Skip to content

Commit

Permalink
Thread executor for running tasks on specific threads. (bevyengine#7087)
Browse files Browse the repository at this point in the history
# Objective

- Spawn tasks from other threads onto an async executor, but limit those tasks to run on a specific thread.
- This is a continuation of trying to break up some of the changes in pipelined rendering.
- Eventually this will be used to allow `NonSend` systems to run on the main thread in pipelined rendering bevyengine#6503 and also to solve bevyengine#6552.
- For this specific PR this allows for us to store a thread executor in a thread local, rather than recreating a scope executor for every scope which should save on a little work.

## Solution

- We create a Executor that does a runtime check for what thread it's on before creating a !Send ticker. The ticker is the only way for the executor to make progress.

---

## Changelog

- create a ThreadExecutor that can only be ticked on one thread.
  • Loading branch information
hymm authored and alradish committed Jan 22, 2023
1 parent 1270ce0 commit 428fed9
Show file tree
Hide file tree
Showing 3 changed files with 186 additions and 50 deletions.
5 changes: 5 additions & 0 deletions crates/bevy_tasks/src/lib.rs
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,11 @@ mod usages;
pub use usages::tick_global_task_pools_on_main_thread;
pub use usages::{AsyncComputeTaskPool, ComputeTaskPool, IoTaskPool};

#[cfg(not(target_arch = "wasm32"))]
mod thread_executor;
#[cfg(not(target_arch = "wasm32"))]
pub use thread_executor::{ThreadExecutor, ThreadExecutorTicker};

mod iter;
pub use iter::ParallelIterator;

Expand Down
103 changes: 53 additions & 50 deletions crates/bevy_tasks/src/task_pool.rs
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@ use async_task::FallibleTask;
use concurrent_queue::ConcurrentQueue;
use futures_lite::{future, pin, FutureExt};

use crate::Task;
use crate::{thread_executor::ThreadExecutor, Task};

struct CallOnDrop(Option<Arc<dyn Fn() + Send + Sync + 'static>>);

Expand Down Expand Up @@ -108,6 +108,7 @@ pub struct TaskPool {
impl TaskPool {
thread_local! {
static LOCAL_EXECUTOR: async_executor::LocalExecutor<'static> = async_executor::LocalExecutor::new();
static THREAD_EXECUTOR: ThreadExecutor<'static> = ThreadExecutor::new();
}

/// Create a `TaskPool` with the default configuration.
Expand Down Expand Up @@ -271,59 +272,61 @@ impl TaskPool {
F: for<'scope> FnOnce(&'scope Scope<'scope, 'env, T>),
T: Send + 'static,
{
// SAFETY: This safety comment applies to all references transmuted to 'env.
// Any futures spawned with these references need to return before this function completes.
// This is guaranteed because we drive all the futures spawned onto the Scope
// to completion in this function. However, rust has no way of knowing this so we
// transmute the lifetimes to 'env here to appease the compiler as it is unable to validate safety.
let executor: &async_executor::Executor = &self.executor;
let executor: &'env async_executor::Executor = unsafe { mem::transmute(executor) };
let task_scope_executor = &async_executor::Executor::default();
let task_scope_executor: &'env async_executor::Executor =
unsafe { mem::transmute(task_scope_executor) };
let spawned: ConcurrentQueue<FallibleTask<T>> = ConcurrentQueue::unbounded();
let spawned_ref: &'env ConcurrentQueue<FallibleTask<T>> =
unsafe { mem::transmute(&spawned) };

let scope = Scope {
executor,
task_scope_executor,
spawned: spawned_ref,
scope: PhantomData,
env: PhantomData,
};

let scope_ref: &'env Scope<'_, 'env, T> = unsafe { mem::transmute(&scope) };

f(scope_ref);

if spawned.is_empty() {
Vec::new()
} else {
let get_results = async {
let mut results = Vec::with_capacity(spawned_ref.len());
while let Ok(task) = spawned_ref.pop() {
results.push(task.await.unwrap());
}

results
Self::THREAD_EXECUTOR.with(|thread_executor| {
// SAFETY: This safety comment applies to all references transmuted to 'env.
// Any futures spawned with these references need to return before this function completes.
// This is guaranteed because we drive all the futures spawned onto the Scope
// to completion in this function. However, rust has no way of knowing this so we
// transmute the lifetimes to 'env here to appease the compiler as it is unable to validate safety.
let executor: &async_executor::Executor = &self.executor;
let executor: &'env async_executor::Executor = unsafe { mem::transmute(executor) };
let thread_executor: &'env ThreadExecutor<'env> =
unsafe { mem::transmute(thread_executor) };
let spawned: ConcurrentQueue<FallibleTask<T>> = ConcurrentQueue::unbounded();
let spawned_ref: &'env ConcurrentQueue<FallibleTask<T>> =
unsafe { mem::transmute(&spawned) };

let scope = Scope {
executor,
thread_executor,
spawned: spawned_ref,
scope: PhantomData,
env: PhantomData,
};

// Pin the futures on the stack.
pin!(get_results);
let scope_ref: &'env Scope<'_, 'env, T> = unsafe { mem::transmute(&scope) };

f(scope_ref);

if spawned.is_empty() {
Vec::new()
} else {
let get_results = async {
let mut results = Vec::with_capacity(spawned_ref.len());
while let Ok(task) = spawned_ref.pop() {
results.push(task.await.unwrap());
}

loop {
if let Some(result) = future::block_on(future::poll_once(&mut get_results)) {
break result;
results
};

std::panic::catch_unwind(|| {
executor.try_tick();
task_scope_executor.try_tick();
})
.ok();
// Pin the futures on the stack.
pin!(get_results);

let thread_ticker = thread_executor.ticker().unwrap();
loop {
if let Some(result) = future::block_on(future::poll_once(&mut get_results)) {
break result;
};

std::panic::catch_unwind(|| {
executor.try_tick();
thread_ticker.try_tick();
})
.ok();
}
}
}
})
}

