Speedup for guarded allocator

- Re-arrange locks, so no actual memory allocation
  (which is relatively slow) happens from inside
  the lock. operation system will take care of locks
  which might be needed there on it's own.

- Use spin lock instead of mutex, since it's just
  list operations happens from inside lock, no need
  in mutex here.

- Use atomic operations for memory in use and total
  used blocks counters.

This makes guarded allocator almost the same speed
as non-guarded one in files from Tube project.

There're still MemHead/MemTail overhead which might
be bad for CPU cache utilization

1 files changed, 3 insertions, 0 deletions

diff --git a/source/blender/windowmanager/intern/wm_init_exit.c b/source/blender/windowmanager/intern/wm_init_exit.c
index 63cf1eeb40d..4e9b849af40 100644
--- a/source/blender/windowmanager/intern/wm_init_exit.c
+++ b/source/blender/windowmanager/intern/wm_init_exit.c
@@ -51,6 +51,7 @@
 #include "BLI_listbase.h"
 #include "BLI_path_util.h"
 #include "BLI_string.h"
+#include "BLI_threads.h"
 #include "BLI_utildefines.h"
 
 #include "BKE_blender.h"
@@ -510,6 +511,8 @@ void WM_exit_ext(bContext *C, const short do_python)
 	
 	GHOST_DisposeSystemPaths();
 
+	BLI_threadapi_exit();
+
 	if (MEM_get_memory_blocks_in_use() != 0) {
 		printf("Error: Not freed memory blocks: %d\n", MEM_get_memory_blocks_in_use());
 		MEM_printmemlist();