pcsx2/common/include/Utilities/Threading.h

/*  PCSX2 - PS2 Emulator for PCs
 *  Copyright (C) 2002-2009  PCSX2 Dev Team
 * 
 *  PCSX2 is free software: you can redistribute it and/or modify it under the terms
 *  of the GNU Lesser General Public License as published by the Free Software Found-
 *  ation, either version 3 of the License, or (at your option) any later version.
 *
 *  PCSX2 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 PCSX2.
 *  If not, see <http://www.gnu.org/licenses/>.
 */
 
#pragma once

#include <semaphore.h>
#include <errno.h> // EBUSY
#include <pthread.h>

#include "Pcsx2Defs.h"
#include "ScopedPtr.h"

#undef Yield		// release th burden of windows.h global namespace spam.
class wxTimeSpan;

#define AllowFromMainThreadOnly() \
	pxAssertMsg( wxThread::IsMain(), "Thread affinity violation: Call allowed from main thread only." )

namespace Threading
{
	// --------------------------------------------------------------------------------------
	//  Platform Specific External APIs
	// --------------------------------------------------------------------------------------
	// The following set of documented functions have Linux/Win32 specific implementations,
	// which are found in WinThreads.cpp and LnxThreads.cpp
	

	// Returns the number of available logical CPUs (cores plus hyperthreaded cpus)
	extern void CountLogicalCores( int LogicalCoresPerPhysicalCPU, int PhysicalCoresPerPhysicalCPU );

	// Releases a timeslice to other threads.
	extern void Timeslice();

	// For use in spin/wait loops.
	extern void SpinWait();

	// sleeps the current thread for the given number of milliseconds.
	extern void Sleep( int ms );


	// pthread Cond is an evil api that is not suited for Pcsx2 needs.
	// Let's not use it. Use mutexes and semaphores instead to create waits. (Air)
#if 0
	struct WaitEvent
	{
		pthread_cond_t cond;
		pthread_mutex_t mutex;

		WaitEvent();
		~WaitEvent() throw();

		void Set();
		void Wait();
	};
#endif

	// --------------------------------------------------------------------------------------
	//  NonblockingMutex
	// --------------------------------------------------------------------------------------
	// This is a very simple non-blocking mutex, which behaves similarly to pthread_mutex's
	// trylock(), but without any of the extra overhead needed to set up a structure capable
	// of blocking waits.  It basically optimizes to a single InterlockedExchange.
	//
	// Simple use: if TryLock() returns false, the Bool is already interlocked by another thread.
	// If TryLock() returns true, you've locked the object and are *responsible* for unlocking
	// it later.
	//
	class NonblockingMutex
	{
	protected:
		volatile long val;

	public:
		NonblockingMutex() : val( false ) {}
		virtual ~NonblockingMutex() throw() {}

		bool TryLock() throw()
		{
			return !_InterlockedExchange( &val, true );
		}

		bool IsLocked()
		{ return !!val; }

		void Release()
		{ val = false; }
	};

	class Semaphore
	{
	protected:
		sem_t m_sema;

	public:
		Semaphore();
		virtual ~Semaphore() throw();

		void Reset();
		void Post();
		void Post( int multiple );
		
		void WaitRaw();
		bool WaitRaw( const wxTimeSpan& timeout );
		void WaitNoCancel();
		int  Count();

#if wxUSE_GUI
		void Wait();
		bool Wait( const wxTimeSpan& timeout );

	protected:
		bool _WaitGui_RecursionGuard();
#endif
	};

	class MutexLock
	{
	protected:
		pthread_mutex_t mutex;

	public:
		MutexLock();
		virtual ~MutexLock() throw();

		void Lock();
		void Unlock();
		bool TryLock();
	
	protected:
		// empty constructor used by MutexLockRecursive
		MutexLock( bool ) {}
	};

	class MutexLockRecursive : public MutexLock
	{
	public:
		MutexLockRecursive();
		virtual ~MutexLockRecursive() throw();
	};

// --------------------------------------------------------------------------------------
// IThread - Interface for the public access to PersistentThread.
// --------------------------------------------------------------------------------------
// Class usage: Can be used for allowing safe nullification of a thread handle.  Rather
// than being NULL'd, the handle can be mapped to an IThread implementation which acts
// as a do-nothing placebo or an assertion generator.
//
	class IThread
	{
		DeclareNoncopyableObject(IThread);

	public:
		IThread() {}
		virtual ~IThread() throw() {}

		virtual bool IsSelf() const { return false; }
		virtual bool IsRunning() { return false; }

		virtual void Start() {}
		virtual void Cancel( bool isBlocking = true ) {}
		virtual void Block() {}
		virtual bool Detach() { return false; }
	};

