aboutsummaryrefslogtreecommitdiffstats
path: root/net/ipv4
diff options
context:
space:
mode:
authorJiri Kosina <jkosina@suse.cz>2010-12-22 12:57:02 -0500
committerJiri Kosina <jkosina@suse.cz>2010-12-22 12:57:02 -0500
commit4b7bd364700d9ac8372eff48832062b936d0793b (patch)
tree0dbf78c95456a0b02d07fcd473281f04a87e266d /net/ipv4
parentc0d8768af260e2cbb4bf659ae6094a262c86b085 (diff)
parent90a8a73c06cc32b609a880d48449d7083327e11a (diff)
Merge branch 'master' into for-next
Conflicts: MAINTAINERS arch/arm/mach-omap2/pm24xx.c drivers/scsi/bfa/bfa_fcpim.c Needed to update to apply fixes for which the old branch was too outdated.
Diffstat (limited to 'net/ipv4')
-rw-r--r--net/ipv4/fib_lookup.h5
-rw-r--r--net/ipv4/fib_trie.c2
-rw-r--r--net/ipv4/icmp.c3
-rw-r--r--net/ipv4/igmp.c4
-rw-r--r--net/ipv4/inet_diag.c27
-rw-r--r--net/ipv4/inet_hashtables.c3
-rw-r--r--net/ipv4/ip_gre.c6
-rw-r--r--net/ipv4/netfilter/arp_tables.c1
-rw-r--r--net/ipv4/netfilter/ip_tables.c1
-rw-r--r--net/ipv4/netfilter/nf_nat_core.c40
-rw-r--r--net/ipv4/proc.c9
-rw-r--r--net/ipv4/sysctl_net_ipv4.c11
-rw-r--r--net/ipv4/tcp.c6
-rw-r--r--net/ipv4/tcp_input.c11
-rw-r--r--net/ipv4/tcp_ipv4.c12
-rw-r--r--net/ipv4/tcp_minisocks.c2
-rw-r--r--net/ipv4/tcp_output.c42
-rw-r--r--net/ipv4/udp.c4
18 files changed, 102 insertions, 87 deletions
diff --git a/net/ipv4/fib_lookup.h b/net/ipv4/fib_lookup.h
index a29edf2219c8..c079cc0ec651 100644
--- a/net/ipv4/fib_lookup.h
+++ b/net/ipv4/fib_lookup.h
@@ -47,11 +47,8 @@ extern int fib_detect_death(struct fib_info *fi, int order,
47static inline void fib_result_assign(struct fib_result *res, 47static inline void fib_result_assign(struct fib_result *res,
48 struct fib_info *fi) 48 struct fib_info *fi)
49{ 49{
50 if (res->fi != NULL) 50 /* we used to play games with refcounts, but we now use RCU */
51 fib_info_put(res->fi);
52 res->fi = fi; 51 res->fi = fi;
53 if (fi != NULL)
54 atomic_inc(&fi->fib_clntref);
55} 52}
56 53
57#endif /* _FIB_LOOKUP_H */ 54#endif /* _FIB_LOOKUP_H */
diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
index 200eb538fbb3..0f280348e0fd 100644
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -365,7 +365,7 @@ static struct tnode *tnode_alloc(size_t size)
365 if (size <= PAGE_SIZE) 365 if (size <= PAGE_SIZE)
366 return kzalloc(size, GFP_KERNEL); 366 return kzalloc(size, GFP_KERNEL);
367 else 367 else
368 return __vmalloc(size, GFP_KERNEL | __GFP_ZERO, PAGE_KERNEL); 368 return vzalloc(size);
369} 369}
370 370
371static void __tnode_vfree(struct work_struct *arg) 371static void __tnode_vfree(struct work_struct *arg)
diff --git a/net/ipv4/icmp.c b/net/ipv4/icmp.c
index 96bc7f9475a3..e5d1a44bcbdf 100644
--- a/net/ipv4/icmp.c
+++ b/net/ipv4/icmp.c
@@ -569,6 +569,9 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
569 /* No need to clone since we're just using its address. */ 569 /* No need to clone since we're just using its address. */
570 rt2 = rt; 570 rt2 = rt;
571 571
572 if (!fl.nl_u.ip4_u.saddr)
573 fl.nl_u.ip4_u.saddr = rt->rt_src;
574
572 err = xfrm_lookup(net, (struct dst_entry **)&rt, &fl, NULL, 0); 575 err = xfrm_lookup(net, (struct dst_entry **)&rt, &fl, NULL, 0);
573 switch (err) { 576 switch (err) {
574 case 0: 577 case 0:
diff --git a/net/ipv4/igmp.c b/net/ipv4/igmp.c
index c8877c6c7216..3c53c2d89e3b 100644
--- a/net/ipv4/igmp.c
+++ b/net/ipv4/igmp.c
@@ -2306,10 +2306,8 @@ void ip_mc_drop_socket(struct sock *sk)
2306 2306
2307 in_dev = inetdev_by_index(net, iml->multi.imr_ifindex); 2307 in_dev = inetdev_by_index(net, iml->multi.imr_ifindex);
2308 (void) ip_mc_leave_src(sk, iml, in_dev); 2308 (void) ip_mc_leave_src(sk, iml, in_dev);
