Merge branch 'listener_refactor'

Eric Dumazet says:

====================
inet: tcp listener refactoring, part 8

These patches prepare request socks being hashed into general ehash
table : We declare 3 aliases (ireq_state, ireq_refcnt, ireq_family)

Note that refcnt is not yet handled, this will be done later.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

include/net/inet_sock.h

@@ -27,6 +27,7 @@
 #include <net/sock.h>
 #include <net/request_sock.h>
 #include <net/netns/hash.h>
+#include <net/tcp_states.h>
 
 /** struct ip_options - IP Options
  *
@@ -79,6 +80,9 @@ struct inet_request_sock {
 #define ir_iif			req.__req_common.skc_bound_dev_if
 #define ir_cookie		req.__req_common.skc_cookie
 #define ireq_net		req.__req_common.skc_net
+#define ireq_state		req.__req_common.skc_state
+#define ireq_refcnt		req.__req_common.skc_refcnt
+#define ireq_family		req.__req_common.skc_family
 
 	kmemcheck_bitfield_begin(flags);
 	u16			snd_wscale : 4,
@@ -249,6 +253,8 @@ static inline struct request_sock *inet_reqsk_alloc(struct request_sock_ops *ops
 	if (req != NULL) {
 		kmemcheck_annotate_bitfield(ireq, flags);
 		ireq->opt = NULL;
+		atomic64_set(&ireq->ir_cookie, 0);
+		ireq->ireq_state = TCP_NEW_SYN_RECV;
 	}
 
 	return req;

include/net/request_sock.h

@@ -49,6 +49,8 @@ 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
+
 	struct request_sock		*dl_next;
 	u16				mss;
 	u8				num_retrans; /* number of retransmits */
@@ -75,6 +77,11 @@ static inline struct request_sock *reqsk_alloc(const struct request_sock_ops *op
 	return req;
 }
 
+static inline struct request_sock *inet_reqsk(struct sock *sk)
+{
+	return (struct request_sock *)sk;
+}
+
 static inline void __reqsk_free(struct request_sock *req)
 {
 	kmem_cache_free(req->rsk_ops->slab, req);
@@ -86,6 +93,12 @@ static inline void reqsk_free(struct request_sock *req)
 	__reqsk_free(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;
 
 /** struct listen_sock - listen state

include/net/sock.h

@@ -1625,7 +1625,7 @@ static inline void sock_put(struct sock *sk)
 		sk_free(sk);
 }
 /* Generic version of sock_put(), dealing with all sockets
- * (TCP_TIMEWAIT, ESTABLISHED...)
+ * (TCP_TIMEWAIT, TCP_NEW_SYN_RECV, ESTABLISHED...)
  */
 void sock_gen_put(struct sock *sk);
 

include/net/tcp_states.h

@@ -25,6 +25,7 @@ enum {
 	TCP_LAST_ACK,
 	TCP_LISTEN,
 	TCP_CLOSING,	/* Now a valid state */
+	TCP_NEW_SYN_RECV,
 
 	TCP_MAX_STATES	/* Leave at the end! */
 };
@@ -44,7 +45,8 @@ enum {
 	TCPF_CLOSE_WAIT	 = (1 << 8),
 	TCPF_LAST_ACK	 = (1 << 9),
 	TCPF_LISTEN	 = (1 << 10),
-	TCPF_CLOSING	 = (1 << 11) 
+	TCPF_CLOSING	 = (1 << 11),
+	TCPF_NEW_SYN_RECV = (1 << 12),
 };
 
 #endif	/* _LINUX_TCP_STATES_H */

net/core/sock.c

@@ -1668,6 +1668,8 @@ void sock_edemux(struct sk_buff *skb)
 
 	if (sk->sk_state == TCP_TIME_WAIT)
 		inet_twsk_put(inet_twsk(sk));
+	else if (sk->sk_state == TCP_NEW_SYN_RECV)
+		reqsk_put(inet_reqsk(sk));
 	else
 		sock_put(sk);
 }
@@ -2726,6 +2728,42 @@ static inline void release_proto_idx(struct proto *prot)
 }
 #endif
 
