Develop Reference

When are ThreadPool threads are released back to the ThreadPool?

Can anyone explain (or have a resource that explains) exactly when ThreadPool threads are released back to the ThreadPool? Here is a small example program (Dotnet fiddle: https://dotnetfiddle.net/XRso3q)
public static async Task Main()
{
Console.WriteLine("Start");
var t1 = ShortWork("SW1");
var t2 = ShortWork("SW2");
await Task.Delay(50);
var t3 = LongWork("LW1");
Console.WriteLine($"After starting LongWork Thread={Thread.CurrentThread.ManagedThreadId}");
await Task.WhenAll(t1, t2);
await t3;
Console.WriteLine("Done");
}
public static async Task ShortWork(string name)
{
Console.WriteLine($"SHORT Start {name} Thread={Thread.CurrentThread.ManagedThreadId}");
await Task.Delay(500);
Console.WriteLine($"SHORT End {name} Thread={Thread.CurrentThread.ManagedThreadId}");
}
public static async Task LongWork(string name)
{
Console.WriteLine($"LONG Start {name} Thread={Thread.CurrentThread.ManagedThreadId}");
await Task.Delay(2500);
Console.WriteLine($"LONG End {name} Thread={Thread.CurrentThread.ManagedThreadId}");
}
Outputs:
Start
SHORT Start SW1 Thread=1
SHORT Start SW2 Thread=1
LONG Start LW1 Thread=5
After starting LongWork Thread=5
SHORT End SW1 Thread=7
SHORT End SW2 Thread=5
LONG End LW1 Thread=5
Done
Long work starts on thread 5, but at some point thread 5 is released back to the threadpool as thread 5 is able to pick up Short SW1 ending. When exactly does 5 get released back to the threadpool after await Task.Delay(2500) in LongWork? Does the await call release it back to the threadpool? I dont think this is the case as if I log the thread id right after the call to LongWork, that is still running on thread 5. await Task.WhenAll is called on thread 5 - which then releases control back up to whatever called 'main' - is this where it gets released as there is no 'caller' to go back to?
My understanding of what happens:
Starts on thread 1, thread 1 executes ShortWork SW1 and SW2.
Task.Delay(50) is awaited and thread 1 gets released (as there is no more work to do?)
Thread 5 is chosen to pick up the continuation after the 50ms delay
Thread 5 kicks off LongWork, and it it gets to the awaited 2500ms delay. Control gets released back up to main, still on thread 5. t1 and t2 are awaited - control gets released back up to whatever called main (and so thread 5's work is done - it gets released to the threadpool)
At this point no threads are 'doing' anything
When the ShortWork delay is done, thread 5 and 7 are selected from the pool for the continuation of each call. Once done with the continuation, these are released to the pool (?)
Another thread picks up the continuation between Task.WhenAll and await t3, which then immediately gets released because it is just awaiting t3
A ThreadPool thread is selected to do the continuation of the LongWork call
Finally, a ThreadPool thread picks up the last work to write done after t3 is done.
Also as a bonus, why does 5 pick up end of SW1, and 7 pick up end of LW1? These are the threads that were just used. Are they somehow kept as 'hot threads and prioritised for continuations that come up?

The way await works is that it first checks its awaitable argument; if it has already completed, then the async method continues synchronously. If it is not complete, then the method returns to its caller.
A second key to understanding is that all async methods begin executing synchronously, on a normal call stack just like any other method.
The third useful piece of information here is that a Console app needs a foreground thread to keep running or it will exit. So when you have an async Main, behind the scenes the runtime blocks the main thread on the returned task. So, in your example, when the first await is hit in Main, it returns a task, and the main thread 1 spends the rest of the time blocked on that task.
In this code, all continuations are run by thread pool threads. It's not specified or guaranteed which thread(s) will run which continuations.
The current implementation uses synchronous continuations, so in your example the thread id for LONG Start LW1 and After starting LongWork will always be the same. You can even place a breakpoint on After starting LongWork and see how a LongWork continuation is actually in the call stack of your Main continuation.

