1 files changed, 86 insertions, 54 deletions

diff --git a/include/net/request_sock.h b/include/net/request_sock.h
index 7f830ff67f08..fe41f3ceb008 100644
--- a/include/net/request_sock.h
+++ b/include/net/request_sock.h
@@ -39,8 +39,7 @@ struct request_sock_ops {
 	void		(*send_reset)(struct sock *sk,
 				      struct sk_buff *skb);
 	void		(*destructor)(struct request_sock *req);
-	void		(*syn_ack_timeout)(struct sock *sk,
-					   struct request_sock *req);
+	void		(*syn_ack_timeout)(const struct request_sock *req);
 };
 
 int inet_rtx_syn_ack(struct sock *parent, struct request_sock *req);
@@ -49,7 +48,11 @@ int inet_rtx_syn_ack(struct sock *parent, struct request_sock *req);
  */
 struct request_sock {
 	struct sock_common		__req_common;
+#define rsk_refcnt			__req_common.skc_refcnt
+#define rsk_hash			__req_common.skc_hash
+
 	struct request_sock		*dl_next;
+	struct sock			*rsk_listener;
 	u16				mss;
 	u8				num_retrans; /* number of retransmits */
 	u8				cookie_ts:1; /* syncookie: encode tcpopts in timestamp */
@@ -58,32 +61,56 @@ struct request_sock {
 	u32				window_clamp; /* window clamp at creation time */
 	u32				rcv_wnd;	  /* rcv_wnd offered first time */
 	u32				ts_recent;
-	unsigned long			expires;
+	struct timer_list		rsk_timer;
 	const struct request_sock_ops	*rsk_ops;
 	struct sock			*sk;
 	u32				secid;
 	u32				peer_secid;
 };
 
-static inline struct request_sock *reqsk_alloc(const struct request_sock_ops *ops)
+static inline struct request_sock *
+reqsk_alloc(const struct request_sock_ops *ops, struct sock *sk_listener)
 {
 	struct request_sock *req = kmem_cache_alloc(ops->slab, GFP_ATOMIC);
 
-	if (req != NULL)
+	if (req) {
 		req->rsk_ops = ops;
-
+		sock_hold(sk_listener);
+		req->rsk_listener = sk_listener;
+
+		/* Following is temporary. It is coupled with debugging
+		 * helpers in reqsk_put() & reqsk_free()
+		 */
+		atomic_set(&req->rsk_refcnt, 0);
+	}
 	return req;
 }
 
-static inline void __reqsk_free(struct request_sock *req)
+static inline struct request_sock *inet_reqsk(struct sock *sk)
 {
-	kmem_cache_free(req->rsk_ops->slab, req);
+	return (struct request_sock *)sk;
+}
+
+static inline struct sock *req_to_sk(struct request_sock *req)
+{
+	return (struct sock *)req;
 }
 
 static inline void reqsk_free(struct request_sock *req)
 {
+	/* temporary debugging */
+	WARN_ON_ONCE(atomic_read(&req->rsk_refcnt) != 0);
+
 	req->rsk_ops->destructor(req);
-	__reqsk_free(req);
+	if (req->rsk_listener)
+		sock_put(req->rsk_listener);
+	kmem_cache_free(req->rsk_ops->slab, req);
+}
+
+static inline void reqsk_put(struct request_sock *req)
+{
+	if (atomic_dec_and_test(&req->rsk_refcnt))
+		reqsk_free(req);
 }
 
 extern int sysctl_max_syn_backlog;
@@ -93,12 +120,16 @@ extern int sysctl_max_syn_backlog;
  * @max_qlen_log - log_2 of maximal queued SYNs/REQUESTs
  */
 struct listen_sock {
-	u8			max_qlen_log;
+	int			qlen_inc; /* protected by listener lock */
+	int			young_inc;/* protected by listener lock */
+
+	/* following fields can be updated by timer */
+	atomic_t		qlen_dec; /* qlen = qlen_inc - qlen_dec */
+	atomic_t		young_dec;
+
+	u8			max_qlen_log ____cacheline_aligned_in_smp;
 	u8			synflood_warned;
 	/* 2 bytes hole, try to use */
-	int			qlen;
-	int			qlen_young;
-	int			clock_hand;
 	u32			hash_rnd;
 	u32			nr_table_entries;
 	struct request_sock	*syn_table[0];
@@ -142,18 +173,11 @@ struct fastopen_queue {
  * %syn_wait_lock is necessary only to avoid proc interface having to grab the main
  * lock sock while browsing the listening hash (otherwise it's deadlock prone).
  *
- * This lock is acquired in read mode only from listening_get_next() seq_file
- * op and it's acquired in write mode _only_ from code that is actively
- * changing rskq_accept_head. All readers that are holding the master sock lock
- * don't need to grab this lock in read mode too as rskq_accept_head. writes
- * are always protected from the main sock lock.
  */
 struct request_sock_queue {
 	struct request_sock	*rskq_accept_head;
 	struct request_sock	*rskq_accept_tail;
-	rwlock_t		syn_wait_lock;
 	u8			rskq_defer_accept;
-	/* 3 bytes hole, try to pack */
 	struct listen_sock	*listen_opt;
 	struct fastopen_queue	*fastopenq; /* This is non-NULL iff TFO has been
 					     * enabled on this listener. Check
@@ -161,6 +185,9 @@ struct request_sock_queue {
 					     * to determine if TFO is enabled
 					     * right at this moment.
 					     */
+
+	/* temporary alignment, our goal is to get rid of this lock */
+	spinlock_t		syn_wait_lock ____cacheline_aligned_in_smp;
 };
 
