runtime support for async JS interop on threadpool threads in multi-threaded WebAssembly.

Conceptually there are a few things here:

* A mechanism for the native runtime to start some threads as "exitable" so that we don't clean up when they return from their thread start function. (we will clean up when the pthreads TLS dtor runs)
* A change to make JSHostImplementation.s_csOwnedObjects a thread-static in multithreaded builds. This is because the map keys on a small integer handle that is allocated per-thread on the JS side as a handle for JS objects that are to be kept alive because managed code depends on them. (This is needed in general, but also makes the new smoke test work)
* A version of the PortableThreadPool.WorkerThread that starts as an exitable thread and uses asynchronous callbacks to wait for available threadpool work items and returns to the JS event loop periodically in order to allow JS promises to settle.
* A smoke test that does a background JS fetch and some async callbacks on it.

---


* [wasm-mt] Enable JSInterop on threadpool workers

   Implement PortableThreadPool loop using semaphore callbacks

   manage emscripten event loop from PortableThreadPool.WorkerThread

   make sure to keep the thread alive after setting up the semaphore wait.
   Cleanup the thread when exiting

   minimal sample - fetch on a background thread works

   Add WebWorkerEventLoop internal class to managed event loop keepalive

   Start threadpool threads with keepalive checks

   Add a flag to mono's thread start wrappers to keep track of threads that may not want cleanup to run after the Start function returns.

   Use the flag when starting threadpool threads.

   make minimal FetchBackground sample more like a unit test

   Set WorkerThread.IsIOPending when the current thread has unsettled JS interop promises.

   When IsIOPending is true, the worker will not exit even if it has no more work to do.  Instead it will repeatedly wait for more work to arrive or for all promises to settle.

   change minimal sample's fetch helper to artificially delay

   the delay is longer that the threadpool worker's semaphore timeout, in order to validate that the worker stays alive while there are unsettled promises

* [wasm-mt] Add background interop to smoketest

* update to use the LowLevelLifoAsyncWaitSemaphore

* adjust to renamed PortableThreadPool helper methods

* adjust to renamed WebWorkerEventLoop.HasJavaScriptInteropDependents

* extend and rationalize the smoke test a bit

   Add a test that just starts a thread and asserts that it has a different thread ID than the main thread.  This should allow us to rule out accidentally having the test pass on a single-threaded runtime

* hide some debug output

* smoke test: dispose of the ImportAsync result after the task is done

* [wasm-mt] make JSHostImplementation.s_csOwnedObjects ThreadStatic

   The integer jsHandles are not global - they are essentially indices into a JS array. So the mapping from a jsHandle to a JSObject must be per-thread.

   This fixes the thread affinity assertions in the smoketest (which were false positives - we looked up a worker's jsHandle and got back the main thread's JSObject - and then asserted that it was accessed from the wrong thread)

* remove locking on JSHostImplementation.CsOwnedObjects

   In single-threaded wasm, there is no need to lock since there is only one caller at a time.

   In multi-threaded wasm, the underlying dictionary is thread-static

* [threads] make the "external eventloop" platform independent

   It only does something on WASM, but in principle if other platforms allow us to run some code after returning from a thread start function, we could do it there, too.

* Add a Thread.HasExternalEventLoop managed property

   Set it from WebWorkerEventLoop.StartExitable.
   In native code, use it to set the `MONO_THREAD_CREATE_FLAGS_EXTERNAL_EVENTLOOP` flag when starting the thread.

* rename JSHostImplementation.ThreadCsOwnedObjects

   (used to be CsOwnedObjects)

   Rename to make it clear that it's objects owned by the current thread, not the runtime globally