What actually happens is that when a pool thread starts, it takes a task from the pool's work queue, waiting if necessary. It then executes the task, and then takes a new one, again waiting if necessary. This is repeated until the thread determines that it has to die and exits.
While the thread is taking a new task, or waiting for one, you could say that it has been "released to the thread pool", but that's not really useful or helpful. Nothing from the pool actually does things to the threads, after starting them up. The threads control themselves.
When you write an async function, the compiler transforms it in a way that divides it into many tasks, each of which will be executed by a function call. Where you write Task.Delay(), what actually happens is that your function schedules the task that represents the remainder of its execution, and then returns all the way out to the thread's main procedure, allowing it to get a new task from the thread pool's work queue.

Two Tasks run on the same thread which invalidates lock

Edit
I find Building Async Coordination Primitives, Part 1: AsyncManualResetEvent might be related to my topic.
In the case of TaskCompletionSource, that means that synchronous continuations can happen as part of a call to {Try}Set*, which means in our AsyncManualResetEvent example, those continuations could execute as part of the Set method. Depending on your needs (and whether callers of Set may be ok with a potentially longer-running Set call as all synchronous continuations execute), this may or may not be what you want.
Many thanks to all of the answers, thank you for your knowledge and patience!
Original Question
I know that Task.Run runs on a threadpool thread and threads can have re-entrancy. But I never knew that 2 tasks can run on the same thread when they are both alive!
My Question is: is that reasonable by design? Does that mean lock inside an async method is meaningless (or say, lock cannot be trusted in async method block, if I'd like a method that doesn't allow reentrancy)?
Code:
namespace TaskHijacking
{
class Program
{
static TaskCompletionSource<bool> tcs = new TaskCompletionSource<bool>();
static object methodLock = new object();
static void MethodNotAllowReetrance(string callerName)
{
lock(methodLock)
{
Console.WriteLine($"Enter MethodNotAllowReetrance, caller: {callerName}, on thread: {Thread.CurrentThread.ManagedThreadId}");
if (callerName == "task1")
{
tcs.SetException(new Exception("Terminate tcs"));
}
Thread.Sleep(1000);
Console.WriteLine($"Exit MethodNotAllowReetrance, caller: {callerName}, on thread: {Thread.CurrentThread.ManagedThreadId}");
}
}
static void Main(string[] args)
{
var task1 = Task.Run(async () =>
{
await Task.Delay(1000);
MethodNotAllowReetrance("task1");
});
var task2 = Task.Run(async () =>
{
try
{
await tcs.Task; // await here until task SetException on tcs
}
catch
{
// Omit the exception
}
MethodNotAllowReetrance("task2");
});
Task.WaitAll(task1, task2);
Console.ReadKey();
}
}
}
Output:
Enter MethodNotAllowReetrance, caller: task1, on thread: 6
Enter MethodNotAllowReetrance, caller: task2, on thread: 6
Exit MethodNotAllowReetrance, caller: task2, on thread: 6
Exit MethodNotAllowReetrance, caller: task1, on thread: 6
The control flow of the thread 6 is shown in the figure:

