kern/rcu: comment the loop in rcu_cpu_data_check_gp_state

1 files changed, 9 insertions, 1 deletions

diff --git a/kern/rcu.c b/kern/rcu.c
index 34ebc0f..973d6d5 100644
--- a/kern/rcu.c
+++ b/kern/rcu.c
@@ -638,7 +638,13 @@ rcu_cpu_data_check_gp_state(struct rcu_cpu_data *cpu_data)
 
     data = &rcu_data;
 
-    for (;;) {
+    /*
+     * A loop is used to optimize the case where a processor is the last to
+     * acknowledge a grace period state change, in which case the latter
+     * also immediately changes and can be acknowleged right away. As a
+     * result, this loop may never run more than twice.
+     */
+    for (unsigned int i = 0; /* no condition */; i++) {
         local_gp_state = cpu_data->gp_state;
         diff = rcu_data_check_gp_state(data, local_gp_state, &global_gp_state);
 
@@ -646,6 +652,8 @@ rcu_cpu_data_check_gp_state(struct rcu_cpu_data *cpu_data)
             break;
         }
 
+        assert(i < 2);
+
         switch (global_gp_state) {
         case RCU_GP_STATE_WORK_WINDOW_FLIP:
             rcu_cpu_data_flip_work_wid(cpu_data);