Commit b9d7c45e authored by gabime's avatar gabime

Use blocking queue

parent cf63bcb8
......@@ -28,6 +28,8 @@ int main(int, char *[])
try
{
async_example();
return 0;
auto console = spdlog::stdout_color_st("console");
console->info("Welcome to spdlog!");
......@@ -82,7 +84,7 @@ int main(int, char *[])
// Asynchronous logging is very fast..
// Just call spdlog::set_async_mode(q_size) and all created loggers from now on will be asynchronous..
async_example();
//async_example();
// Log user-defined types example
user_defined_example();
......@@ -107,15 +109,23 @@ int main(int, char *[])
#include "spdlog/async.h"
void async_example()
{
auto async_file = spd::basic_logger_mt<spdlog::create_async>("async_file_logger", "logs/async_log.txt");
for (int i = 0; i < 100; ++i)
//auto async_file = spd::basic_logger_mt<spdlog::create_async>("async_file_logger", "logs/async_log.txt");
for (int j = 0; j < 1; j++)
{
spdlog::init_thread_pool(1024, 10);
auto async_file = spd::stderr_color_mt<spdlog::create_async>("console");
for (int i = 0; i < 1024; ++i)
{
async_file->info("Async message #{}", i);
async_file->info("{} Async message #{}", j, i);
}
spdlog::drop_all();
}
//std::this_thread::sleep_for(std::chrono::seconds(1));
// you can also modify thread pool settings *before* creating the logger:
// spdlog::init_thread_pool(32768, 4); // queue with 32k of pre allocated items and 4 backing threads.
// if not called a defaults are: preallocated 8192 queue items and 1 worker thread.
// spdlog::init_thread_pool(32768, 4); // queue with max 32k items 4 backing threads.
}
// syslog example (linux/osx/freebsd)
......
......@@ -8,7 +8,7 @@
// async logger implementation
// uses a thread pool to perform the actual logging
#include "../details/thread_pool.h"
#include "spdlog/details/thread_pool.h"
#include <chrono>
#include <memory>
......
#pragma once
//
// Copyright(c) 2018 Gabi Melman.
// Distributed under the MIT License (http://opensource.org/licenses/MIT)
//
// async log helper :
// multi producer-multi consumer blocking queue
// enqueue(..) - will block until room found to put the new message
// enqueue_nowait(..) - will return immediatly with false if no room left in the queue
// dequeue_for(..) - will block until the queue is not empty or timeout passed
#include <condition_variable>
#include <mutex>
#include <queue>
namespace spdlog {
namespace details {
template<typename T>
class mpmc_bounded_queue
{
public:
using item_type = T;
explicit mpmc_bounded_queue(size_t max_items)
: max_items_(max_items)
{
}
// try to enqueue and block if no room left
void enqueue(T &&item)
{
{
std::unique_lock<std::mutex> lock(queue_mutex_);
pop_cv_.wait(lock, [this] { return this->q_.size() < this->max_items_; });
q_.push(std::move(item));
}
push_cv_.notify_one();
}
// try to enqueue and return immdeialty false if no room left
bool enqueue_nowait(T &&item)
{
{
std::unique_lock<std::mutex> lock(queue_mutex_);
if (q_.size() == this->max_items_)
{
return false;
}
q_.push(std::forward<T>(item));
}
push_cv_.notify_one();
return true;
}
// try to dequeue item. if no item found. wait upto timeout and try again
// Return true, if succeeded dequeue item, false otherwise
bool dequeue_for(T &popped_item, std::chrono::milliseconds wait_duration)
{
{
std::unique_lock<std::mutex> lock(queue_mutex_);
if (!push_cv_.wait_for(lock, wait_duration, [this] { return this->q_.size() > 0; }))
{
return false;
}
popped_item = std::move(q_.front());
q_.pop();
}
pop_cv_.notify_one();
return true;
}
private:
size_t max_items_;
std::mutex queue_mutex_;
std::condition_variable push_cv_;
std::condition_variable pop_cv_;
std::queue<T> q_;
};
} // namespace details
} // namespace spdlog
/*
A modified version of Bounded MPMC queue by Dmitry Vyukov.
Original code from:
http://www.1024cores.net/home/lock-free-algorithms/queues/bounded-mpmc-queue
licensed by Dmitry Vyukov under the terms below:
Simplified BSD license
Copyright (c) 2010-2011 Dmitry Vyukov. All rights reserved.
Redistribution and use in source and binary forms, with or without modification,
are permitted provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of
conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list
of conditions and the following disclaimer in the documentation and/or other materials
provided with the distribution.