You already have several solutions. I just want to describe the problem a bit more. There are several factors at play here that combine to cause the observed re-entrancy.
First, lock is re-entrant. lock is strictly about mutual exclusion of threads, which is not the same as mutual exclusion of code. I think re-entrant locks are a bad idea in the 99% case (as described on my blog), since developers generally want mutual exclusion of code and not threads. SemaphoreSlim, since it is not re-entrant, mutually excludes code. IMO re-entrant locks are a holdover from decades ago, when they were introduced as an OS concept, and the OS is just concerned about managing threads.
Next, TaskCompletionSource<T> by default invokes task continuations synchronously.
Also, await will schedule its method continuation as a synchronous task continuation (as described on my blog).
Task continuations will sometimes run asynchronously even if scheduled synchronously, but in this scenario they will run synchronously. The context captured by await is the thread pool context, and the completing thread (the one calling TCS.TrySet*) is a thread pool thread, and in that case the continuation will almost always run synchronously.
So, you end up with a thread that takes a lock, completes a TCS, thus executing the continuations of that task, which includes continuing another method, which is then able to take that same lock.
To repeat the existing solutions in other answers, to solve this you need to break that chain at some point:
(OK) Use a non-reentrant lock. SemaphoreSlim.WaitAsync will still execute the continuations while holding the lock (not a good idea), but since SemaphoreSlim isn't re-entrant, the method continuation will (asynchronously) wait for the lock to be available.
(Best) Use TaskCompletionSource.RunContinuationsAsynchronously, which will force task continuations onto a (different) thread pool thread. This is a better solution because your code is no longer invoking arbitrary code while holding a lock (i.e., the task continuations).
You can also break the chain by using a non-thread-pool context for the method awaiting the TCS. E.g., if that method had to resume on a UI thread, then it could not be run synchronously from a thread pool thread.
From a broader perspective, if you're mixing locks and TaskCompletionSource instances, it sounds like you may be building (or may need) an asynchronous coordination primitive. I have an open-source library that implements a bunch of them, if that helps.

A task is an abstraction over some amount of work. Usually this means that the work is split into parts, where the execution can be paused and resumed between parts. When resuming it may very well run on another thread. But the pausing/resuming may only be done at the await statements. Notably, while the task is 'paused', for example because it is waiting for IO, it does not consume any thread at all, it will only use a thread while it is actually running.
My Question is: is that reasonable by design? Does that mean lock inside an async method is meaningless?
locks inside a async method is far from meaningless since it allows you to ensure a section of code is only run by one thread at a time.
In your first example there can be only one thread that has the lock at a time. While the lock is held that task cannot be paused/resumed since await is not legal while in a lock body. So a single thread will execute the entire lock body, and that thread cannot do anything else until it completes the lock body. So there is no risk of re-entrancy unless you invoke some code that can call back to the same method.
In your updated example the problem occurs due to TaskCompletionSource.SetException, this is allowed to reuse the current thread to run any continuation of the task immediately. To avoid this, and many other issues, make sure you only hold the lock while running a limited amount of code. Any method calls that may run arbitrary code risks causing deadlocks, reentrancy, and many other problems.
You can solve the specific problem by using a ManualResetEvent(Slim) to do the signaling between threads instead of using a TaskCompletionSource.

So your method is basically like this:
static void MethodNotAllowReetrance()
{
lock (methodLock) tcs.SetResult();
}
...and the tcs.Task has a continuation attached that invokes the MethodNotAllowReetrance. What happens then is the same thing that would happen if your method was like this instead:
static void MethodNotAllowReetrance()
{
lock (methodLock) MethodNotAllowReetrance();
}
The moral lesson is that you must be very careful every time you invoke any method inside a lock-protected region. In this particular case you have a couple of options:
Don't complete the TaskCompletionSource while holding the lock. Defer its completion until after you have exited the protected region:
static void MethodNotAllowReetrance()
{
bool doComplete = false;
lock (methodLock) doComplete = true;
if (doComplete) tcs.SetResult();
}
Configure the TaskCompletionSource so that it invokes its continuations asynchronously, by passing the TaskCreationOptions.RunContinuationsAsynchronously in its constructor. This is an option that you don't have very often. For example when you cancel a CancellationTokenSource, you don't have the option to invoke asynchronously the callbacks registered to its associated CancellationToken.
Refactor the MethodNotAllowReetrance method in a way that it can handle reentrancy.

Use SemaphoreSlim instead of lock, since, as the documentation says:
The SemaphoreSlim class doesn't enforce thread or task identity
In your case, it would look something like this:
// Semaphore only allows one request to enter at a time
private static readonly SemaphoreSlim _semaphoreSlim = new SemaphoreSlim(1, 1);
void SyncMethod() {
_semaphoreSlim.Wait();
try {
// Do some sync work
} finally {
_semaphoreSlim.Release();
}
}
The try/finally block is optional, but it makes sure that the semaphore is released even if an exception is thrown somewhere in your code.
Note that SemaphoreSlim also has a WaitAsync() method, if you want to wait asynchronously to enter the semaphore.