+static void req_prot_cleanup(struct request_sock_ops *rsk_prot)
+{
+	if (!rsk_prot)
+		return;
+	kfree(rsk_prot->slab_name);
+	rsk_prot->slab_name = NULL;
+	if (rsk_prot->slab) {
+		kmem_cache_destroy(rsk_prot->slab);
+		rsk_prot->slab = NULL;
+	}
+}
+
+static int req_prot_init(const struct proto *prot)
+{
+	struct request_sock_ops *rsk_prot = prot->rsk_prot;
+
+	if (!rsk_prot)
+		return 0;
+
+	rsk_prot->slab_name = kasprintf(GFP_KERNEL, "request_sock_%s",
+					prot->name);
+	if (!rsk_prot->slab_name)
+		return -ENOMEM;
+
+	rsk_prot->slab = kmem_cache_create(rsk_prot->slab_name,
+					   rsk_prot->obj_size, 0,
+					   SLAB_HWCACHE_ALIGN, NULL);
+
+	if (!rsk_prot->slab) {
+		pr_crit("%s: Can't create request sock SLAB cache!\n",
+			prot->name);
+		return -ENOMEM;
+	}
+	return 0;
+}
+
 int proto_register(struct proto *prot, int alloc_slab)
 {
 	if (alloc_slab) {
@@ -2739,21 +2777,8 @@ int proto_register(struct proto *prot, int alloc_slab)
 			goto out;
 		}
 
-		if (prot->rsk_prot != NULL) {
+		if (req_prot_init(prot))
-			prot->rsk_prot->slab_name = kasprintf(GFP_KERNEL, "request_sock_%s", prot->name);
+			goto out_free_request_sock_slab;
-			if (prot->rsk_prot->slab_name == NULL)
-				goto out_free_sock_slab;
-
-			prot->rsk_prot->slab = kmem_cache_create(prot->rsk_prot->slab_name,
-								 prot->rsk_prot->obj_size, 0,
-								 SLAB_HWCACHE_ALIGN, NULL);
-
-			if (prot->rsk_prot->slab == NULL) {
-				pr_crit("%s: Can't create request sock SLAB cache!\n",
-					prot->name);
-				goto out_free_request_sock_slab_name;
-			}
-		}
 
 		if (prot->twsk_prot != NULL) {
 			prot->twsk_prot->twsk_slab_name = kasprintf(GFP_KERNEL, "tw_sock_%s", prot->name);
@@ -2782,14 +2807,8 @@ int proto_register(struct proto *prot, int alloc_slab)
 out_free_timewait_sock_slab_name:
 	kfree(prot->twsk_prot->twsk_slab_name);
 out_free_request_sock_slab:
-	if (prot->rsk_prot && prot->rsk_prot->slab) {
+	req_prot_cleanup(prot->rsk_prot);
-		kmem_cache_destroy(prot->rsk_prot->slab);
+
-		prot->rsk_prot->slab = NULL;
-	}
-out_free_request_sock_slab_name:
-	if (prot->rsk_prot)
-		kfree(prot->rsk_prot->slab_name);
-out_free_sock_slab:
 	kmem_cache_destroy(prot->slab);
 	prot->slab = NULL;
 out:
@@ -2809,11 +2828,7 @@ void proto_unregister(struct proto *prot)
 		prot->slab = NULL;
 	}
 
-	if (prot->rsk_prot != NULL && prot->rsk_prot->slab != NULL) {
+	req_prot_cleanup(prot->rsk_prot);
-		kmem_cache_destroy(prot->rsk_prot->slab);
-		kfree(prot->rsk_prot->slab_name);
-		prot->rsk_prot->slab = NULL;
-	}
 
 	if (prot->twsk_prot != NULL && prot->twsk_prot->twsk_slab != NULL) {
 		kmem_cache_destroy(prot->twsk_prot->twsk_slab);

net/dccp/ipv4.c

@@ -642,7 +642,7 @@ int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 	ireq->ir_loc_addr = ip_hdr(skb)->daddr;
 	ireq->ir_rmt_addr = ip_hdr(skb)->saddr;
 	write_pnet(&ireq->ireq_net, sock_net(sk));
-	atomic64_set(&ireq->ir_cookie, 0);
+	ireq->ireq_family = AF_INET;
 