2309 if (in_dev != NULL) { 2309 if (in_dev != NULL)
2310 ip_mc_dec_group(in_dev, iml->multi.imr_multiaddr.s_addr); 2310 ip_mc_dec_group(in_dev, iml->multi.imr_multiaddr.s_addr);
2311 in_dev_put(in_dev);
2312 }
2313 /* decrease mem now to avoid the memleak warning */ 2311 /* decrease mem now to avoid the memleak warning */
2314 atomic_sub(sizeof(*iml), &sk->sk_omem_alloc); 2312 atomic_sub(sizeof(*iml), &sk->sk_omem_alloc);
2315 call_rcu(&iml->rcu, ip_mc_socklist_reclaim); 2313 call_rcu(&iml->rcu, ip_mc_socklist_reclaim);
diff --git a/net/ipv4/inet_diag.c b/net/ipv4/inet_diag.c
index ba8042665849..2ada17129fce 100644
--- a/net/ipv4/inet_diag.c
+++ b/net/ipv4/inet_diag.c
@@ -490,9 +490,11 @@ static int inet_csk_diag_dump(struct sock *sk,
490{ 490{
491 struct inet_diag_req *r = NLMSG_DATA(cb->nlh); 491 struct inet_diag_req *r = NLMSG_DATA(cb->nlh);
492 492
493 if (cb->nlh->nlmsg_len > 4 + NLMSG_SPACE(sizeof(*r))) { 493 if (nlmsg_attrlen(cb->nlh, sizeof(*r))) {
494 struct inet_diag_entry entry; 494 struct inet_diag_entry entry;
495 struct rtattr *bc = (struct rtattr *)(r + 1); 495 const struct nlattr *bc = nlmsg_find_attr(cb->nlh,
496 sizeof(*r),
497 INET_DIAG_REQ_BYTECODE);
496 struct inet_sock *inet = inet_sk(sk); 498 struct inet_sock *inet = inet_sk(sk);
497 499
498 entry.family = sk->sk_family; 500 entry.family = sk->sk_family;
@@ -512,7 +514,7 @@ static int inet_csk_diag_dump(struct sock *sk,
512 entry.dport = ntohs(inet->inet_dport); 514 entry.dport = ntohs(inet->inet_dport);
513 entry.userlocks = sk->sk_userlocks; 515 entry.userlocks = sk->sk_userlocks;
514 516
515 if (!inet_diag_bc_run(RTA_DATA(bc), RTA_PAYLOAD(bc), &entry)) 517 if (!inet_diag_bc_run(nla_data(bc), nla_len(bc), &entry))
516 return 0; 518 return 0;
517 } 519 }
518 520
@@ -527,9 +529,11 @@ static int inet_twsk_diag_dump(struct inet_timewait_sock *tw,
527{ 529{
528 struct inet_diag_req *r = NLMSG_DATA(cb->nlh); 530 struct inet_diag_req *r = NLMSG_DATA(cb->nlh);
529 531
530 if (cb->nlh->nlmsg_len > 4 + NLMSG_SPACE(sizeof(*r))) { 532 if (nlmsg_attrlen(cb->nlh, sizeof(*r))) {
531 struct inet_diag_entry entry; 533 struct inet_diag_entry entry;
532 struct rtattr *bc = (struct rtattr *)(r + 1); 534 const struct nlattr *bc = nlmsg_find_attr(cb->nlh,
535 sizeof(*r),
536 INET_DIAG_REQ_BYTECODE);
533 537
534 entry.family = tw->tw_family; 538 entry.family = tw->tw_family;
535#if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE) 539#if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
@@ -548,7 +552,7 @@ static int inet_twsk_diag_dump(struct inet_timewait_sock *tw,
548 entry.dport = ntohs(tw->tw_dport); 552 entry.dport = ntohs(tw->tw_dport);
549 entry.userlocks = 0; 553 entry.userlocks = 0;
550 554
551 if (!inet_diag_bc_run(RTA_DATA(bc), RTA_PAYLOAD(bc), &entry)) 555 if (!inet_diag_bc_run(nla_data(bc), nla_len(bc), &entry))
552 return 0; 556 return 0;
553 } 557 }
554 558
@@ -618,7 +622,7 @@ static int inet_diag_dump_reqs(struct sk_buff *skb, struct sock *sk,
618 struct inet_diag_req *r = NLMSG_DATA(cb->nlh); 622 struct inet_diag_req *r = NLMSG_DATA(cb->nlh);
619 struct inet_connection_sock *icsk = inet_csk(sk); 623 struct inet_connection_sock *icsk = inet_csk(sk);
620 struct listen_sock *lopt; 624 struct listen_sock *lopt;
621 struct rtattr *bc = NULL; 625 const struct nlattr *bc = NULL;
622 struct inet_sock *inet = inet_sk(sk); 626 struct inet_sock *inet = inet_sk(sk);
623 int j, s_j; 627 int j, s_j;
624 int reqnum, s_reqnum; 628 int reqnum, s_reqnum;
@@ -638,8 +642,9 @@ static int inet_diag_dump_reqs(struct sk_buff *skb, struct sock *sk,
638 if (!lopt || !lopt->qlen) 642 if (!lopt || !lopt->qlen)
639 goto out; 643 goto out;