Why this async method don't cause deadlock on a thread?

I need an explanation why this code doesn't cause deadlock on the thread it's running in
(It's a WinForm application and these happens in button_Click):
Task FakeMainThread = Task.Run(async() => // fake main thread
{
async Task asyncMethod() // executed in FakeMainThread
{
await Task.Run(() => // 'await'ing in FakeMainThread
{
Thread.Sleep(5000);
MessageBox.Show("child Task finished");
return 3;
});
MessageBox.Show("asyncMethod finished");
}
asyncMethod().Wait(); // deadlock should appear in FakeMainThread
MessageBox.Show("FakeMainThread finished");
return 4;
});
asyncMethod().Wait(); should block the fakeMainThread Task's thread while it's await another task to finish and since await is happening in fakeMainThread too it shouldn't be able to 'await' and deadlock should appear on fakeMainThread. But it doesn't happen and I see bot MessageBox.Show("asyncMethod finished"); and MessageBox.Show("FakeMainThread finished"); messages.
Note: if I put this:
async Task asyncMethod()
{
await Task.Run(() =>
{
Thread.Sleep(5000);
MessageBox.Show("child Task finished");
return 3;
});
MessageBox.Show("asyncMethod finished");
}
asyncMethod().Wait(); // deadlock appears in Main Thread (UI)
MessageBox.Show("FakeMainThread finished");
in main button1_click() inside directli deadlock appears on UI. Thanks!

You don't experience a deadlock in the first code example because of the SynchronizationContext. This context says what the await must do to restore your code. When you start a new task (Task.Run), you will get the default context. In the button1_click you get the context from the form.
The context for the form allows to execute only a single thread (the one used to paint/update your form).
When you call .Wait(), then you keep that thread 'locked' with waiting until the task is done. That means that the thread is not released for another job and this is one of the reasons that the form goes into the 'not responding' state.
When you do an await [code], and that code is done, then it will ask the synchronization context to schedule the remainder of the code.
In case of the default context, any free thread is taken and your code will continue, the task is marked as done, and thus the 'loop' in the .Wait() gets signaled that the task is done and continues.
In case of the forms context, there is only a single thread, thus the completion of this task is scheduled to run after the current code is done. This is the reason of the deadlock.
Avoiding this deadlock is the main reason that you most often get the suggestion to use ConfigureAwait(false), this tells the await that it should substitute the current synchronization context with the default one, thus allowing you to use any free thread.
The only reason you don't want to use this (or say ConfigureAwait(true)) is if you need to update something on your form (WPF or WinForms) or you need your http context (ASP.NET, not ASP.NET Core). This is because only a single thread has access to your form or http context. Here you don't get an exception on your MessageBox.Show because this is 'outside' of the forms context and doesn't need this special thread/synchronization context.
See https://devblogs.microsoft.com/dotnet/configureawait-faq/ for more information.

What's the difference between starting and awaiting a Task?

What's the difference between starting and awaiting? Code below taken from Stephen Cleary's blog (including comments)
public async Task DoOperationsConcurrentlyAsync()
{
Task[] tasks = new Task[3];
tasks[0] = DoOperation0Async();
tasks[1] = DoOperation1Async();
tasks[2] = DoOperation2Async();
// At this point, all three tasks are running at the same time.
// Now, we await them all.
await Task.WhenAll(tasks);
}
I thought that the tasks begin running when you await them ... but the comments in the code seem to imply otherwise.
Also, how can the tasks be running after I just attributed them to an array of type Task. Isn't that just an attribution, by nature not involving action?