 int reqsk_queue_alloc(struct request_sock_queue *queue,
@@ -186,12 +213,21 @@ static inline int reqsk_queue_empty(struct request_sock_queue *queue)
 }
 
 static inline void reqsk_queue_unlink(struct request_sock_queue *queue,
-				      struct request_sock *req,
-				      struct request_sock **prev_req)
+				      struct request_sock *req)
 {
-	write_lock(&queue->syn_wait_lock);
-	*prev_req = req->dl_next;
-	write_unlock(&queue->syn_wait_lock);
+	struct listen_sock *lopt = queue->listen_opt;
+	struct request_sock **prev;
+
+	spin_lock(&queue->syn_wait_lock);
+
+	prev = &lopt->syn_table[req->rsk_hash];
+	while (*prev != req)
+		prev = &(*prev)->dl_next;
+	*prev = req->dl_next;
+
+	spin_unlock(&queue->syn_wait_lock);
+	if (del_timer(&req->rsk_timer))
+		reqsk_put(req);
 }
 
 static inline void reqsk_queue_add(struct request_sock_queue *queue,
@@ -224,57 +260,53 @@ static inline struct request_sock *reqsk_queue_remove(struct request_sock_queue
 	return req;
 }
 
-static inline int reqsk_queue_removed(struct request_sock_queue *queue,
-				      struct request_sock *req)
+static inline void reqsk_queue_removed(struct request_sock_queue *queue,
+				       const struct request_sock *req)
 {
 	struct listen_sock *lopt = queue->listen_opt;
 
 	if (req->num_timeout == 0)
-		--lopt->qlen_young;
-
-	return --lopt->qlen;
+		atomic_inc(&lopt->young_dec);
+	atomic_inc(&lopt->qlen_dec);
 }
 
-static inline int reqsk_queue_added(struct request_sock_queue *queue)
+static inline void reqsk_queue_added(struct request_sock_queue *queue)
 {
 	struct listen_sock *lopt = queue->listen_opt;
-	const int prev_qlen = lopt->qlen;
 
-	lopt->qlen_young++;
-	lopt->qlen++;
-	return prev_qlen;
+	lopt->young_inc++;
+	lopt->qlen_inc++;
 }
 
-static inline int reqsk_queue_len(const struct request_sock_queue *queue)
+static inline int listen_sock_qlen(const struct listen_sock *lopt)
 {
-	return queue->listen_opt != NULL ? queue->listen_opt->qlen : 0;
+	return lopt->qlen_inc - atomic_read(&lopt->qlen_dec);
 }
 
-static inline int reqsk_queue_len_young(const struct request_sock_queue *queue)
+static inline int listen_sock_young(const struct listen_sock *lopt)
 {
-	return queue->listen_opt->qlen_young;
+	return lopt->young_inc - atomic_read(&lopt->young_dec);
 }
 
-static inline int reqsk_queue_is_full(const struct request_sock_queue *queue)
+static inline int reqsk_queue_len(const struct request_sock_queue *queue)
 {
-	return queue->listen_opt->qlen >> queue->listen_opt->max_qlen_log;
+	const struct listen_sock *lopt = queue->listen_opt;
+
+	return lopt ? listen_sock_qlen(lopt) : 0;
 }
 
-static inline void reqsk_queue_hash_req(struct request_sock_queue *queue,
-					u32 hash, struct request_sock *req,
-					unsigned long timeout)
+static inline int reqsk_queue_len_young(const struct request_sock_queue *queue)
 {
-	struct listen_sock *lopt = queue->listen_opt;
-
-	req->expires = jiffies + timeout;
-	req->num_retrans = 0;
-	req->num_timeout = 0;
-	req->sk = NULL;
-	req->dl_next = lopt->syn_table[hash];
+	return listen_sock_young(queue->listen_opt);
+}
 
-	write_lock(&queue->syn_wait_lock);
-	lopt->syn_table[hash] = req;
-	write_unlock(&queue->syn_wait_lock);
+static inline int reqsk_queue_is_full(const struct request_sock_queue *queue)
+{
+	return reqsk_queue_len(queue) >> queue->listen_opt->max_qlen_log;
 }
 
+void reqsk_queue_hash_req(struct request_sock_queue *queue,
+			  u32 hash, struct request_sock *req,
+			  unsigned long timeout);
+
 #endif /* _REQUEST_SOCK_H */