640 644
641 if (cb->nlh->nlmsg_len > 4 + NLMSG_SPACE(sizeof(*r))) { 645 if (nlmsg_attrlen(cb->nlh, sizeof(*r))) {
642 bc = (struct rtattr *)(r + 1); 646 bc = nlmsg_find_attr(cb->nlh, sizeof(*r),
647 INET_DIAG_REQ_BYTECODE);
643 entry.sport = inet->inet_num; 648 entry.sport = inet->inet_num;
644 entry.userlocks = sk->sk_userlocks; 649 entry.userlocks = sk->sk_userlocks;
645 } 650 }
@@ -672,8 +677,8 @@ static int inet_diag_dump_reqs(struct sk_buff *skb, struct sock *sk,
672 &ireq->rmt_addr; 677 &ireq->rmt_addr;
673 entry.dport = ntohs(ireq->rmt_port); 678 entry.dport = ntohs(ireq->rmt_port);
674 679
675 if (!inet_diag_bc_run(RTA_DATA(bc), 680 if (!inet_diag_bc_run(nla_data(bc),
676 RTA_PAYLOAD(bc), &entry)) 681 nla_len(bc), &entry))
677 continue; 682 continue;
678 } 683 }
679 684
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index 1b344f30b463..3c0369a3a663 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -133,8 +133,7 @@ int __inet_inherit_port(struct sock *sk, struct sock *child)
133 } 133 }
134 } 134 }
135 } 135 }
136 sk_add_bind_node(child, &tb->owners); 136 inet_bind_hash(child, tb, port);
137 inet_csk(child)->icsk_bind_hash = tb;
138 spin_unlock(&head->lock); 137 spin_unlock(&head->lock);
139 138
140 return 0; 139 return 0;
diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index 01087e035b7d..70ff77f02eee 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -1325,7 +1325,6 @@ static void ipgre_fb_tunnel_init(struct net_device *dev)
1325{ 1325{
1326 struct ip_tunnel *tunnel = netdev_priv(dev); 1326 struct ip_tunnel *tunnel = netdev_priv(dev);
1327 struct iphdr *iph = &tunnel->parms.iph; 1327 struct iphdr *iph = &tunnel->parms.iph;
1328 struct ipgre_net *ign = net_generic(dev_net(dev), ipgre_net_id);
1329 1328
1330 tunnel->dev = dev; 1329 tunnel->dev = dev;
1331 strcpy(tunnel->parms.name, dev->name); 1330 strcpy(tunnel->parms.name, dev->name);
@@ -1336,7 +1335,6 @@ static void ipgre_fb_tunnel_init(struct net_device *dev)
1336 tunnel->hlen = sizeof(struct iphdr) + 4; 1335 tunnel->hlen = sizeof(struct iphdr) + 4;
1337 1336
1338 dev_hold(dev); 1337 dev_hold(dev);
1339 rcu_assign_pointer(ign->tunnels_wc[0], tunnel);
1340} 1338}
1341 1339
1342 1340
@@ -1383,10 +1381,12 @@ static int __net_init ipgre_init_net(struct net *net)
1383 if ((err = register_netdev(ign->fb_tunnel_dev))) 1381 if ((err = register_netdev(ign->fb_tunnel_dev)))
1384 goto err_reg_dev; 1382 goto err_reg_dev;
1385 1383
1384 rcu_assign_pointer(ign->tunnels_wc[0],
1385 netdev_priv(ign->fb_tunnel_dev));
1386 return 0; 1386 return 0;
1387 1387
1388err_reg_dev: 1388err_reg_dev:
1389 free_netdev(ign->fb_tunnel_dev); 1389 ipgre_dev_free(ign->fb_tunnel_dev);
1390err_alloc_dev: 1390err_alloc_dev:
1391 return err; 1391 return err;
1392} 1392}
diff --git a/net/ipv4/netfilter/arp_tables.c b/net/ipv4/netfilter/arp_tables.c
index 3cad2591ace0..3fac340a28d5 100644
--- a/net/ipv4/netfilter/arp_tables.c
+++ b/net/ipv4/netfilter/arp_tables.c
@@ -927,6 +927,7 @@ static int get_info(struct net *net, void __user *user,
927 private = &tmp; 927 private = &tmp;
928 } 928 }
929#endif 929#endif
930 memset(&info, 0, sizeof(info));
930 info.valid_hooks = t->valid_hooks; 931 info.valid_hooks = t->valid_hooks;
931 memcpy(info.hook_entry, private->hook_entry, 932 memcpy(info.hook_entry, private->hook_entry,
932 sizeof(info.hook_entry)); 933 sizeof(info.hook_entry));
diff --git a/net/ipv4/netfilter/ip_tables.c b/net/ipv4/netfilter/ip_tables.c
index d31b007a6d80..a846d633b3b6 100644
--- a/net/ipv4/netfilter/ip_tables.c
+++ b/net/ipv4/netfilter/ip_tables.c
@@ -1124,6 +1124,7 @@ static int get_info(struct net *net, void __user *user,
