diff --git a/src/core/sal/perf/BreakpointBuffer.cc b/src/core/sal/perf/BreakpointBuffer.cc
index be222233..3b60244e 100644
--- a/src/core/sal/perf/BreakpointBuffer.cc
+++ b/src/core/sal/perf/BreakpointBuffer.cc
@@ -1,13 +1,14 @@
 #include "BreakpointBuffer.hpp"
 #include "../SALInst.hpp"
+#include "../Listener.hpp"
 
 namespace fail {
 
-// FIXME: can not be inlined this way
+// FIXME: Can not be inlined this way!
 ResultSet& PerfVectorBreakpoints::gather(BPEvent* pData)
 {
 	static ResultSet res;
-	res.clear();
+	res.clear(); // FIXME: This should not free the memory of the underlying std::vector.
 	// Search for all indices of matching listener objects:
 	for(std::vector<index_t>::iterator it = m_BufList.begin(); it != m_BufList.end(); ++it) {
 		BPListener* pLi = static_cast<BPListener*>(simulator.dereference(*it));
diff --git a/src/core/sal/perf/BreakpointBuffer.hpp b/src/core/sal/perf/BreakpointBuffer.hpp
index 7585cf68..9b4537da 100644
--- a/src/core/sal/perf/BreakpointBuffer.hpp
+++ b/src/core/sal/perf/BreakpointBuffer.hpp
@@ -2,58 +2,19 @@
   #define __BREAKPOINT_BUFFER_HPP__
 
 #include "BufferInterface.hpp"
-#include "../Listener.hpp"
-#include <cassert>
-#include <vector>
 
 // TODOs: 
 //  - Make these implementations even faster (see below: continue PerfVecSortedSingleBP).
-//  - The implementation of gather() (see below) in BreakpointBuffer.cc (not inlined in
-//    .hpp) avoids an include cycle. Unfortunately, this may cause a bad performance
-//    because gather() won't be inlined anymore! (The method is called quite often.)
 
 namespace fail {
 
 class BPEvent;
 
 /**
- * \class ResultSet
- *
- * Results (= indices of matching listeners) returned by the "gather"-method,
- * see below. (This class can be seen as a "temporary fire-list".)
+ * Concrete implementation for the \c BPSingleListener class.
  */
-class ResultSet {
-	std::vector<index_t> m_Res;
+class PerfVectorBreakpoints : public DefPerfVector<BPEvent> {
 public:
-	ResultSet() { }
-	bool hasMore() const { return !m_Res.empty(); }
-	index_t getNext() { index_t idx = m_Res.back(); m_Res.pop_back(); return idx; }
-	void add(index_t idx) { m_Res.push_back(idx); }
-	size_t size() const { return m_Res.size(); }
-	void clear() { m_Res.clear(); }
-};
-
-/**
- * Concrete implementation of the PerfBufferBase class for \c std::vector
- * and \c BPSingleListener.
- */
-class PerfVectorBreakpoints : public PerfBufferBase {
-protected:
-	std::vector<index_t> m_BufList;
-public:
-	void add(index_t idx) { m_BufList.push_back(idx); }
-	void remove(index_t idx)
-	{
-		for (std::vector<index_t>::iterator it = m_BufList.begin();
-		     it != m_BufList.end(); ++it) {
-			if (*it == idx) {
-				m_BufList.erase(it);
-				break;
-			}
-		}
-	}
-	void clear() { m_BufList.clear(); }
-	size_t size() const { return m_BufList.size(); }
 	ResultSet& gather(BPEvent* pData);
 };
 
diff --git a/src/core/sal/perf/BreakpointManagerSlice.ah b/src/core/sal/perf/BreakpointManagerSlice.ah
index 4d7266b9..65753990 100644
--- a/src/core/sal/perf/BreakpointManagerSlice.ah
+++ b/src/core/sal/perf/BreakpointManagerSlice.ah
@@ -5,7 +5,9 @@
 
 #ifdef CONFIG_FAST_BREAKPOINTS
 
+#include <cassert>
 #include "BreakpointBuffer.hpp"
+#include "../Listener.hpp"
 
 /**
  * \class BreakpointManagerSlice
diff --git a/src/core/sal/perf/BufferInterface.hpp b/src/core/sal/perf/BufferInterface.hpp
index 30b73ae5..b1fe622c 100644
--- a/src/core/sal/perf/BufferInterface.hpp
+++ b/src/core/sal/perf/BufferInterface.hpp
@@ -2,6 +2,7 @@
   #define __BUFFER_INTERFACE_HPP__
 
 #include <cstddef>
+#include <vector>
 
 namespace fail {
 
@@ -37,6 +38,50 @@ public:
 	virtual std::size_t size() const = 0;
 };
 
+/**
+ * \class ResultSet
+ *
+ * Results (= indices of matching listeners) returned by the "gather"-method,
+ * see below. (This class can be seen as a "temporary fire-list".)
+ */
+class ResultSet {
+private:
+	std::vector<index_t> m_Res; //!< vector of matching listener indices
+public:
+	ResultSet() { }
+	bool hasMore() const { return !m_Res.empty(); }
+	index_t getNext() { index_t idx = m_Res.back(); m_Res.pop_back(); return idx; }
+	void add(index_t idx) { m_Res.push_back(idx); }
+	size_t size() const { return m_Res.size(); }
+	void clear() { m_Res.clear(); }
+};
+
+/**
+ * \class DefPerfVector
+ * 
+ * Default \c std::vector based performance implementation (abstract)
+ */
+template<class T>
+class DefPerfVector : public PerfBufferBase {
+protected:
+	std::vector<index_t> m_BufList; //!< the performance buffer-list
+public:
+	void add(index_t idx) { m_BufList.push_back(idx); }
+	void remove(index_t idx)
+	{
+		for (std::vector<index_t>::iterator it = m_BufList.begin();
+		     it != m_BufList.end(); ++it) {
+			if (*it == idx) {
+				m_BufList.erase(it);
+				break;
+			}
+		}
+	}
+	void clear() { m_BufList.clear(); }
+	size_t size() const { return m_BufList.size(); }
+	virtual ResultSet& gather(T* pData) = 0;
+};
+
 } // end-of-namespace: fail
 
 #endif // __BUFFER_INTERFACE_HPP__