A Task returns "hot" (i.e. already started). await asynchronously waits for the Task to complete.
In your example, where you actually do the await will affect whether the tasks are ran one after the other, or all of them at the same time:
await DoOperation0Async(); // start DoOperation0Async, wait for completion, then move on
await DoOperation1Async(); // start DoOperation1Async, wait for completion, then move on
await DoOperation2Async(); // start DoOperation2Async, wait for completion, then move on
As opposed to:
tasks[0] = DoOperation0Async(); // start DoOperation0Async, move on without waiting for completion
tasks[1] = DoOperation1Async(); // start DoOperation1Async, move on without waiting for completion
tasks[2] = DoOperation2Async(); // start DoOperation2Async, move on without waiting for completion
await Task.WhenAll(tasks); // wait for all of them to complete
Update
"doesn't await make an async operation... behave like sync, in this example (and not only)? Because we can't (!) run anything else in parallel with DoOperation0Async() in the first case. By comparison, in the 2nd case DoOperation0Async() and DoOperation1Async() run in parallel (e.g. concurrency,the main benefits of async?)"
This is a big subject and a question worth being asked as it's own thread on SO as it deviates from the original question of the difference between starting and awaiting tasks - therefore I'll keep this answer short, while referring you to other answers where appropriate.
No, awaiting an async operation does not make it behave like sync; what these keywords do is enabling developers to write asynchronous code that resembles a synchronous workflow (see this answer by Eric Lippert for more).
Calling await DoOperation0Async() will not block the thread executing this code flow, whereas a synchronous version of DoOperation0 (or something like DoOperation0Async.Result) will block the thread until the operation is complete.
Think about this in a web context. Let's say a request arrives in a server application. As part of producing a response to that request, you need to do a long-running operation (e.g. query an external API to get some value needed to produce your response). If the execution of this long-running operation was synchronous, the thread executing your request would block as it would have to wait for the long-running operation to complete. On the other hand, if the execution of this long-running operation was asynchronous, the request thread could be freed up so it could do other things (like service other requests) while the long-running operation was still running. Then, when the long-running operation would eventually complete, the request thread (or possibly another thread from the thread pool) could pick up from where it left off (as the long-running operation would be complete and it's result would now be available) and do whatever work was left to produce the response.
The server application example also addresses the second part of your question about the main benefits of async - async/await is all about freeing up threads.

Isn't that just an attribution, by nature not involving action?
By calling the async method you execute the code within. Usually down the chain one method will create a Task and return it either by using return or by awaiting.
Starting a Task
You can start a Task by using Task.Run(...). This schedules some work on the Task Thread Pool.
Awaiting a Task
To get a Task you usually call some (async) Method that returns a Task. An async method behaves like a regular method until you await (or use Task.Run() ). Note that if you await down a chain of methods and the "final" method only does a Thread.Sleep() or synchronous operation - then you will block the initial calling thread, because no method ever used the Task's Thread Pool.
You can do some actual asynchronous operation in many ways:
using Task.Run
using Task.Delay
using Task.Yield
call a library that offers asynchronous operations
These are the ones that come to my mind, there are probably more.
By example
Let's assume that Thread ID 1 is the main thread where you are calling MethodA() from. Thread IDs 5 and up are Threads to run Tasks on (System.Threading.Tasks provides a default Scheduler for that).
public async Task MethodA()
{
// Thread ID 1, 0s passed total
var a = MethodB(); // takes 1s
// Thread ID 1, 1s passed total
await Task.WhenAll(a); // takes 2s
// Thread ID 5, 3s passed total
// When the method returns, the SynchronizationContext
// can change the Thread - see below
}
public async Task MethodB()
{
// Thread ID 1, 0s passed total
Thread.Sleep(1000); // simulate blocking operation for 1s
// Thread ID 1, 1s passed total
// the await makes MethodB return a Task to MethodA
// this task is run on the Task ThreadPool
await Task.Delay(2000); // simulate async call for 2s
// Thread ID 2 (Task's pool Thread), 3s passed total
}
We can see that MethodA was blocked on the MethodB until we hit an await statement.
Await, SynchronizationContext, and Console Apps
You should be aware of one feature of Tasks. They make sure to invoke back to a SynchronizationContext if one is present (basically non-console apps). You can easily run into a deadlock when using .Result or .Wait() on a Task if the called code does not take measures. See https://blogs.msdn.microsoft.com/pfxteam/2012/01/20/await-synchronizationcontext-and-console-apps/
async/await as syntactic sugar
await basically just schedules following code to run after the call was completed. Let me illustrate the idea of what is happening behind the scenes.
This is the untransformed code using async/await. The Something method is awaited, so all following code (Bye) will be run after Something completed.
public async Task SomethingAsync()
{
Hello();
await Something();
Bye();
}
To explain this I add a utility class Worker that simply takes some action to run and then notify when done.
public class Worker
{
private Action _action;
public event DoneHandler Done;
// skipping defining DoneHandler delegate
// store the action
public Worker(Action action) => _action = action;
public void Run()
{
// execute the action
_action();
// notify so that following code is run
Done?.Invoke();
}
}
Now our transformed code, not using async/await
public Task SomethingAsync()
{
Hello(); // this remains untouched
// create the worker to run the "awaited" method
var worker = new Worker(() => Something());
// register the rest of our method
worker.Done += () => Bye();
// execute it
worker.Run();
// I left out the part where we return something
// or run the action on a threadpool to keep it simple
}