1124 private = &tmp; 1124 private = &tmp;
1125 } 1125 }
1126#endif 1126#endif
1127 memset(&info, 0, sizeof(info));
1127 info.valid_hooks = t->valid_hooks; 1128 info.valid_hooks = t->valid_hooks;
1128 memcpy(info.hook_entry, private->hook_entry, 1129 memcpy(info.hook_entry, private->hook_entry,
1129 sizeof(info.hook_entry)); 1130 sizeof(info.hook_entry));
diff --git a/net/ipv4/netfilter/nf_nat_core.c b/net/ipv4/netfilter/nf_nat_core.c
index 295c97431e43..c04787ce1a71 100644
--- a/net/ipv4/netfilter/nf_nat_core.c
+++ b/net/ipv4/netfilter/nf_nat_core.c
@@ -47,26 +47,6 @@ __nf_nat_proto_find(u_int8_t protonum)
47 return rcu_dereference(nf_nat_protos[protonum]); 47 return rcu_dereference(nf_nat_protos[protonum]);
48} 48}
49 49
50static const struct nf_nat_protocol *
51nf_nat_proto_find_get(u_int8_t protonum)
52{
53 const struct nf_nat_protocol *p;
54
55 rcu_read_lock();
56 p = __nf_nat_proto_find(protonum);
57 if (!try_module_get(p->me))
58 p = &nf_nat_unknown_protocol;
59 rcu_read_unlock();
60
61 return p;
62}
63
64static void
65nf_nat_proto_put(const struct nf_nat_protocol *p)
66{
67 module_put(p->me);
68}
69
70/* We keep an extra hash for each conntrack, for fast searching. */ 50/* We keep an extra hash for each conntrack, for fast searching. */
71static inline unsigned int 51static inline unsigned int
72hash_by_src(const struct net *net, u16 zone, 52hash_by_src(const struct net *net, u16 zone,
@@ -588,6 +568,26 @@ static struct nf_ct_ext_type nat_extend __read_mostly = {
588#include <linux/netfilter/nfnetlink.h> 568#include <linux/netfilter/nfnetlink.h>
589#include <linux/netfilter/nfnetlink_conntrack.h> 569#include <linux/netfilter/nfnetlink_conntrack.h>
590 570
571static const struct nf_nat_protocol *
572nf_nat_proto_find_get(u_int8_t protonum)
573{
574 const struct nf_nat_protocol *p;
575
576 rcu_read_lock();
577 p = __nf_nat_proto_find(protonum);
578 if (!try_module_get(p->me))
579 p = &nf_nat_unknown_protocol;
580 rcu_read_unlock();
581
582 return p;
583}
584
585static void
586nf_nat_proto_put(const struct nf_nat_protocol *p)
587{
588 module_put(p->me);
589}
590
591static const struct nla_policy protonat_nla_policy[CTA_PROTONAT_MAX+1] = { 591static const struct nla_policy protonat_nla_policy[CTA_PROTONAT_MAX+1] = {
592 [CTA_PROTONAT_PORT_MIN] = { .type = NLA_U16 }, 592 [CTA_PROTONAT_PORT_MIN] = { .type = NLA_U16 },
593 [CTA_PROTONAT_PORT_MAX] = { .type = NLA_U16 }, 593 [CTA_PROTONAT_PORT_MAX] = { .type = NLA_U16 },
diff --git a/net/ipv4/proc.c b/net/ipv4/proc.c
index 4ae1f203f7cb..b14ec7d03b6e 100644
--- a/net/ipv4/proc.c
+++ b/net/ipv4/proc.c
@@ -59,13 +59,13 @@ static int sockstat_seq_show(struct seq_file *seq, void *v)
59 local_bh_enable(); 59 local_bh_enable();
60 60
61 socket_seq_show(seq); 61 socket_seq_show(seq);
62 seq_printf(seq, "TCP: inuse %d orphan %d tw %d alloc %d mem %d\n", 62 seq_printf(seq, "TCP: inuse %d orphan %d tw %d alloc %d mem %ld\n",
63 sock_prot_inuse_get(net, &tcp_prot), orphans, 63 sock_prot_inuse_get(net, &tcp_prot), orphans,
64 tcp_death_row.tw_count, sockets, 64 tcp_death_row.tw_count, sockets,
65 atomic_read(&tcp_memory_allocated)); 65 atomic_long_read(&tcp_memory_allocated));
66 seq_printf(seq, "UDP: inuse %d mem %d\n", 66 seq_printf(seq, "UDP: inuse %d mem %ld\n",
67 sock_prot_inuse_get(net, &udp_prot), 67 sock_prot_inuse_get(net, &udp_prot),
68 atomic_read(&udp_memory_allocated)); 68 atomic_long_read(&udp_memory_allocated));
69 seq_printf(seq, "UDPLITE: inuse %d\n", 69 seq_printf(seq, "UDPLITE: inuse %d\n",
70 sock_prot_inuse_get(net, &udplite_prot)); 70 sock_prot_inuse_get(net, &udplite_prot));