* [checked] assert GC Safe mode, when returning to external eventloop

Author: lambdageek

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

@@ -2530,9 +2530,9 @@
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PortableThreadPool.ThreadCounts.cs" />
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PortableThreadPool.WaitThread.cs" />
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PortableThreadPool.WorkerThread.cs" />
-    <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PortableThreadPool.WorkerThread.NonBrowser.cs" />
+    <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PortableThreadPool.WorkerThread.NonBrowser.cs" Condition="'$(TargetsBrowser)' != 'true'"/>
     <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PortableThreadPool.WorkerTracking.cs" />
-    <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PortableThreadPool.Unix.cs" Condition="'$(TargetsUnix)' == 'true' or '$(TargetsBrowser)' == 'true' or '$(TargetsWasi)' == 'true'" />
+    <Compile Include="$(MSBuildThisFileDirectory)System\Threading\PortableThreadPool.Unix.cs" Condition="'$(TargetsUnix)' == 'true' or ('$(TargetsBrowser)' == 'true' and '$(FeatureWasmThreads)' != 'true') or '$(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'" />

src/libraries/System.Private.CoreLib/src/System/Threading/ThreadPool.Portable.cs

@@ -19,7 +19,9 @@ public static partial class ThreadPool
     {
         // Indicates whether the thread pool should yield the thread from the dispatch loop to the runtime periodically so that
         // the runtime may use the thread for processing other work
+#if !(TARGET_BROWSER && FEATURE_WASM_THREADS)
         internal static bool YieldFromDispatchLoop => false;
+#endif
 
 #if NATIVEAOT
         private const bool IsWorkerTrackingEnabledInConfig = false;

src/libraries/System.Runtime.InteropServices.JavaScript/src/System/Runtime/InteropServices/JavaScript/Interop/LegacyExports.cs

@@ -32,18 +32,15 @@ internal static void PreventTrimming()
 
         public static void GetCSOwnedObjectByJSHandleRef(nint jsHandle, int shouldAddInflight, out JSObject? result)
         {
-            lock (JSHostImplementation.s_csOwnedObjects)
+            if (JSHostImplementation.ThreadCsOwnedObjects.TryGetValue((int)jsHandle, out WeakReference<JSObject>? reference))
             {
-                if (JSHostImplementation.s_csOwnedObjects.TryGetValue((int)jsHandle, out WeakReference<JSObject>? reference))
+                reference.TryGetTarget(out JSObject? jsObject);
+                if (shouldAddInflight != 0)
                 {
-                    reference.TryGetTarget(out JSObject? jsObject);
-                    if (shouldAddInflight != 0)
-                    {
-                        jsObject?.AddInFlight();
-                    }
-                    result = jsObject;
-                    return;
+                    jsObject?.AddInFlight();
                 }
+                result = jsObject;
+                return;
             }
             result = null;
         }
@@ -77,14 +74,12 @@ public static void CreateCSOwnedProxyRef(nint jsHandle, LegacyHostImplementation
 
             JSObject? res = null;
 
-            lock (JSHostImplementation.s_csOwnedObjects)
+            if (!JSHostImplementation.ThreadCsOwnedObjects.TryGetValue((int)jsHandle, out WeakReference<JSObject>? reference) ||
+                !reference.TryGetTarget(out res) ||
+                res.IsDisposed)
             {
-                if (!JSHostImplementation.s_csOwnedObjects.TryGetValue((int)jsHandle, out WeakReference<JSObject>? reference) ||
-                    !reference.TryGetTarget(out res) ||
-                    res.IsDisposed)
-                {
 #pragma warning disable CS0612 // Type or member is obsolete
-                    res = mappedType switch
+                res = mappedType switch
                     {
                         LegacyHostImplementation.MappedType.JSObject => new JSObject(jsHandle),
                         LegacyHostImplementation.MappedType.Array => new Array(jsHandle),
@@ -95,8 +90,7 @@ public static void CreateCSOwnedProxyRef(nint jsHandle, LegacyHostImplementation
                         _ => throw new ArgumentOutOfRangeException(nameof(mappedType))
                     };
 #pragma warning restore CS0612 // Type or member is obsolete
-                    JSHostImplementation.s_csOwnedObjects[(int)jsHandle] = new WeakReference<JSObject>(res, trackResurrection: true);
-                }
+                JSHostImplementation.ThreadCsOwnedObjects[(int)jsHandle] = new WeakReference<JSObject>(res, trackResurrection: true);
             }
             if (shouldAddInflight != 0)
             {

src/libraries/System.Runtime.InteropServices.JavaScript/src/System/Runtime/InteropServices/JavaScript/JSHostImplementation.cs