THIS SOFTWARE IS PROVIDED BY DMITRY VYUKOV "AS IS" AND ANY EXPRESS OR IMPLIED
WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT
SHALL DMITRY VYUKOV OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA,
OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
The views and conclusions contained in the software and documentation are those of the authors and
should not be interpreted as representing official policies, either expressed or implied, of Dmitry Vyukov.
*/
/*
The code in its current form adds the license below:
Copyright(c) 2015 Gabi Melman.
Distributed under the MIT License (http://opensource.org/licenses/MIT)
*/
#pragma once
#include "spdlog/common.h"
#include <atomic>
#include <utility>
namespace spdlog {
namespace details {
template<typename T>
class mpmc_bounded_queue
{
public:
using item_type = T;
explicit mpmc_bounded_queue(size_t buffer_size)
: max_size_(buffer_size)
, buffer_(new cell_t[buffer_size])
, buffer_mask_(buffer_size - 1)
{
// queue size must be power of two
if (!((buffer_size >= 2) && ((buffer_size & (buffer_size - 1)) == 0)))
{
throw spdlog_ex("async logger queue size must be power of two");
}
for (size_t i = 0; i != buffer_size; i += 1)
{
buffer_[i].sequence_.store(i, std::memory_order_relaxed);
}
enqueue_pos_.store(0, std::memory_order_relaxed);
dequeue_pos_.store(0, std::memory_order_relaxed);
}
~mpmc_bounded_queue()
{
delete[] buffer_;
}
mpmc_bounded_queue(mpmc_bounded_queue const &) = delete;
void operator=(mpmc_bounded_queue const &) = delete;
bool enqueue(T &&data)
{
cell_t *cell;
size_t pos = enqueue_pos_.load(std::memory_order_relaxed);
for (;;)
{
cell = &buffer_[pos & buffer_mask_];
size_t seq = cell->sequence_.load(std::memory_order_acquire);
intptr_t dif = static_cast<intptr_t>(seq) - static_cast<intptr_t>(pos);
if (dif == 0)
{
if (enqueue_pos_.compare_exchange_weak(pos, pos + 1, std::memory_order_relaxed))
{
break;
}
}
else if (dif < 0)
{
return false;
}
else
{
pos = enqueue_pos_.load(std::memory_order_relaxed);
}
}
cell->data_ = std::move(data);
cell->sequence_.store(pos + 1, std::memory_order_release);
return true;
}
bool dequeue(T &data)
{
cell_t *cell;
size_t pos = dequeue_pos_.load(std::memory_order_relaxed);
for (;;)
{
cell = &buffer_[pos & buffer_mask_];
size_t seq = cell->sequence_.load(std::memory_order_acquire);
intptr_t dif = static_cast<intptr_t>(seq) - static_cast<intptr_t>(pos + 1);
if (dif == 0)
{
if (dequeue_pos_.compare_exchange_weak(pos, pos + 1, std::memory_order_relaxed))
{
break;
}
}
else if (dif < 0)
{
return false;
}
else
{
pos = dequeue_pos_.load(std::memory_order_relaxed);
}
}
data = std::move(cell->data_);
cell->sequence_.store(pos + buffer_mask_ + 1, std::memory_order_release);
return true;
}
bool is_empty()
{
size_t front, front1, back;
// try to take a consistent snapshot of front/tail.
do
{
front = enqueue_pos_.load(std::memory_order_acquire);
back = dequeue_pos_.load(std::memory_order_acquire);
front1 = enqueue_pos_.load(std::memory_order_relaxed);
} while (front != front1);
return back == front;
}
private:
struct cell_t
{
std::atomic<size_t> sequence_;
T data_;
};
size_t const max_size_;
static size_t const cacheline_size = 64;
using cacheline_pad_t = char[cacheline_size];
cacheline_pad_t pad0_;
cell_t *const buffer_;
size_t const buffer_mask_;
cacheline_pad_t pad1_;
std::atomic<size_t> enqueue_pos_;
cacheline_pad_t pad2_;
std::atomic<size_t> dequeue_pos_;
cacheline_pad_t pad3_;
};
} // namespace details
} // namespace spdlog
#pragma once
#include "spdlog/details/log_msg.h"
#include "spdlog/details/mpmc_bounded_q.h"
#include "spdlog/details/mpmc_blocking_q.h"
#include "spdlog/details/os.h"
#include <chrono>
......@@ -10,21 +10,21 @@
#include <vector>
namespace spdlog {
namespace details {
namespace details {
using async_logger_ptr = std::shared_ptr<spdlog::async_logger>;
using async_logger_ptr = std::shared_ptr<spdlog::async_logger>;
enum class async_msg_type
{
enum class async_msg_type
{
log,
flush,
terminate
};
};
// Async msg to move to/from the queue
// Movable only. should never be copied
struct async_msg
{
// Async msg to move to/from the queue
// Movable only. should never be copied
struct async_msg
{
async_msg_type msg_type;
level::level_enum level;
log_clock::time_point time;
......@@ -37,10 +37,10 @@ struct async_msg
async_msg() = default;
~async_msg() = default;
// never copy or assign. should only be move assigned in to the queue..
