aboutsummaryrefslogtreecommitdiffstats
path: root/net/ipv6
diff options
context:
space:
mode:
authorNicholas Bellinger <nab@linux-iscsi.org>2011-01-16 16:21:04 -0500
committerNicholas Bellinger <nab@linux-iscsi.org>2011-01-16 16:21:04 -0500
commitf652f6c5b7cfdf139f4155d78f397e99ae1c4acc (patch)
tree71c6344688bf56ea6aaf18c586ab69ff4f077ade /net/ipv6
parent140e3008e7fe1526cbb12f8f07dbc273ac713b75 (diff)
parentc66ac9db8d4ad9994a02b3e933ea2ccc643e1fe5 (diff)
Merge branch 'master' of /pub/scm/linux/kernel/git/jejb/scsi-post-merge-2.6 into for-linus
Diffstat (limited to 'net/ipv6')
-rw-r--r--net/ipv6/addrconf.c119
-rw-r--r--net/ipv6/af_inet6.c4
-rw-r--r--net/ipv6/ah6.c8
-rw-r--r--net/ipv6/esp6.c32
-rw-r--r--net/ipv6/inet6_connection_sock.c56
-rw-r--r--net/ipv6/ip6_output.c12
-rw-r--r--net/ipv6/ip6_tunnel.c2
-rw-r--r--net/ipv6/ip6mr.c4
-rw-r--r--net/ipv6/mcast.c77
-rw-r--r--net/ipv6/ndisc.c29
-rw-r--r--net/ipv6/netfilter.c6
-rw-r--r--net/ipv6/netfilter/Makefile4
-rw-r--r--net/ipv6/netfilter/ip6_tables.c45
-rw-r--r--net/ipv6/netfilter/ip6t_REJECT.c2
-rw-r--r--net/ipv6/reassembly.c36
-rw-r--r--net/ipv6/route.c163
-rw-r--r--net/ipv6/sit.c14
-rw-r--r--net/ipv6/tcp_ipv6.c151
-rw-r--r--net/ipv6/udp.c11
-rw-r--r--net/ipv6/udplite.c1
-rw-r--r--net/ipv6/xfrm6_mode_tunnel.c3
-rw-r--r--net/ipv6/xfrm6_output.c16
22 files changed, 487 insertions, 308 deletions
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 93b7a933a775..5b189c97c2fc 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -2669,8 +2669,9 @@ static int addrconf_ifdown(struct net_device *dev, int how)
2669 2669
2670 ASSERT_RTNL(); 2670 ASSERT_RTNL();
2671 2671
2672 rt6_ifdown(net, dev); 2672 /* Flush routes if device is being removed or it is not loopback */
2673 neigh_ifdown(&nd_tbl, dev); 2673 if (how || !(dev->flags & IFF_LOOPBACK))
2674 rt6_ifdown(net, dev);
2674 2675
2675 idev = __in6_dev_get(dev); 2676 idev = __in6_dev_get(dev);
2676 if (idev == NULL) 2677 if (idev == NULL)
@@ -3836,6 +3837,15 @@ static inline void ipv6_store_devconf(struct ipv6_devconf *cnf,
3836 array[DEVCONF_FORCE_TLLAO] = cnf->force_tllao; 3837 array[DEVCONF_FORCE_TLLAO] = cnf->force_tllao;
3837} 3838}
3838 3839
3840static inline size_t inet6_ifla6_size(void)
3841{
3842 return nla_total_size(4) /* IFLA_INET6_FLAGS */
3843 + nla_total_size(sizeof(struct ifla_cacheinfo))
3844 + nla_total_size(DEVCONF_MAX * 4) /* IFLA_INET6_CONF */
3845 + nla_total_size(IPSTATS_MIB_MAX * 8) /* IFLA_INET6_STATS */
3846 + nla_total_size(ICMP6_MIB_MAX * 8); /* IFLA_INET6_ICMP6STATS */
3847}
3848
3839static inline size_t inet6_if_nlmsg_size(void) 3849static inline size_t inet6_if_nlmsg_size(void)
3840{ 3850{
3841 return NLMSG_ALIGN(sizeof(struct ifinfomsg)) 3851 return NLMSG_ALIGN(sizeof(struct ifinfomsg))
@@ -3843,13 +3853,7 @@ static inline size_t inet6_if_nlmsg_size(void)
3843 + nla_total_size(MAX_ADDR_LEN) /* IFLA_ADDRESS */ 3853 + nla_total_size(MAX_ADDR_LEN) /* IFLA_ADDRESS */
3844 + nla_total_size(4) /* IFLA_MTU */ 3854 + nla_total_size(4) /* IFLA_MTU */
3845 + nla_total_size(4) /* IFLA_LINK */ 3855 + nla_total_size(4) /* IFLA_LINK */
3846 + nla_total_size( /* IFLA_PROTINFO */ 3856 + nla_total_size(inet6_ifla6_size()); /* IFLA_PROTINFO */
3847 nla_total_size(4) /* IFLA_INET6_FLAGS */
3848 + nla_total_size(sizeof(struct ifla_cacheinfo))
3849 + nla_total_size(DEVCONF_MAX * 4) /* IFLA_INET6_CONF */
3850 + nla_total_size(IPSTATS_MIB_MAX * 8) /* IFLA_INET6_STATS */
3851 + nla_total_size(ICMP6_MIB_MAX * 8) /* IFLA_INET6_ICMP6STATS */
3852 );
3853} 3857}
3854 3858
3855static inline void __snmp6_fill_stats(u64 *stats, void __percpu **mib, 3859static inline void __snmp6_fill_stats(u64 *stats, void __percpu **mib,
@@ -3896,15 +3900,70 @@ static void snmp6_fill_stats(u64 *stats, struct inet6_dev *idev, int attrtype,
3896 } 3900 }
3897} 3901}
3898 3902
3903static int inet6_fill_ifla6_attrs(struct sk_buff *skb, struct inet6_dev *idev)
3904{
3905 struct nlattr *nla;
3906 struct ifla_cacheinfo ci;
3907
3908 NLA_PUT_U32(skb, IFLA_INET6_FLAGS, idev->if_flags);
3909
3910 ci.max_reasm_len = IPV6_MAXPLEN;
3911 ci.tstamp = cstamp_delta(idev->tstamp);
3912 ci.reachable_time = jiffies_to_msecs(idev->nd_parms->reachable_time);
3913 ci.retrans_time = jiffies_to_msecs(idev->nd_parms->retrans_time);
3914 NLA_PUT(skb, IFLA_INET6_CACHEINFO, sizeof(ci), &ci);
3915
3916 nla = nla_reserve(skb, IFLA_INET6_CONF, DEVCONF_MAX * sizeof(s32));
3917 if (nla == NULL)
3918 goto nla_put_failure;
3919 ipv6_store_devconf(&idev->cnf, nla_data(nla), nla_len(nla));
3920
3921 /* XXX - MC not implemented */
3922
3923 nla = nla_reserve(skb, IFLA_INET6_STATS, IPSTATS_MIB_MAX * sizeof(u64));
3924 if (nla == NULL)
3925 goto nla_put_failure;
3926 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_STATS, nla_len(nla));
3927
3928 nla = nla_reserve(skb, IFLA_INET6_ICMP6STATS, ICMP6_MIB_MAX * sizeof(u64));
3929 if (nla == NULL)
3930 goto nla_put_failure;
3931 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_ICMP6STATS, nla_len(nla));
3932
3933 return 0;
3934
3935nla_put_failure:
3936 return -EMSGSIZE;
3937}
3938
3939static size_t inet6_get_link_af_size(const struct net_device *dev)
3940{
3941 if (!__in6_dev_get(dev))
3942 return 0;
3943
3944 return inet6_ifla6_size();
3945}
3946
3947static int inet6_fill_link_af(struct sk_buff *skb, const struct net_device *dev)
3948{
3949 struct inet6_dev *idev = __in6_dev_get(dev);
3950
3951 if (!idev)
3952 return -ENODATA;
3953
3954 if (inet6_fill_ifla6_attrs(skb, idev) < 0)
3955 return -EMSGSIZE;
3956
3957 return 0;
3958}
3959
3899static int inet6_fill_ifinfo(struct sk_buff *skb, struct inet6_dev *idev, 3960static int inet6_fill_ifinfo(struct sk_buff *skb, struct inet6_dev *idev,
3900 u32 pid, u32 seq, int event, unsigned int flags) 3961 u32 pid, u32 seq, int event, unsigned int flags)
3901{ 3962{
3902 struct net_device *dev = idev->dev; 3963 struct net_device *dev = idev->dev;
3903 struct nlattr *nla;
3904 struct ifinfomsg *hdr; 3964 struct ifinfomsg *hdr;
3905 struct nlmsghdr *nlh; 3965 struct nlmsghdr *nlh;
3906 void *protoinfo; 3966 void *protoinfo;
3907 struct ifla_cacheinfo ci;
3908 3967
3909 nlh = nlmsg_put(skb, pid, seq, event, sizeof(*hdr), flags); 3968 nlh = nlmsg_put(skb, pid, seq, event, sizeof(*hdr), flags);
3910 if (nlh == NULL) 3969 if (nlh == NULL)
@@ -3931,30 +3990,8 @@ static int inet6_fill_ifinfo(struct sk_buff *skb, struct inet6_dev *idev,
3931 if (protoinfo == NULL) 3990 if (protoinfo == NULL)
3932 goto nla_put_failure; 3991 goto nla_put_failure;
3933 3992
3934 NLA_PUT_U32(skb, IFLA_INET6_FLAGS, idev->if_flags); 3993 if (inet6_fill_ifla6_attrs(skb, idev) < 0)
3935
3936 ci.max_reasm_len = IPV6_MAXPLEN;
3937 ci.tstamp = cstamp_delta(idev->tstamp);
3938 ci.reachable_time = jiffies_to_msecs(idev->nd_parms->reachable_time);
3939 ci.retrans_time = jiffies_to_msecs(idev->nd_parms->retrans_time);
3940 NLA_PUT(skb, IFLA_INET6_CACHEINFO, sizeof(ci), &ci);
3941
3942 nla = nla_reserve(skb, IFLA_INET6_CONF, DEVCONF_MAX * sizeof(s32));
3943 if (nla == NULL)
3944 goto nla_put_failure;
3945 ipv6_store_devconf(&idev->cnf, nla_data(nla), nla_len(nla));
3946
3947 /* XXX - MC not implemented */
3948
3949 nla = nla_reserve(skb, IFLA_INET6_STATS, IPSTATS_MIB_MAX * sizeof(u64));
3950 if (nla == NULL)
3951 goto nla_put_failure;
3952 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_STATS, nla_len(nla));
3953
3954 nla = nla_reserve(skb, IFLA_INET6_ICMP6STATS, ICMP6_MIB_MAX * sizeof(u64));
3955 if (nla == NULL)
3956 goto nla_put_failure; 3994 goto nla_put_failure;
3957 snmp6_fill_stats(nla_data(nla), idev, IFLA_INET6_ICMP6STATS, nla_len(nla));
3958 3995
3959 nla_nest_end(skb, protoinfo); 3996 nla_nest_end(skb, protoinfo);
3960 return nlmsg_end(skb, nlh); 3997 return nlmsg_end(skb, nlh);
@@ -4625,6 +4662,12 @@ int unregister_inet6addr_notifier(struct notifier_block *nb)
4625} 4662}
4626EXPORT_SYMBOL(unregister_inet6addr_notifier); 4663EXPORT_SYMBOL(unregister_inet6addr_notifier);
4627 4664
4665static struct rtnl_af_ops inet6_ops = {
4666 .family = AF_INET6,
4667 .fill_link_af = inet6_fill_link_af,
4668 .get_link_af_size = inet6_get_link_af_size,
4669};
4670
4628/* 4671/*
4629 * Init / cleanup code 4672 * Init / cleanup code
4630 */ 4673 */
@@ -4676,6 +4719,10 @@ int __init addrconf_init(void)
4676 4719
4677 addrconf_verify(0); 4720 addrconf_verify(0);
4678 4721
4722 err = rtnl_af_register(&inet6_ops);
4723 if (err < 0)
4724 goto errout_af;
4725
4679 err = __rtnl_register(PF_INET6, RTM_GETLINK, NULL, inet6_dump_ifinfo); 4726 err = __rtnl_register(PF_INET6, RTM_GETLINK, NULL, inet6_dump_ifinfo);
4680 if (err < 0) 4727 if (err < 0)
4681 goto errout; 4728 goto errout;
@@ -4691,6 +4738,8 @@ int __init addrconf_init(void)
4691 4738
4692 return 0; 4739 return 0;
4693errout: 4740errout:
4741 rtnl_af_unregister(&inet6_ops);
4742errout_af:
4694 unregister_netdevice_notifier(&ipv6_dev_notf); 4743 unregister_netdevice_notifier(&ipv6_dev_notf);
4695errlo: 4744errlo:
4696 unregister_pernet_subsys(&addrconf_ops); 4745 unregister_pernet_subsys(&addrconf_ops);
@@ -4711,6 +4760,8 @@ void addrconf_cleanup(void)
4711 4760
4712 rtnl_lock(); 4761 rtnl_lock();
4713 4762
4763 __rtnl_af_unregister(&inet6_ops);
4764
4714 /* clean dev list */ 4765 /* clean dev list */
4715 for_each_netdev(&init_net, dev) { 4766 for_each_netdev(&init_net, dev) {
4716 if (__in6_dev_get(dev) == NULL) 4767 if (__in6_dev_get(dev) == NULL)
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index 54e8e42f7a88..978e80e2c4a8 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -300,7 +300,7 @@ int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
300 goto out; 300 goto out;
301 } 301 }
302 302
303 /* Reproduce AF_INET checks to make the bindings consitant */ 303 /* Reproduce AF_INET checks to make the bindings consistent */
304 v4addr = addr->sin6_addr.s6_addr32[3]; 304 v4addr = addr->sin6_addr.s6_addr32[3];
305 chk_addr_ret = inet_addr_type(net, v4addr); 305 chk_addr_ret = inet_addr_type(net, v4addr);
306 if (!sysctl_ip_nonlocal_bind && 306 if (!sysctl_ip_nonlocal_bind &&
@@ -810,7 +810,7 @@ static struct sk_buff *ipv6_gso_segment(struct sk_buff *skb, int features)
810 } 810 }
811 rcu_read_unlock(); 811 rcu_read_unlock();
812 812
813 if (unlikely(IS_ERR(segs))) 813 if (IS_ERR(segs))
814 goto out; 814 goto out;
815 815
816 for (skb = segs; skb; skb = skb->next) { 816 for (skb = segs; skb; skb = skb->next) {
diff --git a/net/ipv6/ah6.c b/net/ipv6/ah6.c
index ee82d4ef26ce..1aba54ae53c4 100644
--- a/net/ipv6/ah6.c
+++ b/net/ipv6/ah6.c
@@ -538,14 +538,16 @@ static int ah6_input(struct xfrm_state *x, struct sk_buff *skb)
538 if (!pskb_may_pull(skb, ah_hlen)) 538 if (!pskb_may_pull(skb, ah_hlen))
539 goto out; 539 goto out;
540 540
541 ip6h = ipv6_hdr(skb);
542
543 skb_push(skb, hdr_len);
544 541
545 if ((err = skb_cow_data(skb, 0, &trailer)) < 0) 542 if ((err = skb_cow_data(skb, 0, &trailer)) < 0)
546 goto out; 543 goto out;
547 nfrags = err; 544 nfrags = err;
548 545
546 ah = (struct ip_auth_hdr *)skb->data;
547 ip6h = ipv6_hdr(skb);
548
549 skb_push(skb, hdr_len);
550
549 work_iph = ah_alloc_tmp(ahash, nfrags, hdr_len + ahp->icv_trunc_len); 551 work_iph = ah_alloc_tmp(ahash, nfrags, hdr_len + ahp->icv_trunc_len);
550 if (!work_iph) 552 if (!work_iph)
551 goto out; 553 goto out;
diff --git a/net/ipv6/esp6.c b/net/ipv6/esp6.c
index ee9b93bdd6a2..1b5c9825743b 100644
--- a/net/ipv6/esp6.c
+++ b/net/ipv6/esp6.c
@@ -49,6 +49,8 @@ struct esp_skb_cb {
49 49
50#define ESP_SKB_CB(__skb) ((struct esp_skb_cb *)&((__skb)->cb[0])) 50#define ESP_SKB_CB(__skb) ((struct esp_skb_cb *)&((__skb)->cb[0]))
51 51
52static u32 esp6_get_mtu(struct xfrm_state *x, int mtu);
53
52/* 54/*
53 * Allocate an AEAD request structure with extra space for SG and IV. 55 * Allocate an AEAD request structure with extra space for SG and IV.
