[PATCH net-next 0/6] inet: tcp listener refactoring, part 13

[PATCH net-next 0/6] inet: tcp listener refactoring, part 13

[Eric Dumazet]

inet_hash functions are in a bad state : Too much IPv6/IPv4 copy/pasting.

Lets refactor a bit.

Idea is that we do not want to have an equivalent of inet_csk(sk)->icsk_af_ops
for request socks in order to be able to use the right variant.

In this patch series, I started to let IPv6/IPv4 converge to common helpers.

Idea is to use ipv6_addr_set_v4mapped() even for AF_INET sockets, so that
we can test
       if (sk->sk_family == AF_INET6 &&
           !ipv6_addr_v4mapped(&sk->sk_v6_daddr))
to tell if we deal with an IPv6 socket, or IPv4 one, at least in slow paths.

Ideally, we could save 8 bytes per struct sock_common, if we
alias skc_daddr & skc_rcv_saddr to skc_v6_daddr[3]/skc_v6_rcv_saddr[3].


Eric Dumazet (6):
  netns: constify net_hash_mix() and various callers
  net: introduce sk_ehashfn() helper
  inet: add IPv6 support to sk_ehashfn()
  ipv6: get rid of __inet6_hash()
  inet: get rid of last __inet_hash_connect() argument
  inet: request sock should init IPv6/IPv4 addresses

 include/linux/udp.h            |  2 +-
 include/net/inet6_hashtables.h |  2 --
 include/net/inet_hashtables.h  | 32 ++++++++++++++++++-----
 include/net/netns/hash.h       |  4 +--
 include/net/request_sock.h     |  5 ++++
 net/dccp/ipv4.c                | 13 +++++-----
 net/dccp/ipv6.c                | 27 +++-----------------
 net/ipv4/devinet.c             |  2 +-
 net/ipv4/inet_diag.c           |  4 +--
 net/ipv4/inet_hashtables.c     | 55 ++++++++++++++++++++-------------------
 net/ipv4/ping.c                |  6 ++---
 net/ipv4/syncookies.c          |  4 +--
 net/ipv4/tcp_ipv4.c            | 16 ++++++------
 net/ipv4/udp.c                 | 10 ++++----
 net/ipv6/inet6_hashtables.c    | 58 +++---------------------------------------
 net/ipv6/tcp_ipv6.c            | 28 +++-----------------
 net/ipv6/udp.c                 | 16 ++++++------
 17 files changed, 110 insertions(+), 174 deletions(-)

-- 
2.2.0.rc0.207.ga3a616c

[PATCH net-next 1/6] netns: constify net_hash_mix() and various callers

[Eric Dumazet]

const qualifiers ease code review by making clear
which objects are not written in a function.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 include/linux/udp.h           |  2 +-
 include/net/inet_hashtables.h |  6 +++---
 include/net/netns/hash.h      |  4 ++--
 net/ipv4/devinet.c            |  2 +-
 net/ipv4/inet_hashtables.c    |  6 +++---
 net/ipv4/ping.c               |  6 +++---
 net/ipv4/udp.c                | 10 +++++-----
 net/ipv6/inet6_hashtables.c   | 10 +++++-----
 net/ipv6/udp.c                | 16 ++++++++--------
 9 files changed, 31 insertions(+), 31 deletions(-)

diff --git a/include/linux/udp.h b/include/linux/udp.h
index 247cfdcc4b08..87c094961bd5 100644
--- a/include/linux/udp.h
+++ b/include/linux/udp.h
@@ -34,7 +34,7 @@ static inline struct udphdr *inner_udp_hdr(const struct sk_buff *skb)
 
 #define UDP_HTABLE_SIZE_MIN		(CONFIG_BASE_SMALL ? 128 : 256)
 
-static inline int udp_hashfn(struct net *net, unsigned num, unsigned mask)
+static inline u32 udp_hashfn(const struct net *net, u32 num, u32 mask)
 {
 	return (num + net_hash_mix(net)) & mask;
 }
diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
index bcd64756e5fe..eb1963af0ebd 100644
--- a/include/net/inet_hashtables.h
+++ b/include/net/inet_hashtables.h
@@ -221,8 +221,8 @@ inet_bind_bucket_create(struct kmem_cache *cachep, struct net *net,
 void inet_bind_bucket_destroy(struct kmem_cache *cachep,
 			      struct inet_bind_bucket *tb);
 
-static inline int inet_bhashfn(struct net *net, const __u16 lport,
-			       const int bhash_size)
+static inline u32 inet_bhashfn(const struct net *net, const __u16 lport,
+			       const u32 bhash_size)
 {
 	return (lport + net_hash_mix(net)) & (bhash_size - 1);
 }
@@ -231,7 +231,7 @@ void inet_bind_hash(struct sock *sk, struct inet_bind_bucket *tb,
 		    const unsigned short snum);
 
 /* These can have wildcards, don't try too hard. */
-static inline int inet_lhashfn(struct net *net, const unsigned short num)
+static inline u32 inet_lhashfn(const struct net *net, const unsigned short num)
 {
 	return (num + net_hash_mix(net)) & (INET_LHTABLE_SIZE - 1);
 }
diff --git a/include/net/netns/hash.h b/include/net/netns/hash.h
index c06ac58ca107..69a6715d9f3f 100644
--- a/include/net/netns/hash.h
+++ b/include/net/netns/hash.h
@@ -5,7 +5,7 @@
 
 struct net;
 