71 seq_printf(seq, "RAW: inuse %d\n", 71 seq_printf(seq, "RAW: inuse %d\n",
@@ -253,6 +253,7 @@ static const struct snmp_mib snmp4_net_list[] = {
253 SNMP_MIB_ITEM("TCPMinTTLDrop", LINUX_MIB_TCPMINTTLDROP), 253 SNMP_MIB_ITEM("TCPMinTTLDrop", LINUX_MIB_TCPMINTTLDROP),
254 SNMP_MIB_ITEM("TCPDeferAcceptDrop", LINUX_MIB_TCPDEFERACCEPTDROP), 254 SNMP_MIB_ITEM("TCPDeferAcceptDrop", LINUX_MIB_TCPDEFERACCEPTDROP),
255 SNMP_MIB_ITEM("IPReversePathFilter", LINUX_MIB_IPRPFILTER), 255 SNMP_MIB_ITEM("IPReversePathFilter", LINUX_MIB_IPRPFILTER),
256 SNMP_MIB_ITEM("TCPTimeWaitOverflow", LINUX_MIB_TCPTIMEWAITOVERFLOW),
256 SNMP_MIB_SENTINEL 257 SNMP_MIB_SENTINEL
257}; 258};
258 259
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index d96c1da4b17c..1b4ec21497a4 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -26,6 +26,8 @@ static int zero;
26static int tcp_retr1_max = 255; 26static int tcp_retr1_max = 255;
27static int ip_local_port_range_min[] = { 1, 1 }; 27static int ip_local_port_range_min[] = { 1, 1 };
28static int ip_local_port_range_max[] = { 65535, 65535 }; 28static int ip_local_port_range_max[] = { 65535, 65535 };
29static int tcp_adv_win_scale_min = -31;
30static int tcp_adv_win_scale_max = 31;
29 31
30/* Update system visible IP port range */ 32/* Update system visible IP port range */
31static void set_local_port_range(int range[2]) 33static void set_local_port_range(int range[2])
@@ -398,7 +400,7 @@ static struct ctl_table ipv4_table[] = {
398 .data = &sysctl_tcp_mem, 400 .data = &sysctl_tcp_mem,
399 .maxlen = sizeof(sysctl_tcp_mem), 401 .maxlen = sizeof(sysctl_tcp_mem),
400 .mode = 0644, 402 .mode = 0644,
401 .proc_handler = proc_dointvec 403 .proc_handler = proc_doulongvec_minmax
402 }, 404 },
403 { 405 {
404 .procname = "tcp_wmem", 406 .procname = "tcp_wmem",
@@ -426,7 +428,9 @@ static struct ctl_table ipv4_table[] = {
426 .data = &sysctl_tcp_adv_win_scale, 428 .data = &sysctl_tcp_adv_win_scale,
427 .maxlen = sizeof(int), 429 .maxlen = sizeof(int),
428 .mode = 0644, 430 .mode = 0644,
429 .proc_handler = proc_dointvec 431 .proc_handler = proc_dointvec_minmax,
432 .extra1 = &tcp_adv_win_scale_min,
433 .extra2 = &tcp_adv_win_scale_max,
430 }, 434 },
431 { 435 {
432 .procname = "tcp_tw_reuse", 436 .procname = "tcp_tw_reuse",
@@ -602,8 +606,7 @@ static struct ctl_table ipv4_table[] = {
602 .data = &sysctl_udp_mem, 606 .data = &sysctl_udp_mem,
603 .maxlen = sizeof(sysctl_udp_mem), 607 .maxlen = sizeof(sysctl_udp_mem),
604 .mode = 0644, 608 .mode = 0644,
605 .proc_handler = proc_dointvec_minmax, 609 .proc_handler = proc_doulongvec_minmax,
606 .extra1 = &zero
607 }, 610 },
608 { 611 {
609 .procname = "udp_rmem_min", 612 .procname = "udp_rmem_min",
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 1664a0590bb8..f15c36a706ec 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -282,7 +282,7 @@ int sysctl_tcp_fin_timeout __read_mostly = TCP_FIN_TIMEOUT;
282struct percpu_counter tcp_orphan_count; 282struct percpu_counter tcp_orphan_count;
283EXPORT_SYMBOL_GPL(tcp_orphan_count); 283EXPORT_SYMBOL_GPL(tcp_orphan_count);
284 284
285int sysctl_tcp_mem[3] __read_mostly; 285long sysctl_tcp_mem[3] __read_mostly;
286int sysctl_tcp_wmem[3] __read_mostly; 286int sysctl_tcp_wmem[3] __read_mostly;
287int sysctl_tcp_rmem[3] __read_mostly; 287int sysctl_tcp_rmem[3] __read_mostly;
288 288
@@ -290,7 +290,7 @@ EXPORT_SYMBOL(sysctl_tcp_mem);
290EXPORT_SYMBOL(sysctl_tcp_rmem); 290EXPORT_SYMBOL(sysctl_tcp_rmem);
291EXPORT_SYMBOL(sysctl_tcp_wmem); 291EXPORT_SYMBOL(sysctl_tcp_wmem);
292 292
293atomic_t tcp_memory_allocated; /* Current allocated memory. */ 293atomic_long_t tcp_memory_allocated; /* Current allocated memory. */
