1
0
Fork 0

Add a Linux-specific event loop

Implementations for Windows and macOS should be fairly similar, but
using the OS' message loop instad of a thread.
This commit is contained in:
Robbert van der Helm 2022-01-31 19:45:11 +01:00
parent 4495064558
commit 2380768bad
3 changed files with 215 additions and 0 deletions

84
src/context.rs Normal file
View file

@ -0,0 +1,84 @@
// nih-plug: plugins, but rewritten in Rust
// Copyright (C) 2022 Robbert van der Helm
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with this program. If not, see <https://www.gnu.org/licenses/>.
//! Different contexts the plugin can use to make callbacks to the host in different...contexts.
use std::sync::Arc;
#[cfg(all(target_family = "unix", not(target_os = "macos")))]
mod linux;
#[cfg(all(target_family = "unix", not(target_os = "macos")))]
pub(crate) use linux::LinuxEventLoop as OsEventLoop;
// TODO: ProcessContext for parameter automation and sending events
// TODO: GuiContext for GUI parameter automation and resizing
pub(crate) const TASK_QUEUE_CAPACITY: usize = 512;
/// General callbacks the plugin can make during its lifetime. This is passed to the plugin during
/// [Plugin::initialize].
///
/// # Safety
///
/// This context is passed to be stored by the plugin, and it may thus not outlive the wrapper.
/// Hence the use of reference counted smart pointers. The implementing wrapper needs to be able to
/// handle concurrent requests, and it should perform the actual callback within
/// [MainThreadQueue::do_maybe_async].
pub unsafe trait ProcessContext {
/// Update the current latency of the plugin. If the plugin is currently processing audio, then
/// this may cause audio playback to be restarted.
fn set_latency_samples(self: &Arc<Self>, samples: u32);
}
/// A trait describing the functionality of the platform-specific event loop that can execute tasks
/// of type `T` in executor `E`. Posting a task to the queue should be realtime safe. This thread
/// queue should be created during the wrapper's initial initialization on the main thread.
///
/// This is never used generically, but having this as a trait will cause any missing functions on
/// an implementation to show up as compiler errors even when using a different platform.
///
/// TODO: At some point rethink the design to make it possible to have a singleton message queue for
/// all instances of a plugin.
pub(crate) trait EventLoop<T, E>
where
T: Send,
E: MainThreadExecutor<T>,
{
/// Create a main thread tasks queue for the given executor. The thread this is called on will
/// be designated as the main thread, so this should be called when constructing the wrapper.
///
/// TODO: Spawn, and update docs
fn new_and_spawn(executor: Arc<E>) -> Self;
/// Either post the function to a queue so it can be run later from the main thread using a
/// timer, or run the function directly if this is the main thread. This needs to be callable at
/// any time withotu blocking.
///
/// If the task queue was full, then this will return false.
#[must_use]
fn do_maybe_async(&self, task: T) -> bool;
/// Whether the calling thread is the even loop's main thread. This is usually the thread the
/// event loop instance wel initialized on.
fn is_main_thread(&self) -> bool;
}
/// Something that can execute tasks of type `T`.
pub(crate) trait MainThreadExecutor<T>: Send + Sync {
/// Execute a task on the current thread.
fn execute(&self, task: T);
}

127
src/context/linux.rs Normal file
View file

@ -0,0 +1,127 @@
// nih-plug: plugins, but rewritten in Rust
// Copyright (C) 2022 Robbert van der Helm
//
// This program is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with this program. If not, see <https://www.gnu.org/licenses/>.
//! An event loop impelemntation for Linux. APIs on Linux are generally thread safe, so the context
//! of a main thread does not exist there. Because of that, this mostly just serves as a way to
//! delegate expensive processing to another thread.
use crossbeam::channel;
use std::sync::Arc;
use std::thread::{self, JoinHandle, ThreadId};
use crate::nih_log;
use super::{EventLoop, MainThreadExecutor};
/// See [super::EventLoop].
pub(crate) struct LinuxEventLoop<T, E> {
/// The thing that ends up executing these tasks. The tasks are usually executed from the worker
/// thread, but if the current thread is the main thread then the task cna also be executed
/// directly.
executor: Arc<E>,
/// The ID of the main thread. In practice this is the ID of the thread that created this task
/// queue.
main_thread_id: ThreadId,
/// A thread that act as our worker thread. When [do_maybe_async] is called, this thread will be
/// woken up to execute the task on the executor. This is wrapped in an `Option` so the thread
/// can be taken out of it and joined when this struct gets dropped.
worker_thread: Option<JoinHandle<()>>,
/// A channel for waking up the worker thread and having it perform one of the tasks from
/// [Message].
worker_thread_channel: channel::Sender<Message<T>>,
}
/// A message for communicating with the worker thread.
enum Message<T> {
/// A new task for the event loop to execute.
Task(T),
/// Shut down the worker thread.
Shutdown,
}
impl<T, E> EventLoop<T, E> for LinuxEventLoop<T, E>
where
T: Send,
E: MainThreadExecutor<T>,
{
fn new_and_spawn(executor: Arc<E>) -> Self {
let (sender, receiver) = channel::bounded(super::TASK_QUEUE_CAPACITY);
Self {
executor: executor.clone(),
main_thread_id: thread::current().id(),
// With our drop implementation we guarentee that this thread never outlives this struct
worker_thread: Some(unsafe {
thread::Builder::new()
.name(String::from("worker"))
// FIXME: Find another way to bind a thread lifetime to this struct without a
// nightly-only fature
.spawn_unchecked(move || worker_thread(receiver, executor))
.expect("Could not spawn worker thread")
}),
worker_thread_channel: sender,
}
}
fn do_maybe_async(&self, task: T) -> bool {
if self.is_main_thread() {
self.executor.execute(task);
true
} else {
self.worker_thread_channel
.try_send(Message::Task(task))
.is_ok()
}
}
fn is_main_thread(&self) -> bool {
thread::current().id() == self.main_thread_id
}
}
impl<T, E> Drop for LinuxEventLoop<T, E> {
fn drop(&mut self) {
self.worker_thread_channel
.send(Message::Shutdown)
.expect("Failed while sending worker thread shutdown request");
if let Some(join_handle) = self.worker_thread.take() {
join_handle.join().expect("Worker thread panicked");
}
}
}
/// The worker thread used in [EventLoop] that executes incmoing tasks on the event loop's executor.
fn worker_thread<T, E>(receiver: channel::Receiver<Message<T>>, executor: Arc<E>)
where
T: Send,
E: MainThreadExecutor<T>,
{
loop {
match receiver.recv() {
Ok(Message::Task(task)) => executor.execute(task),
Ok(Message::Shutdown) => return,
Err(err) => {
nih_log!(
"Worker thread got disconnected unexpectedly, shutting down: {}",
err
);
return;
}
}
}
}

View file

@ -16,6 +16,10 @@
// TODO: Once everything is more fleshed out, document the basic usage of this library
// FIXME: Find an alternative for this
#![feature(thread_spawn_unchecked)]
pub mod context;
#[macro_use]
pub mod debug;
pub mod formatters;