-static inline unsigned int net_hash_mix(struct net *net)
+static inline u32 net_hash_mix(const struct net *net)
 {
 #ifdef CONFIG_NET_NS
 	/*
@@ -13,7 +13,7 @@ static inline unsigned int net_hash_mix(struct net *net)
 	 * always zeroed
 	 */
 
-	return (unsigned)(((unsigned long)net) >> L1_CACHE_SHIFT);
+	return (u32)(((unsigned long)net) >> L1_CACHE_SHIFT);
 #else
 	return 0;
 #endif
diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 5105759e4e00..375dc71b9a64 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -107,7 +107,7 @@ static const struct nla_policy ifa_ipv4_policy[IFA_MAX+1] = {
 
 static struct hlist_head inet_addr_lst[IN4_ADDR_HSIZE];
 
-static u32 inet_addr_hash(struct net *net, __be32 addr)
+static u32 inet_addr_hash(const struct net *net, __be32 addr)
 {
 	u32 val = (__force u32) addr ^ net_hash_mix(net);
 
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index c28bca4cc15b..330a08bcd1c1 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -24,9 +24,9 @@
 #include <net/secure_seq.h>
 #include <net/ip.h>
 
-static unsigned int inet_ehashfn(struct net *net, const __be32 laddr,
-				 const __u16 lport, const __be32 faddr,
-				 const __be16 fport)
+static u32 inet_ehashfn(const struct net *net, const __be32 laddr,
+			const __u16 lport, const __be32 faddr,
+			const __be16 fport)
 {
 	static u32 inet_ehash_secret __read_mostly;
 
diff --git a/net/ipv4/ping.c b/net/ipv4/ping.c
index fd88f868776f..344e7cdfb8d4 100644
--- a/net/ipv4/ping.c
+++ b/net/ipv4/ping.c
@@ -64,11 +64,11 @@ EXPORT_SYMBOL_GPL(pingv6_ops);
 
 static u16 ping_port_rover;
 
-static inline int ping_hashfn(struct net *net, unsigned int num, unsigned int mask)
+static inline u32 ping_hashfn(const struct net *net, u32 num, u32 mask)
 {
-	int res = (num + net_hash_mix(net)) & mask;
+	u32 res = (num + net_hash_mix(net)) & mask;
 
-	pr_debug("hash(%d) = %d\n", num, res);
+	pr_debug("hash(%u) = %u\n", num, res);
 	return res;
 }
 EXPORT_SYMBOL_GPL(ping_hash);
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index f27556e2158b..294af16633af 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -318,8 +318,8 @@ static int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
 		   inet1->inet_rcv_saddr == inet2->inet_rcv_saddr));
 }
 
-static unsigned int udp4_portaddr_hash(struct net *net, __be32 saddr,
-				       unsigned int port)
+static u32 udp4_portaddr_hash(const struct net *net, __be32 saddr,
+			      unsigned int port)
 {
 	return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
 }
@@ -421,9 +421,9 @@ static inline int compute_score2(struct sock *sk, struct net *net,
 	return score;
 }
 
-static unsigned int udp_ehashfn(struct net *net, const __be32 laddr,
-				 const __u16 lport, const __be32 faddr,
-				 const __be16 fport)
+static u32 udp_ehashfn(const struct net *net, const __be32 laddr,
+		       const __u16 lport, const __be32 faddr,
+		       const __be16 fport)
 {
 	static u32 udp_ehash_secret __read_mostly;
 
diff --git a/net/ipv6/inet6_hashtables.c b/net/ipv6/inet6_hashtables.c
index 051dffb49c90..df7df99d1d7e 100644
--- a/net/ipv6/inet6_hashtables.c
+++ b/net/ipv6/inet6_hashtables.c
@@ -23,11 +23,11 @@
 #include <net/secure_seq.h>
 #include <net/ip.h>
 
-static unsigned int inet6_ehashfn(struct net *net,
-				  const struct in6_addr *laddr,
-				  const u16 lport,
-				  const struct in6_addr *faddr,
-				  const __be16 fport)
+static u32 inet6_ehashfn(const struct net *net,
+			 const struct in6_addr *laddr,
+			 const u16 lport,
+			 const struct in6_addr *faddr,
+			 const __be16 fport)
 {
 	static u32 inet6_ehash_secret __read_mostly;
 	static u32 ipv6_hash_secret __read_mostly;
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index 70568a4548e4..7fe0329c0d37 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -53,11 +53,11 @@
 #include <trace/events/skb.h>
 #include "udp_impl.h"
 
-static unsigned int udp6_ehashfn(struct net *net,
-				  const struct in6_addr *laddr,
-				  const u16 lport,
-				  const struct in6_addr *faddr,
-				  const __be16 fport)
+static u32 udp6_ehashfn(const struct net *net,
+			const struct in6_addr *laddr,
+			const u16 lport,
+			const struct in6_addr *faddr,
+			const __be16 fport)
 {
 	static u32 udp6_ehash_secret __read_mostly;
 	static u32 udp_ipv6_hash_secret __read_mostly;
@@ -104,9 +104,9 @@ int ipv6_rcv_saddr_equal(const struct sock *sk, const struct sock *sk2)
 	return 0;
 }
 
-static unsigned int udp6_portaddr_hash(struct net *net,
-				       const struct in6_addr *addr6,
-				       unsigned int port)
+static u32 udp6_portaddr_hash(const struct net *net,
+			      const struct in6_addr *addr6,
+			      unsigned int port)
 {
 	unsigned int hash, mix = net_hash_mix(net);
 
