From: Paul E. McKenney Date: Wed, 11 Apr 2018 21:33:18 +0000 (-0700) Subject: rcu: Avoid __call_rcu_core() root rcu_node ->lock acquisition X-Git-Tag: v4.18-rc1~169^2~1^2^3~11 X-Git-Url: https://asedeno.scripts.mit.edu/gitweb/?a=commitdiff_plain;h=a6058d85a2b24fa40ce7f0d7683989ec47b603b9;p=linux.git rcu: Avoid __call_rcu_core() root rcu_node ->lock acquisition When __call_rcu_core() notices excessive numbers of callbacks pending on the current CPU, we know that at least one of them is not yet classified, namely the one that was just now queued. Therefore, it is not necessary to invoke rcu_start_gp() and thus not necessary to acquire the root rcu_node structure's ->lock. This commit therefore replaces the rcu_start_gp() with rcu_accelerate_cbs(), thus replacing an acquisition of the root rcu_node structure's ->lock with that of this CPU's leaf rcu_node structure. This decreases contention on the root rcu_node structure's ->lock. Reported-by: Nicholas Piggin Signed-off-by: Paul E. McKenney Tested-by: Nicholas Piggin --- diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index f75eb5174021..6396a3d10be9 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -2988,11 +2988,11 @@ static void __call_rcu_core(struct rcu_state *rsp, struct rcu_data *rdp, /* Start a new grace period if one not already started. */ if (!rcu_gp_in_progress(rsp)) { - struct rcu_node *rnp_root = rcu_get_root(rsp); + struct rcu_node *rnp = rdp->mynode; - raw_spin_lock_rcu_node(rnp_root); - needwake = rcu_start_gp(rsp); - raw_spin_unlock_rcu_node(rnp_root); + raw_spin_lock_rcu_node(rnp); + needwake = rcu_accelerate_cbs(rsp, rnp, rdp); + raw_spin_unlock_rcu_node(rnp); if (needwake) rcu_gp_kthread_wake(rsp); } else {