src/ex/thread_pool.cpp

100.0% Lines (140 / 140) 100.0% Functions (29 / 29)

[ ] prev/next file

thread_pool.cpp

f(x) Functions (29)

Function Calls Lines Blocks

Line	Hits	Source Code
1		//
2		// Copyright (c) 2025 Vinnie Falco (vinnie.falco@gmail.com)
3		// Copyright (c) 2026 Michael Vandeberg
4		//
5		// Distributed under the Boost Software License, Version 1.0. (See accompanying
6		// file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
7		//
8		// Official repository: https://github.com/boostorg/capy
9		//
10
11		#include <boost/capy/ex/thread_pool.hpp>
12		#include <boost/capy/continuation.hpp>
13		#include <boost/capy/detail/thread_local_ptr.hpp>
14		#include <boost/capy/ex/frame_allocator.hpp>
15		#include <boost/capy/test/thread_name.hpp>
16		#include <algorithm>
17		#include <atomic>
18		#include <condition_variable>
19		#include <cstdio>
20		#include <mutex>
21		#include <thread>
22		#include <vector>
23
24		/*
25		Thread pool implementation using a shared work queue.
26
27		Work items are continuations linked via their intrusive next pointer,
28		stored in a single queue protected by a mutex. No per-post heap
29		allocation: the continuation is owned by the caller and linked
30		directly. Worker threads wait on a condition_variable until work
31		is available or stop is requested.
32
33		Threads are started lazily on first post() via std::call_once to avoid
34		spawning threads for pools that are constructed but never used. Each
35		thread is named with a configurable prefix plus index for debugger
36		visibility.
37
38		Work tracking: on_work_started/on_work_finished maintain the atomic
39		outstanding_work_ counter. on_work_started is lock-free; the worker
40		that drives the count to zero takes mutex_ and re-reads the count
41		before deciding to stop, so the count and the stop decision stay
42		consistent even if work is started in between. join() blocks until
43		this counter reaches zero, then signals workers to stop and joins
44		threads.
45
46		Two shutdown paths:
47		- join(): waits for outstanding work to drain, then stops workers.
48		- stop(): immediately signals workers to exit; queued work is abandoned.
49		- Destructor: stop() then join() (abandon + wait for threads).
50		*/
51
52		namespace boost {
53		namespace capy {
54
55		//------------------------------------------------------------------------------
56
57		class thread_pool::impl
58		{
59		// Identifies the pool owning the current worker thread, or
60		// nullptr if the calling thread is not a pool worker. Checked
61		// by dispatch() to decide between symmetric transfer (inline
62		// resume) and post.
63		static inline detail::thread_local_ptr<impl const> current_;
64
65		// Intrusive queue of continuations: the next link is stored in
66		// continuation::reserved (typed continuation* round-tripped through
67		// void*). No per-post allocation: the continuation is owned by the caller.
68		continuation* head_ = nullptr;
69		continuation* tail_ = nullptr;
70
71	17962x	void push(continuation* c) noexcept
72		{
73	17962x	c->reserved = nullptr;
74	17962x	if(tail_)
75	6821x	tail_->reserved = c;
76		else
77	11141x	head_ = c;
78	17962x	tail_ = c;
79	17962x	}
80
81	18357x	continuation* pop() noexcept
82		{
83	18357x	if(!head_)
84	395x	return nullptr;
85	17962x	continuation* c = head_;
86	17962x	head_ = static_cast<continuation*>(head_->reserved);
87	17962x	if(!head_)
88	11141x	tail_ = nullptr;
89	17962x	return c;
90		}
91
92	29351x	bool empty() const noexcept
93		{
94	29351x	return head_ == nullptr;
95		}
96
97		std::mutex mutex_;
98		std::condition_variable work_cv_;
99		std::condition_variable done_cv_;
100		std::vector<std::thread> threads_;
101		std::atomic<std::size_t> outstanding_work_{0};
102		bool stop_{false};
103		bool joined_{false};
104		std::size_t num_threads_;
105		char thread_name_prefix_[13]{}; // 12 chars max + null terminator
106		std::once_flag start_flag_;
107
108		public:
109	395x	~impl() = default;
110
111		bool
112	683x	running_in_this_thread() const noexcept
113		{
114	683x	return current_.get() == this;
115		}
116
117		// Destroy abandoned coroutine frames. Must be called
118		// before execution_context::shutdown()/destroy() so
119		// that suspended-frame destructors touching services
120		// (e.g. cancelling registrations) run while those
121		// services are still valid.
122		void
123	395x	drain_abandoned() noexcept
124		{
125	577x	while(auto* c = pop())
126		{
127	182x	auto h = c->h;
128	182x	if(h && h != std::noop_coroutine())
129	131x	h.destroy();
130	182x	}
131	395x	}
132
133	395x	impl(std::size_t num_threads, std::string_view thread_name_prefix)
134	395x	: num_threads_(num_threads)
135		{
136	395x	if(num_threads_ == 0)
137	8x	num_threads_ = std::max(
138	4x	std::thread::hardware_concurrency(), 1u);
139
140		// Truncate prefix to 12 chars, leaving room for up to 3-digit index.