/// Spawns a static future onto the thread pool. The returned Task is a future. It can also be
Expand Down Expand Up @@ -395,7 +398,7 @@ impl Drop for TaskPool {
#[derive(Debug)]
pub struct Scope<'scope, 'env: 'scope, T> {
executor: &'scope async_executor::Executor<'scope>,
task_scope_executor: &'scope async_executor::Executor<'scope>,
thread_executor: &'scope ThreadExecutor<'scope>,
spawned: &'scope ConcurrentQueue<FallibleTask<T>>,
// make `Scope` invariant over 'scope and 'env
scope: PhantomData<&'scope mut &'scope ()>,
Expand Down Expand Up @@ -425,7 +428,7 @@ impl<'scope, 'env, T: Send + 'scope> Scope<'scope, 'env, T> {
///
/// For more information, see [`TaskPool::scope`].
pub fn spawn_on_scope<Fut: Future<Output = T> + 'scope + Send>(&self, f: Fut) {
let task = self.task_scope_executor.spawn(f).fallible();
let task = self.thread_executor.spawn(f).fallible();
// ConcurrentQueue only errors when closed or full, but we never
// close and use an unbounded queue, so it is safe to unwrap
self.spawned.push(task).unwrap();
Expand Down
128 changes: 128 additions & 0 deletions crates/bevy_tasks/src/thread_executor.rs
Original file line number Diff line number Diff line change
@@ -0,0 +1,128 @@
use std::{
marker::PhantomData,
thread::{self, ThreadId},
};

use async_executor::{Executor, Task};
use futures_lite::Future;

/// An executor that can only be ticked on the thread it was instantiated on. But
/// can spawn `Send` tasks from other threads.
///
/// # Example
/// ```rust
/// # use std::sync::{Arc, atomic::{AtomicI32, Ordering}};
/// use bevy_tasks::ThreadExecutor;
///
/// let thread_executor = ThreadExecutor::new();
/// let count = Arc::new(AtomicI32::new(0));
///
/// // create some owned values that can be moved into another thread
/// let count_clone = count.clone();
///
/// std::thread::scope(|scope| {
/// scope.spawn(|| {
/// // we cannot get the ticker from another thread
/// let not_thread_ticker = thread_executor.ticker();
/// assert!(not_thread_ticker.is_none());
///
/// // but we can spawn tasks from another thread
/// thread_executor.spawn(async move {
/// count_clone.fetch_add(1, Ordering::Relaxed);
/// }).detach();
/// });
/// });
///
/// // the tasks do not make progress unless the executor is manually ticked
/// assert_eq!(count.load(Ordering::Relaxed), 0);
///
/// // tick the ticker until task finishes
/// let thread_ticker = thread_executor.ticker().unwrap();
/// thread_ticker.try_tick();
/// assert_eq!(count.load(Ordering::Relaxed), 1);
/// ```
#[derive(Debug)]
pub struct ThreadExecutor<'task> {
executor: Executor<'task>,
thread_id: ThreadId,
}

impl<'task> Default for ThreadExecutor<'task> {
fn default() -> Self {
Self {
executor: Executor::new(),
thread_id: thread::current().id(),
}
}
}

impl<'task> ThreadExecutor<'task> {
/// create a new [`ThreadExecutor`]
pub fn new() -> Self {
Self::default()
}

/// Spawn a task on the thread executor
pub fn spawn<T: Send + 'task>(
&self,
future: impl Future<Output = T> + Send + 'task,
) -> Task<T> {
self.executor.spawn(future)
}

/// Gets the [`ThreadExecutorTicker`] for this executor.
/// Use this to tick the executor.
/// It only returns the ticker if it's on the thread the executor was created on
/// and returns `None` otherwise.
pub fn ticker<'ticker>(&'ticker self) -> Option<ThreadExecutorTicker<'task, 'ticker>> {
if thread::current().id() == self.thread_id {
return Some(ThreadExecutorTicker {
executor: &self.executor,
_marker: PhantomData::default(),
});
}
None
}
}

/// Used to tick the [`ThreadExecutor`]. The executor does not
/// make progress unless it is manually ticked on the thread it was
/// created on.
#[derive(Debug)]
pub struct ThreadExecutorTicker<'task, 'ticker> {
executor: &'ticker Executor<'task>,
// make type not send or sync
_marker: PhantomData<*const ()>,
}
impl<'task, 'ticker> ThreadExecutorTicker<'task, 'ticker> {
/// Tick the thread executor.
pub async fn tick(&self) {
self.executor.tick().await;
}

/// Synchronously try to tick a task on the executor.
/// Returns false if if does not find a task to tick.
pub fn try_tick(&self) -> bool {
self.executor.try_tick()
}
}

#[cfg(test)]
mod tests {
use super::*;
use std::sync::Arc;

#[test]
fn test_ticker() {
let executor = Arc::new(ThreadExecutor::new());
let ticker = executor.ticker();
assert!(ticker.is_some());

std::thread::scope(|s| {
s.spawn(|| {
let ticker = executor.ticker();
assert!(ticker.is_none());
});
});
}
}

0 comments on commit 428fed9

Please sign in to comment.