54 * 56 *
@@ -140,6 +142,8 @@ static int esp6_output(struct xfrm_state *x, struct sk_buff *skb)
140 int blksize; 142 int blksize;
141 int clen; 143 int clen;
142 int alen; 144 int alen;
145 int plen;
146 int tfclen;
143 int nfrags; 147 int nfrags;
144 u8 *iv; 148 u8 *iv;
145 u8 *tail; 149 u8 *tail;
@@ -148,18 +152,26 @@ static int esp6_output(struct xfrm_state *x, struct sk_buff *skb)
148 /* skb is pure payload to encrypt */ 152 /* skb is pure payload to encrypt */
149 err = -ENOMEM; 153 err = -ENOMEM;
150 154
151 /* Round to block size */
152 clen = skb->len;
153
154 aead = esp->aead; 155 aead = esp->aead;
155 alen = crypto_aead_authsize(aead); 156 alen = crypto_aead_authsize(aead);
156 157
158 tfclen = 0;
159 if (x->tfcpad) {
160 struct xfrm_dst *dst = (struct xfrm_dst *)skb_dst(skb);
161 u32 padto;
162
163 padto = min(x->tfcpad, esp6_get_mtu(x, dst->child_mtu_cached));
164 if (skb->len < padto)
165 tfclen = padto - skb->len;
166 }
157 blksize = ALIGN(crypto_aead_blocksize(aead), 4); 167 blksize = ALIGN(crypto_aead_blocksize(aead), 4);
158 clen = ALIGN(clen + 2, blksize); 168 clen = ALIGN(skb->len + 2 + tfclen, blksize);
159 if (esp->padlen) 169 if (esp->padlen)
160 clen = ALIGN(clen, esp->padlen); 170 clen = ALIGN(clen, esp->padlen);
171 plen = clen - skb->len - tfclen;
161 172
162 if ((err = skb_cow_data(skb, clen - skb->len + alen, &trailer)) < 0) 173 err = skb_cow_data(skb, tfclen + plen + alen, &trailer);
174 if (err < 0)
163 goto error; 175 goto error;
164 nfrags = err; 176 nfrags = err;
165 177
@@ -174,13 +186,17 @@ static int esp6_output(struct xfrm_state *x, struct sk_buff *skb)
174 186
175 /* Fill padding... */ 187 /* Fill padding... */
176 tail = skb_tail_pointer(trailer); 188 tail = skb_tail_pointer(trailer);
189 if (tfclen) {
190 memset(tail, 0, tfclen);
191 tail += tfclen;
192 }
177 do { 193 do {
178 int i; 194 int i;
179 for (i=0; i<clen-skb->len - 2; i++) 195 for (i = 0; i < plen - 2; i++)
180 tail[i] = i + 1; 196 tail[i] = i + 1;
181 } while (0); 197 } while (0);
182 tail[clen-skb->len - 2] = (clen - skb->len) - 2; 198 tail[plen - 2] = plen - 2;
183 tail[clen - skb->len - 1] = *skb_mac_header(skb); 199 tail[plen - 1] = *skb_mac_header(skb);
184 pskb_put(skb, trailer, clen - skb->len + alen); 200 pskb_put(skb, trailer, clen - skb->len + alen);
185 201
186 skb_push(skb, -skb_network_offset(skb)); 202 skb_push(skb, -skb_network_offset(skb));
diff --git a/net/ipv6/inet6_connection_sock.c b/net/ipv6/inet6_connection_sock.c
index 8a1628023bd1..d144e629d2b4 100644
--- a/net/ipv6/inet6_connection_sock.c
+++ b/net/ipv6/inet6_connection_sock.c
@@ -44,7 +44,7 @@ int inet6_csk_bind_conflict(const struct sock *sk,
44 !sk2->sk_bound_dev_if || 44 !sk2->sk_bound_dev_if ||
45 sk->sk_bound_dev_if == sk2->sk_bound_dev_if) && 45 sk->sk_bound_dev_if == sk2->sk_bound_dev_if) &&
46 (!sk->sk_reuse || !sk2->sk_reuse || 46 (!sk->sk_reuse || !sk2->sk_reuse ||
47 sk2->sk_state == TCP_LISTEN) && 47 ((1 << sk2->sk_state) & (TCPF_LISTEN | TCPF_CLOSE))) &&
48 ipv6_rcv_saddr_equal(sk, sk2)) 48 ipv6_rcv_saddr_equal(sk, sk2))
49 break; 49 break;
50 } 50 }
@@ -54,24 +54,54 @@ int inet6_csk_bind_conflict(const struct sock *sk,
54 54
55EXPORT_SYMBOL_GPL(inet6_csk_bind_conflict); 55EXPORT_SYMBOL_GPL(inet6_csk_bind_conflict);
56 56
57struct dst_entry *inet6_csk_route_req(struct sock *sk,
58 const struct request_sock *req)
59{
60 struct inet6_request_sock *treq = inet6_rsk(req);
61 struct ipv6_pinfo *np = inet6_sk(sk);
62 struct in6_addr *final_p, final;
63 struct dst_entry *dst;
64 struct flowi fl;
65
66 memset(&fl, 0, sizeof(fl));
67 fl.proto = IPPROTO_TCP;
68 ipv6_addr_copy(&fl.fl6_dst, &treq->rmt_addr);
69 final_p = fl6_update_dst(&fl, np->opt, &final);
70 ipv6_addr_copy(&fl.fl6_src, &treq->loc_addr);
71 fl.oif = sk->sk_bound_dev_if;
72 fl.mark = sk->sk_mark;
73 fl.fl_ip_dport = inet_rsk(req)->rmt_port;
74 fl.fl_ip_sport = inet_rsk(req)->loc_port;
75 security_req_classify_flow(req, &fl);
76
77 if (ip6_dst_lookup(sk, &dst, &fl))
78 return NULL;
79
80 if (final_p)
81 ipv6_addr_copy(&fl.fl6_dst, final_p);
82
83 if ((xfrm_lookup(sock_net(sk), &dst, &fl, sk, 0)) < 0)
84 return NULL;
85
86 return dst;
87}
88
57/* 89/*
58 * request_sock (formerly open request) hash tables. 90 * request_sock (formerly open request) hash tables.
59 */ 91 */
60static u32 inet6_synq_hash(const struct in6_addr *raddr, const __be16 rport, 92static u32 inet6_synq_hash(const struct in6_addr *raddr, const __be16 rport,
61 const u32 rnd, const u16 synq_hsize) 93 const u32 rnd, const u16 synq_hsize)
62{ 94{
63 u32 a = (__force u32)raddr->s6_addr32[0]; 95 u32 c;
64 u32 b = (__force u32)raddr->s6_addr32[1]; 96
65 u32 c = (__force u32)raddr->s6_addr32[2]; 97 c = jhash_3words((__force u32)raddr->s6_addr32[0],
66 98 (__force u32)raddr->s6_addr32[1],
67 a += JHASH_GOLDEN_RATIO; 99 (__force u32)raddr->s6_addr32[2],
68 b += JHASH_GOLDEN_RATIO; 100 rnd);
69 c += rnd; 101
70 __jhash_mix(a, b, c); 102 c = jhash_2words((__force u32)raddr->s6_addr32[3],
71 103 (__force u32)rport,
72 a += (__force u32)raddr->s6_addr32[3]; 104 c);
73 b += (__force u32)rport;
74 __jhash_mix(a, b, c);
75 105
76 return c & (synq_hsize - 1); 106 return c & (synq_hsize - 1);
77} 107}
diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index 99157b4cd56e..94b5bf132b2e 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -56,7 +56,7 @@
56#include <net/checksum.h> 56#include <net/checksum.h>
57#include <linux/mroute6.h> 57#include <linux/mroute6.h>
58 58
59static int ip6_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *)); 59int ip6_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *));
60 60
61int __ip6_local_out(struct sk_buff *skb) 61int __ip6_local_out(struct sk_buff *skb)
62{ 62{
@@ -145,14 +145,6 @@ static int ip6_finish_output2(struct sk_buff *skb)
145 return -EINVAL; 145 return -EINVAL;
146} 146}
147 147
148static inline int ip6_skb_dst_mtu(struct sk_buff *skb)
149{
150 struct ipv6_pinfo *np = skb->sk ? inet6_sk(skb->sk) : NULL;
151
152 return (np && np->pmtudisc == IPV6_PMTUDISC_PROBE) ?