-- 
2.2.0.rc0.207.ga3a616c

[PATCH net-next 2/6] net: introduce sk_ehashfn() helper

[Eric Dumazet]

Goal is to unify IPv4/IPv6 inet_hash handling, and use common helpers
for all kind of sockets (full sockets, timewait and request sockets)

inet_sk_ehashfn() becomes sk_ehashfn() but still only copes with IPv4

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 include/net/inet_hashtables.h |  2 ++
 net/ipv4/inet_hashtables.c    | 17 ++++++-----------
 2 files changed, 8 insertions(+), 11 deletions(-)

diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
index eb1963af0ebd..ef993ef571ea 100644
--- a/include/net/inet_hashtables.h
+++ b/include/net/inet_hashtables.h
@@ -383,6 +383,8 @@ static inline struct sock *__inet_lookup_skb(struct inet_hashinfo *hashinfo,
 				     iph->daddr, dport, inet_iif(skb));
 }
 
+u32 sk_ehashfn(const struct sock *sk);
+
 int __inet_hash_connect(struct inet_timewait_death_row *death_row,
 			struct sock *sk, u32 port_offset,
 			int (*check_established)(struct inet_timewait_death_row *,
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index 330a08bcd1c1..3a86dfd7ae33 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -37,16 +37,11 @@ static u32 inet_ehashfn(const struct net *net, const __be32 laddr,
 }
 
 
-static unsigned int inet_sk_ehashfn(const struct sock *sk)
+u32 sk_ehashfn(const struct sock *sk)
 {
-	const struct inet_sock *inet = inet_sk(sk);
-	const __be32 laddr = inet->inet_rcv_saddr;
-	const __u16 lport = inet->inet_num;
-	const __be32 faddr = inet->inet_daddr;
-	const __be16 fport = inet->inet_dport;
-	struct net *net = sock_net(sk);
-
-	return inet_ehashfn(net, laddr, lport, faddr, fport);
+	return inet_ehashfn(sock_net(sk),
+			    sk->sk_rcv_saddr, sk->sk_num,
+			    sk->sk_daddr, sk->sk_dport);
 }
 
 /*
@@ -407,13 +402,13 @@ int __inet_hash_nolisten(struct sock *sk, struct inet_timewait_sock *tw)
 {
 	struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;
 	struct hlist_nulls_head *list;
-	spinlock_t *lock;
 	struct inet_ehash_bucket *head;
+	spinlock_t *lock;
 	int twrefcnt = 0;
 
 	WARN_ON(!sk_unhashed(sk));
 
-	sk->sk_hash = inet_sk_ehashfn(sk);
+	sk->sk_hash = sk_ehashfn(sk);
 	head = inet_ehash_bucket(hashinfo, sk->sk_hash);
 	list = &head->chain;
 	lock = inet_ehash_lockp(hashinfo, sk->sk_hash);
-- 
2.2.0.rc0.207.ga3a616c

[PATCH net-next 3/6] inet: add IPv6 support to sk_ehashfn()

[Eric Dumazet]

Intent is to converge IPv4 & IPv6 inet_hash functions to
factorize code.

IPv4 sockets initialize sk_rcv_saddr and sk_v6_daddr
in this patch, thanks to new sk_daddr_set() and sk_rcv_saddr_set()
helpers.

__inet6_hash can now use sk_ehashfn() instead of a private
inet6_sk_ehashfn() and will simply use __inet_hash() in a
following patch.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 include/net/inet_hashtables.h | 19 +++++++++++++++++++
 net/dccp/ipv4.c               |  9 ++++-----
 net/dccp/ipv6.c               | 10 ++--------
 net/ipv4/inet_hashtables.c    | 11 ++++++++++-
 net/ipv4/tcp_ipv4.c           |  8 ++++----
 net/ipv6/inet6_hashtables.c   | 22 ++++------------------
 net/ipv6/tcp_ipv6.c           | 11 ++---------
 7 files changed, 45 insertions(+), 45 deletions(-)

diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
index ef993ef571ea..06ad42182ec2 100644
--- a/include/net/inet_hashtables.h
+++ b/include/net/inet_hashtables.h
@@ -384,6 +384,25 @@ static inline struct sock *__inet_lookup_skb(struct inet_hashinfo *hashinfo,
 }
 
 u32 sk_ehashfn(const struct sock *sk);
+u32 inet6_ehashfn(const struct net *net,
+		  const struct in6_addr *laddr, const u16 lport,
+		  const struct in6_addr *faddr, const __be16 fport);
+
+static inline void sk_daddr_set(struct sock *sk, __be32 addr)
+{
+	sk->sk_daddr = addr; /* alias of inet_daddr */
+#if IS_ENABLED(CONFIG_IPV6)
+	ipv6_addr_set_v4mapped(addr, &sk->sk_v6_daddr);
+#endif
+}
+
+static inline void sk_rcv_saddr_set(struct sock *sk, __be32 addr)
+{
+	sk->sk_rcv_saddr = addr; /* alias of inet_rcv_saddr */
+#if IS_ENABLED(CONFIG_IPV6)
+	ipv6_addr_set_v4mapped(addr, &sk->sk_v6_rcv_saddr);
+#endif
+}
 
