+/*
+ * Beware: xchg on x86 has an implied lock prefix. There will be the cost of
+ * full lock semantics even though they are not needed.
+ */
+#define percpu_xchg_op(var, nval) \
+({ \
+ typeof(var) pxo_ret__; \
+ typeof(var) pxo_new__ = (nval); \
+ switch (sizeof(var)) { \
+ case 1: \
+ asm("xchgb %2, "__percpu_arg(1) \
+ : "=a" (pxo_ret__), "+m" (var) \
+ : "q" (pxo_new__) \
+ : "memory"); \
+ break; \
+ case 2: \
+ asm("xchgw %2, "__percpu_arg(1) \
+ : "=a" (pxo_ret__), "+m" (var) \
+ : "r" (pxo_new__) \
+ : "memory"); \
+ break; \
+ case 4: \
+ asm("xchgl %2, "__percpu_arg(1) \
+ : "=a" (pxo_ret__), "+m" (var) \
+ : "r" (pxo_new__) \
+ : "memory"); \
+ break; \
+ case 8: \
+ asm("xchgq %2, "__percpu_arg(1) \
+ : "=a" (pxo_ret__), "+m" (var) \
+ : "r" (pxo_new__) \
+ : "memory"); \
+ break; \
+ default: __bad_percpu_size(); \
+ } \
+ pxo_ret__; \
+})
+
+/*
+ * cmpxchg has no such implied lock semantics as a result it is much
+ * more efficient for cpu local operations.
+ */
+#define percpu_cmpxchg_op(var, oval, nval) \
+({ \
+ typeof(var) pco_ret__; \
+ typeof(var) pco_old__ = (oval); \
+ typeof(var) pco_new__ = (nval); \
+ switch (sizeof(var)) { \
+ case 1: \
+ asm("cmpxchgb %2, "__percpu_arg(1) \
+ : "=a" (pco_ret__), "+m" (var) \
+ : "q" (pco_new__), "0" (pco_old__) \
+ : "memory"); \
+ break; \
+ case 2: \
+ asm("cmpxchgw %2, "__percpu_arg(1) \
+ : "=a" (pco_ret__), "+m" (var) \
+ : "r" (pco_new__), "0" (pco_old__) \
+ : "memory"); \
+ break; \
+ case 4: \
+ asm("cmpxchgl %2, "__percpu_arg(1) \
+ : "=a" (pco_ret__), "+m" (var) \
+ : "r" (pco_new__), "0" (pco_old__) \
+ : "memory"); \
+ break; \
+ case 8: \
+ asm("cmpxchgq %2, "__percpu_arg(1) \
+ : "=a" (pco_ret__), "+m" (var) \
+ : "r" (pco_new__), "0" (pco_old__) \
+ : "memory"); \
+ break; \
+ default: __bad_percpu_size(); \
+ } \
+ pco_ret__; \
+})
+