@@ -15,7 +15,20 @@ internal static partial class JSHostImplementation
         private const string TaskGetResultName = "get_Result";
         private static MethodInfo? s_taskGetResultMethodInfo;
         // we use this to maintain identity of JSHandle for a JSObject proxy
-        public static readonly Dictionary<int, WeakReference<JSObject>> s_csOwnedObjects = new Dictionary<int, WeakReference<JSObject>>();
+#if FEATURE_WASM_THREADS
+        [ThreadStatic]
+#endif
+        private static Dictionary<int, WeakReference<JSObject>>? s_csOwnedObjects;
+
+        public static Dictionary<int, WeakReference<JSObject>> ThreadCsOwnedObjects
+        {
+            get
+            {
+                s_csOwnedObjects ??= new ();
+                return s_csOwnedObjects;
+            }
+        }
+
         // we use this to maintain identity of GCHandle for a managed object
         public static Dictionary<object, IntPtr> s_gcHandleFromJSOwnedObject = new Dictionary<object, IntPtr>(ReferenceEqualityComparer.Instance);
 
@@ -24,10 +37,7 @@ public static void ReleaseCSOwnedObject(nint jsHandle)
         {
             if (jsHandle != IntPtr.Zero)
             {
-                lock (s_csOwnedObjects)
-                {
-                    s_csOwnedObjects.Remove((int)jsHandle);
-                }
+                ThreadCsOwnedObjects.Remove((int)jsHandle);
                 Interop.Runtime.ReleaseCSOwnedObject(jsHandle);
             }
         }
@@ -175,17 +185,14 @@ public static unsafe void FreeMethodSignatureBuffer(JSFunctionBinding signature)
 
         public static JSObject CreateCSOwnedProxy(nint jsHandle)
         {
-            JSObject? res = null;
+            JSObject? res;
 
-            lock (s_csOwnedObjects)
+            if (!ThreadCsOwnedObjects.TryGetValue((int)jsHandle, out WeakReference<JSObject>? reference) ||
+                !reference.TryGetTarget(out res) ||
+                res.IsDisposed)
             {
-                if (!s_csOwnedObjects.TryGetValue((int)jsHandle, out WeakReference<JSObject>? reference) ||
-                    !reference.TryGetTarget(out res) ||
-                    res.IsDisposed)
-                {
-                    res = new JSObject(jsHandle);
-                    s_csOwnedObjects[(int)jsHandle] = new WeakReference<JSObject>(res, trackResurrection: true);
-                }
+                res = new JSObject(jsHandle);
+                ThreadCsOwnedObjects[(int)jsHandle] = new WeakReference<JSObject>(res, trackResurrection: true);
             }
             return res;
         }

src/libraries/System.Runtime.InteropServices.JavaScript/src/System/Runtime/InteropServices/JavaScript/Legacy/LegacyHostImplementation.cs

@@ -21,10 +21,7 @@ public static void ReleaseInFlight(object obj)
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static void RegisterCSOwnedObject(JSObject proxy)
         {
-            lock (JSHostImplementation.s_csOwnedObjects)
-            {
-                JSHostImplementation.s_csOwnedObjects[(int)proxy.JSHandle] = new WeakReference<JSObject>(proxy, trackResurrection: true);
-            }
+            JSHostImplementation.ThreadCsOwnedObjects[(int)proxy.JSHandle] = new WeakReference<JSObject>(proxy, trackResurrection: true);
         }
 
         public static MarshalType GetMarshalTypeFromType(Type type)

