Spinlock removed, causes long standing issues on single core machines.

node/Mutex.hpp

@@ -24,69 +24,6 @@
 
 namespace ZeroTier {
 
-#if defined(__GNUC__) && (defined(__amd64) || defined(__amd64__) || defined(__x86_64) || defined(__x86_64__) || defined(__AMD64) || defined(__AMD64__) || defined(_M_X64))
-
-// Inline ticket lock on x64 systems with GCC and CLANG (Mac, Linux) -- this is really fast as long as locking durations are very short
-class Mutex
-{
-public:
-	Mutex() :
-		nextTicket(0),
-		nowServing(0)
-	{
-	}
-
-	inline void lock() const
-	{
-		const uint16_t myTicket = __sync_fetch_and_add(&(const_cast<Mutex *>(this)->nextTicket),1);
-		while (nowServing != myTicket) {
-			__asm__ __volatile__("rep;nop"::);
-			__asm__ __volatile__("":::"memory");
-		}
-	}
-
-	inline void unlock() const
-	{
-		++(const_cast<Mutex *>(this)->nowServing);
-	}
-
-	/**
-	 * Uses C++ contexts and constructor/destructor to lock/unlock automatically
-	 */
-	class Lock
-	{
-	public:
-		Lock(Mutex &m) :
-			_m(&m)
-		{
-			m.lock();
-		}
-
-		Lock(const Mutex &m) :
-			_m(const_cast<Mutex *>(&m))
-		{
-			_m->lock();
-		}
-
-		~Lock()
-		{
-			_m->unlock();
-		}
-
-	private:
-		Mutex *const _m;
-	};
-
-private:
-	Mutex(const Mutex &) {}
-	const Mutex &operator=(const Mutex &) { return *this; }
-
-	uint16_t nextTicket;
-	uint16_t nowServing;
-};
-
-#else
-
 // libpthread based mutex lock
 class Mutex
 {
@@ -142,11 +79,9 @@ private:
 	pthread_mutex_t _mh;
 };
 
-#endif
-
 } // namespace ZeroTier
 
-#endif // Apple / Linux
+#endif
 
 #ifdef __WINDOWS__