141	395x	auto n = thread_name_prefix.copy(thread_name_prefix_, 12);
142	395x	thread_name_prefix_[n] = '\0';
143	395x	}
144
145		void
146	17962x	post(continuation& c)
147		{
148	17962x	ensure_started();
149		{
150	17962x	std::lock_guard<std::mutex> lock(mutex_);
151	17962x	push(&c);
152		// Under the lock so the pool cannot drain, join, and
153		// destroy the condition variable mid-signal.
154	17962x	work_cv_.notify_one();
155	17962x	}
156	17962x	}
157
158		void
159	682x	on_work_started() noexcept
160		{
161	682x	outstanding_work_.fetch_add(1, std::memory_order_acq_rel);
162	682x	}
163
164		void
165	682x	on_work_finished() noexcept
166		{
167	682x	if(outstanding_work_.fetch_sub(
168	682x	1, std::memory_order_acq_rel) == 1)
169		{
170		// fetch_sub's result can be stale: a concurrent
171		// on_work_started() may raise the count before we take the
172		// lock, so re-read it here rather than trust the decrement.
173	319x	std::lock_guard<std::mutex> lock(mutex_);
174	319x	if(outstanding_work_.load(
175	319x	std::memory_order_acquire) == 0 && joined_ && !stop_)
176		{
177	194x	stop_ = true;
178	194x	done_cv_.notify_all();
179	194x	work_cv_.notify_all();
180		}
181	319x	}
182	682x	}
183
184		void
185	656x	join() noexcept
186		{
187		{
188	656x	std::unique_lock<std::mutex> lock(mutex_);
189	656x	if(joined_)
190	261x	return;
191	395x	joined_ = true;
192
193	395x	if(outstanding_work_.load(
194	395x	std::memory_order_acquire) == 0)
195		{
196	151x	stop_ = true;
197	151x	work_cv_.notify_all();
198		}
199		else
200		{
201	244x	done_cv_.wait(lock, [this]{
202	439x	return stop_;
203		});
204		}
205	656x	}
206
207	844x	for(auto& t : threads_)
208	449x	if(t.joinable())
209	449x	t.join();
210		}
211
212		void
213	397x	stop() noexcept
214		{
215		{
216	397x	std::lock_guard<std::mutex> lock(mutex_);
217	397x	stop_ = true;
218	397x	}
219	397x	work_cv_.notify_all();
220	397x	done_cv_.notify_all();
221	397x	}
222
223		private:
224		void
225	17962x	ensure_started()
226		{
227	17962x	std::call_once(start_flag_, [this]{
228	345x	threads_.reserve(num_threads_);
229	794x	for(std::size_t i = 0; i < num_threads_; ++i)
230	898x	threads_.emplace_back([this, i]{ run(i); });
231	345x	});
232	17962x	}
233
234		void
235	449x	run(std::size_t index)
236		{
237		// Build name; set_current_thread_name truncates to platform limits.
238		char name[16];
239	449x	std::snprintf(name, sizeof(name), "%s%zu", thread_name_prefix_, index);
240	449x	set_current_thread_name(name);
241
242		// Mark this thread as a worker of this pool so dispatch()
243		// can symmetric-transfer when called from within pool work.
244		struct scoped_pool
245		{
246	449x	scoped_pool(impl const* p) noexcept { current_.set(p); }
247	449x	~scoped_pool() noexcept { current_.set(nullptr); }
248	449x	} guard(this);
249
250		for(;;)
251		{
252	18229x	continuation* c = nullptr;
253		{
254	18229x	std::unique_lock<std::mutex> lock(mutex_);
255	18229x	work_cv_.wait(lock, [this]{
256	40829x	return !empty() \|\|
257	40829x	stop_;
258		});
259	18229x	if(stop_)
260	898x	return;
261	17780x	c = pop();
262	18229x	}
263	17780x	if(c)
264	17780x	safe_resume(c->h);
265	17780x	}
266	449x	}
267		};
268
269		//------------------------------------------------------------------------------
270
271	395x	thread_pool::
272		~thread_pool()
273		{
274	395x	impl_->stop();
275	395x	impl_->join();
276	395x	impl_->drain_abandoned();
277	395x	shutdown();
278	395x	destroy();
279	395x	delete impl_;
280	395x	}
281
282	395x	thread_pool::
283	395x	thread_pool(std::size_t num_threads, std::string_view thread_name_prefix)
284	395x	: impl_(new impl(num_threads, thread_name_prefix))
285		{
286	395x	this->set_frame_allocator(std::allocator<void>{});
287	395x	}
288
289		void
290	261x	thread_pool::
291		join() noexcept
292		{
293	261x	impl_->join();
294	261x	}
295
296		void
297	2x	thread_pool::
298		stop() noexcept
299		{
300	2x	impl_->stop();
301	2x	}
302
303		//------------------------------------------------------------------------------
304
305		thread_pool::executor_type
306	11807x	thread_pool::
307		get_executor() const noexcept
308		{
309	11807x	return executor_type(
310	11807x	const_cast<thread_pool&>(*this));
311		}
312
313		void
314	682x	thread_pool::executor_type::
315		on_work_started() const noexcept
316		{
317	682x	pool_->impl_->on_work_started();
318	682x	}
319
320		void
321	682x	thread_pool::executor_type::
322		on_work_finished() const noexcept
323		{
324	682x	pool_->impl_->on_work_finished();
325	682x	}
326
327		void
328	17286x	thread_pool::executor_type::
329		post(continuation& c) const
330		{
331	17286x	pool_->impl_->post(c);
332	17286x	}
333
334		std::coroutine_handle<>
335	683x	thread_pool::executor_type::
336		dispatch(continuation& c) const
337		{
338	683x	if(pool_->impl_->running_in_this_thread())
339	7x	return c.h;
340	676x	pool_->impl_->post(c);
341	676x	return std::noop_coroutine();
342		}
343
344		} // capy
345		} // boost
346