完善mutex_v2代码

README.md

@@ -1,7 +1,12 @@
-# librf 2.4
+# librf 2.8
 
 ### librf  - 协程库
 
+2020-03-18 更新：
+    更新event/mutex/when_all/when_any实现。至此，2.x版本完整恢复1.x版本的所有功能。
+    版本号提升至 2.8.0。
+    3.0之前，只打算做修复BUG相关的工作。
+    3.0的目标，是根据executor的设计，重写scheduler代码。
 2020-03-08 更新：
     更新channel实现，效率提高了近三倍。
     channel的新的实现方法，为event/mutex指明了新的修改方向。
@@ -13,17 +18,18 @@
 
 目前仅支持:
 
-    Windows (使用VS2017/VS2019/clang编译)
+    Windows (使用VS2017/VS2019/clang 9编译)
     Android (使用NDK 20.1 自带的clang编译)
 
 
 librf有以下特点：
 
- *   1.基于C++17提案'Stackless Resumable Functions'编写的非对称stackless协程库，可以以同步的方式编写简单的代码，同时获得异步的性能
- *   2.理论上支持海量协程, 创建100万个协程只需使用<430M>物理内存
- *   3.提供协程锁(mutex), 定时器, channel等特性, 帮助用户更加容易地编写程序 
- *   4.可以很好的跟asio,libuv等库结合，能跟现有的callback范式的异步/延迟代码很好的结合
- *   5.目前还处于实验状态，不对今后正式的C++ Coroutines支持有任何正式的承诺
+ *   1.基于C++20提案'Stackless Resumable Functions'编写的非对称stackless协程库，可以以同步的方式编写简单的代码，同时获得异步的性能
+ *   2.理论上支持海量协程, 创建1000万个协程只需使用<2.2G>物理内存(使用clang编译)
+ *   3.拥有极小的协程调度，在I7 8100 3.6GHz的CPU上，1000个协程的平均切换开销是32纳秒
+ *   4.提供协程锁(mutex), 定时器, channel, event等特性, 帮助用户更加容易地编写程序
+ *   5.可以很好的跟asio, libuv等库结合，能跟现有的callback范式的异步/延迟代码很好的结合
+ *   6.目前已处于较为完善状态，已经小规模在生产项目中使用。不出意外，2.8以上版本就是C++20 Coroutines对应的版本
 
  *   如果你发现了任何bug、有好的建议、或使用上有不明之处，可以提交到issue，也可以直接联系作者:
       email: tearshark@163.net  QQ交流群: 296561497

librf/src/mutex_v2.cpp

@@ -50,7 +50,7 @@ RESUMEF_NS
 			}
 		}
 
-		bool mutex_v2_impl::try_lock(void* sch) noexcept
+		bool mutex_v2_impl::try_lock(void* sch)
 		{
 			scoped_lock<detail::mutex_v2_impl::lock_type> lock_(_lock);
 			return try_lock_lockless(sch);
@@ -59,11 +59,15 @@ RESUMEF_NS
 		bool mutex_v2_impl::try_lock_lockless(void* sch) noexcept
 		{
 			void* oldValue = _owner.load(std::memory_order_relaxed);
-			if (oldValue == nullptr || oldValue == sch)
+			if (oldValue == nullptr)
 			{
 				_owner.store(sch, std::memory_order_relaxed);
 				_counter.fetch_add(1, std::memory_order_relaxed);
-
+				return true;
+			}
+			if (oldValue == sch)
+			{
+				_counter.fetch_add(1, std::memory_order_relaxed);
 				return true;
 			}
 			return false;

librf/src/mutex_v2.h

@@ -11,6 +11,7 @@ RESUMEF_NS
 	{
 		struct scoped_lock_mutex_t;
 
+		//支持递归的锁
 		struct mutex_t
 		{
 			typedef std::shared_ptr<detail::mutex_v2_impl> mutex_impl_ptr;
@@ -22,11 +23,12 @@ RESUMEF_NS
 
 			struct [[nodiscard]] awaiter;
 
+			awaiter operator co_await() const noexcept;
 			awaiter lock() const noexcept;
 
-			scoped_lock_mutex_t lock(scheduler_t* sch) const noexcept;
-			bool try_lock(scheduler_t* sch) const noexcept;
-			void unlock(scheduler_t* sch) const noexcept;
+			scoped_lock_mutex_t lock(void* unique_address) const;
+			bool try_lock(void* unique_address) const;
+			void unlock(void* unique_address) const;
 
 			mutex_t(const mutex_t&) = default;
 			mutex_t(mutex_t&&) = default;

librf/src/mutex_v2.inl

@@ -39,7 +39,7 @@ RESUMEF_NS
 				return _owner.load(std::memory_order_relaxed);
 			}
 
