percpu: Optimize __get_cpu_var()

Redefine __get_cpu_var() using this_cpu_ptr() which can be
arch-optimized.
Signed-off-by: Brian Gerst <brgerst@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>

include/asm-generic/percpu.h

@@ -55,10 +55,6 @@ extern unsigned long __per_cpu_offset[NR_CPUS];
  */
 #define per_cpu(var, cpu) \
 	(*SHIFT_PERCPU_PTR(&(var), per_cpu_offset(cpu)))
-#define __get_cpu_var(var) \
-	(*SHIFT_PERCPU_PTR(&(var), my_cpu_offset))
-#define __raw_get_cpu_var(var) \
-	(*SHIFT_PERCPU_PTR(&(var), __my_cpu_offset))
 
 #ifndef __this_cpu_ptr
 #define __this_cpu_ptr(ptr) SHIFT_PERCPU_PTR(ptr, __my_cpu_offset)
@@ -69,6 +65,9 @@ extern unsigned long __per_cpu_offset[NR_CPUS];
 #define this_cpu_ptr(ptr) __this_cpu_ptr(ptr)
 #endif
 
+#define __get_cpu_var(var) (*this_cpu_ptr(&(var)))
+#define __raw_get_cpu_var(var) (*__this_cpu_ptr(&(var)))
+
 #ifdef CONFIG_HAVE_SETUP_PER_CPU_AREA
 extern void setup_per_cpu_areas(void);
 #endif