rcu: Parameterize rcu_check_gp_start_stall()

In order to debug forward-progress stalls, it is necessary to check
for excessively delayed grace-period starts.  This is currently done
for RCU CPU stall warnings by rcu_check_gp_start_stall(), which checks
to see if the start of a requested grace period has been delayed by an
RCU CPU stall warning period.  Because rcutorture will need to check
for the time consumed by an RCU forward-progress delay, this commit
promotes gpssdelay from a local variable to a formal parameter.  It is
not necessary to export rcu_check_gp_start_stall() because rcutorture
will access it via a wrapper function.

Signed-off-by: Paul E. McKenney <paulmck@linux.ibm.com>

diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 4933f5d..36e3015 100644 (file)
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -2600,10 +2600,10 @@ static void force_quiescent_state(void)
  * This function checks for grace-period requests that fail to motivate
  * RCU to come out of its idle mode.
  */
-static void
-rcu_check_gp_start_stall(struct rcu_node *rnp, struct rcu_data *rdp)
+void
+rcu_check_gp_start_stall(struct rcu_node *rnp, struct rcu_data *rdp,
+                        const unsigned long gpssdelay)
 {
-       const unsigned long gpssdelay = rcu_jiffies_till_stall_check();
        unsigned long flags;
        unsigned long j;
        struct rcu_node *rnp_root = rcu_get_root();
@@ -2690,7 +2690,7 @@ static __latent_entropy void rcu_process_callbacks(struct softirq_action *unused
                local_irq_restore(flags);
        }
 
-       rcu_check_gp_start_stall(rnp, rdp);
+       rcu_check_gp_start_stall(rnp, rdp, rcu_jiffies_till_stall_check());
 
        /* If there are callbacks ready, invoke them. */
        if (rcu_segcblist_ready_cbs(&rdp->cblist))