-			bool try_lock(void* sch) noexcept;			//内部加锁
+			bool try_lock(void* sch);					//内部加锁
 			bool unlock(void* sch);						//内部加锁
 			void lock_until_succeed(void* sch);			//内部加锁
 		public:
@@ -74,26 +74,26 @@ RESUMEF_NS
 
 			//此函数，应该在try_lock()获得锁后使用
 			//或者在协程里，由awaiter使用
-			scoped_lock_mutex_t(std::adopt_lock_t, detail::mutex_v2_impl* mtx, void* sch)
-				: _mutex(mtx->shared_from_this())
+			scoped_lock_mutex_t(std::adopt_lock_t, mutex_impl_ptr mtx, void* sch)
+				: _mutex(std::move(mtx))
 				, _owner(sch)
 			{}
 
 			//此函数，适合在非协程里使用
-			scoped_lock_mutex_t(detail::mutex_v2_impl* mtx, void* sch)
-				: _mutex(mtx->shared_from_this())
+			scoped_lock_mutex_t(mutex_impl_ptr mtx, void* sch)
+				: _mutex(std::move(mtx))
 				, _owner(sch)
 			{
-				if (sch != nullptr)
+				if (_mutex != nullptr)
 					_mutex->lock_until_succeed(sch);
 			}
 
 
 			scoped_lock_mutex_t(std::adopt_lock_t, const mutex_t& mtx, void* sch)
-				: scoped_lock_mutex_t(std::adopt_lock, mtx._mutex.get(), sch)
+				: scoped_lock_mutex_t(std::adopt_lock, mtx._mutex, sch)
 			{}
 			scoped_lock_mutex_t(const mutex_t& mtx, void* sch)
-				: scoped_lock_mutex_t(mtx._mutex.get(), sch)
+				: scoped_lock_mutex_t(mtx._mutex, sch)
 			{}
 
 			~scoped_lock_mutex_t()
@@ -123,9 +123,10 @@ RESUMEF_NS
 
 		struct [[nodiscard]] mutex_t::awaiter
 		{
-			awaiter(detail::mutex_v2_impl* evt) noexcept
-				: _mutex(evt)
+			awaiter(detail::mutex_v2_impl* mtx) noexcept
+				: _mutex(mtx)
 			{
+				assert(_mutex != nullptr);
 			}
 
 			~awaiter() noexcept(false)
@@ -163,7 +164,7 @@ RESUMEF_NS
 
 			scoped_lock_mutex_t await_resume() noexcept
 			{
-				detail::mutex_v2_impl* mtx = _mutex;
+				mutex_impl_ptr mtx = _root ? _mutex->shared_from_this() : nullptr;
 				_mutex = nullptr;
 
 				return { std::adopt_lock, mtx, _root };
@@ -174,30 +175,30 @@ RESUMEF_NS
 			state_base_t* _root = nullptr;
 		};
 
-		inline mutex_t::awaiter mutex_t::lock() const noexcept
+		inline mutex_t::awaiter mutex_t::operator co_await() const noexcept
 		{
 			return { _mutex.get() };
 		}
 
-		inline scoped_lock_mutex_t mutex_t::lock(scheduler_t* sch) const noexcept
+		inline mutex_t::awaiter mutex_t::lock() const noexcept
 		{
-			if (sch != nullptr)
-				_mutex->lock_until_succeed(sch);
+			return { _mutex.get() };
+		}
 
-			return { std::adopt_lock, _mutex.get(), sch };
+		inline scoped_lock_mutex_t mutex_t::lock(void* unique_address) const
+		{
+			_mutex->lock_until_succeed(unique_address);
+			return { std::adopt_lock, _mutex, unique_address };
 		}
 
-		inline bool mutex_t::try_lock(scheduler_t* sch) const noexcept
+		inline bool mutex_t::try_lock(void* unique_address) const
 		{
-			if (sch == nullptr)
-				return false;
-			return _mutex->try_lock(sch);
+			return _mutex->try_lock(unique_address);
 		}
 
-		inline void mutex_t::unlock(scheduler_t* sch) const noexcept
+		inline void mutex_t::unlock(void* unique_address) const
 		{
-			assert(sch != nullptr);
-			_mutex->unlock(sch);
+			_mutex->unlock(unique_address);
 		}
 	}
 }

tutorial/test_async_mutex.cpp

@@ -9,47 +9,102 @@
 #include "librf.h"
 
 using namespace resumef;
+using namespace std::chrono;
 
-mutex_t g_lock;
-std::deque<size_t> g_queue;
+static mutex_t g_lock;
+static intptr_t g_counter = 0;
 
-future_t<> test_mutex_pop(size_t idx)
-{
-	using namespace std::chrono;
+static const size_t N = 10;
 