 	/*
 	 * Step 3: Process LISTEN state

net/dccp/ipv6.c

@@ -403,6 +403,8 @@ static int dccp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
 	ireq = inet_rsk(req);
 	ireq->ir_v6_rmt_addr = ipv6_hdr(skb)->saddr;
 	ireq->ir_v6_loc_addr = ipv6_hdr(skb)->daddr;
+	write_pnet(&ireq->ireq_net, sock_net(sk));
+	ireq->ireq_family = AF_INET6;
 
 	if (ipv6_opt_accepted(sk, skb, IP6CB(skb)) ||
 	    np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo ||

net/ipv4/inet_diag.c

@@ -718,7 +718,6 @@ static int inet_diag_fill_req(struct sk_buff *skb, struct sock *sk,
 			      const struct nlmsghdr *unlh)
 {
 	const struct inet_request_sock *ireq = inet_rsk(req);
-	struct inet_sock *inet = inet_sk(sk);
 	struct inet_diag_msg *r;
 	struct nlmsghdr *nlh;
 	long tmo;
@@ -729,7 +728,7 @@ static int inet_diag_fill_req(struct sk_buff *skb, struct sock *sk,
 		return -EMSGSIZE;
 
 	r = nlmsg_data(nlh);
-	r->idiag_family = sk->sk_family;
+	r->idiag_family = ireq->ireq_family;
 	r->idiag_state = TCP_SYN_RECV;
 	r->idiag_timer = 1;
 	r->idiag_retrans = req->num_retrans;
@@ -744,7 +743,7 @@ static int inet_diag_fill_req(struct sk_buff *skb, struct sock *sk,
 	if (tmo < 0)
 		tmo = 0;
 
-	r->id.idiag_sport = inet->inet_sport;
+	r->id.idiag_sport = htons(ireq->ir_num);
 	r->id.idiag_dport = ireq->ir_rmt_port;
 
 	memset(&r->id.idiag_src, 0, sizeof(r->id.idiag_src));

net/ipv4/inet_hashtables.c

@@ -262,6 +262,8 @@ void sock_gen_put(struct sock *sk)
 
 	if (sk->sk_state == TCP_TIME_WAIT)
 		inet_twsk_free(inet_twsk(sk));
+	else if (sk->sk_state == TCP_NEW_SYN_RECV)
+		reqsk_free(inet_reqsk(sk));
 	else
 		sk_free(sk);
 }

net/ipv4/syncookies.c

@@ -347,6 +347,7 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb)
 	treq->snt_synack	= tcp_opt.saw_tstamp ? tcp_opt.rcv_tsecr : 0;
 	treq->listener		= NULL;
 	write_pnet(&ireq->ireq_net, sock_net(sk));
+	ireq->ireq_family = AF_INET;
 
 	/* We throwed the options of the initial SYN away, so we hope
 	 * the ACK carries the same options again (see RFC1122 4.2.3.8)

net/ipv4/tcp_input.c

@@ -5966,7 +5966,6 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
 	tmp_opt.tstamp_ok = tmp_opt.saw_tstamp;
 	tcp_openreq_init(req, &tmp_opt, skb, sk);
 	write_pnet(&inet_rsk(req)->ireq_net, sock_net(sk));
-	atomic64_set(&inet_rsk(req)->ir_cookie, 0);
 
 	af_ops->init_req(req, sk, skb);
 

