diff --git a/pjlib/include/pj/stack.h b/pjlib/include/pj/stack.h
index 6e9cf32611..da68ce5258 100644
--- a/pjlib/include/pj/stack.h
+++ b/pjlib/include/pj/stack.h
@@ -47,8 +47,6 @@
 
 #endif // PJ_STACK_IMPLEMENTATION
 
-#include <pj/stack.h>
-
 
 PJ_BEGIN_DECL
 
@@ -64,7 +62,7 @@ PJ_BEGIN_DECL
  * Stack in PJLIB is single-linked list with First In Last Out logic. 
  * Stack is thread safe. Common PJLIB stack implementation uses internal locking mechanism so is thread-safe.
  * Implementation for Windows platform uses locking free Windows embeded single linked list implementation.
- * The performance of pj_stack implementation for Windows platform is 2-5x higher than cross-platform.
+ * The performance of pj_stack implementation for Windows platform is considerably higher than cross-platform.
  * 
  * By default pjlib compile and link os independent "cross-platform" implementation. 
  * To select implementation you may optionaly define PJ_STACK_IMPLEMENTATION as PJ_STACK_WIN32 
@@ -77,13 +75,14 @@ PJ_BEGIN_DECL
  * 
  * Windows single linked list implementation requires aligned data, both stack item and stack itself should 
  * be aligned by 8 (for x86) or 16 (for x64) byte. 
- * pjsip build system define PJ_POOL_ALIGNMENT macro to corresponding value.
  * winnt.h define MEMORY_ALLOCATION_ALIGNMENT macro for this purpose.
- * To use this macro in build system we recomend (this is optional) to add #include <windows.h> 
- * to your config_site.h.
+ * pjsip build system define PJ_POOL_ALIGNMENT macro to corresponding value.
  * You may redefine PJ_POOL_ALIGNMENT in your config_site.h but to use PJ_STACK_WIN32 implementation 
  * PJ_POOL_ALIGNMENTshould not be less then MEMORY_ALLOCATION_ALIGNMENT
  * 
+ * To use MEMORY_ALLOCATION_ALIGNMENT macro in build system we recomend (this is optional) to add
+ * #include <windows.h> to your config_site.h.
+ * 
  * Stack won't require dynamic memory allocation (just as all PJLIB data structures). The stack here
  * should be viewed more like a low level C stack instead of high level C++ stack
  * (which normally are easier to use but require dynamic memory allocations),
diff --git a/pjlib/src/pj/stack.c b/pjlib/src/pj/stack.c
index f0d99db880..624f53ecde 100644
--- a/pjlib/src/pj/stack.c
+++ b/pjlib/src/pj/stack.c
@@ -143,12 +143,6 @@ PJ_DEF(pj_stack_t*) pj_stack_pop(pj_stack_type *stack)
  * Traverse the stack and get it's elements quantity.
  * The return value of pj_stack_size should not be relied upon in multithreaded applications
  * because the item count can be changed at any time by another thread.
- * For Windows platform returns the number of entries in the stack modulo 65535. For example,
- * if the specified stack contains 65536 entries, pj_stack_size returns zero.
- *
- * @param stack     The target stack.
- *
- * @return          Number of elements.
  */
 PJ_DEF(pj_size_t) pj_stack_size(/*const*/ pj_stack_type *stack)
 {
diff --git a/pjlib/src/pj/stack_win32.c b/pjlib/src/pj/stack_win32.c
index 0335e3adba..f315dcb6f8 100644
--- a/pjlib/src/pj/stack_win32.c
+++ b/pjlib/src/pj/stack_win32.c
@@ -140,10 +140,6 @@ PJ_DEF(pj_stack_t*) pj_stack_pop(pj_stack_type *stack)
  * because the item count can be changed at any time by another thread.
  * For Windows platform returns the number of entries in the stack modulo 65535. For example,
  * if the specified stack contains 65536 entries, pj_stack_size returns zero.
- *
- * @param stack	    The target stack.
- *
- * @return	    Number of elements.
  */
 PJ_DEF(pj_size_t) pj_stack_size(/*const*/ pj_stack_type *stack)
 {
diff --git a/pjlib/src/pjlib-test/stack.c b/pjlib/src/pjlib-test/stack.c
index 3bb6c811a8..ed16d5f5a0 100644
--- a/pjlib/src/pjlib-test/stack.c
+++ b/pjlib/src/pjlib-test/stack.c
@@ -51,6 +51,15 @@
 
 #endif
 
+/* define this variable to control multithreaded testing for other platforms */
+#ifndef HAS_MT_STACK_STRESS_TEST
+#   ifdef PJ_WIN32
+#       define HAS_MT_STACK_STRESS_TEST 1
+#   endif
+#       define HAS_MT_STACK_STRESS_TEST 0
+#   endif
+#endif
+
 #define THIS_FILE       "stack.c"
 #define MAX_RESERVED    16
 #define MAX_SLOTS       100
@@ -215,10 +224,12 @@ int stack_test()
         rc = -55;
     }
 
+#ifdef HAS_MT_STACK_STRESS_TEST
     for (i = 0; !rc && i < PJ_ARRAY_SIZE(tests); ++i) {
         tests[i].state.pool = pool;
         rc = stack_stress_test(&tests[i]);
     }
+#endif //HAS_MT_STACK_STRESS_TEST
 
     if (pool)
         pj_pool_release(pool);