-	for (size_t i = 0; i < 10; ++i)
+//🔒-50ms-🔒🗝🗝-150ms-|
+//-------------.........
+static future_t<> test_mutex_pop(size_t idx)
+{
+	for (size_t i = 0; i < N / 2; ++i)
 	{
-		auto _locker = co_await g_lock.lock();
-
-		if (g_queue.size() > 0)
 		{
-			size_t val = g_queue.front();
-			g_queue.pop_front();
+			auto _locker = co_await g_lock.lock();	//_locker析构后，会调用对应的unlock()函数。
+
+			--g_counter;
+			std::cout << "pop :" << g_counter << " on " << idx << std::endl;
 
-			std::cout << val << " on " << idx << std::endl;
+			co_await 50ms;
+
+			auto _locker_2 = co_await g_lock.lock();
+
+			--g_counter;
+			std::cout << "pop :" << g_counter << " on " << idx << std::endl;
 		}
-		co_await sleep_for(500ms);
+		co_await 150ms;
 	}
 }
 
-future_t<> test_mutex_push(size_t idx)
+//🔒-50ms-🗝-50ms-🔒-50ms-🗝-50ms-|
+//---------........---------.......
+static future_t<> test_mutex_push(size_t idx)
 {
-	using namespace std::chrono;
-
-	for (size_t i = 0; i < 10; ++i)
+	for (size_t i = 0; i < N; ++i)
 	{
-		auto _locker = co_await g_lock.lock();
-		g_queue.push_back(i);
-		std::cout << i << " on " << idx << std::endl;
+		{
+			auto _locker = co_await g_lock;
 
-		co_await sleep_for(500ms);
+			++g_counter;
+			std::cout << "push:" << g_counter << " on " << idx << std::endl;
+
+			co_await 50ms;
+		}
+		co_await 50ms;
 	}
 }
 
-void resumable_main_mutex()
+//🔒-50ms-🗝-50ms-🔒-50ms-🗝-50ms-|
+//---------........---------.......
+static std::thread test_mutex_async_push(size_t idx)
+{
+	return std::thread([=]
+	{
+		char provide_unique_address = 0;
+		for (size_t i = 0; i < N; ++i)
+		{
+			{
+				auto _locker = g_lock.lock(&provide_unique_address);
+
+				++g_counter;
+				std::cout << "push:" << g_counter << " on " << idx << std::endl;
+				std::this_thread::sleep_for(50ms);
+			}
+
+			std::this_thread::sleep_for(50ms);
+		}
+	});
+}
+
+static void resumable_mutex_synch()
 {
 	go test_mutex_push(0);
 	go test_mutex_pop(1);
 
 	this_scheduler()->run_until_notask();
+
+	std::cout << "result:" << g_counter << std::endl;
+}
+
+static void resumable_mutex_async()
+{
+	auto th = test_mutex_async_push(0);
+	std::this_thread::sleep_for(25ms);
+	go test_mutex_pop(1);
+
+	this_scheduler()->run_until_notask();
+	th.join();
+
+	std::cout << "result:" << g_counter << std::endl;
+}
+
+void resumable_main_mutex()
+{
+	resumable_mutex_synch();
+	std::cout << std::endl;
+
+	resumable_mutex_async();
 }

vs_proj/librf.cpp

@@ -47,8 +47,8 @@ int main(int argc, const char* argv[])
 	//resumable_main_event();
 	//resumable_main_event_timeout();
 	//resumable_main_sleep();
-	resumable_main_mutex();
-	return 0;
+	resumable_main_resumable();
+	//return 0;
 
 	//if (argc > 1)
 	//	resumable_main_benchmark_asio_client(atoi(argv[1]));

vs_proj/librf.vcxproj

@@ -23,7 +23,7 @@
     <ProjectGuid>{C1D4A6BD-592F-4E48-8178-7C87219BF80E}</ProjectGuid>
     <Keyword>Win32Proj</Keyword>
     <RootNamespace>librf</RootNamespace>
-    <WindowsTargetPlatformVersion>10.0</WindowsTargetPlatformVersion>
+    <WindowsTargetPlatformVersion>10.0.18362.0</WindowsTargetPlatformVersion>
   </PropertyGroup>
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
@@ -46,7 +46,7 @@
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
     <ConfigurationType>Application</ConfigurationType>
     <UseDebugLibraries>false</UseDebugLibraries>
-    <PlatformToolset>v142</PlatformToolset>
+    <PlatformToolset>v141</PlatformToolset>
     <WholeProgramOptimization>true</WholeProgramOptimization>
     <CharacterSet>NotSet</CharacterSet>
   </PropertyGroup>
@@ -154,7 +154,7 @@
       <AdditionalOptions>/await</AdditionalOptions>
       <FavorSizeOrSpeed>Size</FavorSizeOrSpeed>
       <StringPooling>true</StringPooling>
-      <LanguageStandard>stdcpplatest</LanguageStandard>
+      <LanguageStandard>stdcpp17</LanguageStandard>
       <MultiProcessorCompilation>true</MultiProcessorCompilation>
       <SDLCheck>
       </SDLCheck>