// --------------------------------------------------------------------------------------
// PersistentThread - Helper class for the basics of starting/managing persistent threads.
// --------------------------------------------------------------------------------------
// This class is meant to be a helper for the typical threading model of "start once and 
// reuse many times."  This class incorporates a lot of extra overhead in stopping and
// starting threads, but in turn provides most of the basic thread-safety and event-handling
// functionality needed for a threaded operation.  In practice this model is usually an
// ideal one for efficiency since Operating Systems themselves typically subscribe to a
// design where sleeping, suspending, and resuming threads is very efficient, but starting
// new threads has quite a bit of overhead.
//
// To use this as a base class for your threaded procedure, overload the following virtual
// methods:
//  void OnStart();
//  void ExecuteTaskInThread();
//  void OnCleanupInThread();
//  
// Use the public methods Start() and Cancel() to start and shutdown the thread, and use
// m_sem_event internally to post/receive events for the thread (make a public accessor for
// it in your derived class if your thread utilizes the post).
//
// Notes:
//  * Constructing threads as static global vars isn't recommended since it can potentially
//    confuse w32pthreads, if the static initializers are executed out-of-order (C++ offers
//    no dependency options for ensuring correct static var initializations).  Use heap
//    allocation to create thread objects instead.
//
	class PersistentThread : public virtual IThread
	{
		DeclareNoncopyableObject(PersistentThread);

	protected:
		typedef int (*PlainJoeFP)();
		
		wxString	m_name;				// diagnostic name for our thread.

		pthread_t	m_thread;
		Semaphore	m_sem_event;		// general wait event that's needed by most threads.
		Semaphore	m_sem_finished;		// used for canceling and closing threads in a deadlock-safe manner
		MutexLockRecursive	m_lock_start;	// used to lock the Start() code from starting simultaneous threads accidentally.
		
		volatile long m_detached;		// a boolean value which indicates if the m_thread handle is valid
		volatile long m_running;		// set true by Start(), and set false by Cancel(), Block(), etc.

		// exception handle, set non-NULL if the thread terminated with an exception
		// Use RethrowException() to re-throw the exception using its original exception type.
		ScopedPtr<Exception::BaseException> m_except;
		
	public:
		virtual ~PersistentThread() throw();
		PersistentThread();
		PersistentThread( const char* name );

		virtual void Start();
		virtual void Cancel( bool isBlocking = true );
		virtual bool Detach();
		virtual void Block();
		virtual void RethrowException() const;

		bool IsRunning() const;
		bool IsSelf() const;
		wxString GetName() const;

	protected:
		// Extending classes should always implement your own OnStart(), which is called by
		// Start() once necessary locks have been obtained.  Do not override Start() directly
		// unless you're really sure that's what you need to do. ;)
		virtual void OnStart()=0;
		virtual void OnCleanupInThread()=0;

		// Implemented by derived class to handle threading actions!
		virtual void ExecuteTaskInThread()=0;

		// Inserts a thread cancellation point.  If the thread has received a cancel request, this
		// function will throw an SEH exception designed to exit the thread (so make sure to use C++
		// object encapsulation for anything that could leak resources, to ensure object unwinding
		// and cleanup, or use the DoThreadCleanup() override to perform resource cleanup).
		void TestCancel();

		// Yields this thread to other threads and checks for cancellation.  A sleeping thread should
		// always test for cancellation, however if you really don't want to, you can use Threading::Sleep()
		// or better yet, disable cancellation of the thread completely with DisableCancellation().
		//
		// Parameters:
		//   ms - 'minimum' yield time in milliseconds (rough -- typically yields are longer by 1-5ms
		//         depending on operating system/platform).  If ms is 0 or unspecified, then a single
		//         timeslice is yielded to other contending threads.  If no threads are contending for
		//         time when ms==0, then no yield is done, but cancellation is still tested.
		void Yield( int ms = 0 )
		{
			pxAssert( IsSelf() );
			Threading::Sleep( ms );
			TestCancel();
		}

		// ----------------------------------------------------------------------------
		// Section of methods for internal use only.

		void _DoSetThreadName( const wxString& name );
		void _DoSetThreadName( __unused const char* name );
		void _internal_execute();
		void _try_virtual_invoke( void (PersistentThread::*method)() );
		void _ThreadCleanup();

		static void* _internal_callback( void* func );
		static void _pt_callback_cleanup( void* handle );
	};
	
	//////////////////////////////////////////////////////////////////////////////////////////
	// ScopedLock: Helper class for using Mutexes.
	// Using this class provides an exception-safe (and generally clean) method of locking
	// code inside a function or conditional block.
	//
	class ScopedLock
	{
		DeclareNoncopyableObject(ScopedLock);

	protected:
		MutexLock&	m_lock;
		bool		m_IsLocked;

	public:
		virtual ~ScopedLock() throw()
		{
			if( m_IsLocked )
				m_lock.Unlock();
		}