 int __inet_hash_connect(struct inet_timewait_death_row *death_row,
 			struct sock *sk, u32 port_offset,
diff --git a/net/dccp/ipv4.c b/net/dccp/ipv4.c
index bf897829f4f0..f3f8906f482e 100644
--- a/net/dccp/ipv4.c
+++ b/net/dccp/ipv4.c
@@ -89,10 +89,9 @@ int dccp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 
 	if (inet->inet_saddr == 0)
 		inet->inet_saddr = fl4->saddr;
-	inet->inet_rcv_saddr = inet->inet_saddr;
-
+	sk_rcv_saddr_set(sk, inet->inet_saddr);
 	inet->inet_dport = usin->sin_port;
-	inet->inet_daddr = daddr;
+	sk_daddr_set(sk, daddr);
 
 	inet_csk(sk)->icsk_ext_hdr_len = 0;
 	if (inet_opt)
@@ -408,8 +407,8 @@ struct sock *dccp_v4_request_recv_sock(struct sock *sk, struct sk_buff *skb,
 
 	newinet		   = inet_sk(newsk);
 	ireq		   = inet_rsk(req);
-	newinet->inet_daddr	= ireq->ir_rmt_addr;
-	newinet->inet_rcv_saddr = ireq->ir_loc_addr;
+	sk_daddr_set(newsk, ireq->ir_rmt_addr);
+	sk_rcv_saddr_set(newsk, ireq->ir_loc_addr);
 	newinet->inet_saddr	= ireq->ir_loc_addr;
 	newinet->inet_opt	= ireq->opt;
 	ireq->opt	   = NULL;
diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c
index d7e7c7b0a3f1..9216d173dd5f 100644
--- a/net/dccp/ipv6.c
+++ b/net/dccp/ipv6.c
@@ -470,11 +470,7 @@ static struct sock *dccp_v6_request_recv_sock(struct sock *sk,
 
 		memcpy(newnp, np, sizeof(struct ipv6_pinfo));
 
-		ipv6_addr_set_v4mapped(newinet->inet_daddr, &newsk->sk_v6_daddr);
-
-		ipv6_addr_set_v4mapped(newinet->inet_saddr, &newnp->saddr);
-
-		newsk->sk_v6_rcv_saddr = newnp->saddr;
+		newnp->saddr = newsk->sk_v6_rcv_saddr;
 
 		inet_csk(newsk)->icsk_af_ops = &dccp_ipv6_mapped;
 		newsk->sk_backlog_rcv = dccp_v4_do_rcv;
@@ -917,9 +913,7 @@ static int dccp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
 			sk->sk_backlog_rcv = dccp_v6_do_rcv;
 			goto failure;
 		}
-		ipv6_addr_set_v4mapped(inet->inet_saddr, &np->saddr);
-		ipv6_addr_set_v4mapped(inet->inet_rcv_saddr, &sk->sk_v6_rcv_saddr);
-
+		np->saddr = sk->sk_v6_rcv_saddr;
 		return err;
 	}
 
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index 3a86dfd7ae33..ab7f677a97db 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -36,9 +36,18 @@ static u32 inet_ehashfn(const struct net *net, const __be32 laddr,
 			      inet_ehash_secret + net_hash_mix(net));
 }
 
-
+/* This function handles inet_sock, but also timewait and request sockets
+ * for IPv4/IPv6.
+ */
 u32 sk_ehashfn(const struct sock *sk)
 {
+#if IS_ENABLED(CONFIG_IPV6)
+	if (sk->sk_family == AF_INET6 &&
+	    !ipv6_addr_v4mapped(&sk->sk_v6_daddr))
+		return inet6_ehashfn(sock_net(sk),
+				     &sk->sk_v6_rcv_saddr, sk->sk_num,
+				     &sk->sk_v6_daddr, sk->sk_dport);
+#endif
 	return inet_ehashfn(sock_net(sk),
 			    sk->sk_rcv_saddr, sk->sk_num,
 			    sk->sk_daddr, sk->sk_dport);
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 80067d5858b4..ca207df4af1c 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -189,7 +189,7 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 
 	if (!inet->inet_saddr)
 		inet->inet_saddr = fl4->saddr;
-	inet->inet_rcv_saddr = inet->inet_saddr;
+	sk_rcv_saddr_set(sk, inet->inet_saddr);
 
 	if (tp->rx_opt.ts_recent_stamp && inet->inet_daddr != daddr) {
 		/* Reset inherited state */
@@ -204,7 +204,7 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 		tcp_fetch_timewait_stamp(sk, &rt->dst);
 
 	inet->inet_dport = usin->sin_port;
-	inet->inet_daddr = daddr;
+	sk_daddr_set(sk, daddr);
 
 	inet_csk(sk)->icsk_ext_hdr_len = 0;
 	if (inet_opt)
@@ -1319,8 +1319,8 @@ struct sock *tcp_v4_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
 	newtp		      = tcp_sk(newsk);
 	newinet		      = inet_sk(newsk);
 	ireq		      = inet_rsk(req);
-	newinet->inet_daddr   = ireq->ir_rmt_addr;
-	newinet->inet_rcv_saddr = ireq->ir_loc_addr;
+	sk_daddr_set(newsk, ireq->ir_rmt_addr);
+	sk_rcv_saddr_set(newsk, ireq->ir_loc_addr);
 	newinet->inet_saddr	      = ireq->ir_loc_addr;
 	inet_opt	      = ireq->opt;
 	rcu_assign_pointer(newinet->inet_opt, inet_opt);
diff --git a/net/ipv6/inet6_hashtables.c b/net/ipv6/inet6_hashtables.c
index df7df99d1d7e..ed5787b20192 100644
--- a/net/ipv6/inet6_hashtables.c
+++ b/net/ipv6/inet6_hashtables.c
@@ -23,11 +23,9 @@
 #include <net/secure_seq.h>
 #include <net/ip.h>
 