Here's the short answer:
To answer this you just need to understand what the async / await keywords do.
We know a single thread can only do one thing at a time and we also know that a single thread bounces all over the application to various method calls and events, ETC. This means that where the thread needs to go next is most likely scheduled or queued up somewhere behind the scenes (it is but I won't explain that part here.) When a thread calls a method, that method is ran to completion before any other methods can be ran which is why long running methods are preferred to be dispatched to other threads to prevent the application from freezing. In order to break a single method up into separate queues we need to do some fancy programming OR you can put the async signature on the method. This tells the compiler that at some point the method can be broken up into other methods and placed in a queue to be ran later.
If that makes sense then you're already figuring out what await does... await tells the compiler that this is where the method is going to be broken up and scheduled to run later. This is why you can use the async keyword without the await keyword; although the compiler knows this and warns you. await does all this for you by use of a Task.
How does await use a Task tell the compiler to schedule the rest of the method? When you call await Task the compilers calls the Task.GetAwaiter() method on that Task for you. GetAwaiter() return a TaskAwaiter. The TaskAwaiter implements two interfaces ICriticalNotifyCompletion, INotifyCompletion. Each has one method, UnsafeOnCompleted(Action continuation) and OnCompleted(Action continuation). The compiler then wraps the rest of the method (after the await keyword) and puts it in an Action and then it calls the OnCompleted and UnsafeOnCompleted methods and passes that Action in as a parameter. Now when the Task is complete, if successful it calls OnCompleted and if not it calls UnsafeOnCompleted and it calls those on the same thread context used to start the Task. It uses the ThreadContext to dispatch the thread to the original thread.
Now you can understand that neither async or await execute any Tasks. They simply tell the compiler to use some prewritten code to schedule all of it for you. In fact; you can await a Task that's not running and it will await until the Task is executed and completed or until the application ends.
Knowing this; lets get hacky and understand it deeper by doing what async await does manually.
Using async await
using System;
using System.Threading.Tasks;
namespace Question_Answer_Console_App
{
class Program
{
static void Main(string[] args)
{
Test();
Console.ReadKey();
}
public static async void Test()
{
Console.WriteLine($"Before Task");
await DoWorkAsync();
Console.WriteLine($"After Task");
}
static public Task DoWorkAsync()
{
return Task.Run(() =>
{
Console.WriteLine($"{nameof(DoWorkAsync)} starting...");
Task.Delay(1000).Wait();
Console.WriteLine($"{nameof(DoWorkAsync)} ending...");
});
}
}
}
//OUTPUT
//Before Task
//DoWorkAsync starting...
//DoWorkAsync ending...
//After Task
Doing what the compiler does manually (sort of)
Note: Although this code works it is meant to help you understand async await from a top down point of view. It DOES NOT encompass or execute the same way the compiler does verbatim.
using System;
using System.Threading.Tasks;
namespace Question_Answer_Console_App
{
class Program
{
static void Main(string[] args)
{
Test();
Console.ReadKey();
}
public static void Test()
{
Console.WriteLine($"Before Task");
var task = DoWorkAsync();
var taskAwaiter = task.GetAwaiter();
taskAwaiter.OnCompleted(() => Console.WriteLine($"After Task"));
}
static public Task DoWorkAsync()
{
return Task.Run(() =>
{
Console.WriteLine($"{nameof(DoWorkAsync)} starting...");
Task.Delay(1000).Wait();
Console.WriteLine($"{nameof(DoWorkAsync)} ending...");
});
}
}
}
//OUTPUT
//Before Task
//DoWorkAsync starting...
//DoWorkAsync ending...
//After Task
LESSON SUMMARY:
Note that the method in my example DoWorkAsync() is just a function that returns a Task. In my example the Task is running because in the method I use return Task.Run(() =>…. Using the keyword await does not change that logic. It's exactly the same; await only does what I mentioned above.
If you have any questions just ask and I'll be happy to answer them.

