diff --git a/Makefile.darwin b/Makefile.darwin
index 61df6ec..d6f61eb 100755
--- a/Makefile.darwin
+++ b/Makefile.darwin
@@ -1,6 +1,6 @@
 include config.darwin
 
-OBJS= src/clocks.o src/core_profiler.o src/function_meter.o src/stack.o src/lua50_profiler.o
+OBJS= src/clocks.o src/cache.o src/core_profiler.o src/function_meter.o src/stack.o src/lua50_profiler.o
 
 profiler: $(OBJS)
 	export MACOSX_DEPLOYMENT_TARGET=10.3 && mkdir -p bin && gcc -bundle -undefined dynamic_lookup -o $(PROFILER_OUTPUT) $(OBJS)
diff --git a/Makefile.linux b/Makefile.linux
index 0a2c2c9..c57d60f 100755
--- a/Makefile.linux
+++ b/Makefile.linux
@@ -1,7 +1,6 @@
 include config.linux
 
-OBJS= src/clocks.o src/core_profiler.o src/function_meter.o src/stack.o src/lua50_profiler.o
-
+OBJS= src/clocks.o src/cache.o src/core_profiler.o src/function_meter.o src/stack.o src/lua50_profiler.o
 
 profiler: $(OBJS)
 	mkdir -p bin && $(LD) -Bshareable -o $(PROFILER_OUTPUT) $(OBJS)
diff --git a/Makefile.win b/Makefile.win
index 3304422..41870b1 100755
--- a/Makefile.win
+++ b/Makefile.win
@@ -5,7 +5,7 @@ V= 2.0.1
 
 include config.win
 
-OBJS= src\clocks.obj src\core_profiler.obj src\function_meter.obj src\stack.obj src\lua50_profiler.obj
+OBJS= src\clocks.obj src\cache.obj src\core_profiler.obj src\function_meter.obj src\stack.obj src\lua50_profiler.obj
 
 lib: src\$(LIBNAME)
 
diff --git a/src/cache.c b/src/cache.c
new file mode 100644
index 0000000..9762694
--- /dev/null
+++ b/src/cache.c
@@ -0,0 +1,79 @@
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "cache.h"
+#include <stdlib.h>
+#include <string.h>
+
+#define LPROF_CACHE_SIZE (1024*1024)
+#define LPROF_LINE_SIZE (256)
+#define LPROF_OUTPUT_LINE_SIZE (4096)
+
+struct Cache
+{
+	char line_buf[LPROF_LINE_SIZE];
+	char data[LPROF_CACHE_SIZE];
+	size_t pos;
+};
+
+static struct Cache* cache;
+
+void lprofCache_append(FILE* outfile, const char* fmt, ...)
+{
+	va_list ap;
+	va_start(ap, fmt);
+	lprofCache_vappend(outfile, fmt, ap);
+	va_end(ap);
+}
+
+void lprofCache_vappend(FILE* outfile, const char* fmt, va_list args)
+{
+	if (cache == NULL)
+	{
+		cache = (struct Cache*)malloc(sizeof(struct Cache));
+	}
+
+	vsprintf(cache->line_buf, fmt, args);
+	size_t len = strlen(cache->line_buf);
+	if (cache->pos + len >= LPROF_CACHE_SIZE - 1) /* leave a space for '\0' */
+	{
+		lprofCache_write_all(outfile);
+	}
+	
+	memcpy(cache->data + cache->pos, cache->line_buf, len + 1); /* len+1 copy '\0' */
+	cache->pos += len;
+}
+
+void lprofCache_write_all(FILE* outfile)
+{
+	size_t start = 0;
+	while(cache->pos > 0)
+	{
+		size_t line_size = cache->pos < LPROF_OUTPUT_LINE_SIZE ?
+								 cache->pos : LPROF_OUTPUT_LINE_SIZE;
+		fwrite(cache->data + start, sizeof(char), line_size, outfile);
+		cache->pos -= line_size;
+		start += line_size;
+	}
+}
+
+/* test cache */
+/*
+int main(int argc, char const *argv[])
+{
+	FILE* file = fopen("/tmp/test.out", "a");
+	lprofCache_append(file, "abcdefg");
+	lprofCache_append(file, "b\n");
+	lprofCache_append(file, "c\n");
+
+	lprofCache_write_all(file);
+	fclose(file);
+
+	return 0;
+}
+*/
+
+#ifdef __cplusplus
+}
+#endif
\ No newline at end of file
diff --git a/src/cache.h b/src/cache.h
new file mode 100644
index 0000000..d2ad295
--- /dev/null
+++ b/src/cache.h
@@ -0,0 +1,30 @@
+/*
+** LuaProfiler
+** Jennal
+** 2014-11-15
+*/
+
+/*****************************************************************************
+cache.h:
+   Module to faster write file
+
+Design:
+   'lprofCache_append()' append string to cache
+   'lprofCache_vappend()' append string to cache with va_list
+   'lprofCache_write_all()' write all cache data to output and clear cache
+*****************************************************************************/
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <stdio.h>
+#include <stdarg.h>
+
+void lprofCache_append(FILE* outfile, const char* fmt, ...);
+void lprofCache_vappend(FILE* outfile, const char* fmt, va_list args);
+void lprofCache_write_all(FILE* outfile);
+    
+#ifdef __cplusplus
+}
+#endif
diff --git a/src/clocks.c b/src/clocks.c
index 132e1c8..db8837d 100755
--- a/src/clocks.c
+++ b/src/clocks.c
@@ -14,6 +14,11 @@ clocks.c:
                           was started
 *****************************************************************************/
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+    
+
 #include <stdio.h>
 #include "clocks.h"
 