-static u32 inet6_ehashfn(const struct net *net,
-			 const struct in6_addr *laddr,
-			 const u16 lport,
-			 const struct in6_addr *faddr,
-			 const __be16 fport)
+u32 inet6_ehashfn(const struct net *net,
+		  const struct in6_addr *laddr, const u16 lport,
+		  const struct in6_addr *faddr, const __be16 fport)
 {
 	static u32 inet6_ehash_secret __read_mostly;
 	static u32 ipv6_hash_secret __read_mostly;
@@ -44,18 +42,6 @@ static u32 inet6_ehashfn(const struct net *net,
 			       inet6_ehash_secret + net_hash_mix(net));
 }
 
-static int inet6_sk_ehashfn(const struct sock *sk)
-{
-	const struct inet_sock *inet = inet_sk(sk);
-	const struct in6_addr *laddr = &sk->sk_v6_rcv_saddr;
-	const struct in6_addr *faddr = &sk->sk_v6_daddr;
-	const __u16 lport = inet->inet_num;
-	const __be16 fport = inet->inet_dport;
-	struct net *net = sock_net(sk);
-
-	return inet6_ehashfn(net, laddr, lport, faddr, fport);
-}
-
 int __inet6_hash(struct sock *sk, struct inet_timewait_sock *tw)
 {
 	struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;
@@ -75,7 +61,7 @@ int __inet6_hash(struct sock *sk, struct inet_timewait_sock *tw)
 		struct hlist_nulls_head *list;
 		spinlock_t *lock;
 
-		sk->sk_hash = hash = inet6_sk_ehashfn(sk);
+		sk->sk_hash = hash = sk_ehashfn(sk);
 		list = &inet_ehash_bucket(hashinfo, hash)->chain;
 		lock = inet_ehash_lockp(hashinfo, hash);
 		spin_lock(lock);
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index e4761b22307b..5546df074583 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -233,11 +233,8 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
 			tp->af_specific = &tcp_sock_ipv6_specific;
 #endif
 			goto failure;
-		} else {
-			ipv6_addr_set_v4mapped(inet->inet_saddr, &np->saddr);
-			ipv6_addr_set_v4mapped(inet->inet_rcv_saddr,
-					       &sk->sk_v6_rcv_saddr);
 		}
+		np->saddr = sk->sk_v6_rcv_saddr;
 
 		return err;
 	}
@@ -1078,11 +1075,7 @@ static struct sock *tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
 
 		memcpy(newnp, np, sizeof(struct ipv6_pinfo));
 
-		ipv6_addr_set_v4mapped(newinet->inet_daddr, &newsk->sk_v6_daddr);
-
-		ipv6_addr_set_v4mapped(newinet->inet_saddr, &newnp->saddr);
-
-		newsk->sk_v6_rcv_saddr = newnp->saddr;
+		newnp->saddr = newsk->sk_v6_rcv_saddr;
 
 		inet_csk(newsk)->icsk_af_ops = &ipv6_mapped;
 		newsk->sk_backlog_rcv = tcp_v4_do_rcv;
-- 
2.2.0.rc0.207.ga3a616c

[PATCH net-next 4/6] ipv6: get rid of __inet6_hash()

[Eric Dumazet]

We can now use inet_hash() and __inet_hash() instead of private
functions.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 include/net/inet6_hashtables.h |  2 --
 include/net/inet_hashtables.h  |  1 +
 net/dccp/ipv6.c                | 17 ++---------------
 net/ipv4/inet_hashtables.c     | 12 ++++++------
 net/ipv6/inet6_hashtables.c    | 38 +-------------------------------------
 net/ipv6/tcp_ipv6.c            | 17 ++---------------
 6 files changed, 12 insertions(+), 75 deletions(-)

diff --git a/include/net/inet6_hashtables.h b/include/net/inet6_hashtables.h
index 9201afe083fa..7ff588ca6817 100644
--- a/include/net/inet6_hashtables.h
+++ b/include/net/inet6_hashtables.h
@@ -38,8 +38,6 @@ static inline unsigned int __inet6_ehashfn(const u32 lhash,
 	return jhash_3words(lhash, fhash, ports, initval);
 }
 
-int __inet6_hash(struct sock *sk, struct inet_timewait_sock *twp);
-
 /*
  * Sockets in TCP_CLOSE state are _always_ taken out of the hash, so
  * we need not check it for TCP lookups anymore, thanks Alexey. -DaveM
diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
index 06ad42182ec2..eee6c3399990 100644
--- a/include/net/inet_hashtables.h
+++ b/include/net/inet_hashtables.h
@@ -249,6 +249,7 @@ void inet_put_port(struct sock *sk);
 void inet_hashinfo_init(struct inet_hashinfo *h);
 
 int __inet_hash_nolisten(struct sock *sk, struct inet_timewait_sock *tw);
+int __inet_hash(struct sock *sk, struct inet_timewait_sock *tw);
 void inet_hash(struct sock *sk);
 void inet_unhash(struct sock *sk);
 
diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c
index 9216d173dd5f..c655de5f67c9 100644
--- a/net/dccp/ipv6.c
+++ b/net/dccp/ipv6.c
@@ -40,19 +40,6 @@
 static const struct inet_connection_sock_af_ops dccp_ipv6_mapped;
 static const struct inet_connection_sock_af_ops dccp_ipv6_af_ops;
 