// should only be moved in or out of the queue..
async_msg(const async_msg &) = delete;
async_msg &operator=(const async_msg &other) = delete;
async_msg(async_msg &&other) = delete;
async_msg(async_msg &&other) = default;
async_msg &operator=(async_msg &&other) = default;
// construct from log_msg with given type
async_msg(async_logger_ptr &&worker, async_msg_type the_type, details::log_msg &&m)
......@@ -64,19 +64,6 @@ struct async_msg
{
}
// used to move to the message queue
async_msg &operator=(async_msg &&other) SPDLOG_NOEXCEPT
{
msg_type = other.msg_type;
level = other.level;
time = other.time;
thread_id = other.thread_id;
raw = std::move(other.raw);
msg_id = other.msg_id;
worker_ptr = std::move(other.worker_ptr);
return *this;
}
// copy into log_msg
void to_log_msg(log_msg &&msg)
{
......@@ -90,17 +77,17 @@ struct async_msg
msg.color_range_start = 0;
msg.color_range_end = 0;
}
};
};
class thread_pool
{
public:
class thread_pool
{
public:
using item_type = async_msg;
using q_type = details::mpmc_bounded_queue<item_type>;
using clock_type = std::chrono::steady_clock;
thread_pool(size_t q_size_bytes, size_t threads_n)
: _msg_counter(0)
: msg_counter_(0)
, _q(q_size_bytes)
{
// std::cout << "thread_pool() q_size_bytes: " << q_size_bytes << "\tthreads_n: " << threads_n << std::endl;
......@@ -110,7 +97,7 @@ public:
}
for (size_t i = 0; i < threads_n; i++)
{
_threads.emplace_back(std::bind(&thread_pool::_worker_loop, this));
_threads.emplace_back(std::bind(&thread_pool::worker_loop, this));
}
}
......@@ -121,14 +108,14 @@ public:
{
for (size_t i = 0; i < _threads.size(); i++)
{
_post_async_msg(async_msg(async_msg_type::terminate), async_overflow_policy::block_retry);
post_async_msg(async_msg(async_msg_type::terminate), async_overflow_policy::block_retry);
}
for (auto &t : _threads)
{
t.join();
}
// std::cout << "~thread_pool() _msg_counter: " << _msg_counter << std::endl;
// std::cout << "~thread_pool() msg_counter_: " << msg_counter_ << std::endl;
}
catch (...)
{
......@@ -137,106 +124,82 @@ public:
void post_log(async_logger_ptr &&worker_ptr, details::log_msg &&msg, async_overflow_policy overflow_policy)
{
async_msg as_m(std::forward<async_logger_ptr>(worker_ptr), async_msg_type::log, std::forward<log_msg>(msg));
_post_async_msg(std::move(as_m), overflow_policy);
async_msg async_m(std::forward<async_logger_ptr>(worker_ptr), async_msg_type::log, std::forward<log_msg>(msg));
post_async_msg(std::move(async_m), overflow_policy);
}
void post_flush(async_logger_ptr &&worker_ptr, async_overflow_policy overflow_policy)
{
_post_async_msg(async_msg(std::forward<async_logger_ptr>(worker_ptr), async_msg_type::flush), overflow_policy);
post_async_msg(async_msg(std::forward<async_logger_ptr>(worker_ptr), async_msg_type::flush), overflow_policy);
}
size_t msg_counter()
{
return _msg_counter.load(std::memory_order_relaxed);
return msg_counter_.load(std::memory_order_relaxed);
}
private:
std::atomic<size_t> _msg_counter; // total # of messages processed in this pool
private:
std::atomic<size_t> msg_counter_; // total # of messages processed in this pool
