kernel - Improve vm.vmmeter sysctl performance

* Improve performance by reducing the rate at which the marker is moved.

* Fix a bug, noting that lwkt_yield() can cause us to lose the token,
  by always using the marker for both it and the collision case.

diff --git a/sys/vm/vm_meter.c b/sys/vm/vm_meter.c
index bd09a92..2c5093b 100644 (file)
--- a/sys/vm/vm_meter.c
+++ b/sys/vm/vm_meter.c
@@ -98,6 +98,7 @@ do_vmtotal(SYSCTL_HANDLER_ARGS)
        struct vm_object marker;
        vm_object_t object;
        long collisions;
+       int burst;
 
        bzero(&total, sizeof(total));
        totalp = &total;
@@ -130,6 +131,7 @@ do_vmtotal(SYSCTL_HANDLER_ARGS)
         */
        lwkt_gettoken(&vmobj_token);
        TAILQ_INSERT_HEAD(&vm_object_list, &marker, object_list);
+       burst = 0;
 
        for (object = TAILQ_FIRST(&vm_object_list);
            object != NULL;
@@ -171,14 +173,21 @@ do_vmtotal(SYSCTL_HANDLER_ARGS)
                }
 
                /*
+                * Don't waste time unnecessarily
+                */
+               if (++burst < 25)
+                       continue;
+               burst = 0;
+
+               /*
                 * Don't hog the vmobj_token if someone else wants it.
                 */
+               TAILQ_REMOVE(&vm_object_list, &marker, object_list);
+               TAILQ_INSERT_AFTER(&vm_object_list, object,
+                                  &marker, object_list);
+               object = &marker;
                if (collisions != vmobj_token.t_collisions) {
-                       TAILQ_REMOVE(&vm_object_list, &marker, object_list);
-                       TAILQ_INSERT_AFTER(&vm_object_list, object,
-                                          &marker, object_list);
                        tsleep(&vm_object_list, 0, "breath", 1);
-                       object = &marker;
                        collisions = vmobj_token.t_collisions;
                } else {
                        lwkt_yield();