		ScopedLock( MutexLock& locker ) :
			m_lock( locker )
		,	m_IsLocked( true )
		{
			m_lock.Lock();
		}

		// Provides manual unlocking of a scoped lock prior to object destruction.
		void Unlock()
		{
			if( !m_IsLocked ) return;
			m_IsLocked = false;
			m_lock.Unlock();
		}

		// provides manual locking of a scoped lock, to re-lock after a manual unlocking.
		void Lock()
		{
			if( m_IsLocked ) return;
			m_lock.Lock();
			m_IsLocked = true;
		}
		
		bool IsLocked() const { return m_IsLocked; }

	protected:
		// Special constructor used by ScopedTryLock
		ScopedLock( MutexLock& locker, bool isTryLock ) :
			m_lock( locker )
		,	m_IsLocked( isTryLock ? m_lock.TryLock() : false )
		{
		}
			
	};
	
	class ScopedTryLock : public ScopedLock
	{
	public:
		ScopedTryLock( MutexLock& locker ) : ScopedLock( locker, true ) { }
		virtual ~ScopedTryLock() throw() {}
		bool Failed() const { return !m_IsLocked; }
	};

	//////////////////////////////////////////////////////////////////////////////////////////
	// BaseTaskThread - an abstract base class which provides simple parallel execution of
	// single tasks.
	//
	// Implementation:
	//   To use this class your derived class will need to implement its own Task() function
	//   and also a "StartTask( parameters )" function which suits the need of your task, along
	//   with any local variables your task needs to do its job.  You may additionally want to
	//   implement a "GetResult()" function, which would be a combination of WaitForResult()
	//   and a return value of the computational result.
	//
	// Thread Safety:
	//   If operating on local variables, you must execute WaitForResult() before leaving the
	//   variable scope -- or alternatively have your StartTask() implementation make full
	//   copies of dependent data.  Also, by default PostTask() always assumes the previous
	//   task has completed.  If your system can post a new task before the previous one has
	//   completed, then it needs to explicitly call WaitForResult() or provide a mechanism
	//   to cancel the previous task (which is probably more work than it's worth).
	//
	// Performance notes:
	//  * Remember that thread creation is generally slow, so you should make your object
	//    instance once early and then feed it tasks repeatedly over the course of program
	//    execution.
	//
	//  * For threading to be a successful speedup, the task being performed should be as lock
	//    free as possible.  For example using STL containers in parallel usually fails to
	//    yield any speedup due to the gratuitous amount of locking that the STL performs
	//    internally.
	//
	//  * The best application of tasking threads is to divide a large loop over a linear array
	//    into smaller sections.  For example, if you have 20,000 items to process, the task
	//    can be divided into two threads of 10,000 items each.
	//
	class BaseTaskThread : public PersistentThread
	{
	protected:
		volatile bool m_Done;
		volatile bool m_TaskPending;
		Semaphore m_post_TaskComplete;
		MutexLock m_lock_TaskComplete;

	public:
		virtual ~BaseTaskThread() throw() {}
		BaseTaskThread() :
			m_Done( false )
		,	m_TaskPending( false )
		,	m_post_TaskComplete()
		{
		}

		void Block();
		void PostTask();
		void WaitForResult();

	protected:
		// Abstract method run when a task has been posted.  Implementing classes should do
		// all your necessary processing work here.
		virtual void Task()=0;

		virtual void ExecuteTaskInThread();
	};

	//////////////////////////////////////////////////////////////////////////////////////////
	// Our fundamental interlocking functions.  All other useful interlocks can be derived
	// from these little beasties!

	extern void AtomicExchange( volatile u32& Target, u32 value );
	extern void AtomicExchangeAdd( volatile u32& Target, u32 value );
	extern void AtomicIncrement( volatile u32& Target );
	extern void AtomicDecrement( volatile u32& Target );
	extern void AtomicExchange( volatile s32& Target, s32 value );
	extern void AtomicExchangeAdd( volatile s32& Target, u32 value );
	extern void AtomicIncrement( volatile s32& Target );
	extern void AtomicDecrement( volatile s32& Target );

	extern void _AtomicExchangePointer( const void ** target, const void* value );
	extern void _AtomicCompareExchangePointer( const void ** target, const void* value, const void* comparand );

	#define AtomicExchangePointer( target, value ) \
		_AtomicExchangePointer( (const void**)(&target), (const void*)(value) )

	#define AtomicCompareExchangePointer( target, value, comparand ) \
		_AtomicCompareExchangePointer( (const void**)(&target), (const void*)(value), (const void*)(comparand) )

}