q_type _q;
std::vector<std::thread> _threads;
void _post_async_msg(async_msg &&new_msg, async_overflow_policy overflow_policy)
void post_async_msg(async_msg &&new_msg, async_overflow_policy overflow_policy)
{
if (!_q.enqueue(std::forward<async_msg>(new_msg)) && overflow_policy == async_overflow_policy::block_retry)
if (overflow_policy == async_overflow_policy::block_retry)
{
auto last_op_time = clock_type::now();
auto now = last_op_time;
do
_q.enqueue(std::move(new_msg));
}
else
{
now = clock_type::now();
sleep_or_yield(now, last_op_time);
} while (!_q.enqueue(std::move(new_msg)));
_q.enqueue_nowait(std::move(new_msg));
}
}
// pop log messages from the queue and send to the logger worker
void _worker_loop()
void worker_loop()
{
async_msg popped_async_msg;
log_msg msg;
bool active = true;
auto last_pop_time = clock_type::now();
while (active)
{
if (_q.dequeue(popped_async_msg))
while (process_next_msg())
{
last_pop_time = clock_type::now();
switch (popped_async_msg.msg_type)
{
case async_msg_type::flush:
{
auto worker = std::move(popped_async_msg.worker_ptr);
worker->_backend_flush();
break;
};
}
case async_msg_type::terminate:
active = false;
break;
default:
// process next message in the queue
// return true if this thread should still be active (while no terminate msg was received)
bool process_next_msg()
{
popped_async_msg.to_log_msg(std::move(msg));
auto worker = std::move(popped_async_msg.worker_ptr);
worker->_backend_log(msg);
_msg_counter.fetch_add(1, std::memory_order_relaxed);
}
}
}
else // queue is empty - the only place we can terminate the thread if needed.
async_msg incoming_async_msg;
bool dequeued = _q.dequeue_for(incoming_async_msg, std::chrono::seconds(10));
if (!dequeued)
{
sleep_or_yield(clock_type::now(), last_pop_time);
}
}
return true;
}
// spin, yield or sleep. use the time passed since last message as a hint
static void sleep_or_yield(const clock_type::time_point &now, const clock_type::time_point &last_op_time)
switch (incoming_async_msg.msg_type)
{
using std::chrono::microseconds;
using std::chrono::milliseconds;
auto time_since_op = now - last_op_time;
// yield upto 150 micros
if (time_since_op <= microseconds(150))
case async_msg_type::flush:
{
return std::this_thread::yield();
incoming_async_msg.worker_ptr->_backend_flush();
return true;
}
// sleep for 20 ms upto 200 ms
if (time_since_op <= milliseconds(200))
case async_msg_type::terminate:
{
return details::os::sleep_for_millis(20);
return false;
}
// sleep for 500 ms
return details::os::sleep_for_millis(500);
default:
{
log_msg msg;
incoming_async_msg.to_log_msg(std::move(msg));
incoming_async_msg.worker_ptr->_backend_log(msg);
msg_counter_.fetch_add(1, std::memory_order_relaxed);
return true;
}
}
assert(false);
return true; // should not be reached
}
};
};
} // namespace details
} // namespace details
} // namespace spdlog
......@@ -7,7 +7,8 @@ set(SPDLOG_UTESTS_SOURCES
file_helper.cpp
file_log.cpp
test_misc.cpp
test_pattern_formatter
test_pattern_formatter.cpp
test_async.cpp
includes.h
registry.cpp
test_macros.cpp
......