294EXPORT_SYMBOL(tcp_memory_allocated); 294EXPORT_SYMBOL(tcp_memory_allocated);
295 295
296/* 296/*
@@ -2246,7 +2246,7 @@ static int do_tcp_setsockopt(struct sock *sk, int level,
2246 /* Values greater than interface MTU won't take effect. However 2246 /* Values greater than interface MTU won't take effect. However
2247 * at the point when this call is done we typically don't yet 2247 * at the point when this call is done we typically don't yet
2248 * know which interface is going to be used */ 2248 * know which interface is going to be used */
2249 if (val < 8 || val > MAX_TCP_WINDOW) { 2249 if (val < TCP_MIN_MSS || val > MAX_TCP_WINDOW) {
2250 err = -EINVAL; 2250 err = -EINVAL;
2251 break; 2251 break;
2252 } 2252 }
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 3357f69e353d..6d8ab1c4efc3 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -259,8 +259,11 @@ static void tcp_fixup_sndbuf(struct sock *sk)
259 int sndmem = tcp_sk(sk)->rx_opt.mss_clamp + MAX_TCP_HEADER + 16 + 259 int sndmem = tcp_sk(sk)->rx_opt.mss_clamp + MAX_TCP_HEADER + 16 +
260 sizeof(struct sk_buff); 260 sizeof(struct sk_buff);
261 261
262 if (sk->sk_sndbuf < 3 * sndmem) 262 if (sk->sk_sndbuf < 3 * sndmem) {
263 sk->sk_sndbuf = min(3 * sndmem, sysctl_tcp_wmem[2]); 263 sk->sk_sndbuf = 3 * sndmem;
264 if (sk->sk_sndbuf > sysctl_tcp_wmem[2])
265 sk->sk_sndbuf = sysctl_tcp_wmem[2];
266 }
264} 267}
265 268
266/* 2. Tuning advertised window (window_clamp, rcv_ssthresh) 269/* 2. Tuning advertised window (window_clamp, rcv_ssthresh)
@@ -396,7 +399,7 @@ static void tcp_clamp_window(struct sock *sk)
396 if (sk->sk_rcvbuf < sysctl_tcp_rmem[2] && 399 if (sk->sk_rcvbuf < sysctl_tcp_rmem[2] &&
397 !(sk->sk_userlocks & SOCK_RCVBUF_LOCK) && 400 !(sk->sk_userlocks & SOCK_RCVBUF_LOCK) &&
398 !tcp_memory_pressure && 401 !tcp_memory_pressure &&
399 atomic_read(&tcp_memory_allocated) < sysctl_tcp_mem[0]) { 402 atomic_long_read(&tcp_memory_allocated) < sysctl_tcp_mem[0]) {
400 sk->sk_rcvbuf = min(atomic_read(&sk->sk_rmem_alloc), 403 sk->sk_rcvbuf = min(atomic_read(&sk->sk_rmem_alloc),
401 sysctl_tcp_rmem[2]); 404 sysctl_tcp_rmem[2]);
402 } 405 }
@@ -4861,7 +4864,7 @@ static int tcp_should_expand_sndbuf(struct sock *sk)
4861 return 0; 4864 return 0;
4862 4865
4863 /* If we are under soft global TCP memory pressure, do not expand. */ 4866 /* If we are under soft global TCP memory pressure, do not expand. */
4864 if (atomic_read(&tcp_memory_allocated) >= sysctl_tcp_mem[0]) 4867 if (atomic_long_read(&tcp_memory_allocated) >= sysctl_tcp_mem[0])
4865 return 0; 4868 return 0;
4866 4869
4867 /* If we filled the congestion window, do not expand. */ 4870 /* If we filled the congestion window, do not expand. */
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 8f8527d41682..e13da6de1fc7 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -415,6 +415,9 @@ void tcp_v4_err(struct sk_buff *icmp_skb, u32 info)
415 !icsk->icsk_backoff) 415 !icsk->icsk_backoff)
416 break; 416 break;
417 417
418 if (sock_owned_by_user(sk))
419 break;
420
418 icsk->icsk_backoff--; 421 icsk->icsk_backoff--;
419 inet_csk(sk)->icsk_rto = __tcp_set_rto(tp) << 422 inet_csk(sk)->icsk_rto = __tcp_set_rto(tp) <<
420 icsk->icsk_backoff; 423 icsk->icsk_backoff;
@@ -429,11 +432,6 @@ void tcp_v4_err(struct sk_buff *icmp_skb, u32 info)
429 if (remaining) { 432 if (remaining) {
430 inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS, 433 inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
431 remaining, TCP_RTO_MAX); 434 remaining, TCP_RTO_MAX);
432 } else if (sock_owned_by_user(sk)) {
433 /* RTO revert clocked out retransmission,