@@ -61,3 +66,6 @@ clock_t clocks;
         return (float)clocks / (float)CLOCKS_PER_SEC;
 }
 
+#ifdef __cplusplus
+}
+#endif
diff --git a/src/clocks.h b/src/clocks.h
index a230e2a..f5c695a 100755
--- a/src/clocks.h
+++ b/src/clocks.h
@@ -14,7 +14,16 @@ clocks.h:
                           was started
 *****************************************************************************/
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+    
+
 #include <time.h>
 
 void lprofC_start_timer(clock_t *time_marker);
 float lprofC_get_seconds(clock_t time_marker);
+    
+#ifdef __cplusplus
+}
+#endif
diff --git a/src/core_profiler.c b/src/core_profiler.c
index c879b9c..c8f30da 100755
--- a/src/core_profiler.c
+++ b/src/core_profiler.c
@@ -34,14 +34,19 @@ the vertex is "printed" only after all his descendents have been processed (in
 a depth-first search recursive algorithm).
 *****************************************************************************/
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+    
+
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 #include <stdarg.h>
 
 #include "function_meter.h"
-
 #include "core_profiler.h"
+#include "cache.h" /* Jennal added */
 
     /* default log name (%s is used to place a random string) */
 #define OUT_FILENAME "lprof_%s.out"
@@ -57,13 +62,19 @@ static float function_call_time;
 /* output a line to the log file, using 'printf()' syntax */
 /* assume the timer is off */
 static void output(const char *format, ...) {
+  /* Jennal modified */
+  // va_list ap;
+  // va_start(ap, format);
+  // vfprintf(outf, format, ap);
+  // va_end(ap);
+
+  // /* write now to avoid delays when the timer is on */
+  // fflush(outf);
+
   va_list ap;
   va_start(ap, format);
-  vfprintf(outf, format, ap);
+  lprofCache_vappend(outf, format, ap);
   va_end(ap);
-
-  /* write now to avoid delays when the timer is on */
-  fflush(outf);
 }
 
 
@@ -97,10 +108,18 @@ int lprofP_callhookOUT(lprofP_STATE* S) {
 
   S->stack_level--;
 
-  /* 0: do not resume the parent function's timer yet... */
-  info = lprofM_leave_function(S, 0);
+  //Jennal move up
+    /* original code is
+     // 0: do not resume the parent function's timer yet...
+     info = lprofM_leave_function(S, 0);
+     // writing a log may take too long to be computed with the function's time ...
+     lprofM_pause_total_time(S);
+    */
   /* writing a log may take too long to be computed with the function's time ...*/
   lprofM_pause_total_time(S);
+  /* 0: do not resume the parent function's timer yet... */
+  info = lprofM_leave_function(S, 0);
+
   info->local_time += function_call_time;
   info->total_time += function_call_time;
   
@@ -169,6 +188,7 @@ lprofP_STATE* lprofP_init_core_profiler(const char *_out_filename, int isto_prin
   /* initialize the 'function_meter' */
   S = lprofM_init();
   if(!S) {
+    lprofCache_write_all(outf);
     fclose(outf);
     return 0;
   }
@@ -177,7 +197,11 @@ lprofP_STATE* lprofP_init_core_profiler(const char *_out_filename, int isto_prin
 }
 
 void lprofP_close_core_profiler(lprofP_STATE* S) {
-  if(outf) fclose(outf);
+  if(outf) {
+    lprofCache_write_all(outf);
+    fclose(outf);
+  }
+
   if(S) free(S);
 }
 
@@ -195,3 +219,6 @@ lprofP_STATE* lprofP_create_profiler(float _function_call_time) {
   return S;
 }
 
+#ifdef __cplusplus
+}
+#endif
diff --git a/src/core_profiler.h b/src/core_profiler.h
index 447ec88..e63e40f 100755
--- a/src/core_profiler.h
+++ b/src/core_profiler.h
@@ -16,6 +16,10 @@ Design (using the Lua callhook mechanism) :
    'lprofP_callhookOUT'        called whenever Lua leaves a function
 *****************************************************************************/
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+    
 #include "stack.h"
 
 /* computes new stack and new timer */
@@ -25,7 +29,11 @@ void lprofP_callhookIN(lprofP_STATE* S, char *func_name, char *file, int linedef
 /* returns if there is another function in the stack */
 int  lprofP_callhookOUT(lprofP_STATE* S);
 