#include "includes.h"
#include "test_sink.h"
#include "spdlog/async.h"
#include "spdlog/sinks/simple_file_sink.h"
//std::unique_ptr<spdlog::async_logger> create_logger(size_t tp_queue_size, size_t tp_threads)
//{
// auto tp = std::make_shared<details::thread_pool>(8192, 1);
// auto logger = std::make_shared<async_logger>("as", test_sink, tp, async_overflow_policy::block_retry);
//}
TEST_CASE("basic async test ", "[async]")
{
using namespace spdlog;
auto test_sink = std::make_shared<sinks::test_sink_mt>();
size_t queue_size = 128;
size_t messages = 256;
{
auto tp = std::make_shared<details::thread_pool>(queue_size, 1);
auto logger = std::make_shared<async_logger>("as", test_sink, tp, async_overflow_policy::block_retry);
for (size_t i = 0; i < messages; i++)
{
logger->info("Hello message #{}", i);
}
logger->flush();
}
REQUIRE(test_sink->msg_counter() == messages);
REQUIRE(test_sink->flushed_msg_counter() == messages);
}
TEST_CASE("discard policy ", "[async]")
{
using namespace spdlog;
auto test_sink = std::make_shared<sinks::test_sink_mt>();
size_t queue_size = 2;
size_t messages = 1024;
{
auto tp = std::make_shared<details::thread_pool>(queue_size, 1);
auto logger = std::make_shared<async_logger>("as", test_sink, tp, async_overflow_policy::discard_log_msg);
for (size_t i = 0; i < messages; i++)
{
logger->info("Hello message #{}", i);
}
}
REQUIRE(test_sink->msg_counter() < messages);
REQUIRE(test_sink->flushed_msg_counter() < messages);
}
TEST_CASE("flush", "[async]")
{
using namespace spdlog;
auto test_sink = std::make_shared<sinks::test_sink_mt>();
size_t queue_size = 256;
size_t messages = 256;
{
auto tp = std::make_shared<details::thread_pool>(queue_size, 1);
auto logger = std::make_shared<async_logger>("as", test_sink, tp, async_overflow_policy::block_retry);
for (size_t i = 0; i < messages; i++)
{
logger->info("Hello message #{}", i);
}
logger->flush();
}
std::this_thread::sleep_for(std::chrono::milliseconds(250));
REQUIRE(test_sink->msg_counter() == messages);
REQUIRE(test_sink->flushed_msg_counter() == messages);
}
TEST_CASE("multi threads", "[async]")
{
using namespace spdlog;
auto test_sink = std::make_shared<sinks::test_sink_mt>();
size_t queue_size = 128;
size_t messages = 256;
size_t n_threads = 10;
{
auto tp = std::make_shared<details::thread_pool>(queue_size, 1);
auto logger = std::make_shared<async_logger>("as", test_sink, tp, async_overflow_policy::block_retry);
std::vector<std::thread> threads;
for (size_t i = 0; i < n_threads; i++)
{
threads.emplace_back([logger, messages] {
for (size_t j = 0; j < messages; j++)
{
logger->info("Hello message #{}", j);
}
});
}
for (auto &t : threads)
{
t.join();
}
logger->flush();
}
REQUIRE(test_sink->msg_counter() == messages * n_threads);
REQUIRE(test_sink->flushed_msg_counter() == messages * n_threads);
}
TEST_CASE("to_file", "[async]")
{
prepare_logdir();
size_t queue_size = 512;
size_t messages = 512;
size_t n_threads = 4;
spdlog::init_thread_pool(queue_size, n_threads);
auto logger= spdlog::basic_logger_mt<spdlog::create_async>("as", "logs/async_test.log", true);
std::vector<std::thread> threads;
for (size_t i = 0; i < n_threads; i++)
{
threads.emplace_back([logger, messages] {
for (size_t j = 0; j < messages; j++)
{
logger->info("Hello message #{}", j);
}
});
}
for (auto &t : threads)
{
t.join();
}
logger.reset();
spdlog::drop("as");
std::this_thread::sleep_for(std::chrono::seconds(1));
REQUIRE(count_lines("logs/async_test.log") == messages * n_threads);
}
//
// Copyright(c) 2018 Gabi Melman.
// Distributed under the MIT License (http://opensource.org/licenses/MIT)
//
#pragma once
#include "spdlog/details/null_mutex.h"
#include "spdlog/sinks/base_sink.h"
#include <mutex>
namespace spdlog {
namespace sinks {
template<class Mutex>
class test_sink : public base_sink<Mutex>
{
public:
size_t msg_counter()
{
return msg_counter_;
}
size_t flushed_msg_counter()
{
return flushed_msg_counter_;
}
protected:
void _sink_it(const details::log_msg &) override
{
msg_counter_++;
}
void _flush() override
{
flushed_msg_counter_ += msg_counter_;
}
size_t msg_counter_{0};
size_t flushed_msg_counter_{0};
};
using test_sink_mt = test_sink<std::mutex>;
using test_sink_st = test_sink<details::null_mutex>;
} // namespace sinks
} // namespace spdlog
......@@ -129,6 +129,7 @@
<ClCompile Include="errors.cpp" />
<ClCompile Include="file_helper.cpp" />
<ClCompile Include="file_log.cpp" />
<ClCompile Include="test_async.cpp" />
<ClCompile Include="test_misc.cpp" />
<ClCompile Include="main.cpp" />
<ClCompile Include="registry.cpp" />
......
......@@ -42,6 +42,9 @@
<ClCompile Include="test_misc.cpp">
<Filter>Source Files</Filter>
</ClCompile>
<ClCompile Include="test_async.cpp">
<Filter>Source Files</Filter>
</ClCompile>
</ItemGroup>
<ItemGroup>
<ClInclude Include="includes.h">
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment