MultiThreaded Demo:

- fixing various race conditions throughout (usage of static vars, etc)
 - addition of a few lightweight mutexes (which are compiled out by default)
 - slight code rearrangement in discreteDynamicsWorld to facilitate multithreading
 - PoolAllocator::allocate() can now be called when pool is full without
     crashing (null pointer returned)
 - PoolAllocator allocate and freeMemory, are OPTIONALLY threadsafe
     (default is un-threadsafe)
 - CollisionDispatcher no longer checks if the pool allocator is full
     before calling allocate(), instead it just calls allocate() and
     checks if the return is null -- this avoids a race condition
 - SequentialImpulseConstraintSolver OPTIONALLY uses different logic in
     getOrInitSolverBody() to avoid a race condition with kinematic bodies
 - addition of 2 classes which together allow simulation islands to be run
   in parallel:
    - btSimulationIslandManagerMt
    - btDiscreteDynamicsWorldMt
 - MultiThreadedDemo example in the example browser demonstrating use of
   OpenMP, Microsoft PPL, and Intel TBB
 - use multithreading for other demos
 - benchmark demo: add parallel raycasting

src/BulletCollision/BroadphaseCollision/btDbvt.h

@@ -267,7 +267,6 @@ struct	btDbvt
 
 	
 	btAlignedObjectArray<sStkNN>	m_stkStack;
-	mutable btAlignedObjectArray<const btDbvtNode*>	m_rayTestStack;
 
 
 	// Methods
@@ -357,6 +356,7 @@ struct	btDbvt
 								btScalar lambda_max,
 								const btVector3& aabbMin,
 								const btVector3& aabbMax,
+                                btAlignedObjectArray<const btDbvtNode*>& stack,
 								DBVT_IPOLICY) const;
 
 	DBVT_PREFIX
@@ -1006,7 +1006,8 @@ inline void		btDbvt::rayTestInternal(	const btDbvtNode* root,
 								btScalar lambda_max,
 								const btVector3& aabbMin,
 								const btVector3& aabbMax,
-								DBVT_IPOLICY) const
+                                btAlignedObjectArray<const btDbvtNode*>& stack,
+                                DBVT_IPOLICY ) const
 {
         (void) rayTo;
 	DBVT_CHECKTYPE
@@ -1016,7 +1017,6 @@ inline void		btDbvt::rayTestInternal(	const btDbvtNode* root,
 
 		int								depth=1;
 		int								treshold=DOUBLE_STACKSIZE-2;
-		btAlignedObjectArray<const btDbvtNode*>&	stack = m_rayTestStack;
 		stack.resize(DOUBLE_STACKSIZE);
 		stack[0]=root;
 		btVector3 bounds[2];

src/BulletCollision/BroadphaseCollision/btDbvtBroadphase.cpp

@@ -16,6 +16,7 @@ subject to the following restrictions:
 ///btDbvtBroadphase implementation by Nathanael Presson
 
 #include "btDbvtBroadphase.h"
+#include "LinearMath/btThreads.h"
 
 //
 // Profiling
@@ -142,6 +143,11 @@ btDbvtBroadphase::btDbvtBroadphase(btOverlappingPairCache* paircache)
 	{
 		m_stageRoots[i]=0;
 	}
+#if BT_THREADSAFE
+    m_rayTestStacks.resize(BT_MAX_THREAD_COUNT);
+#else
+    m_rayTestStacks.resize(1);
+#endif
 #if DBVT_BP_PROFILE
 	clear(m_profiling);
 #endif
@@ -227,6 +233,23 @@ struct	BroadphaseRayTester : btDbvt::ICollide
 void	btDbvtBroadphase::rayTest(const btVector3& rayFrom,const btVector3& rayTo, btBroadphaseRayCallback& rayCallback,const btVector3& aabbMin,const btVector3& aabbMax)
 {
 	BroadphaseRayTester callback(rayCallback);
+    btAlignedObjectArray<const btDbvtNode*>* stack = &m_rayTestStacks[0];
+#if BT_THREADSAFE
+    // for this function to be threadsafe, each thread must have a separate copy
+    // of this stack.  This could be thread-local static to avoid dynamic allocations,
+    // instead of just a local.
+    int threadIndex = btGetCurrentThreadIndex();
+    btAlignedObjectArray<const btDbvtNode*> localStack;
+    if (threadIndex < m_rayTestStacks.size())
+    {
+        // use per-thread preallocated stack if possible to avoid dynamic allocations
+        stack = &m_rayTestStacks[threadIndex];
+    }
+    else
+    {
+        stack = &localStack;
+    }
+#endif
 
 	m_sets[0].rayTestInternal(	m_sets[0].m_root,
 		rayFrom,
@@ -236,6 +259,7 @@ void	btDbvtBroadphase::rayTest(const btVector3& rayFrom,const btVector3& rayTo,
 		rayCallback.m_lambda_max,
 		aabbMin,
 		aabbMax,
+        *stack,
 		callback);
 
 	m_sets[1].rayTestInternal(	m_sets[1].m_root,
@@ -246,6 +270,7 @@ void	btDbvtBroadphase::rayTest(const btVector3& rayFrom,const btVector3& rayTo,
 		rayCallback.m_lambda_max,
 		aabbMin,
 		aabbMax,
+        *stack,
 		callback);
 
 }

src/BulletCollision/BroadphaseCollision/btDbvtBroadphase.h

@@ -87,6 +87,7 @@ struct	btDbvtBroadphase : btBroadphaseInterface
 	bool					m_releasepaircache;			// Release pair cache on delete
 	bool					m_deferedcollide;			// Defere dynamic/static collision to collide call
 	bool					m_needcleanup;				// Need to run cleanup?
+    btAlignedObjectArray< btAlignedObjectArray<const btDbvtNode*> > m_rayTestStacks;
 #if DBVT_BP_PROFILE
 	btClock					m_clock;
 	struct	{