434 * but socket is locked. Will defer. */
435 inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
436 HZ/20, TCP_RTO_MAX);
437 } else { 435 } else {
438 /* RTO revert clocked out retransmission. 436 /* RTO revert clocked out retransmission.
439 * Will retransmit now */ 437 * Will retransmit now */
@@ -2045,7 +2043,9 @@ get_req:
2045 } 2043 }
2046get_sk: 2044get_sk:
2047 sk_nulls_for_each_from(sk, node) { 2045 sk_nulls_for_each_from(sk, node) {
2048 if (sk->sk_family == st->family && net_eq(sock_net(sk), net)) { 2046 if (!net_eq(sock_net(sk), net))
2047 continue;
2048 if (sk->sk_family == st->family) {
2049 cur = sk; 2049 cur = sk;
2050 goto out; 2050 goto out;
2051 } 2051 }
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index 43cf901d7659..a66735f75963 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -347,7 +347,7 @@ void tcp_time_wait(struct sock *sk, int state, int timeo)
347 * socket up. We've got bigger problems than 347 * socket up. We've got bigger problems than
348 * non-graceful socket closings. 348 * non-graceful socket closings.
349 */ 349 */
350 LIMIT_NETDEBUG(KERN_INFO "TCP: time wait bucket table overflow\n"); 350 NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPTIMEWAITOVERFLOW);
351 } 351 }
352 352
353 tcp_update_metrics(sk); 353 tcp_update_metrics(sk);
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index e96152207164..8750d4050b48 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -231,11 +231,10 @@ void tcp_select_initial_window(int __space, __u32 mss,
231 /* when initializing use the value from init_rcv_wnd 231 /* when initializing use the value from init_rcv_wnd
232 * rather than the default from above 232 * rather than the default from above
233 */ 233 */
234 if (init_rcv_wnd && 234 if (init_rcv_wnd)
235 (*rcv_wnd > init_rcv_wnd * mss)) 235 *rcv_wnd = min(*rcv_wnd, init_rcv_wnd * mss);
236 *rcv_wnd = init_rcv_wnd * mss; 236 else
237 else if (*rcv_wnd > init_cwnd * mss) 237 *rcv_wnd = min(*rcv_wnd, init_cwnd * mss);
238 *rcv_wnd = init_cwnd * mss;
239 } 238 }
240 239
241 /* Set the clamp no higher than max representable value */ 240 /* Set the clamp no higher than max representable value */
@@ -386,27 +385,30 @@ struct tcp_out_options {
386 */ 385 */
387static u8 tcp_cookie_size_check(u8 desired) 386static u8 tcp_cookie_size_check(u8 desired)
388{ 387{
389 if (desired > 0) { 388 int cookie_size;
389
390 if (desired > 0)
390 /* previously specified */ 391 /* previously specified */
391 return desired; 392 return desired;
392 } 393
393 if (sysctl_tcp_cookie_size <= 0) { 394 cookie_size = ACCESS_ONCE(sysctl_tcp_cookie_size);
395 if (cookie_size <= 0)
394 /* no default specified */ 396 /* no default specified */
395 return 0; 397 return 0;
396 } 398
397 if (sysctl_tcp_cookie_size <= TCP_COOKIE_MIN) { 399 if (cookie_size <= TCP_COOKIE_MIN)
398 /* value too small, specify minimum */ 400 /* value too small, specify minimum */
399 return TCP_COOKIE_MIN; 401 return TCP_COOKIE_MIN;
400 } 402
401 if (sysctl_tcp_cookie_size >= TCP_COOKIE_MAX) { 403 if (cookie_size >= TCP_COOKIE_MAX)
402 /* value too large, specify maximum */ 404 /* value too large, specify maximum */
403 return TCP_COOKIE_MAX; 405 return TCP_COOKIE_MAX;
404 } 406
405 if (0x1 & sysctl_tcp_cookie_size) { 407 if (cookie_size & 1)
406 /* 8-bit multiple, illegal, fix it */ 408 /* 8-bit multiple, illegal, fix it */
407 return (u8)(sysctl_tcp_cookie_size + 0x1); 409 cookie_size++;
408 } 410
409 return (u8)sysctl_tcp_cookie_size; 411 return (u8)cookie_size;
410} 412}
411 413
412/* Write previously computed TCP options to the packet. 414/* Write previously computed TCP options to the packet.