src/mono/System.Private.CoreLib/System.Private.CoreLib.csproj

@@ -281,6 +281,9 @@
   <ItemGroup Condition="('$(TargetsBrowser)' == 'true'  or '$(TargetsWasi)' == 'true') and '$(FeatureWasmThreads)' == 'true'">
       <Compile Include="$(BclSourcesRoot)\System\Threading\ThreadPoolBoundHandle.Browser.Threads.Mono.cs" />
       <Compile Include="$(BclSourcesRoot)\System\Threading\LowLevelLifoAsyncWaitSemaphore.Browser.Threads.Mono.cs" />
+      <Compile Include="$(BclSourcesRoot)\System\Threading\PortableThreadPool.Browser.Threads.Mono.cs" />
+      <Compile Include="$(BclSourcesRoot)\System\Threading\PortableThreadPool.WorkerThread.Browser.Threads.Mono.cs" />
+      <Compile Include="$(BclSourcesRoot)\System\Threading\ThreadPool.Browser.Threads.Mono.cs" />
       <Compile Include="$(BclSourcesRoot)\System\Threading\WebWorkerEventLoop.Browser.Threads.Mono.cs" />
   </ItemGroup>
   <ItemGroup Condition="('$(TargetsBrowser)' == 'true' or '$(TargetsWasi)' == 'true') and '$(FeatureWasmThreads)' != 'true'">

src/mono/System.Private.CoreLib/src/System/Threading/PortableThreadPool.Browser.Threads.Mono.cs

@@ -0,0 +1,19 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+
+namespace System.Threading;
+
+internal sealed partial class PortableThreadPool
+{
+    private static partial class WorkerThread
+    {
+        private static bool IsIOPending => WebWorkerEventLoop.HasJavaScriptInteropDependents;
+    }
+
+    private struct CpuUtilizationReader
+    {
+#pragma warning disable CA1822
+        public double CurrentUtilization => 0.0; // FIXME: can we do better
+#pragma warning restore CA1822
+    }
+}

src/mono/System.Private.CoreLib/src/System/Threading/PortableThreadPool.WorkerThread.Browser.Threads.Mono.cs

@@ -0,0 +1,122 @@
+// 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.Diagnostics.CodeAnalysis;
+using System.Diagnostics.Tracing;
+using System.Runtime.CompilerServices;
+
+namespace System.Threading
+{
+    internal sealed partial class PortableThreadPool
+    {
+        /// <summary>
+        /// The worker thread infastructure for the CLR thread pool.
+        /// </summary>
+        private static partial class WorkerThread
+        {
+            /// <summary>
+            /// Semaphore for controlling how many threads are currently working.
+            /// </summary>
+            private static readonly LowLevelLifoAsyncWaitSemaphore s_semaphore =
+                new LowLevelLifoAsyncWaitSemaphore(
+                    0,
+                    MaxPossibleThreadCount,
+                    AppContextConfigHelper.GetInt32Config(
+                        "System.Threading.ThreadPool.UnfairSemaphoreSpinLimit",
+                        SemaphoreSpinCountDefault,
+                        false),
+                    onWait: () =>
+                    {
+                        if (NativeRuntimeEventSource.Log.IsEnabled())
+                        {
+                            NativeRuntimeEventSource.Log.ThreadPoolWorkerThreadWait(
+                                (uint)ThreadPoolInstance._separated.counts.VolatileRead().NumExistingThreads);
+                        }
+                    });
+
+            private static readonly ThreadStart s_workerThreadStart = WorkerThreadStart;
+
+            private sealed record SemaphoreWaitState(PortableThreadPool ThreadPoolInstance, LowLevelLock ThreadAdjustmentLock, WebWorkerEventLoop.KeepaliveToken KeepaliveToken)
+            {
+                public bool SpinWait = true;
+
+                public void ResetIteration() {
+                    SpinWait = true;
+                }
+            }
+
+            private static void WorkerThreadStart()
+            {
+                Thread.CurrentThread.SetThreadPoolWorkerThreadName();
+
+                PortableThreadPool threadPoolInstance = ThreadPoolInstance;
+
+                if (NativeRuntimeEventSource.Log.IsEnabled())
+                {
+                    NativeRuntimeEventSource.Log.ThreadPoolWorkerThreadStart(
+                        (uint)threadPoolInstance._separated.counts.VolatileRead().NumExistingThreads);
+                }
+
+                LowLevelLock threadAdjustmentLock = threadPoolInstance._threadAdjustmentLock;
+                var keepaliveToken = WebWorkerEventLoop.KeepalivePush();
+                SemaphoreWaitState state = new(threadPoolInstance, threadAdjustmentLock, keepaliveToken) { SpinWait = true };
+                // set up the callbacks for semaphore waits, tell
+                // emscripten to keep the thread alive, and return to
+                // the JS event loop.
+                WaitForWorkLoop(s_semaphore, state);
+                // return from thread start with keepalive - the thread will stay alive in the JS event loop
+            }
+
+            private static readonly Action<LowLevelLifoAsyncWaitSemaphore, object?> s_WorkLoopSemaphoreSuccess = new(WorkLoopSemaphoreSuccess);
+            private static readonly Action<LowLevelLifoAsyncWaitSemaphore, object?> s_WorkLoopSemaphoreTimedOut = new(WorkLoopSemaphoreTimedOut);
+
+            private static void WaitForWorkLoop(LowLevelLifoAsyncWaitSemaphore semaphore, SemaphoreWaitState state)
+            {
+                semaphore.PrepareAsyncWait(ThreadPoolThreadTimeoutMs, s_WorkLoopSemaphoreSuccess, s_WorkLoopSemaphoreTimedOut, state);
+                // thread should still be kept alive
+                Debug.Assert(state.KeepaliveToken.Valid);
+            }
+
+            private static void WorkLoopSemaphoreSuccess(LowLevelLifoAsyncWaitSemaphore semaphore, object? stateObject)
+            {
+                SemaphoreWaitState state = (SemaphoreWaitState)stateObject!;
+                WorkerDoWork(state.ThreadPoolInstance, ref state.SpinWait);
+                // Go around the loop one more time, keeping existing mutated state
+                WaitForWorkLoop(semaphore, state);
+            }
+
+            private static void WorkLoopSemaphoreTimedOut(LowLevelLifoAsyncWaitSemaphore semaphore, object? stateObject)
+            {
+                SemaphoreWaitState state = (SemaphoreWaitState)stateObject!;
+                if (ShouldExitWorker(state.ThreadPoolInstance, state.ThreadAdjustmentLock)) {
+                    // we're done, kill the thread.
+
+                    // we're wrapped in an emscripten eventloop handler which will consult the
+                    // keepalive count, destroy the thread and run the TLS dtor which will
+                    // unregister the thread from Mono
+                    state.KeepaliveToken.Pop();
+                    return;
+                } else {
+                    // more work showed up while we were shutting down, go around one more time
+                    state.ResetIteration();
+                    WaitForWorkLoop(semaphore, state);
+                }
+            }
+
+            private static void CreateWorkerThread()
+            {
+                // Thread pool threads must start in the default execution context without transferring the context, so
+                // using captureContext: false.
+                Thread workerThread = new Thread(s_workerThreadStart);
+                workerThread.IsThreadPoolThread = true;
+                workerThread.IsBackground = true;
+                // thread name will be set in thread proc
+
+                // This thread will return to the JS event loop - tell the runtime not to cleanup
+                // after the start function returns, if the Emscripten keepalive is non-zero.
+                WebWorkerEventLoop.StartExitable(workerThread, captureContext: false);
+            }
+        }
+    }
+}

src/mono/System.Private.CoreLib/src/System/Threading/Thread.Mono.cs

@@ -37,6 +37,7 @@ public partial class Thread
         private int interruption_requested;
         private IntPtr longlived;
         internal bool threadpool_thread;
+        internal bool external_eventloop; // browser-wasm: thread will return to the JS eventloop
         /* These are used from managed code */
         internal byte apartment_state;
         internal int managed_id;
@@ -352,5 +353,17 @@ private static void SpinWait_nop()
         private static extern void SetPriority(Thread thread, int priority);
 
         internal int GetSmallId() => small_id;
+
+        internal bool HasExternalEventLoop
+        {
+            get
+            {
+                return external_eventloop;
+            }
+            set
+            {
+                external_eventloop = value;
+            }
+        }
     }
 }

src/mono/System.Private.CoreLib/src/System/Threading/ThreadPool.Browser.Threads.Mono.cs

@@ -0,0 +1,13 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+
+namespace System.Threading
+{
+    public static partial class ThreadPool
+    {
+        // Indicates that the threadpool should yield the thread from the dispatch loop to the
+        // runtime periodically.  We use this to return back to the JS event loop so that the JS
+        // event queue can be drained
+        internal static bool YieldFromDispatchLoop => true;
+    }
+}