Adapt primary inbound calls in x64 interpreter.

Change argument order for lj_vm_cpcall() to simplify x64 interpreter.

src/lj_vm.h

@@ -12,8 +12,8 @@
 LJ_ASMF void lj_vm_call(lua_State *L, TValue *base, int nres1);
 LJ_ASMF int lj_vm_pcall(lua_State *L, TValue *base, int nres1, ptrdiff_t ef);
 typedef TValue *(*lua_CPFunction)(lua_State *L, lua_CFunction func, void *ud);
-LJ_ASMF int lj_vm_cpcall(lua_State *L, lua_CPFunction cp, lua_CFunction func,
-			 void *ud);
+LJ_ASMF int lj_vm_cpcall(lua_State *L, lua_CFunction func, void *ud,
+			 lua_CPFunction cp);
 LJ_ASMF int lj_vm_resume(lua_State *L, TValue *base, int nres1, ptrdiff_t ef);
 LJ_ASMF_NORET void lj_vm_unwind_c(void *cframe, int errcode);
 LJ_ASMF_NORET void lj_vm_unwind_ff(void *cframe);