153 skb_dst(skb)->dev->mtu : dst_mtu(skb_dst(skb));
154}
155
156static int ip6_finish_output(struct sk_buff *skb) 148static int ip6_finish_output(struct sk_buff *skb)
157{ 149{
158 if ((skb->len > ip6_skb_dst_mtu(skb) && !skb_is_gso(skb)) || 150 if ((skb->len > ip6_skb_dst_mtu(skb) && !skb_is_gso(skb)) ||
@@ -601,7 +593,7 @@ int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr)
601 return offset; 593 return offset;
602} 594}
603 595
604static int ip6_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *)) 596int ip6_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *))
605{ 597{
606 struct sk_buff *frag; 598 struct sk_buff *frag;
607 struct rt6_info *rt = (struct rt6_info*)skb_dst(skb); 599 struct rt6_info *rt = (struct rt6_info*)skb_dst(skb);
diff --git a/net/ipv6/ip6_tunnel.c b/net/ipv6/ip6_tunnel.c
index 70e891a20fb9..4f4483e697bd 100644
--- a/net/ipv6/ip6_tunnel.c
+++ b/net/ipv6/ip6_tunnel.c
@@ -58,8 +58,6 @@ MODULE_AUTHOR("Ville Nuorvala");
58MODULE_DESCRIPTION("IPv6 tunneling device"); 58MODULE_DESCRIPTION("IPv6 tunneling device");
59MODULE_LICENSE("GPL"); 59MODULE_LICENSE("GPL");
60 60
61#define IPV6_TLV_TEL_DST_SIZE 8
62
63#ifdef IP6_TNL_DEBUG 61#ifdef IP6_TNL_DEBUG
64#define IP6_TNL_TRACE(x...) printk(KERN_DEBUG "%s:" x "\n", __func__) 62#define IP6_TNL_TRACE(x...) printk(KERN_DEBUG "%s:" x "\n", __func__)
65#else 63#else
diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index 6f32ffce7022..9fab274019c0 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -1843,9 +1843,7 @@ static int ip6mr_forward2(struct net *net, struct mr6_table *mrt,
1843 1843
1844 fl = (struct flowi) { 1844 fl = (struct flowi) {
1845 .oif = vif->link, 1845 .oif = vif->link,
1846 .nl_u = { .ip6_u = 1846 .fl6_dst = ipv6h->daddr,
1847 { .daddr = ipv6h->daddr, }
1848 }
1849 }; 1847 };
1850 1848
1851 dst = ip6_route_output(net, NULL, &fl); 1849 dst = ip6_route_output(net, NULL, &fl);
diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
index d1444b95ad7e..49f986d626a0 100644
--- a/net/ipv6/mcast.c
+++ b/net/ipv6/mcast.c
@@ -82,7 +82,7 @@ static void *__mld2_query_bugs[] __attribute__((__unused__)) = {
82static struct in6_addr mld2_all_mcr = MLD2_ALL_MCR_INIT; 82static struct in6_addr mld2_all_mcr = MLD2_ALL_MCR_INIT;
83 83
84/* Big mc list lock for all the sockets */ 84/* Big mc list lock for all the sockets */
85static DEFINE_RWLOCK(ipv6_sk_mc_lock); 85static DEFINE_SPINLOCK(ipv6_sk_mc_lock);
86 86
87static void igmp6_join_group(struct ifmcaddr6 *ma); 87static void igmp6_join_group(struct ifmcaddr6 *ma);
88static void igmp6_leave_group(struct ifmcaddr6 *ma); 88static void igmp6_leave_group(struct ifmcaddr6 *ma);
@@ -123,6 +123,11 @@ int sysctl_mld_max_msf __read_mostly = IPV6_MLD_MAX_MSF;
123 * socket join on multicast group 123 * socket join on multicast group
124 */ 124 */
125 125
126#define for_each_pmc_rcu(np, pmc) \
127 for (pmc = rcu_dereference(np->ipv6_mc_list); \
128 pmc != NULL; \
129 pmc = rcu_dereference(pmc->next))
130
126int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr) 131int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
127{ 132{
128 struct net_device *dev = NULL; 133 struct net_device *dev = NULL;
@@ -134,15 +139,15 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
134 if (!ipv6_addr_is_multicast(addr)) 139 if (!ipv6_addr_is_multicast(addr))
135 return -EINVAL; 140 return -EINVAL;
136 141
137 read_lock_bh(&ipv6_sk_mc_lock); 142 rcu_read_lock();
138 for (mc_lst=np->ipv6_mc_list; mc_lst; mc_lst=mc_lst->next) { 143 for_each_pmc_rcu(np, mc_lst) {
139 if ((ifindex == 0 || mc_lst->ifindex == ifindex) && 144 if ((ifindex == 0 || mc_lst->ifindex == ifindex) &&
140 ipv6_addr_equal(&mc_lst->addr, addr)) { 145 ipv6_addr_equal(&mc_lst->addr, addr)) {
141 read_unlock_bh(&ipv6_sk_mc_lock); 146 rcu_read_unlock();
142 return -EADDRINUSE; 147 return -EADDRINUSE;
143 } 148 }
144 } 149 }
145 read_unlock_bh(&ipv6_sk_mc_lock); 150 rcu_read_unlock();
146 151
147 mc_lst = sock_kmalloc(sk, sizeof(struct ipv6_mc_socklist), GFP_KERNEL); 152 mc_lst = sock_kmalloc(sk, sizeof(struct ipv6_mc_socklist), GFP_KERNEL);
148 153
@@ -186,33 +191,41 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, const struct in6_addr *addr)
186 return err; 191 return err;
187 } 192 }
188 193
189 write_lock_bh(&ipv6_sk_mc_lock); 194 spin_lock(&ipv6_sk_mc_lock);
190 mc_lst->next = np->ipv6_mc_list; 195 mc_lst->next = np->ipv6_mc_list;
191 np->ipv6_mc_list = mc_lst; 196 rcu_assign_pointer(np->ipv6_mc_list, mc_lst);
192 write_unlock_bh(&ipv6_sk_mc_lock); 197 spin_unlock(&ipv6_sk_mc_lock);
193 198
194 rcu_read_unlock(); 199 rcu_read_unlock();
195 200
196 return 0; 201 return 0;
197} 202}
198 203
204static void ipv6_mc_socklist_reclaim(struct rcu_head *head)
205{
206 kfree(container_of(head, struct ipv6_mc_socklist, rcu));
207}
199/* 208/*
200 * socket leave on multicast group 209 * socket leave on multicast group
201 */ 210 */
202int ipv6_sock_mc_drop(struct sock *sk, int ifindex, const struct in6_addr *addr) 211int ipv6_sock_mc_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
203{ 212{
204 struct ipv6_pinfo *np = inet6_sk(sk); 213 struct ipv6_pinfo *np = inet6_sk(sk);
205 struct ipv6_mc_socklist *mc_lst, **lnk; 214 struct ipv6_mc_socklist *mc_lst;
215 struct ipv6_mc_socklist __rcu **lnk;
206 struct net *net = sock_net(sk); 216 struct net *net = sock_net(sk);
207 217
208 write_lock_bh(&ipv6_sk_mc_lock); 218 spin_lock(&ipv6_sk_mc_lock);
209 for (lnk = &np->ipv6_mc_list; (mc_lst = *lnk) !=NULL ; lnk = &mc_lst->next) { 219 for (lnk = &np->ipv6_mc_list;
220 (mc_lst = rcu_dereference_protected(*lnk,
221 lockdep_is_held(&ipv6_sk_mc_lock))) !=NULL ;
222 lnk = &mc_lst->next) {
210 if ((ifindex == 0 || mc_lst->ifindex == ifindex) && 223 if ((ifindex == 0 || mc_lst->ifindex == ifindex) &&
211 ipv6_addr_equal(&mc_lst->addr, addr)) { 224 ipv6_addr_equal(&mc_lst->addr, addr)) {
212 struct net_device *dev; 225 struct net_device *dev;
213 226
214 *lnk = mc_lst->next; 227 *lnk = mc_lst->next;
215 write_unlock_bh(&ipv6_sk_mc_lock); 228 spin_unlock(&ipv6_sk_mc_lock);
216 229
217 rcu_read_lock(); 230 rcu_read_lock();
218 dev = dev_get_by_index_rcu(net, mc_lst->ifindex); 231 dev = dev_get_by_index_rcu(net, mc_lst->ifindex);
@@ -225,11 +238,12 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, const struct in6_addr *addr)
225 } else 238 } else
226 (void) ip6_mc_leave_src(sk, mc_lst, NULL); 239 (void) ip6_mc_leave_src(sk, mc_lst, NULL);
227 rcu_read_unlock(); 240 rcu_read_unlock();
228 sock_kfree_s(sk, mc_lst, sizeof(*mc_lst)); 241 atomic_sub(sizeof(*mc_lst), &sk->sk_omem_alloc);
242 call_rcu(&mc_lst->rcu, ipv6_mc_socklist_reclaim);
229 return 0; 243 return 0;
230 } 244 }
231 } 245 }
232 write_unlock_bh(&ipv6_sk_mc_lock); 246 spin_unlock(&ipv6_sk_mc_lock);
233 247
234 return -EADDRNOTAVAIL; 248 return -EADDRNOTAVAIL;
235} 249}
@@ -257,7 +271,7 @@ static struct inet6_dev *ip6_mc_find_dev_rcu(struct net *net,
257 return NULL; 271 return NULL;
258 idev = __in6_dev_get(dev); 272 idev = __in6_dev_get(dev);
259 if (!idev) 273 if (!idev)
260 return NULL;; 274 return NULL;
261 read_lock_bh(&idev->lock); 275 read_lock_bh(&idev->lock);
262 if (idev->dead) { 276 if (idev->dead) {
263 read_unlock_bh(&idev->lock); 277 read_unlock_bh(&idev->lock);
@@ -272,12 +286,13 @@ void ipv6_sock_mc_close(struct sock *sk)
272 struct ipv6_mc_socklist *mc_lst; 286 struct ipv6_mc_socklist *mc_lst;
273 struct net *net = sock_net(sk); 287 struct net *net = sock_net(sk);
274 288
275 write_lock_bh(&ipv6_sk_mc_lock); 289 spin_lock(&ipv6_sk_mc_lock);
276 while ((mc_lst = np->ipv6_mc_list) != NULL) { 290 while ((mc_lst = rcu_dereference_protected(np->ipv6_mc_list,
291 lockdep_is_held(&ipv6_sk_mc_lock))) != NULL) {
277 struct net_device *dev; 292 struct net_device *dev;
278 293
279 np->ipv6_mc_list = mc_lst->next; 294 np->ipv6_mc_list = mc_lst->next;
280 write_unlock_bh(&ipv6_sk_mc_lock); 295 spin_unlock(&ipv6_sk_mc_lock);
281 296
282 rcu_read_lock(); 297 rcu_read_lock();
283 dev = dev_get_by_index_rcu(net, mc_lst->ifindex); 298 dev = dev_get_by_index_rcu(net, mc_lst->ifindex);
@@ -290,11 +305,13 @@ void ipv6_sock_mc_close(struct sock *sk)
290 } else 305 } else
291 (void) ip6_mc_leave_src(sk, mc_lst, NULL); 306 (void) ip6_mc_leave_src(sk, mc_lst, NULL);
292 rcu_read_unlock(); 307 rcu_read_unlock();
293 sock_kfree_s(sk, mc_lst, sizeof(*mc_lst));
294 308
295 write_lock_bh(&ipv6_sk_mc_lock); 309 atomic_sub(sizeof(*mc_lst), &sk->sk_omem_alloc);
310 call_rcu(&mc_lst->rcu, ipv6_mc_socklist_reclaim);
311
312 spin_lock(&ipv6_sk_mc_lock);
296 } 313 }
297 write_unlock_bh(&ipv6_sk_mc_lock); 314 spin_unlock(&ipv6_sk_mc_lock);
298} 315}
299 316
300int ip6_mc_source(int add, int omode, struct sock *sk, 317int ip6_mc_source(int add, int omode, struct sock *sk,
@@ -328,8 +345,7 @@ int ip6_mc_source(int add, int omode, struct sock *sk,
328 345
329 err = -EADDRNOTAVAIL; 346 err = -EADDRNOTAVAIL;
330 347
331 read_lock(&ipv6_sk_mc_lock); 348 for_each_pmc_rcu(inet6, pmc) {
332 for (pmc=inet6->ipv6_mc_list; pmc; pmc=pmc->next) {
333 if (pgsr->gsr_interface && pmc->ifindex != pgsr->gsr_interface) 349 if (pgsr->gsr_interface && pmc->ifindex != pgsr->gsr_interface)
334 continue; 350 continue;
335 if (ipv6_addr_equal(&pmc->addr, group)) 351 if (ipv6_addr_equal(&pmc->addr, group))
@@ -428,7 +444,6 @@ int ip6_mc_source(int add, int omode, struct sock *sk,
428done: 444done:
429 if (pmclocked) 445 if (pmclocked)
430 write_unlock(&pmc->sflock); 446 write_unlock(&pmc->sflock);
431 read_unlock(&ipv6_sk_mc_lock);
432 read_unlock_bh(&idev->lock); 447 read_unlock_bh(&idev->lock);
433 rcu_read_unlock(); 448 rcu_read_unlock();
434 if (leavegroup) 449 if (leavegroup)
@@ -466,14 +481,13 @@ int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf)
466 dev = idev->dev; 481 dev = idev->dev;
467 482
468 err = 0; 483 err = 0;
469 read_lock(&ipv6_sk_mc_lock);
470 484
471 if (gsf->gf_fmode == MCAST_INCLUDE && gsf->gf_numsrc == 0) { 485 if (gsf->gf_fmode == MCAST_INCLUDE && gsf->gf_numsrc == 0) {
472 leavegroup = 1; 486 leavegroup = 1;
473 goto done; 487 goto done;
474 } 488 }
475 489
476 for (pmc=inet6->ipv6_mc_list; pmc; pmc=pmc->next) { 490 for_each_pmc_rcu(inet6, pmc) {
477 if (pmc->ifindex != gsf->gf_interface) 491 if (pmc->ifindex != gsf->gf_interface)
478 continue; 492 continue;
479 if (ipv6_addr_equal(&pmc->addr, group)) 493 if (ipv6_addr_equal(&pmc->addr, group))
@@ -521,7 +535,6 @@ int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf)
521 write_unlock(&pmc->sflock); 535 write_unlock(&pmc->sflock);
522 err = 0; 536 err = 0;
523done: 537done:
524 read_unlock(&ipv6_sk_mc_lock);
525 read_unlock_bh(&idev->lock); 538 read_unlock_bh(&idev->lock);
526 rcu_read_unlock(); 539 rcu_read_unlock();
527 if (leavegroup) 540 if (leavegroup)
@@ -562,7 +575,7 @@ int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf,
562 * so reading the list is safe. 575 * so reading the list is safe.
563 */ 576 */
564 577
565 for (pmc=inet6->ipv6_mc_list; pmc; pmc=pmc->next) { 578 for_each_pmc_rcu(inet6, pmc) {
566 if (pmc->ifindex != gsf->gf_interface) 579 if (pmc->ifindex != gsf->gf_interface)
567 continue; 580 continue;
568 if (ipv6_addr_equal(group, &pmc->addr)) 581 if (ipv6_addr_equal(group, &pmc->addr))
@@ -612,13 +625,13 @@ int inet6_mc_check(struct sock *sk, const struct in6_addr *mc_addr,
612 struct ip6_sf_socklist *psl; 625 struct ip6_sf_socklist *psl;
613 int rv = 1; 626 int rv = 1;
614 627
615 read_lock(&ipv6_sk_mc_lock); 628 rcu_read_lock();
616 for (mc = np->ipv6_mc_list; mc; mc = mc->next) { 629 for_each_pmc_rcu(np, mc) {
617 if (ipv6_addr_equal(&mc->addr, mc_addr)) 630 if (ipv6_addr_equal(&mc->addr, mc_addr))
618 break; 631 break;
619 } 632 }
620 if (!mc) { 633 if (!mc) {
621 read_unlock(&ipv6_sk_mc_lock); 634 rcu_read_unlock();
622 return 1; 635 return 1;
623 } 636 }
624 read_lock(&mc->sflock); 637 read_lock(&mc->sflock);
@@ -638,7 +651,7 @@ int inet6_mc_check(struct sock *sk, const struct in6_addr *mc_addr,
638 rv = 0; 651 rv = 0;
639 } 652 }
640 read_unlock(&mc->sflock); 653 read_unlock(&mc->sflock);
641 read_unlock(&ipv6_sk_mc_lock); 654 rcu_read_unlock();
642 655
643 return rv; 656 return rv;
644} 657}
diff --git a/net/ipv6/ndisc.c b/net/ipv6/ndisc.c
index 998d6d27e7cf..2342545a5ee9 100644
--- a/net/ipv6/ndisc.c
+++ b/net/ipv6/ndisc.c
@@ -141,18 +141,18 @@ struct neigh_table nd_tbl = {
141 .proxy_redo = pndisc_redo, 141 .proxy_redo = pndisc_redo,
142 .id = "ndisc_cache", 142 .id = "ndisc_cache",
143 .parms = { 143 .parms = {
144 .tbl = &nd_tbl, 144 .tbl = &nd_tbl,
145 .base_reachable_time = 30 * HZ, 145 .base_reachable_time = ND_REACHABLE_TIME,
146 .retrans_time = 1 * HZ, 146 .retrans_time = ND_RETRANS_TIMER,
147 .gc_staletime = 60 * HZ, 147 .gc_staletime = 60 * HZ,
148 .reachable_time = 30 * HZ, 148 .reachable_time = ND_REACHABLE_TIME,
149 .delay_probe_time = 5 * HZ, 149 .delay_probe_time = 5 * HZ,
150 .queue_len = 3, 150 .queue_len = 3,
151 .ucast_probes = 3, 151 .ucast_probes = 3,
152 .mcast_probes = 3, 152 .mcast_probes = 3,
153 .anycast_delay = 1 * HZ, 153 .anycast_delay = 1 * HZ,
154 .proxy_delay = (8 * HZ) / 10, 154 .proxy_delay = (8 * HZ) / 10,
155 .proxy_qlen = 64, 155 .proxy_qlen = 64,
156 }, 156 },
157 .gc_interval = 30 * HZ, 157 .gc_interval = 30 * HZ,
158 .gc_thresh1 = 128, 158 .gc_thresh1 = 128,
@@ -1259,7 +1259,8 @@ static void ndisc_router_discovery(struct sk_buff *skb)
1259 if (ra_msg->icmph.icmp6_hop_limit) { 1259 if (ra_msg->icmph.icmp6_hop_limit) {
1260 in6_dev->cnf.hop_limit = ra_msg->icmph.icmp6_hop_limit; 1260 in6_dev->cnf.hop_limit = ra_msg->icmph.icmp6_hop_limit;
1261 if (rt) 1261 if (rt)
1262 rt->dst.metrics[RTAX_HOPLIMIT-1] = ra_msg->icmph.icmp6_hop_limit; 1262 dst_metric_set(&rt->dst, RTAX_HOPLIMIT,
1263 ra_msg->icmph.icmp6_hop_limit);
1263 } 1264 }
1264 1265
1265skip_defrtr: 1266skip_defrtr:
@@ -1377,7 +1378,7 @@ skip_linkparms:
1377 in6_dev->cnf.mtu6 = mtu; 1378 in6_dev->cnf.mtu6 = mtu;
1378 1379
1379 if (rt) 1380 if (rt)
1380 rt->dst.metrics[RTAX_MTU-1] = mtu; 1381 dst_metric_set(&rt->dst, RTAX_MTU, mtu);
1381 1382
1382 rt6_mtu_change(skb->dev, mtu); 1383 rt6_mtu_change(skb->dev, mtu);
1383 } 1384 }
diff --git a/net/ipv6/netfilter.c b/net/ipv6/netfilter.c
index 7155b2451d7c..35915e8617f0 100644
--- a/net/ipv6/netfilter.c
+++ b/net/ipv6/netfilter.c
@@ -18,10 +18,8 @@ int ip6_route_me_harder(struct sk_buff *skb)
18 struct flowi fl = { 18 struct flowi fl = {
19 .oif = skb->sk ? skb->sk->sk_bound_dev_if : 0, 19 .oif = skb->sk ? skb->sk->sk_bound_dev_if : 0,
20 .mark = skb->mark, 20 .mark = skb->mark,
21 .nl_u = 21 .fl6_dst = iph->daddr,
22 { .ip6_u = 22 .fl6_src = iph->saddr,
23 { .daddr = iph->daddr,
24 .saddr = iph->saddr, } },
25 }; 23 };
26 24
27 dst = ip6_route_output(net, skb->sk, &fl); 25 dst = ip6_route_output(net, skb->sk, &fl);
diff --git a/net/ipv6/netfilter/Makefile b/net/ipv6/netfilter/Makefile
index 0a432c9b0795..abfee91ce816 100644
--- a/net/ipv6/netfilter/Makefile
+++ b/net/ipv6/netfilter/Makefile
@@ -11,13 +11,13 @@ obj-$(CONFIG_IP6_NF_RAW) += ip6table_raw.o
11obj-$(CONFIG_IP6_NF_SECURITY) += ip6table_security.o 11obj-$(CONFIG_IP6_NF_SECURITY) += ip6table_security.o
12 12
13# objects for l3 independent conntrack 13# objects for l3 independent conntrack
14nf_conntrack_ipv6-objs := nf_conntrack_l3proto_ipv6.o nf_conntrack_proto_icmpv6.o 14nf_conntrack_ipv6-y := nf_conntrack_l3proto_ipv6.o nf_conntrack_proto_icmpv6.o
15 15
16# l3 independent conntrack 16# l3 independent conntrack
17obj-$(CONFIG_NF_CONNTRACK_IPV6) += nf_conntrack_ipv6.o nf_defrag_ipv6.o 17obj-$(CONFIG_NF_CONNTRACK_IPV6) += nf_conntrack_ipv6.o nf_defrag_ipv6.o
18 18
19# defrag 19# defrag
20nf_defrag_ipv6-objs := nf_defrag_ipv6_hooks.o nf_conntrack_reasm.o 20nf_defrag_ipv6-y := nf_defrag_ipv6_hooks.o nf_conntrack_reasm.o
21obj-$(CONFIG_NF_DEFRAG_IPV6) += nf_defrag_ipv6.o 21obj-$(CONFIG_NF_DEFRAG_IPV6) += nf_defrag_ipv6.o
22 22
23# matches 23# matches
diff --git a/net/ipv6/netfilter/ip6_tables.c b/net/ipv6/netfilter/ip6_tables.c
index 455582384ece..7d227c644f72 100644
--- a/net/ipv6/netfilter/ip6_tables.c
+++ b/net/ipv6/netfilter/ip6_tables.c
@@ -897,42 +897,25 @@ get_counters(const struct xt_table_info *t,
897 struct ip6t_entry *iter; 897 struct ip6t_entry *iter;
898 unsigned int cpu; 898 unsigned int cpu;
899 unsigned int i; 899 unsigned int i;
900 unsigned int curcpu = get_cpu();
901
902 /* Instead of clearing (by a previous call to memset())
903 * the counters and using adds, we set the counters
904 * with data used by 'current' CPU
905 *
906 * Bottom half has to be disabled to prevent deadlock
907 * if new softirq were to run and call ipt_do_table
908 */
909 local_bh_disable();
910 i = 0;
911 xt_entry_foreach(iter, t->entries[curcpu], t->size) {
912 SET_COUNTER(counters[i], iter->counters.bcnt,
913 iter->counters.pcnt);
914 ++i;
915 }
916 local_bh_enable();
917 /* Processing counters from other cpus, we can let bottom half enabled,
918 * (preemption is disabled)
919 */
920 900
921 for_each_possible_cpu(cpu) { 901 for_each_possible_cpu(cpu) {
922 if (cpu == curcpu) 902 seqlock_t *lock = &per_cpu(xt_info_locks, cpu).lock;
923 continue; 903
924 i = 0; 904 i = 0;
925 local_bh_disable();
926 xt_info_wrlock(cpu);
927 xt_entry_foreach(iter, t->entries[cpu], t->size) { 905 xt_entry_foreach(iter, t->entries[cpu], t->size) {
928 ADD_COUNTER(counters[i], iter->counters.bcnt, 906 u64 bcnt, pcnt;
929 iter->counters.pcnt); 907 unsigned int start;
908
909 do {
910 start = read_seqbegin(lock);
911 bcnt = iter->counters.bcnt;
912 pcnt = iter->counters.pcnt;
913 } while (read_seqretry(lock, start));
914
915 ADD_COUNTER(counters[i], bcnt, pcnt);
930 ++i; 916 ++i;
931 } 917 }
932 xt_info_wrunlock(cpu);
933 local_bh_enable();
934 } 918 }
935 put_cpu();
936} 919}
937 920
938static struct xt_counters *alloc_counters(const struct xt_table *table) 921static struct xt_counters *alloc_counters(const struct xt_table *table)
@@ -945,7 +928,7 @@ static struct xt_counters *alloc_counters(const struct xt_table *table)
945 (other than comefrom, which userspace doesn't care 928 (other than comefrom, which userspace doesn't care
946 about). */ 929 about). */
947 countersize = sizeof(struct xt_counters) * private->number; 930 countersize = sizeof(struct xt_counters) * private->number;
948 counters = vmalloc(countersize); 931 counters = vzalloc(countersize);
949 932
950 if (counters == NULL) 933 if (counters == NULL)
951 return ERR_PTR(-ENOMEM); 934 return ERR_PTR(-ENOMEM);
@@ -1216,7 +1199,7 @@ __do_replace(struct net *net, const char *name, unsigned int valid_hooks,
1216 struct ip6t_entry *iter; 1199 struct ip6t_entry *iter;
1217 1200
1218 ret = 0; 1201 ret = 0;
1219 counters = vmalloc(num_counters * sizeof(struct xt_counters)); 1202 counters = vzalloc(num_counters * sizeof(struct xt_counters));
1220 if (!counters) { 1203 if (!counters) {
1221 ret = -ENOMEM; 1204 ret = -ENOMEM;
1222 goto out; 1205 goto out;
diff --git a/net/ipv6/netfilter/ip6t_REJECT.c b/net/ipv6/netfilter/ip6t_REJECT.c
index 2933396e0281..bf998feac14e 100644
--- a/net/ipv6/netfilter/ip6t_REJECT.c
+++ b/net/ipv6/netfilter/ip6t_REJECT.c
@@ -124,7 +124,7 @@ static void send_reset(struct net *net, struct sk_buff *oldskb)
124 skb_reset_network_header(nskb); 124 skb_reset_network_header(nskb);
125 ip6h = ipv6_hdr(nskb); 125 ip6h = ipv6_hdr(nskb);
126 ip6h->version = 6; 126 ip6h->version = 6;
127 ip6h->hop_limit = dst_metric(dst, RTAX_HOPLIMIT); 127 ip6h->hop_limit = ip6_dst_hoplimit(dst);
128 ip6h->nexthdr = IPPROTO_TCP; 128 ip6h->nexthdr = IPPROTO_TCP;
129 ipv6_addr_copy(&ip6h->saddr, &oip6h->daddr); 129 ipv6_addr_copy(&ip6h->saddr, &oip6h->daddr);
130 ipv6_addr_copy(&ip6h->daddr, &oip6h->saddr); 130 ipv6_addr_copy(&ip6h->daddr, &oip6h->saddr);
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 0f2766453759..07beeb06f752 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -104,26 +104,22 @@ static int ip6_frag_reasm(struct frag_queue *fq, struct sk_buff *prev,
104unsigned int inet6_hash_frag(__be32 id, const struct in6_addr *saddr, 104unsigned int inet6_hash_frag(__be32 id, const struct in6_addr *saddr,
105 const struct in6_addr *daddr, u32 rnd) 105 const struct in6_addr *daddr, u32 rnd)
106{ 106{
107 u32 a, b, c; 107 u32 c;
108 108
109 a = (__force u32)saddr->s6_addr32[0]; 109 c = jhash_3words((__force u32)saddr->s6_addr32[0],
110 b = (__force u32)saddr->s6_addr32[1]; 110 (__force u32)saddr->s6_addr32[1],
111 c = (__force u32)saddr->s6_addr32[2]; 111 (__force u32)saddr->s6_addr32[2],
112 112 rnd);
113 a += JHASH_GOLDEN_RATIO; 113
114 b += JHASH_GOLDEN_RATIO; 114 c = jhash_3words((__force u32)saddr->s6_addr32[3],
115 c += rnd; 115 (__force u32)daddr->s6_addr32[0],
116 __jhash_mix(a, b, c); 116 (__force u32)daddr->s6_addr32[1],
117 117 c);
118 a += (__force u32)saddr->s6_addr32[3]; 118
119 b += (__force u32)daddr->s6_addr32[0]; 119 c = jhash_3words((__force u32)daddr->s6_addr32[2],
120 c += (__force u32)daddr->s6_addr32[1]; 120 (__force u32)daddr->s6_addr32[3],
121 __jhash_mix(a, b, c); 121 (__force u32)id,
122 122 c);
123 a += (__force u32)daddr->s6_addr32[2];
124 b += (__force u32)daddr->s6_addr32[3];
125 c += (__force u32)id;
126 __jhash_mix(a, b, c);
127 123
128 return c & (INETFRAGS_HASHSZ - 1); 124 return c & (INETFRAGS_HASHSZ - 1);
129} 125}
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 96455ffb76fb..373bd0416f69 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -76,6 +76,8 @@
76 76
77static struct rt6_info * ip6_rt_copy(struct rt6_info *ort); 77static struct rt6_info * ip6_rt_copy(struct rt6_info *ort);
78static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie); 78static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie);
79static unsigned int ip6_default_advmss(const struct dst_entry *dst);
80static unsigned int ip6_default_mtu(const struct dst_entry *dst);
79static struct dst_entry *ip6_negative_advice(struct dst_entry *); 81static struct dst_entry *ip6_negative_advice(struct dst_entry *);
80static void ip6_dst_destroy(struct dst_entry *); 82static void ip6_dst_destroy(struct dst_entry *);
81static void ip6_dst_ifdown(struct dst_entry *, 83static void ip6_dst_ifdown(struct dst_entry *,
@@ -103,6 +105,8 @@ static struct dst_ops ip6_dst_ops_template = {
103 .gc = ip6_dst_gc, 105 .gc = ip6_dst_gc,
104 .gc_thresh = 1024, 106 .gc_thresh = 1024,
105 .check = ip6_dst_check, 107 .check = ip6_dst_check,
108 .default_advmss = ip6_default_advmss,
109 .default_mtu = ip6_default_mtu,
106 .destroy = ip6_dst_destroy, 110 .destroy = ip6_dst_destroy,
107 .ifdown = ip6_dst_ifdown, 111 .ifdown = ip6_dst_ifdown,
108 .negative_advice = ip6_negative_advice, 112 .negative_advice = ip6_negative_advice,
@@ -129,7 +133,6 @@ static struct rt6_info ip6_null_entry_template = {
129 .__use = 1, 133 .__use = 1,
130 .obsolete = -1, 134 .obsolete = -1,
131 .error = -ENETUNREACH, 135 .error = -ENETUNREACH,
132 .metrics = { [RTAX_HOPLIMIT - 1] = 255, },
133 .input = ip6_pkt_discard, 136 .input = ip6_pkt_discard,
134 .output = ip6_pkt_discard_out, 137 .output = ip6_pkt_discard_out,
135 }, 138 },
@@ -150,7 +153,6 @@ static struct rt6_info ip6_prohibit_entry_template = {
150 .__use = 1, 153 .__use = 1,
151 .obsolete = -1, 154 .obsolete = -1,
152 .error = -EACCES, 155 .error = -EACCES,
153 .metrics = { [RTAX_HOPLIMIT - 1] = 255, },
154 .input = ip6_pkt_prohibit, 156 .input = ip6_pkt_prohibit,
155 .output = ip6_pkt_prohibit_out, 157 .output = ip6_pkt_prohibit_out,
156 }, 158 },
@@ -166,7 +168,6 @@ static struct rt6_info ip6_blk_hole_entry_template = {
166 .__use = 1, 168 .__use = 1,
167 .obsolete = -1, 169 .obsolete = -1,
168 .error = -EINVAL, 170 .error = -EINVAL,
169 .metrics = { [RTAX_HOPLIMIT - 1] = 255, },
170 .input = dst_discard, 171 .input = dst_discard,
171 .output = dst_discard, 172 .output = dst_discard,
172 }, 173 },
@@ -188,11 +189,29 @@ static void ip6_dst_destroy(struct dst_entry *dst)
188{ 189{
189 struct rt6_info *rt = (struct rt6_info *)dst; 190 struct rt6_info *rt = (struct rt6_info *)dst;
190 struct inet6_dev *idev = rt->rt6i_idev; 191 struct inet6_dev *idev = rt->rt6i_idev;
192 struct inet_peer *peer = rt->rt6i_peer;
191 193
192 if (idev != NULL) { 194 if (idev != NULL) {
193 rt->rt6i_idev = NULL; 195 rt->rt6i_idev = NULL;
194 in6_dev_put(idev); 196 in6_dev_put(idev);
195 } 197 }
198 if (peer) {
199 BUG_ON(!(rt->rt6i_flags & RTF_CACHE));
200 rt->rt6i_peer = NULL;
201 inet_putpeer(peer);
202 }
203}
204
205void rt6_bind_peer(struct rt6_info *rt, int create)
206{
207 struct inet_peer *peer;
208
209 if (WARN_ON(!(rt->rt6i_flags & RTF_CACHE)))
210 return;
211
212 peer = inet_getpeer_v6(&rt->rt6i_dst.addr, create);
213 if (peer && cmpxchg(&rt->rt6i_peer, NULL, peer) != NULL)
214 inet_putpeer(peer);
196} 215}
197 216
198static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev, 217static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
@@ -558,11 +577,7 @@ struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
558{ 577{
559 struct flowi fl = { 578 struct flowi fl = {
560 .oif = oif, 579 .oif = oif,
561 .nl_u = { 580 .fl6_dst = *daddr,
562 .ip6_u = {
563 .daddr = *daddr,
564 },
565 },
566 }; 581 };
567 struct dst_entry *dst; 582 struct dst_entry *dst;
568 int flags = strict ? RT6_LOOKUP_F_IFACE : 0; 583 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
@@ -778,13 +793,9 @@ void ip6_route_input(struct sk_buff *skb)
778 int flags = RT6_LOOKUP_F_HAS_SADDR; 793 int flags = RT6_LOOKUP_F_HAS_SADDR;
779 struct flowi fl = { 794 struct flowi fl = {
780 .iif = skb->dev->ifindex, 795 .iif = skb->dev->ifindex,
781 .nl_u = { 796 .fl6_dst = iph->daddr,
782 .ip6_u = { 797 .fl6_src = iph->saddr,
783 .daddr = iph->daddr, 798 .fl6_flowlabel = (* (__be32 *) iph)&IPV6_FLOWINFO_MASK,
784 .saddr = iph->saddr,
785 .flowlabel = (* (__be32 *) iph)&IPV6_FLOWINFO_MASK,
786 },
787 },
788 .mark = skb->mark, 799 .mark = skb->mark,
789 .proto = iph->nexthdr, 800 .proto = iph->nexthdr,
790 }; 801 };
@@ -834,7 +845,7 @@ int ip6_dst_blackhole(struct sock *sk, struct dst_entry **dstp, struct flowi *fl
834 new->input = dst_discard; 845 new->input = dst_discard;
835 new->output = dst_discard; 846 new->output = dst_discard;
836 847
837 memcpy(new->metrics, ort->dst.metrics, RTAX_MAX*sizeof(u32)); 848 dst_copy_metrics(new, &ort->dst);
838 new->dev = ort->dst.dev; 849 new->dev = ort->dst.dev;
839 if (new->dev) 850 if (new->dev)
840 dev_hold(new->dev); 851 dev_hold(new->dev);
@@ -918,18 +929,22 @@ static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu)
918 if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) { 929 if (mtu < dst_mtu(dst) && rt6->rt6i_dst.plen == 128) {
919 rt6->rt6i_flags |= RTF_MODIFIED; 930 rt6->rt6i_flags |= RTF_MODIFIED;
920 if (mtu < IPV6_MIN_MTU) { 931 if (mtu < IPV6_MIN_MTU) {
932 u32 features = dst_metric(dst, RTAX_FEATURES);
921 mtu = IPV6_MIN_MTU; 933 mtu = IPV6_MIN_MTU;
922 dst->metrics[RTAX_FEATURES-1] |= RTAX_FEATURE_ALLFRAG; 934 features |= RTAX_FEATURE_ALLFRAG;
935 dst_metric_set(dst, RTAX_FEATURES, features);
923 } 936 }
924 dst->metrics[RTAX_MTU-1] = mtu; 937 dst_metric_set(dst, RTAX_MTU, mtu);
925 call_netevent_notifiers(NETEVENT_PMTU_UPDATE, dst); 938 call_netevent_notifiers(NETEVENT_PMTU_UPDATE, dst);
926 } 939 }
927} 940}
928 941
929static int ipv6_get_mtu(struct net_device *dev); 942static unsigned int ip6_default_advmss(const struct dst_entry *dst)
930
931static inline unsigned int ipv6_advmss(struct net *net, unsigned int mtu)
932{ 943{
944 struct net_device *dev = dst->dev;
945 unsigned int mtu = dst_mtu(dst);
946 struct net *net = dev_net(dev);
947
933 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr); 948 mtu -= sizeof(struct ipv6hdr) + sizeof(struct tcphdr);
934 949
935 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss) 950 if (mtu < net->ipv6.sysctl.ip6_rt_min_advmss)
@@ -946,6 +961,20 @@ static inline unsigned int ipv6_advmss(struct net *net, unsigned int mtu)
946 return mtu; 961 return mtu;
947} 962}
948 963
964static unsigned int ip6_default_mtu(const struct dst_entry *dst)
965{
966 unsigned int mtu = IPV6_MIN_MTU;
967 struct inet6_dev *idev;
968
969 rcu_read_lock();
970 idev = __in6_dev_get(dst->dev);
971 if (idev)
972 mtu = idev->cnf.mtu6;
973 rcu_read_unlock();
974
975 return mtu;
976}
977
949static struct dst_entry *icmp6_dst_gc_list; 978static struct dst_entry *icmp6_dst_gc_list;
950static DEFINE_SPINLOCK(icmp6_dst_lock); 979static DEFINE_SPINLOCK(icmp6_dst_lock);
951 980
@@ -979,9 +1008,7 @@ struct dst_entry *icmp6_dst_alloc(struct net_device *dev,
979 rt->rt6i_idev = idev; 1008 rt->rt6i_idev = idev;
980 rt->rt6i_nexthop = neigh; 1009 rt->rt6i_nexthop = neigh;
981 atomic_set(&rt->dst.__refcnt, 1); 1010 atomic_set(&rt->dst.__refcnt, 1);
982 rt->dst.metrics[RTAX_HOPLIMIT-1] = 255; 1011 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, 255);
983 rt->dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(rt->rt6i_dev);
984 rt->dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, dst_mtu(&rt->dst));
985 rt->dst.output = ip6_output; 1012 rt->dst.output = ip6_output;
986 1013
987#if 0 /* there's no chance to use these for ndisc */ 1014#if 0 /* there's no chance to use these for ndisc */
@@ -1080,23 +1107,10 @@ out:
1080 Remove it only when all the things will work! 1107 Remove it only when all the things will work!
1081 */ 1108 */
1082 1109
1083static int ipv6_get_mtu(struct net_device *dev)
1084{
1085 int mtu = IPV6_MIN_MTU;
1086 struct inet6_dev *idev;
1087
1088 rcu_read_lock();
1089 idev = __in6_dev_get(dev);
1090 if (idev)
1091 mtu = idev->cnf.mtu6;
1092 rcu_read_unlock();
1093 return mtu;
1094}
1095
1096int ip6_dst_hoplimit(struct dst_entry *dst) 1110int ip6_dst_hoplimit(struct dst_entry *dst)
1097{ 1111{
1098 int hoplimit = dst_metric(dst, RTAX_HOPLIMIT); 1112 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
1099 if (hoplimit < 0) { 1113 if (hoplimit == 0) {
1100 struct net_device *dev = dst->dev; 1114 struct net_device *dev = dst->dev;
1101 struct inet6_dev *idev; 1115 struct inet6_dev *idev;
1102 1116
@@ -1110,6 +1124,7 @@ int ip6_dst_hoplimit(struct dst_entry *dst)
1110 } 1124 }
1111 return hoplimit; 1125 return hoplimit;
1112} 1126}
1127EXPORT_SYMBOL(ip6_dst_hoplimit);
1113 1128
1114/* 1129/*
1115 * 1130 *
@@ -1295,17 +1310,11 @@ install_route:
1295 goto out; 1310 goto out;
1296 } 1311 }
1297 1312
1298 rt->dst.metrics[type - 1] = nla_get_u32(nla); 1313 dst_metric_set(&rt->dst, type, nla_get_u32(nla));
1299 } 1314 }
1300 } 1315 }
1301 } 1316 }
1302 1317
1303 if (dst_metric(&rt->dst, RTAX_HOPLIMIT) == 0)
1304 rt->dst.metrics[RTAX_HOPLIMIT-1] = -1;
1305 if (!dst_mtu(&rt->dst))
1306 rt->dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(dev);
1307 if (!dst_metric(&rt->dst, RTAX_ADVMSS))
1308 rt->dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, dst_mtu(&rt->dst));
1309 rt->dst.dev = dev; 1318 rt->dst.dev = dev;
1310 rt->rt6i_idev = idev; 1319 rt->rt6i_idev = idev;
1311 rt->rt6i_table = table; 1320 rt->rt6i_table = table;
@@ -1463,12 +1472,8 @@ static struct rt6_info *ip6_route_redirect(struct in6_addr *dest,
1463 struct ip6rd_flowi rdfl = { 1472 struct ip6rd_flowi rdfl = {
1464 .fl = { 1473 .fl = {
1465 .oif = dev->ifindex, 1474 .oif = dev->ifindex,
1466 .nl_u = { 1475 .fl6_dst = *dest,
1467 .ip6_u = { 1476 .fl6_src = *src,
1468 .daddr = *dest,
1469 .saddr = *src,
1470 },
1471 },
1472 }, 1477 },
1473 }; 1478 };
1474 1479
@@ -1534,10 +1539,6 @@ void rt6_redirect(struct in6_addr *dest, struct in6_addr *src,
1534 1539
1535 ipv6_addr_copy(&nrt->rt6i_gateway, (struct in6_addr*)neigh->primary_key); 1540 ipv6_addr_copy(&nrt->rt6i_gateway, (struct in6_addr*)neigh->primary_key);
1536 nrt->rt6i_nexthop = neigh_clone(neigh); 1541 nrt->rt6i_nexthop = neigh_clone(neigh);
1537 /* Reset pmtu, it may be better */
1538 nrt->dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(neigh->dev);
1539 nrt->dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(dev_net(neigh->dev),
1540 dst_mtu(&nrt->dst));
1541 1542
1542 if (ip6_ins_rt(nrt)) 1543 if (ip6_ins_rt(nrt))
1543 goto out; 1544 goto out;
@@ -1565,11 +1566,16 @@ static void rt6_do_pmtu_disc(struct in6_addr *daddr, struct in6_addr *saddr,
1565{ 1566{
1566 struct rt6_info *rt, *nrt; 1567 struct rt6_info *rt, *nrt;
1567 int allfrag = 0; 1568 int allfrag = 0;
1568 1569again:
1569 rt = rt6_lookup(net, daddr, saddr, ifindex, 0); 1570 rt = rt6_lookup(net, daddr, saddr, ifindex, 0);
1570 if (rt == NULL) 1571 if (rt == NULL)
1571 return; 1572 return;
1572 1573
1574 if (rt6_check_expired(rt)) {
1575 ip6_del_rt(rt);
1576 goto again;
1577 }
1578
1573 if (pmtu >= dst_mtu(&rt->dst)) 1579 if (pmtu >= dst_mtu(&rt->dst))
1574 goto out; 1580 goto out;
1575 1581
@@ -1596,9 +1602,12 @@ static void rt6_do_pmtu_disc(struct in6_addr *daddr, struct in6_addr *saddr,
1596 would return automatically. 1602 would return automatically.
1597 */ 1603 */
1598 if (rt->rt6i_flags & RTF_CACHE) { 1604 if (rt->rt6i_flags & RTF_CACHE) {
1599 rt->dst.metrics[RTAX_MTU-1] = pmtu; 1605 dst_metric_set(&rt->dst, RTAX_MTU, pmtu);
1600 if (allfrag) 1606 if (allfrag) {
1601 rt->dst.metrics[RTAX_FEATURES-1] |= RTAX_FEATURE_ALLFRAG; 1607 u32 features = dst_metric(&rt->dst, RTAX_FEATURES);
1608 features |= RTAX_FEATURE_ALLFRAG;
1609 dst_metric_set(&rt->dst, RTAX_FEATURES, features);
1610 }
1602 dst_set_expires(&rt->dst, net->ipv6.sysctl.ip6_rt_mtu_expires); 1611 dst_set_expires(&rt->dst, net->ipv6.sysctl.ip6_rt_mtu_expires);
1603 rt->rt6i_flags |= RTF_MODIFIED|RTF_EXPIRES; 1612 rt->rt6i_flags |= RTF_MODIFIED|RTF_EXPIRES;
1604 goto out; 1613 goto out;
@@ -1615,9 +1624,12 @@ static void rt6_do_pmtu_disc(struct in6_addr *daddr, struct in6_addr *saddr,
1615 nrt = rt6_alloc_clone(rt, daddr); 1624 nrt = rt6_alloc_clone(rt, daddr);
1616 1625
1617 if (nrt) { 1626 if (nrt) {
1618 nrt->dst.metrics[RTAX_MTU-1] = pmtu; 1627 dst_metric_set(&nrt->dst, RTAX_MTU, pmtu);
1619 if (allfrag) 1628 if (allfrag) {
1620 nrt->dst.metrics[RTAX_FEATURES-1] |= RTAX_FEATURE_ALLFRAG; 1629 u32 features = dst_metric(&nrt->dst, RTAX_FEATURES);
1630 features |= RTAX_FEATURE_ALLFRAG;
1631 dst_metric_set(&nrt->dst, RTAX_FEATURES, features);
1632 }
1621 1633
1622 /* According to RFC 1981, detecting PMTU increase shouldn't be 1634 /* According to RFC 1981, detecting PMTU increase shouldn't be
1623 * happened within 5 mins, the recommended timer is 10 mins. 1635 * happened within 5 mins, the recommended timer is 10 mins.
@@ -1668,7 +1680,7 @@ static struct rt6_info * ip6_rt_copy(struct rt6_info *ort)
1668 rt->dst.input = ort->dst.input; 1680 rt->dst.input = ort->dst.input;
1669 rt->dst.output = ort->dst.output; 1681 rt->dst.output = ort->dst.output;
1670 1682
1671 memcpy(rt->dst.metrics, ort->dst.metrics, RTAX_MAX*sizeof(u32)); 1683 dst_copy_metrics(&rt->dst, &ort->dst);
1672 rt->dst.error = ort->dst.error; 1684 rt->dst.error = ort->dst.error;
1673 rt->dst.dev = ort->dst.dev; 1685 rt->dst.dev = ort->dst.dev;
1674 if (rt->dst.dev) 1686 if (rt->dst.dev)
@@ -1960,9 +1972,7 @@ struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
1960 rt->dst.output = ip6_output; 1972 rt->dst.output = ip6_output;
1961 rt->rt6i_dev = net->loopback_dev; 1973 rt->rt6i_dev = net->loopback_dev;
1962 rt->rt6i_idev = idev; 1974 rt->rt6i_idev = idev;
1963 rt->dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(rt->rt6i_dev); 1975 dst_metric_set(&rt->dst, RTAX_HOPLIMIT, -1);
1964 rt->dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, dst_mtu(&rt->dst));
1965 rt->dst.metrics[RTAX_HOPLIMIT-1] = -1;
1966 rt->dst.obsolete = -1; 1976 rt->dst.obsolete = -1;
1967 1977
1968 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP; 1978 rt->rt6i_flags = RTF_UP | RTF_NONEXTHOP;
@@ -1999,11 +2009,11 @@ struct arg_dev_net {
1999 2009
2000static int fib6_ifdown(struct rt6_info *rt, void *arg) 2010static int fib6_ifdown(struct rt6_info *rt, void *arg)
2001{ 2011{
2002 struct net_device *dev = ((struct arg_dev_net *)arg)->dev; 2012 const struct arg_dev_net *adn = arg;
2003 struct net *net = ((struct arg_dev_net *)arg)->net; 2013 const struct net_device *dev = adn->dev;
2004 2014
2005 if (((void *)rt->rt6i_dev == dev || dev == NULL) && 2015 if ((rt->rt6i_dev == dev || dev == NULL) &&
2006 rt != net->ipv6.ip6_null_entry) { 2016 rt != adn->net->ipv6.ip6_null_entry) {
2007 RT6_TRACE("deleted by ifdown %p\n", rt); 2017 RT6_TRACE("deleted by ifdown %p\n", rt);
2008 return -1; 2018 return -1;
2009 } 2019 }
@@ -2031,7 +2041,6 @@ static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2031{ 2041{
2032 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg; 2042 struct rt6_mtu_change_arg *arg = (struct rt6_mtu_change_arg *) p_arg;
2033 struct inet6_dev *idev; 2043 struct inet6_dev *idev;
2034 struct net *net = dev_net(arg->dev);
2035 2044
2036 /* In IPv6 pmtu discovery is not optional, 2045 /* In IPv6 pmtu discovery is not optional,
2037 so that RTAX_MTU lock cannot disable it. 2046 so that RTAX_MTU lock cannot disable it.
@@ -2062,8 +2071,7 @@ static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
2062 (dst_mtu(&rt->dst) >= arg->mtu || 2071 (dst_mtu(&rt->dst) >= arg->mtu ||
2063 (dst_mtu(&rt->dst) < arg->mtu && 2072 (dst_mtu(&rt->dst) < arg->mtu &&
2064 dst_mtu(&rt->dst) == idev->cnf.mtu6))) { 2073 dst_mtu(&rt->dst) == idev->cnf.mtu6))) {
2065 rt->dst.metrics[RTAX_MTU-1] = arg->mtu; 2074 dst_metric_set(&rt->dst, RTAX_MTU, arg->mtu);
2066 rt->dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, arg->mtu);
2067 } 2075 }
2068 return 0; 2076 return 0;
2069} 2077}
@@ -2289,7 +2297,7 @@ static int rt6_fill_node(struct net *net,
2289 NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf); 2297 NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf);
2290 } 2298 }
2291 2299
2292 if (rtnetlink_put_metrics(skb, rt->dst.metrics) < 0) 2300 if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0)
2293 goto nla_put_failure; 2301 goto nla_put_failure;
2294 2302
2295 if (rt->dst.neighbour) 2303 if (rt->dst.neighbour)
@@ -2465,8 +2473,6 @@ static int ip6_route_dev_notify(struct notifier_block *this,
2465 2473
2466#ifdef CONFIG_PROC_FS 2474#ifdef CONFIG_PROC_FS
2467 2475
2468#define RT6_INFO_LEN (32 + 4 + 32 + 4 + 32 + 40 + 5 + 1)
2469
2470struct rt6_proc_arg 2476struct rt6_proc_arg
2471{ 2477{
2472 char *buffer; 2478 char *buffer;
@@ -2682,6 +2688,7 @@ static int __net_init ip6_route_net_init(struct net *net)
2682 net->ipv6.ip6_null_entry->dst.path = 2688 net->ipv6.ip6_null_entry->dst.path =
2683 (struct dst_entry *)net->ipv6.ip6_null_entry; 2689 (struct dst_entry *)net->ipv6.ip6_null_entry;
2684 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops; 2690 net->ipv6.ip6_null_entry->dst.ops = &net->ipv6.ip6_dst_ops;
2691 dst_metric_set(&net->ipv6.ip6_null_entry->dst, RTAX_HOPLIMIT, 255);
2685 2692
2686#ifdef CONFIG_IPV6_MULTIPLE_TABLES 2693#ifdef CONFIG_IPV6_MULTIPLE_TABLES
2687 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template, 2694 net->ipv6.ip6_prohibit_entry = kmemdup(&ip6_prohibit_entry_template,
@@ -2692,6 +2699,7 @@ static int __net_init ip6_route_net_init(struct net *net)
2692 net->ipv6.ip6_prohibit_entry->dst.path = 2699 net->ipv6.ip6_prohibit_entry->dst.path =
2693 (struct dst_entry *)net->ipv6.ip6_prohibit_entry; 2700 (struct dst_entry *)net->ipv6.ip6_prohibit_entry;
2694 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops; 2701 net->ipv6.ip6_prohibit_entry->dst.ops = &net->ipv6.ip6_dst_ops;
2702 dst_metric_set(&net->ipv6.ip6_prohibit_entry->dst, RTAX_HOPLIMIT, 255);
2695 2703
2696 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template, 2704 net->ipv6.ip6_blk_hole_entry = kmemdup(&ip6_blk_hole_entry_template,
2697 sizeof(*net->ipv6.ip6_blk_hole_entry), 2705 sizeof(*net->ipv6.ip6_blk_hole_entry),
@@ -2701,6 +2709,7 @@ static int __net_init ip6_route_net_init(struct net *net)
2701 net->ipv6.ip6_blk_hole_entry->dst.path = 2709 net->ipv6.ip6_blk_hole_entry->dst.path =
2702 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry; 2710 (struct dst_entry *)net->ipv6.ip6_blk_hole_entry;
2703 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops; 2711 net->ipv6.ip6_blk_hole_entry->dst.ops = &net->ipv6.ip6_dst_ops;
2712 dst_metric_set(&net->ipv6.ip6_blk_hole_entry->dst, RTAX_HOPLIMIT, 255);
2704#endif 2713#endif
2705 2714
2706 net->ipv6.sysctl.flush_delay = 0; 2715 net->ipv6.sysctl.flush_delay = 0;
diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index 8c4d00c7cd2b..8ce38f10a547 100644
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -731,10 +731,9 @@ static netdev_tx_t ipip6_tunnel_xmit(struct sk_buff *skb,
731 } 731 }
732 732
733 { 733 {
734 struct flowi fl = { .nl_u = { .ip4_u = 734 struct flowi fl = { .fl4_dst = dst,
735 { .daddr = dst, 735 .fl4_src = tiph->saddr,
736 .saddr = tiph->saddr, 736 .fl4_tos = RT_TOS(tos),
737 .tos = RT_TOS(tos) } },
738 .oif = tunnel->parms.link, 737 .oif = tunnel->parms.link,
739 .proto = IPPROTO_IPV6 }; 738 .proto = IPPROTO_IPV6 };
740 if (ip_route_output_key(dev_net(dev), &rt, &fl)) { 739 if (ip_route_output_key(dev_net(dev), &rt, &fl)) {
@@ -856,10 +855,9 @@ static void ipip6_tunnel_bind_dev(struct net_device *dev)
856 iph = &tunnel->parms.iph; 855 iph = &tunnel->parms.iph;
857 856
858 if (iph->daddr) { 857 if (iph->daddr) {
859 struct flowi fl = { .nl_u = { .ip4_u = 858 struct flowi fl = { .fl4_dst = iph->daddr,
860 { .daddr = iph->daddr, 859 .fl4_src = iph->saddr,
861 .saddr = iph->saddr, 860 .fl4_tos = RT_TOS(iph->tos),
862 .tos = RT_TOS(iph->tos) } },
863 .oif = tunnel->parms.link, 861 .oif = tunnel->parms.link,
864 .proto = IPPROTO_IPV6 }; 862 .proto = IPPROTO_IPV6 };
865 struct rtable *rt; 863 struct rtable *rt;
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 7e41e2cbb85e..20aa95e37359 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -130,6 +130,7 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
130 struct ipv6_pinfo *np = inet6_sk(sk); 130 struct ipv6_pinfo *np = inet6_sk(sk);
131 struct tcp_sock *tp = tcp_sk(sk); 131 struct tcp_sock *tp = tcp_sk(sk);
132 struct in6_addr *saddr = NULL, *final_p, final; 132 struct in6_addr *saddr = NULL, *final_p, final;
133 struct rt6_info *rt;
133 struct flowi fl; 134 struct flowi fl;
134 struct dst_entry *dst; 135 struct dst_entry *dst;
135 int addr_type; 136 int addr_type;
@@ -280,6 +281,26 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
280 sk->sk_gso_type = SKB_GSO_TCPV6; 281 sk->sk_gso_type = SKB_GSO_TCPV6;
281 __ip6_dst_store(sk, dst, NULL, NULL); 282 __ip6_dst_store(sk, dst, NULL, NULL);
282 283
284 rt = (struct rt6_info *) dst;
285 if (tcp_death_row.sysctl_tw_recycle &&
286 !tp->rx_opt.ts_recent_stamp &&
287 ipv6_addr_equal(&rt->rt6i_dst.addr, &np->daddr)) {
288 struct inet_peer *peer = rt6_get_peer(rt);
289 /*
290 * VJ's idea. We save last timestamp seen from
291 * the destination in peer table, when entering state
292 * TIME-WAIT * and initialize rx_opt.ts_recent from it,
293 * when trying new connection.
294 */
295 if (peer) {
296 inet_peer_refcheck(peer);
297 if ((u32)get_seconds() - peer->tcp_ts_stamp <= TCP_PAWS_MSL) {
298 tp->rx_opt.ts_recent_stamp = peer->tcp_ts_stamp;
299 tp->rx_opt.ts_recent = peer->tcp_ts;
300 }
301 }
302 }
303
283 icsk->icsk_ext_hdr_len = 0; 304 icsk->icsk_ext_hdr_len = 0;
284 if (np->opt) 305 if (np->opt)
285 icsk->icsk_ext_hdr_len = (np->opt->opt_flen + 306 icsk->icsk_ext_hdr_len = (np->opt->opt_flen +
@@ -906,12 +927,6 @@ static const struct tcp_request_sock_ops tcp_request_sock_ipv6_ops = {
906}; 927};
907#endif 928#endif
908 929
909static struct timewait_sock_ops tcp6_timewait_sock_ops = {
910 .twsk_obj_size = sizeof(struct tcp6_timewait_sock),
911 .twsk_unique = tcp_twsk_unique,
912 .twsk_destructor= tcp_twsk_destructor,
913};
914
915static void __tcp_v6_send_check(struct sk_buff *skb, 930static void __tcp_v6_send_check(struct sk_buff *skb,
916 struct in6_addr *saddr, struct in6_addr *daddr) 931 struct in6_addr *saddr, struct in6_addr *daddr)
917{ 932{
@@ -1176,6 +1191,7 @@ static int tcp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
1176 struct ipv6_pinfo *np = inet6_sk(sk); 1191 struct ipv6_pinfo *np = inet6_sk(sk);
1177 struct tcp_sock *tp = tcp_sk(sk); 1192 struct tcp_sock *tp = tcp_sk(sk);
1178 __u32 isn = TCP_SKB_CB(skb)->when; 1193 __u32 isn = TCP_SKB_CB(skb)->when;
1194 struct dst_entry *dst = NULL;
1179#ifdef CONFIG_SYN_COOKIES 1195#ifdef CONFIG_SYN_COOKIES
1180 int want_cookie = 0; 1196 int want_cookie = 0;
1181#else 1197#else
@@ -1273,6 +1289,8 @@ static int tcp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
1273 TCP_ECN_create_request(req, tcp_hdr(skb)); 1289 TCP_ECN_create_request(req, tcp_hdr(skb));
1274 1290
1275 if (!isn) { 1291 if (!isn) {
1292 struct inet_peer *peer = NULL;
1293
1276 if (ipv6_opt_accepted(sk, skb) || 1294 if (ipv6_opt_accepted(sk, skb) ||
1277 np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo || 1295 np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo ||
1278 np->rxopt.bits.rxhlim || np->rxopt.bits.rxohlim) { 1296 np->rxopt.bits.rxhlim || np->rxopt.bits.rxohlim) {
@@ -1285,13 +1303,57 @@ static int tcp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
1285 if (!sk->sk_bound_dev_if && 1303 if (!sk->sk_bound_dev_if &&
1286 ipv6_addr_type(&treq->rmt_addr) & IPV6_ADDR_LINKLOCAL) 1304 ipv6_addr_type(&treq->rmt_addr) & IPV6_ADDR_LINKLOCAL)
1287 treq->iif = inet6_iif(skb); 1305 treq->iif = inet6_iif(skb);
1288 if (!want_cookie) { 1306
1289 isn = tcp_v6_init_sequence(skb); 1307 if (want_cookie) {
1290 } else {
1291 isn = cookie_v6_init_sequence(sk, skb, &req->mss); 1308 isn = cookie_v6_init_sequence(sk, skb, &req->mss);
1292 req->cookie_ts = tmp_opt.tstamp_ok; 1309 req->cookie_ts = tmp_opt.tstamp_ok;
1310 goto have_isn;
1293 } 1311 }
1312
1313 /* VJ's idea. We save last timestamp seen
1314 * from the destination in peer table, when entering
1315 * state TIME-WAIT, and check against it before
1316 * accepting new connection request.
1317 *
1318 * If "isn" is not zero, this request hit alive
1319 * timewait bucket, so that all the necessary checks
1320 * are made in the function processing timewait state.
1321 */
1322 if (tmp_opt.saw_tstamp &&
1323 tcp_death_row.sysctl_tw_recycle &&
1324 (dst = inet6_csk_route_req(sk, req)) != NULL &&
1325 (peer = rt6_get_peer((struct rt6_info *)dst)) != NULL &&
1326 ipv6_addr_equal((struct in6_addr *)peer->daddr.a6,
1327 &treq->rmt_addr)) {
1328 inet_peer_refcheck(peer);
1329 if ((u32)get_seconds() - peer->tcp_ts_stamp < TCP_PAWS_MSL &&
1330 (s32)(peer->tcp_ts - req->ts_recent) >
1331 TCP_PAWS_WINDOW) {
1332 NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_PAWSPASSIVEREJECTED);
1333 goto drop_and_release;
1334 }
1335 }
1336 /* Kill the following clause, if you dislike this way. */
1337 else if (!sysctl_tcp_syncookies &&
1338 (sysctl_max_syn_backlog - inet_csk_reqsk_queue_len(sk) <
1339 (sysctl_max_syn_backlog >> 2)) &&
1340 (!peer || !peer->tcp_ts_stamp) &&
1341 (!dst || !dst_metric(dst, RTAX_RTT))) {
1342 /* Without syncookies last quarter of
1343 * backlog is filled with destinations,
1344 * proven to be alive.
1345 * It means that we continue to communicate
1346 * to destinations, already remembered
1347 * to the moment of synflood.