With starting you start a task. That means it might be picked up for execution by whatever Multitasaking system is in place.
With waiting, you wait for one task to actually finish before you continue.
There is no such thing as a Fire and Forget Thread. You always need to come back, to react to exceptions or do somethings with the result of the asynchronous operation (Database Query or WebQuery result, FileSystem operation finished, Dokument send to the nearest printer pool).
You can start and have as many task running in paralell as you want. But sooner or later you will require the results before you can go on.

If an async method is single threaded how can it be run in the background?

I'm trying to understand async/await and have read a number of articles but am still confused about the synchronous/asynchronous nature.
I have the following test console app:
static void Main(string[] args)
{
var test = FooAsync();
Console.WriteLine("After FooAsync");
for (int i = 0; i < 100; i++)
Console.WriteLine("After that");
Console.ReadKey();
}
private static async Task FooAsync()
{
Console.WriteLine("Before delay");
await Task.Delay(1);
Console.WriteLine("After delay");
}
The code gives output along the lines of:
Before delay
After FooAsync
After that
After that
After that
After that
After delay
After that
.
.
I understand that async/await will not create a separate thread for processing and that at the point FooAsync reaches the await Task.Delay(1) line it will return back to Main as the task will not yet have completed, however, as we are only running on a single thread can someone explain what triggers the FooAsync method to resume at some arbitrary point within Main before Main can then continue?
Update
I take it back and i3arnon and dariogriffo are correct. The code does use multiple threads (as I'd have seen before had looked in the debugger or done the obvious as kha suggested). I'd been confused by the Threads section on the following page https://msdn.microsoft.com/en-us/library/hh191443.aspx#BKMK_Threads not realising that a "continuation" actually refers to a continuation task schedule to run as soon as the task being "awaited" finishes.

This isn't single threaded.
When the delay task completes the rest of the method is posted to the ThreadPool and runs concurrently with your main thread. The "trigger" here is the callback of the internal System.Threading.Timer being used inside Task.Delay.
This behaviour depends on the SynchronizationContext. In a UI environment this would have been posted to the same UI thread and would have to wait until that thread is free.
If you would have been waiting for the task returned from FooAsync then you would only have a single thread running each time.

Async/await may create new threads OR NOT, it depends of the nature of the operation.
If the operation is an IO (for example disks/network operations) probably is coded in a way it will not spin a new thread. You can read from here:
The async and await keywords don't cause additional threads to be created?
If you create your own Async operation and you create a thread, that's a different story, that's why you shouldn't do async over sync
http://blogs.msdn.com/b/pfxteam/archive/2012/04/13/10293638.aspx
You can check this also but using Thread.CurrentThread to get the Id of the process. (Add that to a Console.WriteLine)

It's a pretty common misconception that the async or await keywords create new threads. They don't.
The threads are created by running a Task. In this case, the thread is created by the Task.Delay call.