-static void dccp_v6_hash(struct sock *sk)
-{
-	if (sk->sk_state != DCCP_CLOSED) {
-		if (inet_csk(sk)->icsk_af_ops == &dccp_ipv6_mapped) {
-			inet_hash(sk);
-			return;
-		}
-		local_bh_disable();
-		__inet6_hash(sk, NULL);
-		local_bh_enable();
-	}
-}
-
 /* add pseudo-header to DCCP checksum stored in skb->csum */
 static inline __sum16 dccp_v6_csum_finish(struct sk_buff *skb,
 				      const struct in6_addr *saddr,
@@ -588,7 +575,7 @@ static struct sock *dccp_v6_request_recv_sock(struct sock *sk,
 		dccp_done(newsk);
 		goto out;
 	}
-	__inet6_hash(newsk, NULL);
+	__inet_hash(newsk, NULL);
 
 	return newsk;
 
@@ -1056,7 +1043,7 @@ static struct proto dccp_v6_prot = {
 	.sendmsg	   = dccp_sendmsg,
 	.recvmsg	   = dccp_recvmsg,
 	.backlog_rcv	   = dccp_v6_do_rcv,
-	.hash		   = dccp_v6_hash,
+	.hash		   = inet_hash,
 	.unhash		   = inet_unhash,
 	.accept		   = inet_csk_accept,
 	.get_port	   = inet_csk_get_port,
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index ab7f677a97db..82753bd57e79 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -434,15 +434,13 @@ int __inet_hash_nolisten(struct sock *sk, struct inet_timewait_sock *tw)
 }
 EXPORT_SYMBOL_GPL(__inet_hash_nolisten);
 
-static void __inet_hash(struct sock *sk)
+int __inet_hash(struct sock *sk, struct inet_timewait_sock *tw)
 {
 	struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;
 	struct inet_listen_hashbucket *ilb;
 
-	if (sk->sk_state != TCP_LISTEN) {
-		__inet_hash_nolisten(sk, NULL);
-		return;
-	}
+	if (sk->sk_state != TCP_LISTEN)
+		return __inet_hash_nolisten(sk, tw);
 
 	WARN_ON(!sk_unhashed(sk));
 	ilb = &hashinfo->listening_hash[inet_sk_listen_hashfn(sk)];
@@ -451,13 +449,15 @@ static void __inet_hash(struct sock *sk)
 	__sk_nulls_add_node_rcu(sk, &ilb->head);
 	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
 	spin_unlock(&ilb->lock);
+	return 0;
 }
+EXPORT_SYMBOL(__inet_hash);
 
 void inet_hash(struct sock *sk)
 {
 	if (sk->sk_state != TCP_CLOSE) {
 		local_bh_disable();
-		__inet_hash(sk);
+		__inet_hash(sk, NULL);
 		local_bh_enable();
 	}
 }
diff --git a/net/ipv6/inet6_hashtables.c b/net/ipv6/inet6_hashtables.c
index ed5787b20192..b86b429f5f81 100644
--- a/net/ipv6/inet6_hashtables.c
+++ b/net/ipv6/inet6_hashtables.c
@@ -42,42 +42,6 @@ u32 inet6_ehashfn(const struct net *net,
 			       inet6_ehash_secret + net_hash_mix(net));
 }
 
-int __inet6_hash(struct sock *sk, struct inet_timewait_sock *tw)
-{
-	struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;
-	int twrefcnt = 0;
-
-	WARN_ON(!sk_unhashed(sk));
-
-	if (sk->sk_state == TCP_LISTEN) {
-		struct inet_listen_hashbucket *ilb;
-
-		ilb = &hashinfo->listening_hash[inet_sk_listen_hashfn(sk)];
-		spin_lock(&ilb->lock);
-		__sk_nulls_add_node_rcu(sk, &ilb->head);
-		spin_unlock(&ilb->lock);
-	} else {
-		unsigned int hash;
-		struct hlist_nulls_head *list;
-		spinlock_t *lock;
-
-		sk->sk_hash = hash = sk_ehashfn(sk);
-		list = &inet_ehash_bucket(hashinfo, hash)->chain;
-		lock = inet_ehash_lockp(hashinfo, hash);
-		spin_lock(lock);
-		__sk_nulls_add_node_rcu(sk, list);
-		if (tw) {
-			WARN_ON(sk->sk_hash != tw->tw_hash);
-			twrefcnt = inet_twsk_unhash(tw);
-		}
-		spin_unlock(lock);
-	}
-
-	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
-	return twrefcnt;
-}
-EXPORT_SYMBOL(__inet6_hash);
-
 /*
  * Sockets in TCP_CLOSE state are _always_ taken out of the hash, so
  * we need not check it for TCP lookups anymore, thanks Alexey. -DaveM
@@ -306,6 +270,6 @@ int inet6_hash_connect(struct inet_timewait_death_row *death_row,
 		       struct sock *sk)
 {
 	return __inet_hash_connect(death_row, sk, inet6_sk_port_offset(sk),
-			__inet6_check_established, __inet6_hash);
+			__inet6_check_established, __inet_hash_nolisten);
 }
 EXPORT_SYMBOL_GPL(inet6_hash_connect);
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 5546df074583..720676d073d9 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -104,19 +104,6 @@ static void inet6_sk_rx_dst_set(struct sock *sk, const struct sk_buff *skb)
 	}
 }
 
-static void tcp_v6_hash(struct sock *sk)
-{
-	if (sk->sk_state != TCP_CLOSE) {
-		if (inet_csk(sk)->icsk_af_ops == &ipv6_mapped) {
-			tcp_prot.hash(sk);
-			return;
-		}
-		local_bh_disable();
-		__inet6_hash(sk, NULL);
-		local_bh_enable();
-	}
-}
-
 static __u32 tcp_v6_init_sequence(const struct sk_buff *skb)
 {
 	return secure_tcpv6_sequence_number(ipv6_hdr(skb)->daddr.s6_addr32,
@@ -1224,7 +1211,7 @@ static struct sock *tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
 		tcp_done(newsk);
 		goto out;
 	}
-	__inet6_hash(newsk, NULL);
+	__inet_hash(newsk, NULL);
 
 	return newsk;
 
@@ -1883,7 +1870,7 @@ struct proto tcpv6_prot = {
 	.sendpage		= tcp_sendpage,
 	.backlog_rcv		= tcp_v6_do_rcv,
 	.release_cb		= tcp_release_cb,
-	.hash			= tcp_v6_hash,
+	.hash			= inet_hash,
 	.unhash			= inet_unhash,
 	.get_port		= inet_csk_get_port,
 	.enter_memory_pressure	= tcp_enter_memory_pressure,
-- 
2.2.0.rc0.207.ga3a616c

[PATCH net-next 5/6] inet: get rid of last __inet_hash_connect() argument

[Eric Dumazet]

We now always call __inet_hash_nolisten(), no need to pass it
as an argument.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 include/net/inet_hashtables.h | 4 +---
 net/ipv4/inet_hashtables.c    | 9 ++++-----
 net/ipv6/inet6_hashtables.c   | 2 +-
 3 files changed, 6 insertions(+), 9 deletions(-)

diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
index eee6c3399990..73fe0f9525d9 100644
--- a/include/net/inet_hashtables.h
+++ b/include/net/inet_hashtables.h
@@ -409,9 +409,7 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row,
 			struct sock *sk, u32 port_offset,
 			int (*check_established)(struct inet_timewait_death_row *,
 						 struct sock *, __u16,
-						 struct inet_timewait_sock **),
-			int (*hash)(struct sock *sk,
-				    struct inet_timewait_sock *twp));
+						 struct inet_timewait_sock **));
 
 int inet_hash_connect(struct inet_timewait_death_row *death_row,
 		      struct sock *sk);
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index 82753bd57e79..0fb841b9d834 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -488,8 +488,7 @@ EXPORT_SYMBOL_GPL(inet_unhash);
 int __inet_hash_connect(struct inet_timewait_death_row *death_row,
 		struct sock *sk, u32 port_offset,
 		int (*check_established)(struct inet_timewait_death_row *,
-			struct sock *, __u16, struct inet_timewait_sock **),
-		int (*hash)(struct sock *sk, struct inet_timewait_sock *twp))
+			struct sock *, __u16, struct inet_timewait_sock **))
 {
 	struct inet_hashinfo *hinfo = death_row->hashinfo;
 	const unsigned short snum = inet_sk(sk)->inet_num;
@@ -559,7 +558,7 @@ ok:
 		inet_bind_hash(sk, tb, port);
 		if (sk_unhashed(sk)) {
 			inet_sk(sk)->inet_sport = htons(port);
-			twrefcnt += hash(sk, tw);
+			twrefcnt += __inet_hash_nolisten(sk, tw);
 		}
 		if (tw)
 			twrefcnt += inet_twsk_bind_unhash(tw, hinfo);
@@ -581,7 +580,7 @@ ok:
 	tb  = inet_csk(sk)->icsk_bind_hash;
 	spin_lock_bh(&head->lock);
 	if (sk_head(&tb->owners) == sk && !sk->sk_bind_node.next) {
-		hash(sk, NULL);
+		__inet_hash_nolisten(sk, NULL);
 		spin_unlock_bh(&head->lock);
 		return 0;
 	} else {
@@ -601,7 +600,7 @@ int inet_hash_connect(struct inet_timewait_death_row *death_row,
 		      struct sock *sk)
 {
 	return __inet_hash_connect(death_row, sk, inet_sk_port_offset(sk),
-			__inet_check_established, __inet_hash_nolisten);
+				   __inet_check_established);
 }
 EXPORT_SYMBOL_GPL(inet_hash_connect);
 
diff --git a/net/ipv6/inet6_hashtables.c b/net/ipv6/inet6_hashtables.c
index b86b429f5f81..033f17816ef4 100644
--- a/net/ipv6/inet6_hashtables.c
+++ b/net/ipv6/inet6_hashtables.c
@@ -270,6 +270,6 @@ int inet6_hash_connect(struct inet_timewait_death_row *death_row,
 		       struct sock *sk)
 {
 	return __inet_hash_connect(death_row, sk, inet6_sk_port_offset(sk),
-			__inet6_check_established, __inet_hash_nolisten);
+				   __inet6_check_established);
 }
 EXPORT_SYMBOL_GPL(inet6_hash_connect);
-- 
2.2.0.rc0.207.ga3a616c

[PATCH net-next 6/6] inet: request sock should init IPv6/IPv4 addresses

[Eric Dumazet]

In order to be able to use sk_ehashfn() for request socks,
we need to initialize their IPv6/IPv4 addresses.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 include/net/request_sock.h | 5 +++++
 net/dccp/ipv4.c            | 4 ++--
 net/ipv4/inet_diag.c       | 4 ++--
 net/ipv4/syncookies.c      | 4 ++--
 net/ipv4/tcp_ipv4.c        | 8 ++++----
 5 files changed, 15 insertions(+), 10 deletions(-)

diff --git a/include/net/request_sock.h b/include/net/request_sock.h
index 3fa4f824900a..e7ef86340514 100644
--- a/include/net/request_sock.h
+++ b/include/net/request_sock.h
@@ -91,6 +91,11 @@ static inline struct request_sock *inet_reqsk(struct sock *sk)
 	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 */
diff --git a/net/dccp/ipv4.c b/net/dccp/ipv4.c
index f3f8906f482e..e7ad291cd96b 100644
--- a/net/dccp/ipv4.c
+++ b/net/dccp/ipv4.c
@@ -638,8 +638,8 @@ int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 		goto drop_and_free;
 
 	ireq = inet_rsk(req);
-	ireq->ir_loc_addr = ip_hdr(skb)->daddr;
-	ireq->ir_rmt_addr = ip_hdr(skb)->saddr;
+	sk_rcv_saddr_set(req_to_sk(req), ip_hdr(skb)->daddr);
+	sk_daddr_set(req_to_sk(req), ip_hdr(skb)->saddr);
 	ireq->ireq_family = AF_INET;
 	ireq->ir_iif = sk->sk_bound_dev_if;
 
diff --git a/net/ipv4/inet_diag.c b/net/ipv4/inet_diag.c
index e7ba59038c8d..74c39c9f3e11 100644
--- a/net/ipv4/inet_diag.c
+++ b/net/ipv4/inet_diag.c
@@ -742,14 +742,14 @@ static int inet_diag_dump_reqs(struct sk_buff *skb, struct sock *sk,
 
 			if (bc) {
 				/* Note: entry.sport and entry.userlocks are already set */
-				entry_fill_addrs(&entry, (struct sock *)req);
+				entry_fill_addrs(&entry, req_to_sk(req));
 				entry.dport = ntohs(ireq->ir_rmt_port);
 
 				if (!inet_diag_bc_run(bc, &entry))
 					continue;
 			}
 
-			err = inet_req_diag_fill((struct sock *)req, skb,
+			err = inet_req_diag_fill(req_to_sk(req), skb,
 						 NETLINK_CB(cb->skb).portid,
 						 cb->nlh->nlmsg_seq,
 						 NLM_F_MULTI, cb->nlh);
diff --git a/net/ipv4/syncookies.c b/net/ipv4/syncookies.c
index 34e755403715..ef01d8570358 100644
--- a/net/ipv4/syncookies.c
+++ b/net/ipv4/syncookies.c
@@ -337,8 +337,8 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb)
 	req->mss		= mss;
 	ireq->ir_num		= ntohs(th->dest);
 	ireq->ir_rmt_port	= th->source;
-	ireq->ir_loc_addr	= ip_hdr(skb)->daddr;
-	ireq->ir_rmt_addr	= ip_hdr(skb)->saddr;
+	sk_rcv_saddr_set(req_to_sk(req), ip_hdr(skb)->daddr);
+	sk_daddr_set(req_to_sk(req), ip_hdr(skb)->saddr);
 	ireq->ir_mark		= inet_request_mark(sk, skb);
 	ireq->snd_wscale	= tcp_opt.snd_wscale;
 	ireq->sack_ok		= tcp_opt.sack_ok;
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index ca207df4af1c..ddd0b1f25b96 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -1219,14 +1219,14 @@ static bool tcp_v4_inbound_md5_hash(struct sock *sk, const struct sk_buff *skb)
 
 #endif
 
-static void tcp_v4_init_req(struct request_sock *req, struct sock *sk,
+static void tcp_v4_init_req(struct request_sock *req, struct sock *sk_listener,
 			    struct sk_buff *skb)
 {
 	struct inet_request_sock *ireq = inet_rsk(req);
 
-	ireq->ir_loc_addr = ip_hdr(skb)->daddr;
-	ireq->ir_rmt_addr = ip_hdr(skb)->saddr;
-	ireq->no_srccheck = inet_sk(sk)->transparent;
+	sk_rcv_saddr_set(req_to_sk(req), ip_hdr(skb)->daddr);
+	sk_daddr_set(req_to_sk(req), ip_hdr(skb)->saddr);
+	ireq->no_srccheck = inet_sk(sk_listener)->transparent;
 	ireq->opt = tcp_v4_save_options(skb);
 	ireq->ireq_family = AF_INET;
 }
-- 
2.2.0.rc0.207.ga3a616c

Re: [PATCH net-next 0/6] inet: tcp listener refactoring, part 13

[David Miller]

From: Eric Dumazet <edumazet@google.com>
Date: Wed, 18 Mar 2015 14:05:32 -0700

> inet_hash functions are in a bad state : Too much IPv6/IPv4 copy/pasting.
> 
> Lets refactor a bit.
> 
> Idea is that we do not want to have an equivalent of inet_csk(sk)->icsk_af_ops
> for request socks in order to be able to use the right variant.
> 
> In this patch series, I started to let IPv6/IPv4 converge to common helpers.
> 
> Idea is to use ipv6_addr_set_v4mapped() even for AF_INET sockets, so that
> we can test
>        if (sk->sk_family == AF_INET6 &&
>            !ipv6_addr_v4mapped(&sk->sk_v6_daddr))
> to tell if we deal with an IPv6 socket, or IPv4 one, at least in slow paths.
> 
> Ideally, we could save 8 bytes per struct sock_common, if we
> alias skc_daddr & skc_rcv_saddr to skc_v6_daddr[3]/skc_v6_rcv_saddr[3].

Looks good, applied, thanks Eric.