1348 */
1349 LIMIT_NETDEBUG(KERN_DEBUG "TCP: drop open request from %pI6/%u\n",
1350 &treq->rmt_addr, ntohs(tcp_hdr(skb)->source));
1351 goto drop_and_release;
1352 }
1353
1354 isn = tcp_v6_init_sequence(skb);
1294 } 1355 }
1356have_isn:
1295 tcp_rsk(req)->snt_isn = isn; 1357 tcp_rsk(req)->snt_isn = isn;
1296 1358
1297 security_inet_conn_request(sk, skb, req); 1359 security_inet_conn_request(sk, skb, req);
@@ -1304,6 +1366,8 @@ static int tcp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
1304 inet6_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT); 1366 inet6_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
1305 return 0; 1367 return 0;
1306 1368
1369drop_and_release:
1370 dst_release(dst);
1307drop_and_free: 1371drop_and_free:
1308 reqsk_free(req); 1372 reqsk_free(req);
1309drop: 1373drop:
@@ -1382,28 +1446,9 @@ static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
1382 if (sk_acceptq_is_full(sk)) 1446 if (sk_acceptq_is_full(sk))
1383 goto out_overflow; 1447 goto out_overflow;
1384 1448
1385 if (dst == NULL) { 1449 if (!dst) {
1386 struct in6_addr *final_p, final; 1450 dst = inet6_csk_route_req(sk, req);
1387 struct flowi fl; 1451 if (!dst)
1388
1389 memset(&fl, 0, sizeof(fl));
1390 fl.proto = IPPROTO_TCP;
1391 ipv6_addr_copy(&fl.fl6_dst, &treq->rmt_addr);
1392 final_p = fl6_update_dst(&fl, opt, &final);
1393 ipv6_addr_copy(&fl.fl6_src, &treq->loc_addr);
1394 fl.oif = sk->sk_bound_dev_if;
1395 fl.mark = sk->sk_mark;
1396 fl.fl_ip_dport = inet_rsk(req)->rmt_port;
1397 fl.fl_ip_sport = inet_rsk(req)->loc_port;
1398 security_req_classify_flow(req, &fl);
1399
1400 if (ip6_dst_lookup(sk, &dst, &fl))
1401 goto out;
1402
1403 if (final_p)
1404 ipv6_addr_copy(&fl.fl6_dst, final_p);
1405
1406 if ((xfrm_lookup(sock_net(sk), &dst, &fl, sk, 0)) < 0)
1407 goto out; 1452 goto out;
1408 } 1453 }
1409 1454
@@ -1476,7 +1521,7 @@ static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
1476 1521
1477 tcp_mtup_init(newsk); 1522 tcp_mtup_init(newsk);
1478 tcp_sync_mss(newsk, dst_mtu(dst)); 1523 tcp_sync_mss(newsk, dst_mtu(dst));
1479 newtp->advmss = dst_metric(dst, RTAX_ADVMSS); 1524 newtp->advmss = dst_metric_advmss(dst);
1480 tcp_initialize_rcv_mss(newsk); 1525 tcp_initialize_rcv_mss(newsk);
1481 1526
1482 newinet->inet_daddr = newinet->inet_saddr = LOOPBACK4_IPV6; 1527 newinet->inet_daddr = newinet->inet_saddr = LOOPBACK4_IPV6;
@@ -1818,19 +1863,51 @@ do_time_wait:
1818 goto discard_it; 1863 goto discard_it;
1819} 1864}
1820 1865
1821static int tcp_v6_remember_stamp(struct sock *sk) 1866static struct inet_peer *tcp_v6_get_peer(struct sock *sk, bool *release_it)
1822{ 1867{
1823 /* Alas, not yet... */ 1868 struct rt6_info *rt = (struct rt6_info *) __sk_dst_get(sk);
1824 return 0; 1869 struct ipv6_pinfo *np = inet6_sk(sk);
1870 struct inet_peer *peer;
1871
1872 if (!rt ||
1873 !ipv6_addr_equal(&np->daddr, &rt->rt6i_dst.addr)) {
1874 peer = inet_getpeer_v6(&np->daddr, 1);
1875 *release_it = true;
1876 } else {
1877 if (!rt->rt6i_peer)
1878 rt6_bind_peer(rt, 1);
1879 peer = rt->rt6i_peer;
1880 *release_it = false;
1881 }
1882
1883 return peer;
1825} 1884}
1826 1885
1886static void *tcp_v6_tw_get_peer(struct sock *sk)
1887{
1888 struct inet6_timewait_sock *tw6 = inet6_twsk(sk);
1889 struct inet_timewait_sock *tw = inet_twsk(sk);
1890
1891 if (tw->tw_family == AF_INET)
1892 return tcp_v4_tw_get_peer(sk);
1893
1894 return inet_getpeer_v6(&tw6->tw_v6_daddr, 1);
1895}
1896
1897static struct timewait_sock_ops tcp6_timewait_sock_ops = {
1898 .twsk_obj_size = sizeof(struct tcp6_timewait_sock),
1899 .twsk_unique = tcp_twsk_unique,
1900 .twsk_destructor= tcp_twsk_destructor,
1901 .twsk_getpeer = tcp_v6_tw_get_peer,
1902};
1903
1827static const struct inet_connection_sock_af_ops ipv6_specific = { 1904static const struct inet_connection_sock_af_ops ipv6_specific = {
1828 .queue_xmit = inet6_csk_xmit, 1905 .queue_xmit = inet6_csk_xmit,
1829 .send_check = tcp_v6_send_check, 1906 .send_check = tcp_v6_send_check,
1830 .rebuild_header = inet6_sk_rebuild_header, 1907 .rebuild_header = inet6_sk_rebuild_header,
1831 .conn_request = tcp_v6_conn_request, 1908 .conn_request = tcp_v6_conn_request,
1832 .syn_recv_sock = tcp_v6_syn_recv_sock, 1909 .syn_recv_sock = tcp_v6_syn_recv_sock,
1833 .remember_stamp = tcp_v6_remember_stamp, 1910 .get_peer = tcp_v6_get_peer,
1834 .net_header_len = sizeof(struct ipv6hdr), 1911 .net_header_len = sizeof(struct ipv6hdr),
1835 .setsockopt = ipv6_setsockopt, 1912 .setsockopt = ipv6_setsockopt,
1836 .getsockopt = ipv6_getsockopt, 1913 .getsockopt = ipv6_getsockopt,
@@ -1862,7 +1939,7 @@ static const struct inet_connection_sock_af_ops ipv6_mapped = {
1862 .rebuild_header = inet_sk_rebuild_header, 1939 .rebuild_header = inet_sk_rebuild_header,
1863 .conn_request = tcp_v6_conn_request, 1940 .conn_request = tcp_v6_conn_request,
1864 .syn_recv_sock = tcp_v6_syn_recv_sock, 1941 .syn_recv_sock = tcp_v6_syn_recv_sock,
1865 .remember_stamp = tcp_v4_remember_stamp, 1942 .get_peer = tcp_v4_get_peer,
1866 .net_header_len = sizeof(struct iphdr), 1943 .net_header_len = sizeof(struct iphdr),
1867 .setsockopt = ipv6_setsockopt, 1944 .setsockopt = ipv6_setsockopt,
1868 .getsockopt = ipv6_getsockopt, 1945 .getsockopt = ipv6_getsockopt,
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index 91def93bec85..9a009c66c8a3 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -54,8 +54,8 @@ int ipv6_rcv_saddr_equal(const struct sock *sk, const struct sock *sk2)
54{ 54{
55 const struct in6_addr *sk_rcv_saddr6 = &inet6_sk(sk)->rcv_saddr; 55 const struct in6_addr *sk_rcv_saddr6 = &inet6_sk(sk)->rcv_saddr;
56 const struct in6_addr *sk2_rcv_saddr6 = inet6_rcv_saddr(sk2); 56 const struct in6_addr *sk2_rcv_saddr6 = inet6_rcv_saddr(sk2);
57 __be32 sk1_rcv_saddr = inet_sk(sk)->inet_rcv_saddr; 57 __be32 sk1_rcv_saddr = sk_rcv_saddr(sk);
58 __be32 sk2_rcv_saddr = inet_rcv_saddr(sk2); 58 __be32 sk2_rcv_saddr = sk_rcv_saddr(sk2);
59 int sk_ipv6only = ipv6_only_sock(sk); 59 int sk_ipv6only = ipv6_only_sock(sk);
60 int sk2_ipv6only = inet_v6_ipv6only(sk2); 60 int sk2_ipv6only = inet_v6_ipv6only(sk2);
61 int addr_type = ipv6_addr_type(sk_rcv_saddr6); 61 int addr_type = ipv6_addr_type(sk_rcv_saddr6);
@@ -227,7 +227,7 @@ begin:
227 227
228 if (result) { 228 if (result) {
229exact_match: 229exact_match:
230 if (unlikely(!atomic_inc_not_zero(&result->sk_refcnt))) 230 if (unlikely(!atomic_inc_not_zero_hint(&result->sk_refcnt, 2)))
231 result = NULL; 231 result = NULL;
232 else if (unlikely(compute_score2(result, net, saddr, sport, 232 else if (unlikely(compute_score2(result, net, saddr, sport,
233 daddr, hnum, dif) < badness)) { 233 daddr, hnum, dif) < badness)) {
@@ -294,7 +294,7 @@ begin:
294 goto begin; 294 goto begin;
295 295
296 if (result) { 296 if (result) {
297 if (unlikely(!atomic_inc_not_zero(&result->sk_refcnt))) 297 if (unlikely(!atomic_inc_not_zero_hint(&result->sk_refcnt, 2)))
298 result = NULL; 298 result = NULL;
299 else if (unlikely(compute_score(result, net, hnum, saddr, sport, 299 else if (unlikely(compute_score(result, net, hnum, saddr, sport,
300 daddr, dport, dif) < badness)) { 300 daddr, dport, dif) < badness)) {
@@ -602,7 +602,7 @@ static void flush_stack(struct sock **stack, unsigned int count,
602 602
603 sk = stack[i]; 603 sk = stack[i];
604 if (skb1) { 604 if (skb1) {
605 if (sk_rcvqueues_full(sk, skb)) { 605 if (sk_rcvqueues_full(sk, skb1)) {
606 kfree_skb(skb1); 606 kfree_skb(skb1);
607 goto drop; 607 goto drop;
608 } 608 }
@@ -1477,6 +1477,7 @@ struct proto udpv6_prot = {
1477 .compat_setsockopt = compat_udpv6_setsockopt, 1477 .compat_setsockopt = compat_udpv6_setsockopt,
1478 .compat_getsockopt = compat_udpv6_getsockopt, 1478 .compat_getsockopt = compat_udpv6_getsockopt,
1479#endif 1479#endif
1480 .clear_sk = sk_prot_clear_portaddr_nulls,
1480}; 1481};
1481 1482
1482static struct inet_protosw udpv6_protosw = { 1483static struct inet_protosw udpv6_protosw = {
diff --git a/net/ipv6/udplite.c b/net/ipv6/udplite.c
index 5f48fadc27f7..986c4de5292e 100644
--- a/net/ipv6/udplite.c
+++ b/net/ipv6/udplite.c
@@ -55,6 +55,7 @@ struct proto udplitev6_prot = {
55 .compat_setsockopt = compat_udpv6_setsockopt, 55 .compat_setsockopt = compat_udpv6_setsockopt,
56 .compat_getsockopt = compat_udpv6_getsockopt, 56 .compat_getsockopt = compat_udpv6_getsockopt,
57#endif 57#endif
58 .clear_sk = sk_prot_clear_portaddr_nulls,
58}; 59};
59 60
60static struct inet_protosw udplite6_protosw = { 61static struct inet_protosw udplite6_protosw = {
diff --git a/net/ipv6/xfrm6_mode_tunnel.c b/net/ipv6/xfrm6_mode_tunnel.c
index b809812c8d30..645cb968d450 100644
--- a/net/ipv6/xfrm6_mode_tunnel.c
+++ b/net/ipv6/xfrm6_mode_tunnel.c
@@ -14,6 +14,7 @@
14#include <net/dsfield.h> 14#include <net/dsfield.h>
15#include <net/dst.h> 15#include <net/dst.h>
16#include <net/inet_ecn.h> 16#include <net/inet_ecn.h>
17#include <net/ip6_route.h>
17#include <net/ipv6.h> 18#include <net/ipv6.h>
18#include <net/xfrm.h> 19#include <net/xfrm.h>
19 20
@@ -53,7 +54,7 @@ static int xfrm6_mode_tunnel_output(struct xfrm_state *x, struct sk_buff *skb)
53 if (x->props.flags & XFRM_STATE_NOECN) 54 if (x->props.flags & XFRM_STATE_NOECN)
54 dsfield &= ~INET_ECN_MASK; 55 dsfield &= ~INET_ECN_MASK;
55 ipv6_change_dsfield(top_iph, 0, dsfield); 56 ipv6_change_dsfield(top_iph, 0, dsfield);
56 top_iph->hop_limit = dst_metric(dst->child, RTAX_HOPLIMIT); 57 top_iph->hop_limit = ip6_dst_hoplimit(dst->child);
57 ipv6_addr_copy(&top_iph->saddr, (struct in6_addr *)&x->props.saddr); 58 ipv6_addr_copy(&top_iph->saddr, (struct in6_addr *)&x->props.saddr);
58 ipv6_addr_copy(&top_iph->daddr, (struct in6_addr *)&x->id.daddr); 59 ipv6_addr_copy(&top_iph->daddr, (struct in6_addr *)&x->id.daddr);
59 return 0; 60 return 0;
diff --git a/net/ipv6/xfrm6_output.c b/net/ipv6/xfrm6_output.c
index 6434bd5ce088..8e688b3de9ab 100644
--- a/net/ipv6/xfrm6_output.c
+++ b/net/ipv6/xfrm6_output.c
@@ -17,6 +17,7 @@
17#include <linux/netfilter_ipv6.h> 17#include <linux/netfilter_ipv6.h>
18#include <net/dst.h> 18#include <net/dst.h>
19#include <net/ipv6.h> 19#include <net/ipv6.h>
20#include <net/ip6_route.h>
20#include <net/xfrm.h> 21#include <net/xfrm.h>
21 22
22int xfrm6_find_1stfragopt(struct xfrm_state *x, struct sk_buff *skb, 23int xfrm6_find_1stfragopt(struct xfrm_state *x, struct sk_buff *skb,
@@ -88,8 +89,21 @@ static int xfrm6_output_finish(struct sk_buff *skb)
88 return xfrm_output(skb); 89 return xfrm_output(skb);
89} 90}
90 91
92static int __xfrm6_output(struct sk_buff *skb)
93{
94 struct dst_entry *dst = skb_dst(skb);
95 struct xfrm_state *x = dst->xfrm;
96
97 if ((x && x->props.mode == XFRM_MODE_TUNNEL) &&
98 ((skb->len > ip6_skb_dst_mtu(skb) && !skb_is_gso(skb)) ||
99 dst_allfrag(skb_dst(skb)))) {
100 return ip6_fragment(skb, xfrm6_output_finish);
101 }
102 return xfrm6_output_finish(skb);
103}
104
91int xfrm6_output(struct sk_buff *skb) 105int xfrm6_output(struct sk_buff *skb)
92{ 106{
93 return NF_HOOK(NFPROTO_IPV6, NF_INET_POST_ROUTING, skb, NULL, 107 return NF_HOOK(NFPROTO_IPV6, NF_INET_POST_ROUTING, skb, NULL,
94 skb_dst(skb)->dev, xfrm6_output_finish); 108 skb_dst(skb)->dev, __xfrm6_output);
95} 109}
generated by cgit v1.2.2 (git 2.25.0) at 2025-03-05 16:29:26 -0500