+void lprofP_close_core_profiler(lprofP_STATE* S);
 /* opens the log file */
 /* returns true if the file could be opened */
 lprofP_STATE* lprofP_init_core_profiler(const char *_out_filename, int isto_printheader, float _function_call_time);
 
+#ifdef __cplusplus
+}
+#endif
diff --git a/src/function_meter.c b/src/function_meter.c
index 709846f..7c7404e 100755
--- a/src/function_meter.c
+++ b/src/function_meter.c
@@ -25,6 +25,11 @@ function_meter.c:
                               another one or when the function terminates
 *****************************************************************************/
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
@@ -201,3 +206,8 @@ lprofP_STATE* lprofM_init() {
     return S;
   } else return NULL;
 }
+    
+    
+#ifdef __cplusplus
+}
+#endif
diff --git a/src/function_meter.h b/src/function_meter.h
index 39afd3a..7445de3 100755
--- a/src/function_meter.h
+++ b/src/function_meter.h
@@ -25,6 +25,14 @@ function_meter.c:
                               another one or when the function terminates
 *****************************************************************************/
 
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+    
+
+
+
 #include "stack.h"
 
 
@@ -62,3 +70,7 @@ lprofS_STACK_RECORD *lprofM_leave_function(lprofP_STATE* S, int isto_resume);
 
 /* init stack */
 lprofP_STATE* lprofM_init();
+    
+#ifdef __cplusplus
+}
+#endif
diff --git a/src/lua50_profiler.c b/src/lua50_profiler.c
index 4e64db1..92e8562 100755
--- a/src/lua50_profiler.c
+++ b/src/lua50_profiler.c
@@ -9,6 +9,7 @@ lua50_profiler.c:
    Lua version dependent profiler interface
 *****************************************************************************/
 
+#include "lua50_profiler.h"
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
@@ -17,9 +18,14 @@ lua50_profiler.c:
 #include "core_profiler.h"
 #include "function_meter.h"
 
-#include "lua.h"
 #include "lauxlib.h"
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+    
+
+
 /* Indices for the main profiler stack and for the original exit function */
 static int exit_id;
 static int profstate_id;
@@ -36,6 +42,7 @@ static void callhook(lua_State *L, lua_Debug *ar) {
   lua_pushlightuserdata(L, &profstate_id);
   lua_gettable(L, LUA_REGISTRYINDEX);
   S = (lprofP_STATE*)lua_touserdata(L, -1);
+  lua_pop(L, 1);
 
   if (lua_getstack(L, 1, &previous_ar) == 0) {
     currentline = -1;
@@ -238,3 +245,6 @@ int luaopen_profiler(lua_State *L) {
   lua_settable (L, -3);
   return 1;
 }
+#ifdef __cplusplus
+}
+#endif
diff --git a/src/lua50_profiler.h b/src/lua50_profiler.h
new file mode 100644
index 0000000..a58e895
--- /dev/null
+++ b/src/lua50_profiler.h
@@ -0,0 +1,25 @@
+/*
+ *  lua50_profiler.h
+ *  RDTower
+ *
+ *  Created by changshuai on 14/11/10.
+ *
+ */
+
+#ifndef RDTower_lua50_profiler_h
+#define RDTower_lua50_profiler_h
+
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "lua.h"
+
+int luaopen_profiler(lua_State *L);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/src/luaprofiler.h b/src/luaprofiler.h
index e435610..c9bd62c 100755
--- a/src/luaprofiler.h
+++ b/src/luaprofiler.h
@@ -8,7 +8,15 @@
 luaprofiler.h:
     Must be included by your main module, in order to profile Lua programs
 *****************************************************************************/
+#ifdef __cplusplus
+extern "C" {
+#endif
+  
+    void init_profiler(void *);
+    
+#ifdef __cplusplus
+}
+#endif
 
 
-void init_profiler(void *);
 
diff --git a/src/stack.c b/src/stack.c
index eff6198..20e9937 100755
--- a/src/stack.c
+++ b/src/stack.c
@@ -9,6 +9,13 @@ stack.c:
    Simple stack manipulation
 *****************************************************************************/
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+    
+
+    
+
 #include <stdio.h>
 #include <stdlib.h>
 #include "stack.h"
@@ -31,3 +38,6 @@ lprofS_STACK q;
         free(q);
         return r;
 }
+#ifdef __cplusplus
+}
+#endif
diff --git a/src/stack.h b/src/stack.h
index b513f92..423c3cc 100755
--- a/src/stack.h
+++ b/src/stack.h
@@ -12,6 +12,11 @@ stack.h:
 #ifndef _STACK_H
 #define _STACK_H
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+    
+
 #include <time.h>
 
 typedef struct lprofS_sSTACK_RECORD lprofS_STACK_RECORD;
@@ -42,3 +47,7 @@ void lprofS_push(lprofS_STACK *p, lprofS_STACK_RECORD r);
 lprofS_STACK_RECORD lprofS_pop(lprofS_STACK *p);
 
 #endif
+    
+#ifdef __cplusplus
+}
+#endif