@@ -1513,6 +1515,7 @@ static int tcp_tso_should_defer(struct sock *sk, struct sk_buff *skb)
1513 struct tcp_sock *tp = tcp_sk(sk); 1515 struct tcp_sock *tp = tcp_sk(sk);
1514 const struct inet_connection_sock *icsk = inet_csk(sk); 1516 const struct inet_connection_sock *icsk = inet_csk(sk);
1515 u32 send_win, cong_win, limit, in_flight; 1517 u32 send_win, cong_win, limit, in_flight;
1518 int win_divisor;
1516 1519
1517 if (TCP_SKB_CB(skb)->flags & TCPHDR_FIN) 1520 if (TCP_SKB_CB(skb)->flags & TCPHDR_FIN)
1518 goto send_now; 1521 goto send_now;
@@ -1544,13 +1547,14 @@ static int tcp_tso_should_defer(struct sock *sk, struct sk_buff *skb)
1544 if ((skb != tcp_write_queue_tail(sk)) && (limit >= skb->len)) 1547 if ((skb != tcp_write_queue_tail(sk)) && (limit >= skb->len))
1545 goto send_now; 1548 goto send_now;
1546 1549
1547 if (sysctl_tcp_tso_win_divisor) { 1550 win_divisor = ACCESS_ONCE(sysctl_tcp_tso_win_divisor);
1551 if (win_divisor) {
1548 u32 chunk = min(tp->snd_wnd, tp->snd_cwnd * tp->mss_cache); 1552 u32 chunk = min(tp->snd_wnd, tp->snd_cwnd * tp->mss_cache);
1549 1553
1550 /* If at least some fraction of a window is available, 1554 /* If at least some fraction of a window is available,
1551 * just use it. 1555 * just use it.
1552 */ 1556 */
1553 chunk /= sysctl_tcp_tso_win_divisor; 1557 chunk /= win_divisor;
1554 if (limit >= chunk) 1558 if (limit >= chunk)
1555 goto send_now; 1559 goto send_now;
1556 } else { 1560 } else {
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index 28cb2d733a3c..5e0a3a582a59 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -110,7 +110,7 @@
110struct udp_table udp_table __read_mostly; 110struct udp_table udp_table __read_mostly;
111EXPORT_SYMBOL(udp_table); 111EXPORT_SYMBOL(udp_table);
112 112
113int sysctl_udp_mem[3] __read_mostly; 113long sysctl_udp_mem[3] __read_mostly;
114EXPORT_SYMBOL(sysctl_udp_mem); 114EXPORT_SYMBOL(sysctl_udp_mem);
115 115
116int sysctl_udp_rmem_min __read_mostly; 116int sysctl_udp_rmem_min __read_mostly;
@@ -119,7 +119,7 @@ EXPORT_SYMBOL(sysctl_udp_rmem_min);
119int sysctl_udp_wmem_min __read_mostly; 119int sysctl_udp_wmem_min __read_mostly;
120EXPORT_SYMBOL(sysctl_udp_wmem_min); 120EXPORT_SYMBOL(sysctl_udp_wmem_min);
121 121
122atomic_t udp_memory_allocated; 122atomic_long_t udp_memory_allocated;
123EXPORT_SYMBOL(udp_memory_allocated); 123EXPORT_SYMBOL(udp_memory_allocated);
124 124
125#define MAX_UDP_PORTS 65536 125#define MAX_UDP_PORTS 65536