Commit d20162e0 authored by Paul E. McKenney's avatar Paul E. McKenney
Browse files

srcu: Move srcu_barrier() fields from srcu_struct to srcu_usage



This commit moves the ->srcu_barrier_seq, ->srcu_barrier_mutex,
->srcu_barrier_completion, and ->srcu_barrier_cpu_cnt fields from the
srcu_struct structure to the srcu_usage structure to reduce the size of
the former in order to improve cache locality.

Suggested-by: default avatarChristoph Hellwig <hch@lst.de>
Tested-by: default avatarSachin Sant <sachinp@linux.ibm.com>
Tested-by: default avatar"Zhang, Qiang1" <qiang1.zhang@intel.com>
Tested-by: default avatarJoel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: default avatarPaul E. McKenney <paulmck@kernel.org>
parent 660349ac
Loading
Loading
Loading
Loading
+7 −7
Original line number Diff line number Diff line
@@ -77,6 +77,13 @@ struct srcu_usage {
	unsigned long srcu_n_lock_retries;	/* Contention events in current interval. */
	unsigned long srcu_n_exp_nodelay;	/* # expedited no-delays in current GP phase. */
	bool sda_is_static;			/* May ->sda be passed to free_percpu()? */
	unsigned long srcu_barrier_seq;		/* srcu_barrier seq #. */
	struct mutex srcu_barrier_mutex;	/* Serialize barrier ops. */
	struct completion srcu_barrier_completion;
						/* Awaken barrier rq at end. */
	atomic_t srcu_barrier_cpu_cnt;		/* # CPUs not yet posting a */
						/*  callback for the barrier */
						/*  operation. */
};

/*
@@ -85,13 +92,6 @@ struct srcu_usage {
struct srcu_struct {
	unsigned int srcu_idx;			/* Current rdr array element. */
	struct srcu_data __percpu *sda;		/* Per-CPU srcu_data array. */
	unsigned long srcu_barrier_seq;		/* srcu_barrier seq #. */
	struct mutex srcu_barrier_mutex;	/* Serialize barrier ops. */
	struct completion srcu_barrier_completion;
						/* Awaken barrier rq at end. */
	atomic_t srcu_barrier_cpu_cnt;		/* # CPUs not yet posting a */
						/*  callback for the barrier */
						/*  operation. */
	unsigned long reschedule_jiffies;
	unsigned long reschedule_count;
	struct delayed_work work;
+19 −19
Original line number Diff line number Diff line
@@ -248,9 +248,9 @@ static int init_srcu_struct_fields(struct srcu_struct *ssp, bool is_static)
	mutex_init(&ssp->srcu_sup->srcu_gp_mutex);
	ssp->srcu_idx = 0;
	ssp->srcu_sup->srcu_gp_seq = 0;
	ssp->srcu_barrier_seq = 0;
	mutex_init(&ssp->srcu_barrier_mutex);
	atomic_set(&ssp->srcu_barrier_cpu_cnt, 0);
	ssp->srcu_sup->srcu_barrier_seq = 0;
	mutex_init(&ssp->srcu_sup->srcu_barrier_mutex);
	atomic_set(&ssp->srcu_sup->srcu_barrier_cpu_cnt, 0);
	INIT_DELAYED_WORK(&ssp->work, process_srcu);
	ssp->srcu_sup->sda_is_static = is_static;
	if (!is_static)
@@ -1496,8 +1496,8 @@ static void srcu_barrier_cb(struct rcu_head *rhp)

	sdp = container_of(rhp, struct srcu_data, srcu_barrier_head);
	ssp = sdp->ssp;
	if (atomic_dec_and_test(&ssp->srcu_barrier_cpu_cnt))
		complete(&ssp->srcu_barrier_completion);
	if (atomic_dec_and_test(&ssp->srcu_sup->srcu_barrier_cpu_cnt))
		complete(&ssp->srcu_sup->srcu_barrier_completion);
}

/*
@@ -1511,13 +1511,13 @@ static void srcu_barrier_cb(struct rcu_head *rhp)
static void srcu_barrier_one_cpu(struct srcu_struct *ssp, struct srcu_data *sdp)
{
	spin_lock_irq_rcu_node(sdp);
	atomic_inc(&ssp->srcu_barrier_cpu_cnt);
	atomic_inc(&ssp->srcu_sup->srcu_barrier_cpu_cnt);
	sdp->srcu_barrier_head.func = srcu_barrier_cb;
	debug_rcu_head_queue(&sdp->srcu_barrier_head);
	if (!rcu_segcblist_entrain(&sdp->srcu_cblist,
				   &sdp->srcu_barrier_head)) {
		debug_rcu_head_unqueue(&sdp->srcu_barrier_head);
		atomic_dec(&ssp->srcu_barrier_cpu_cnt);
		atomic_dec(&ssp->srcu_sup->srcu_barrier_cpu_cnt);
	}
	spin_unlock_irq_rcu_node(sdp);
}
@@ -1530,20 +1530,20 @@ void srcu_barrier(struct srcu_struct *ssp)
{
	int cpu;
	int idx;
	unsigned long s = rcu_seq_snap(&ssp->srcu_barrier_seq);
	unsigned long s = rcu_seq_snap(&ssp->srcu_sup->srcu_barrier_seq);

	check_init_srcu_struct(ssp);
	mutex_lock(&ssp->srcu_barrier_mutex);
	if (rcu_seq_done(&ssp->srcu_barrier_seq, s)) {
	mutex_lock(&ssp->srcu_sup->srcu_barrier_mutex);
	if (rcu_seq_done(&ssp->srcu_sup->srcu_barrier_seq, s)) {
		smp_mb(); /* Force ordering following return. */
		mutex_unlock(&ssp->srcu_barrier_mutex);
		mutex_unlock(&ssp->srcu_sup->srcu_barrier_mutex);
		return; /* Someone else did our work for us. */
	}
	rcu_seq_start(&ssp->srcu_barrier_seq);
	init_completion(&ssp->srcu_barrier_completion);
	rcu_seq_start(&ssp->srcu_sup->srcu_barrier_seq);
	init_completion(&ssp->srcu_sup->srcu_barrier_completion);

	/* Initial count prevents reaching zero until all CBs are posted. */
	atomic_set(&ssp->srcu_barrier_cpu_cnt, 1);
	atomic_set(&ssp->srcu_sup->srcu_barrier_cpu_cnt, 1);

	idx = __srcu_read_lock_nmisafe(ssp);
	if (smp_load_acquire(&ssp->srcu_sup->srcu_size_state) < SRCU_SIZE_WAIT_BARRIER)
@@ -1554,12 +1554,12 @@ void srcu_barrier(struct srcu_struct *ssp)
	__srcu_read_unlock_nmisafe(ssp, idx);

	/* Remove the initial count, at which point reaching zero can happen. */
	if (atomic_dec_and_test(&ssp->srcu_barrier_cpu_cnt))
		complete(&ssp->srcu_barrier_completion);
	wait_for_completion(&ssp->srcu_barrier_completion);
	if (atomic_dec_and_test(&ssp->srcu_sup->srcu_barrier_cpu_cnt))
		complete(&ssp->srcu_sup->srcu_barrier_completion);
	wait_for_completion(&ssp->srcu_sup->srcu_barrier_completion);

	rcu_seq_end(&ssp->srcu_barrier_seq);
	mutex_unlock(&ssp->srcu_barrier_mutex);
	rcu_seq_end(&ssp->srcu_sup->srcu_barrier_seq);
	mutex_unlock(&ssp->srcu_sup->srcu_barrier_mutex);
}
EXPORT_SYMBOL_GPL(srcu_barrier);