net/ipv4/tcp_ipv4.c

@@ -1228,6 +1228,7 @@ static void tcp_v4_init_req(struct request_sock *req, struct sock *sk,
 	ireq->ir_rmt_addr = ip_hdr(skb)->saddr;
 	ireq->no_srccheck = inet_sk(sk)->transparent;
 	ireq->opt = tcp_v4_save_options(skb);
+	ireq->ireq_family = AF_INET;
 }
 
 static struct dst_entry *tcp_v4_route_req(struct sock *sk, struct flowi *fl,
@@ -2204,7 +2205,7 @@ void tcp_proc_unregister(struct net *net, struct tcp_seq_afinfo *afinfo)
 }
 EXPORT_SYMBOL(tcp_proc_unregister);
 
-static void get_openreq4(const struct sock *sk, const struct request_sock *req,
+static void get_openreq4(const struct request_sock *req,
 			 struct seq_file *f, int i, kuid_t uid)
 {
 	const struct inet_request_sock *ireq = inet_rsk(req);
@@ -2214,7 +2215,7 @@ static void get_openreq4(const struct sock *sk, const struct request_sock *req,
 		" %02X %08X:%08X %02X:%08lX %08X %5u %8d %u %d %pK",
 		i,
 		ireq->ir_loc_addr,
-		ntohs(inet_sk(sk)->inet_sport),
+		ireq->ir_num,
 		ireq->ir_rmt_addr,
 		ntohs(ireq->ir_rmt_port),
 		TCP_SYN_RECV,
@@ -2225,7 +2226,7 @@ static void get_openreq4(const struct sock *sk, const struct request_sock *req,
 		from_kuid_munged(seq_user_ns(f), uid),
 		0,  /* non standard timer */
 		0, /* open_requests have no inode */
-		atomic_read(&sk->sk_refcnt),
+		0,
 		req);
 }
 
@@ -2332,7 +2333,7 @@ static int tcp4_seq_show(struct seq_file *seq, void *v)
 			get_tcp4_sock(v, seq, st->num);
 		break;
 	case TCP_SEQ_STATE_OPENREQ:
-		get_openreq4(st->syn_wait_sk, v, seq, st->num, st->uid);
+		get_openreq4(v, seq, st->num, st->uid);
 		break;
 	}
 out:

net/ipv6/syncookies.c

@@ -196,6 +196,8 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb)
 	ireq = inet_rsk(req);
 	treq = tcp_rsk(req);
 	treq->listener = NULL;
+	write_pnet(&ireq->ireq_net, sock_net(sk));
+	ireq->ireq_family = AF_INET6;
 
 	if (security_inet_conn_request(sk, skb, req))
 		goto out_free;

net/ipv6/tcp_ipv6.c

@@ -749,6 +749,7 @@ static void tcp_v6_init_req(struct request_sock *req, struct sock *sk,
 		atomic_inc(&skb->users);
 		ireq->pktopts = skb;
 	}
+	ireq->ireq_family = AF_INET6;
 }
 
 static struct dst_entry *tcp_v6_route_req(struct sock *sk, struct flowi *fl,
@@ -1689,7 +1690,7 @@ static void tcp_v6_destroy_sock(struct sock *sk)
 #ifdef CONFIG_PROC_FS
 /* Proc filesystem TCPv6 sock list dumping. */
 static void get_openreq6(struct seq_file *seq,
-			 const struct sock *sk, struct request_sock *req, int i, kuid_t uid)
+			 struct request_sock *req, int i, kuid_t uid)
 {
 	int ttd = req->expires - jiffies;
 	const struct in6_addr *src = &inet_rsk(req)->ir_v6_loc_addr;
@@ -1827,7 +1828,7 @@ static int tcp6_seq_show(struct seq_file *seq, void *v)
 			get_tcp6_sock(seq, v, st->num);
 		break;
 	case TCP_SEQ_STATE_OPENREQ:
-		get_openreq6(seq, st->syn_wait_sk, v, st->num, st->uid);
+		get_openreq6(seq, v, st->num, st->uid);
 		break;
 	}
 out: