aboutsummaryrefslogtreecommitdiffstats
path: root/net
diff options
context:
space:
mode:
Diffstat (limited to 'net')
-rw-r--r--net/8021q/vlan.c4
-rw-r--r--net/8021q/vlan_core.c4
-rw-r--r--net/9p/client.c114
-rw-r--r--net/9p/protocol.c74
-rw-r--r--net/9p/protocol.h6
-rw-r--r--net/9p/trans_virtio.c142
-rw-r--r--net/Kconfig5
-rw-r--r--net/atm/proc.c10
-rw-r--r--net/bluetooth/bnep/netdev.c10
-rw-r--r--net/bluetooth/hci_sysfs.c41
-rw-r--r--net/bluetooth/l2cap.c4
-rw-r--r--net/bluetooth/rfcomm/core.c4
-rw-r--r--net/bluetooth/rfcomm/sock.c4
-rw-r--r--net/bluetooth/sco.c4
-rw-r--r--net/bridge/Kconfig1
-rw-r--r--net/bridge/br_device.c45
-rw-r--r--net/bridge/br_forward.c18
-rw-r--r--net/bridge/br_if.c6
-rw-r--r--net/bridge/br_input.c10
-rw-r--r--net/bridge/br_multicast.c39
-rw-r--r--net/bridge/br_notify.c4
-rw-r--r--net/bridge/br_private.h20
-rw-r--r--net/bridge/br_sysfs_if.c2
-rw-r--r--net/core/dev.c422
-rw-r--r--net/core/dev_mcast.c5
-rw-r--r--net/core/ethtool.c103
-rw-r--r--net/core/fib_rules.c10
-rw-r--r--net/core/neighbour.c2
-rw-r--r--net/core/net-sysfs.c230
-rw-r--r--net/core/netpoll.c4
-rw-r--r--net/core/pktgen.c58
-rw-r--r--net/core/rtnetlink.c44
-rw-r--r--net/core/skbuff.c2
-rw-r--r--net/core/sock.c19
-rw-r--r--net/dccp/ccids/ccid3.c2
-rw-r--r--net/dccp/dccp.h2
-rw-r--r--net/dccp/input.c2
-rw-r--r--net/dccp/ipv4.c8
-rw-r--r--net/dccp/ipv6.c8
-rw-r--r--net/dccp/minisocks.c2
-rw-r--r--net/dccp/proto.c16
-rw-r--r--net/ipv4/Kconfig8
-rw-r--r--net/ipv4/af_inet.c4
-rw-r--r--net/ipv4/devinet.c4
-rw-r--r--net/ipv4/ip_gre.c7
-rw-r--r--net/ipv4/ipconfig.c59
-rw-r--r--net/ipv4/ipmr.c3
-rw-r--r--net/ipv4/proc.c3
-rw-r--r--net/ipv4/route.c80
-rw-r--r--net/ipv4/tcp.c65
-rw-r--r--net/ipv4/tcp_input.c5
-rw-r--r--net/ipv4/tcp_ipv4.c20
-rw-r--r--net/ipv4/tcp_minisocks.c3
-rw-r--r--net/ipv4/tcp_output.c18
-rw-r--r--net/ipv4/tcp_timer.c2
-rw-r--r--net/ipv4/udp.c6
-rw-r--r--net/ipv4/xfrm4_policy.c5
-rw-r--r--net/ipv6/addrconf.c873
-rw-r--r--net/ipv6/fib6_rules.c11
-rw-r--r--net/ipv6/ip6mr.c3
-rw-r--r--net/ipv6/netfilter/ip6t_hbh.c4
-rw-r--r--net/ipv6/route.c13
-rw-r--r--net/ipv6/tcp_ipv6.c7
-rw-r--r--net/ipv6/udp.c28
-rw-r--r--net/ipv6/xfrm6_policy.c3
-rw-r--r--net/irda/ircomm/ircomm_param.c2
-rw-r--r--net/llc/llc_c_ac.c2
-rw-r--r--net/llc/llc_conn.c3
-rw-r--r--net/llc/llc_core.c6
-rw-r--r--net/mac80211/debugfs_netdev.c10
-rw-r--r--net/mac80211/mesh_plink.c2
-rw-r--r--net/mac80211/mlme.c13
-rw-r--r--net/mac80211/sta_info.c1
-rw-r--r--net/netfilter/nf_conntrack_netlink.c4
-rw-r--r--net/netfilter/nf_conntrack_sip.c4
-rw-r--r--net/netfilter/nfnetlink.c4
-rw-r--r--net/netfilter/xt_hashlimit.c2
-rw-r--r--net/netlink/af_netlink.c19
-rw-r--r--net/packet/af_packet.c6
-rw-r--r--net/phonet/pn_dev.c3
-rw-r--r--net/phonet/pn_netlink.c3
-rw-r--r--net/rds/af_rds.c9
-rw-r--r--net/rds/cong.c2
-rw-r--r--net/rds/ib_cm.c3
-rw-r--r--net/rds/ib_rdma.c5
-rw-r--r--net/rds/ib_recv.c4
-rw-r--r--net/rds/ib_send.c20
-rw-r--r--net/rds/iw_cm.c4
-rw-r--r--net/rds/iw_recv.c4
-rw-r--r--net/rds/iw_send.c3
-rw-r--r--net/rds/loop.c7
-rw-r--r--net/rds/rdma.c4
-rw-r--r--net/rds/rdma_transport.c5
-rw-r--r--net/rds/rds.h2
-rw-r--r--net/rds/send.c38
-rw-r--r--net/rds/tcp_recv.c1
-rw-r--r--net/rds/tcp_send.c4
-rw-r--r--net/rds/threads.c2
-rw-r--r--net/sched/act_api.c45
-rw-r--r--net/sched/cls_api.c30
-rw-r--r--net/sched/cls_u32.c4
-rw-r--r--net/sched/sch_api.c112
-rw-r--r--net/sctp/input.c42
-rw-r--r--net/sctp/ipv6.c2
-rw-r--r--net/sctp/sm_sideeffect.c2
-rw-r--r--net/sctp/socket.c3
-rw-r--r--net/sunrpc/addr.c8
-rw-r--r--net/sunrpc/auth_gss/auth_gss.c11
-rw-r--r--net/sunrpc/auth_gss/gss_spkm3_token.c2
-rw-r--r--net/sunrpc/bc_svc.c2
-rw-r--r--net/sunrpc/rpc_pipe.c9
-rw-r--r--net/sunrpc/svc.c6
-rw-r--r--net/sunrpc/svc_xprt.c27
-rw-r--r--net/sunrpc/svcauth_unix.c49
-rw-r--r--net/sunrpc/svcsock.c3
-rw-r--r--net/sunrpc/xprtrdma/transport.c7
-rw-r--r--net/sunrpc/xprtsock.c18
-rw-r--r--net/tipc/bcast.c35
-rw-r--r--net/tipc/bearer.c37
-rw-r--r--net/tipc/bearer.h2
-rw-r--r--net/tipc/link.c28
-rw-r--r--net/tipc/net.c27
-rw-r--r--net/tipc/node.c2
-rw-r--r--net/tipc/ref.c26
-rw-r--r--net/tipc/socket.c6
-rw-r--r--net/tipc/subscr.c57
-rw-r--r--net/tipc/subscr.h2
-rw-r--r--net/wimax/op-reset.c2
-rw-r--r--net/wimax/op-state-get.c2
-rw-r--r--net/x25/x25_dev.c2
-rw-r--r--net/xfrm/xfrm_policy.c7
131 files changed, 2288 insertions, 1274 deletions
diff --git a/net/8021q/vlan.c b/net/8021q/vlan.c
index 453512266ea1..c39a5f41169c 100644
--- a/net/8021q/vlan.c
+++ b/net/8021q/vlan.c
@@ -530,6 +530,10 @@ static int vlan_device_event(struct notifier_block *unused, unsigned long event,
530 } 530 }
531 unregister_netdevice_many(&list); 531 unregister_netdevice_many(&list);
532 break; 532 break;
533
534 case NETDEV_PRE_TYPE_CHANGE:
535 /* Forbid underlaying device to change its type. */
536 return NOTIFY_BAD;
533 } 537 }
534 538
535out: 539out:
diff --git a/net/8021q/vlan_core.c b/net/8021q/vlan_core.c
index c0316e0ca6e8..c584a0af77d3 100644
--- a/net/8021q/vlan_core.c
+++ b/net/8021q/vlan_core.c
@@ -11,7 +11,7 @@ int __vlan_hwaccel_rx(struct sk_buff *skb, struct vlan_group *grp,
11 if (netpoll_rx(skb)) 11 if (netpoll_rx(skb))
12 return NET_RX_DROP; 12 return NET_RX_DROP;
13 13
14 if (skb_bond_should_drop(skb)) 14 if (skb_bond_should_drop(skb, ACCESS_ONCE(skb->dev->master)))
15 goto drop; 15 goto drop;
16 16
17 skb->skb_iif = skb->dev->ifindex; 17 skb->skb_iif = skb->dev->ifindex;
@@ -83,7 +83,7 @@ vlan_gro_common(struct napi_struct *napi, struct vlan_group *grp,
83{ 83{
84 struct sk_buff *p; 84 struct sk_buff *p;
85 85
86 if (skb_bond_should_drop(skb)) 86 if (skb_bond_should_drop(skb, ACCESS_ONCE(skb->dev->master)))
87 goto drop; 87 goto drop;
88 88
89 skb->skb_iif = skb->dev->ifindex; 89 skb->skb_iif = skb->dev->ifindex;
diff --git a/net/9p/client.c b/net/9p/client.c
index 09d4f1e2e4a8..e3e5bf4469ce 100644
--- a/net/9p/client.c
+++ b/net/9p/client.c
@@ -46,6 +46,7 @@ enum {
46 Opt_msize, 46 Opt_msize,
47 Opt_trans, 47 Opt_trans,
48 Opt_legacy, 48 Opt_legacy,
49 Opt_version,
49 Opt_err, 50 Opt_err,
50}; 51};
51 52
@@ -53,9 +54,42 @@ static const match_table_t tokens = {
53 {Opt_msize, "msize=%u"}, 54 {Opt_msize, "msize=%u"},
54 {Opt_legacy, "noextend"}, 55 {Opt_legacy, "noextend"},
55 {Opt_trans, "trans=%s"}, 56 {Opt_trans, "trans=%s"},
57 {Opt_version, "version=%s"},
56 {Opt_err, NULL}, 58 {Opt_err, NULL},
57}; 59};
58 60
61inline int p9_is_proto_dotl(struct p9_client *clnt)
62{
63 return (clnt->proto_version == p9_proto_2000L);
64}
65EXPORT_SYMBOL(p9_is_proto_dotl);
66
67inline int p9_is_proto_dotu(struct p9_client *clnt)
68{
69 return (clnt->proto_version == p9_proto_2000u);
70}
71EXPORT_SYMBOL(p9_is_proto_dotu);
72
73/* Interpret mount option for protocol version */
74static unsigned char get_protocol_version(const substring_t *name)
75{
76 unsigned char version = -EINVAL;
77 if (!strncmp("9p2000", name->from, name->to-name->from)) {
78 version = p9_proto_legacy;
79 P9_DPRINTK(P9_DEBUG_9P, "Protocol version: Legacy\n");
80 } else if (!strncmp("9p2000.u", name->from, name->to-name->from)) {
81 version = p9_proto_2000u;
82 P9_DPRINTK(P9_DEBUG_9P, "Protocol version: 9P2000.u\n");
83 } else if (!strncmp("9p2000.L", name->from, name->to-name->from)) {
84 version = p9_proto_2000L;
85 P9_DPRINTK(P9_DEBUG_9P, "Protocol version: 9P2000.L\n");
86 } else {
87 P9_DPRINTK(P9_DEBUG_ERROR, "Unknown protocol version %s. ",
88 name->from);
89 }
90 return version;
91}
92
59static struct p9_req_t * 93static struct p9_req_t *
60p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...); 94p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...);
61 95
@@ -75,7 +109,7 @@ static int parse_opts(char *opts, struct p9_client *clnt)
75 int option; 109 int option;
76 int ret = 0; 110 int ret = 0;
77 111
78 clnt->dotu = 1; 112 clnt->proto_version = p9_proto_2000u;
79 clnt->msize = 8192; 113 clnt->msize = 8192;
80 114
81 if (!opts) 115 if (!opts)
@@ -118,7 +152,13 @@ static int parse_opts(char *opts, struct p9_client *clnt)
118 } 152 }
119 break; 153 break;
120 case Opt_legacy: 154 case Opt_legacy:
121 clnt->dotu = 0; 155 clnt->proto_version = p9_proto_legacy;
156 break;
157 case Opt_version:
158 ret = get_protocol_version(&args[0]);
159 if (ret == -EINVAL)
160 goto free_and_return;
161 clnt->proto_version = ret;
122 break; 162 break;
123 default: 163 default:
124 continue; 164 continue;
@@ -410,14 +450,15 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req)
410 int ecode; 450 int ecode;
411 char *ename; 451 char *ename;
412 452
413 err = p9pdu_readf(req->rc, c->dotu, "s?d", &ename, &ecode); 453 err = p9pdu_readf(req->rc, c->proto_version, "s?d",
454 &ename, &ecode);
414 if (err) { 455 if (err) {
415 P9_DPRINTK(P9_DEBUG_ERROR, "couldn't parse error%d\n", 456 P9_DPRINTK(P9_DEBUG_ERROR, "couldn't parse error%d\n",
416 err); 457 err);
417 return err; 458 return err;
418 } 459 }
419 460
420 if (c->dotu) 461 if (p9_is_proto_dotu(c))
421 err = -ecode; 462 err = -ecode;
422 463
423 if (!err || !IS_ERR_VALUE(err)) 464 if (!err || !IS_ERR_VALUE(err))
@@ -515,7 +556,7 @@ p9_client_rpc(struct p9_client *c, int8_t type, const char *fmt, ...)
515 /* marshall the data */ 556 /* marshall the data */
516 p9pdu_prepare(req->tc, tag, type); 557 p9pdu_prepare(req->tc, tag, type);
517 va_start(ap, fmt); 558 va_start(ap, fmt);
518 err = p9pdu_vwritef(req->tc, c->dotu, fmt, ap); 559 err = p9pdu_vwritef(req->tc, c->proto_version, fmt, ap);
519 va_end(ap); 560 va_end(ap);
520 p9pdu_finalize(req->tc); 561 p9pdu_finalize(req->tc);
521 562
@@ -627,14 +668,31 @@ int p9_client_version(struct p9_client *c)
627 char *version; 668 char *version;
628 int msize; 669 int msize;
629 670
630 P9_DPRINTK(P9_DEBUG_9P, ">>> TVERSION msize %d extended %d\n", 671 P9_DPRINTK(P9_DEBUG_9P, ">>> TVERSION msize %d protocol %d\n",
631 c->msize, c->dotu); 672 c->msize, c->proto_version);
632 req = p9_client_rpc(c, P9_TVERSION, "ds", c->msize, 673
633 c->dotu ? "9P2000.u" : "9P2000"); 674 switch (c->proto_version) {
675 case p9_proto_2000L:
676 req = p9_client_rpc(c, P9_TVERSION, "ds",
677 c->msize, "9P2000.L");
678 break;
679 case p9_proto_2000u:
680 req = p9_client_rpc(c, P9_TVERSION, "ds",
681 c->msize, "9P2000.u");
682 break;
683 case p9_proto_legacy:
684 req = p9_client_rpc(c, P9_TVERSION, "ds",
685 c->msize, "9P2000");
686 break;
687 default:
688 return -EINVAL;
689 break;
690 }
691
634 if (IS_ERR(req)) 692 if (IS_ERR(req))
635 return PTR_ERR(req); 693 return PTR_ERR(req);
636 694
637 err = p9pdu_readf(req->rc, c->dotu, "ds", &msize, &version); 695 err = p9pdu_readf(req->rc, c->proto_version, "ds", &msize, &version);
638 if (err) { 696 if (err) {
639 P9_DPRINTK(P9_DEBUG_9P, "version error %d\n", err); 697 P9_DPRINTK(P9_DEBUG_9P, "version error %d\n", err);
640 p9pdu_dump(1, req->rc); 698 p9pdu_dump(1, req->rc);
@@ -642,10 +700,12 @@ int p9_client_version(struct p9_client *c)
642 } 700 }
643 701
644 P9_DPRINTK(P9_DEBUG_9P, "<<< RVERSION msize %d %s\n", msize, version); 702 P9_DPRINTK(P9_DEBUG_9P, "<<< RVERSION msize %d %s\n", msize, version);
645 if (!memcmp(version, "9P2000.u", 8)) 703 if (!strncmp(version, "9P2000.L", 8))
646 c->dotu = 1; 704 c->proto_version = p9_proto_2000L;
647 else if (!memcmp(version, "9P2000", 6)) 705 else if (!strncmp(version, "9P2000.u", 8))
648 c->dotu = 0; 706 c->proto_version = p9_proto_2000u;
707 else if (!strncmp(version, "9P2000", 6))
708 c->proto_version = p9_proto_legacy;
649 else { 709 else {
650 err = -EREMOTEIO; 710 err = -EREMOTEIO;
651 goto error; 711 goto error;
@@ -700,8 +760,8 @@ struct p9_client *p9_client_create(const char *dev_name, char *options)
700 goto put_trans; 760 goto put_trans;
701 } 761 }
702 762
703 P9_DPRINTK(P9_DEBUG_MUX, "clnt %p trans %p msize %d dotu %d\n", 763 P9_DPRINTK(P9_DEBUG_MUX, "clnt %p trans %p msize %d protocol %d\n",
704 clnt, clnt->trans_mod, clnt->msize, clnt->dotu); 764 clnt, clnt->trans_mod, clnt->msize, clnt->proto_version);
705 765
706 err = clnt->trans_mod->create(clnt, dev_name, options); 766 err = clnt->trans_mod->create(clnt, dev_name, options);
707 if (err) 767 if (err)
@@ -784,7 +844,7 @@ struct p9_fid *p9_client_attach(struct p9_client *clnt, struct p9_fid *afid,
784 goto error; 844 goto error;
785 } 845 }
786 846
787 err = p9pdu_readf(req->rc, clnt->dotu, "Q", &qid); 847 err = p9pdu_readf(req->rc, clnt->proto_version, "Q", &qid);
788 if (err) { 848 if (err) {
789 p9pdu_dump(1, req->rc); 849 p9pdu_dump(1, req->rc);
790 p9_free_req(clnt, req); 850 p9_free_req(clnt, req);
@@ -833,7 +893,7 @@ p9_client_auth(struct p9_client *clnt, char *uname, u32 n_uname, char *aname)
833 goto error; 893 goto error;
834 } 894 }
835 895
836 err = p9pdu_readf(req->rc, clnt->dotu, "Q", &qid); 896 err = p9pdu_readf(req->rc, clnt->proto_version, "Q", &qid);
837 if (err) { 897 if (err) {
838 p9pdu_dump(1, req->rc); 898 p9pdu_dump(1, req->rc);
839 p9_free_req(clnt, req); 899 p9_free_req(clnt, req);
@@ -891,7 +951,7 @@ struct p9_fid *p9_client_walk(struct p9_fid *oldfid, int nwname, char **wnames,
891 goto error; 951 goto error;
892 } 952 }
893 953
894 err = p9pdu_readf(req->rc, clnt->dotu, "R", &nwqids, &wqids); 954 err = p9pdu_readf(req->rc, clnt->proto_version, "R", &nwqids, &wqids);
895 if (err) { 955 if (err) {
896 p9pdu_dump(1, req->rc); 956 p9pdu_dump(1, req->rc);
897 p9_free_req(clnt, req); 957 p9_free_req(clnt, req);
@@ -952,7 +1012,7 @@ int p9_client_open(struct p9_fid *fid, int mode)
952 goto error; 1012 goto error;
953 } 1013 }
954 1014
955 err = p9pdu_readf(req->rc, clnt->dotu, "Qd", &qid, &iounit); 1015 err = p9pdu_readf(req->rc, clnt->proto_version, "Qd", &qid, &iounit);
956 if (err) { 1016 if (err) {
957 p9pdu_dump(1, req->rc); 1017 p9pdu_dump(1, req->rc);
958 goto free_and_error; 1018 goto free_and_error;
@@ -997,7 +1057,7 @@ int p9_client_fcreate(struct p9_fid *fid, char *name, u32 perm, int mode,
997 goto error; 1057 goto error;
998 } 1058 }
999 1059
1000 err = p9pdu_readf(req->rc, clnt->dotu, "Qd", &qid, &iounit); 1060 err = p9pdu_readf(req->rc, clnt->proto_version, "Qd", &qid, &iounit);
1001 if (err) { 1061 if (err) {
1002 p9pdu_dump(1, req->rc); 1062 p9pdu_dump(1, req->rc);
1003 goto free_and_error; 1063 goto free_and_error;
@@ -1098,7 +1158,7 @@ p9_client_read(struct p9_fid *fid, char *data, char __user *udata, u64 offset,
1098 goto error; 1158 goto error;
1099 } 1159 }
1100 1160
1101 err = p9pdu_readf(req->rc, clnt->dotu, "D", &count, &dataptr); 1161 err = p9pdu_readf(req->rc, clnt->proto_version, "D", &count, &dataptr);
1102 if (err) { 1162 if (err) {
1103 p9pdu_dump(1, req->rc); 1163 p9pdu_dump(1, req->rc);
1104 goto free_and_error; 1164 goto free_and_error;
@@ -1159,7 +1219,7 @@ p9_client_write(struct p9_fid *fid, char *data, const char __user *udata,
1159 goto error; 1219 goto error;
1160 } 1220 }
1161 1221
1162 err = p9pdu_readf(req->rc, clnt->dotu, "d", &count); 1222 err = p9pdu_readf(req->rc, clnt->proto_version, "d", &count);
1163 if (err) { 1223 if (err) {
1164 p9pdu_dump(1, req->rc); 1224 p9pdu_dump(1, req->rc);
1165 goto free_and_error; 1225 goto free_and_error;
@@ -1199,7 +1259,7 @@ struct p9_wstat *p9_client_stat(struct p9_fid *fid)
1199 goto error; 1259 goto error;
1200 } 1260 }
1201 1261
1202 err = p9pdu_readf(req->rc, clnt->dotu, "wS", &ignored, ret); 1262 err = p9pdu_readf(req->rc, clnt->proto_version, "wS", &ignored, ret);
1203 if (err) { 1263 if (err) {
1204 p9pdu_dump(1, req->rc); 1264 p9pdu_dump(1, req->rc);
1205 p9_free_req(clnt, req); 1265 p9_free_req(clnt, req);
@@ -1226,7 +1286,7 @@ error:
1226} 1286}
1227EXPORT_SYMBOL(p9_client_stat); 1287EXPORT_SYMBOL(p9_client_stat);
1228 1288
1229static int p9_client_statsize(struct p9_wstat *wst, int optional) 1289static int p9_client_statsize(struct p9_wstat *wst, int proto_version)
1230{ 1290{
1231 int ret; 1291 int ret;
1232 1292
@@ -1245,7 +1305,7 @@ static int p9_client_statsize(struct p9_wstat *wst, int optional)
1245 if (wst->muid) 1305 if (wst->muid)
1246 ret += strlen(wst->muid); 1306 ret += strlen(wst->muid);
1247 1307
1248 if (optional) { 1308 if (proto_version == p9_proto_2000u) {
1249 ret += 2+4+4+4; /* extension[s] n_uid[4] n_gid[4] n_muid[4] */ 1309 ret += 2+4+4+4; /* extension[s] n_uid[4] n_gid[4] n_muid[4] */
1250 if (wst->extension) 1310 if (wst->extension)
1251 ret += strlen(wst->extension); 1311 ret += strlen(wst->extension);
@@ -1262,7 +1322,7 @@ int p9_client_wstat(struct p9_fid *fid, struct p9_wstat *wst)
1262 1322
1263 err = 0; 1323 err = 0;
1264 clnt = fid->clnt; 1324 clnt = fid->clnt;
1265 wst->size = p9_client_statsize(wst, clnt->dotu); 1325 wst->size = p9_client_statsize(wst, clnt->proto_version);
1266 P9_DPRINTK(P9_DEBUG_9P, ">>> TWSTAT fid %d\n", fid->fid); 1326 P9_DPRINTK(P9_DEBUG_9P, ">>> TWSTAT fid %d\n", fid->fid);
1267 P9_DPRINTK(P9_DEBUG_9P, 1327 P9_DPRINTK(P9_DEBUG_9P,
1268 " sz=%x type=%x dev=%x qid=%x.%llx.%x\n" 1328 " sz=%x type=%x dev=%x qid=%x.%llx.%x\n"
diff --git a/net/9p/protocol.c b/net/9p/protocol.c
index fc70147c771e..94f5a8f65e9c 100644
--- a/net/9p/protocol.c
+++ b/net/9p/protocol.c
@@ -52,7 +52,7 @@
52#endif 52#endif
53 53
54static int 54static int
55p9pdu_writef(struct p9_fcall *pdu, int optional, const char *fmt, ...); 55p9pdu_writef(struct p9_fcall *pdu, int proto_version, const char *fmt, ...);
56 56
57#ifdef CONFIG_NET_9P_DEBUG 57#ifdef CONFIG_NET_9P_DEBUG
58void 58void
@@ -144,7 +144,8 @@ pdu_write_u(struct p9_fcall *pdu, const char __user *udata, size_t size)
144*/ 144*/
145 145
146static int 146static int
147p9pdu_vreadf(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap) 147p9pdu_vreadf(struct p9_fcall *pdu, int proto_version, const char *fmt,
148 va_list ap)
148{ 149{
149 const char *ptr; 150 const char *ptr;
150 int errcode = 0; 151 int errcode = 0;
@@ -194,7 +195,8 @@ p9pdu_vreadf(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
194 int16_t len; 195 int16_t len;
195 int size; 196 int size;
196 197
197 errcode = p9pdu_readf(pdu, optional, "w", &len); 198 errcode = p9pdu_readf(pdu, proto_version,
199 "w", &len);
198 if (errcode) 200 if (errcode)
199 break; 201 break;
200 202
@@ -217,7 +219,7 @@ p9pdu_vreadf(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
217 struct p9_qid *qid = 219 struct p9_qid *qid =
218 va_arg(ap, struct p9_qid *); 220 va_arg(ap, struct p9_qid *);
219 221
220 errcode = p9pdu_readf(pdu, optional, "bdq", 222 errcode = p9pdu_readf(pdu, proto_version, "bdq",
221 &qid->type, &qid->version, 223 &qid->type, &qid->version,
222 &qid->path); 224 &qid->path);
223 } 225 }
@@ -230,7 +232,7 @@ p9pdu_vreadf(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
230 stbuf->n_uid = stbuf->n_gid = stbuf->n_muid = 232 stbuf->n_uid = stbuf->n_gid = stbuf->n_muid =
231 -1; 233 -1;
232 errcode = 234 errcode =
233 p9pdu_readf(pdu, optional, 235 p9pdu_readf(pdu, proto_version,
234 "wwdQdddqssss?sddd", 236 "wwdQdddqssss?sddd",
235 &stbuf->size, &stbuf->type, 237 &stbuf->size, &stbuf->type,
236 &stbuf->dev, &stbuf->qid, 238 &stbuf->dev, &stbuf->qid,
@@ -250,7 +252,7 @@ p9pdu_vreadf(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
250 void **data = va_arg(ap, void **); 252 void **data = va_arg(ap, void **);
251 253
252 errcode = 254 errcode =
253 p9pdu_readf(pdu, optional, "d", count); 255 p9pdu_readf(pdu, proto_version, "d", count);
254 if (!errcode) { 256 if (!errcode) {
255 *count = 257 *count =
256 MIN(*count, 258 MIN(*count,
@@ -263,8 +265,8 @@ p9pdu_vreadf(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
263 int16_t *nwname = va_arg(ap, int16_t *); 265 int16_t *nwname = va_arg(ap, int16_t *);
264 char ***wnames = va_arg(ap, char ***); 266 char ***wnames = va_arg(ap, char ***);
265 267
266 errcode = 268 errcode = p9pdu_readf(pdu, proto_version,
267 p9pdu_readf(pdu, optional, "w", nwname); 269 "w", nwname);
268 if (!errcode) { 270 if (!errcode) {
269 *wnames = 271 *wnames =
270 kmalloc(sizeof(char *) * *nwname, 272 kmalloc(sizeof(char *) * *nwname,
@@ -278,7 +280,8 @@ p9pdu_vreadf(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
278 280
279 for (i = 0; i < *nwname; i++) { 281 for (i = 0; i < *nwname; i++) {
280 errcode = 282 errcode =
281 p9pdu_readf(pdu, optional, 283 p9pdu_readf(pdu,
284 proto_version,
282 "s", 285 "s",
283 &(*wnames)[i]); 286 &(*wnames)[i]);
284 if (errcode) 287 if (errcode)
@@ -306,7 +309,7 @@ p9pdu_vreadf(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
306 *wqids = NULL; 309 *wqids = NULL;
307 310
308 errcode = 311 errcode =
309 p9pdu_readf(pdu, optional, "w", nwqid); 312 p9pdu_readf(pdu, proto_version, "w", nwqid);
310 if (!errcode) { 313 if (!errcode) {
311 *wqids = 314 *wqids =
312 kmalloc(*nwqid * 315 kmalloc(*nwqid *
@@ -321,7 +324,8 @@ p9pdu_vreadf(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
321 324
322 for (i = 0; i < *nwqid; i++) { 325 for (i = 0; i < *nwqid; i++) {
323 errcode = 326 errcode =
324 p9pdu_readf(pdu, optional, 327 p9pdu_readf(pdu,
328 proto_version,
325 "Q", 329 "Q",
326 &(*wqids)[i]); 330 &(*wqids)[i]);
327 if (errcode) 331 if (errcode)
@@ -336,7 +340,7 @@ p9pdu_vreadf(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
336 } 340 }
337 break; 341 break;
338 case '?': 342 case '?':
339 if (!optional) 343 if (proto_version != p9_proto_2000u)
340 return 0; 344 return 0;
341 break; 345 break;
342 default: 346 default:
@@ -352,7 +356,8 @@ p9pdu_vreadf(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
352} 356}
353 357
354int 358int
355p9pdu_vwritef(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap) 359p9pdu_vwritef(struct p9_fcall *pdu, int proto_version, const char *fmt,
360 va_list ap)
356{ 361{
357 const char *ptr; 362 const char *ptr;
358 int errcode = 0; 363 int errcode = 0;
@@ -389,7 +394,8 @@ p9pdu_vwritef(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
389 if (sptr) 394 if (sptr)
390 len = MIN(strlen(sptr), USHORT_MAX); 395 len = MIN(strlen(sptr), USHORT_MAX);
391 396
392 errcode = p9pdu_writef(pdu, optional, "w", len); 397 errcode = p9pdu_writef(pdu, proto_version,
398 "w", len);
393 if (!errcode && pdu_write(pdu, sptr, len)) 399 if (!errcode && pdu_write(pdu, sptr, len))
394 errcode = -EFAULT; 400 errcode = -EFAULT;
395 } 401 }
@@ -398,7 +404,7 @@ p9pdu_vwritef(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
398 const struct p9_qid *qid = 404 const struct p9_qid *qid =
399 va_arg(ap, const struct p9_qid *); 405 va_arg(ap, const struct p9_qid *);
400 errcode = 406 errcode =
401 p9pdu_writef(pdu, optional, "bdq", 407 p9pdu_writef(pdu, proto_version, "bdq",
402 qid->type, qid->version, 408 qid->type, qid->version,
403 qid->path); 409 qid->path);
404 } break; 410 } break;
@@ -406,7 +412,7 @@ p9pdu_vwritef(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
406 const struct p9_wstat *stbuf = 412 const struct p9_wstat *stbuf =
407 va_arg(ap, const struct p9_wstat *); 413 va_arg(ap, const struct p9_wstat *);
408 errcode = 414 errcode =
409 p9pdu_writef(pdu, optional, 415 p9pdu_writef(pdu, proto_version,
410 "wwdQdddqssss?sddd", 416 "wwdQdddqssss?sddd",
411 stbuf->size, stbuf->type, 417 stbuf->size, stbuf->type,
412 stbuf->dev, &stbuf->qid, 418 stbuf->dev, &stbuf->qid,
@@ -421,8 +427,8 @@ p9pdu_vwritef(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
421 int32_t count = va_arg(ap, int32_t); 427 int32_t count = va_arg(ap, int32_t);
422 const void *data = va_arg(ap, const void *); 428 const void *data = va_arg(ap, const void *);
423 429
424 errcode = 430 errcode = p9pdu_writef(pdu, proto_version, "d",
425 p9pdu_writef(pdu, optional, "d", count); 431 count);
426 if (!errcode && pdu_write(pdu, data, count)) 432 if (!errcode && pdu_write(pdu, data, count))
427 errcode = -EFAULT; 433 errcode = -EFAULT;
428 } 434 }
@@ -431,8 +437,8 @@ p9pdu_vwritef(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
431 int32_t count = va_arg(ap, int32_t); 437 int32_t count = va_arg(ap, int32_t);
432 const char __user *udata = 438 const char __user *udata =
433 va_arg(ap, const void __user *); 439 va_arg(ap, const void __user *);
434 errcode = 440 errcode = p9pdu_writef(pdu, proto_version, "d",
435 p9pdu_writef(pdu, optional, "d", count); 441 count);
436 if (!errcode && pdu_write_u(pdu, udata, count)) 442 if (!errcode && pdu_write_u(pdu, udata, count))
437 errcode = -EFAULT; 443 errcode = -EFAULT;
438 } 444 }
@@ -441,14 +447,15 @@ p9pdu_vwritef(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
441 int16_t nwname = va_arg(ap, int); 447 int16_t nwname = va_arg(ap, int);
442 const char **wnames = va_arg(ap, const char **); 448 const char **wnames = va_arg(ap, const char **);
443 449
444 errcode = 450 errcode = p9pdu_writef(pdu, proto_version, "w",
445 p9pdu_writef(pdu, optional, "w", nwname); 451 nwname);
446 if (!errcode) { 452 if (!errcode) {
447 int i; 453 int i;
448 454
449 for (i = 0; i < nwname; i++) { 455 for (i = 0; i < nwname; i++) {
450 errcode = 456 errcode =
451 p9pdu_writef(pdu, optional, 457 p9pdu_writef(pdu,
458 proto_version,
452 "s", 459 "s",
453 wnames[i]); 460 wnames[i]);
454 if (errcode) 461 if (errcode)
@@ -462,14 +469,15 @@ p9pdu_vwritef(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
462 struct p9_qid *wqids = 469 struct p9_qid *wqids =
463 va_arg(ap, struct p9_qid *); 470 va_arg(ap, struct p9_qid *);
464 471
465 errcode = 472 errcode = p9pdu_writef(pdu, proto_version, "w",
466 p9pdu_writef(pdu, optional, "w", nwqid); 473 nwqid);
467 if (!errcode) { 474 if (!errcode) {
468 int i; 475 int i;
469 476
470 for (i = 0; i < nwqid; i++) { 477 for (i = 0; i < nwqid; i++) {
471 errcode = 478 errcode =
472 p9pdu_writef(pdu, optional, 479 p9pdu_writef(pdu,
480 proto_version,
473 "Q", 481 "Q",
474 &wqids[i]); 482 &wqids[i]);
475 if (errcode) 483 if (errcode)
@@ -479,7 +487,7 @@ p9pdu_vwritef(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
479 } 487 }
480 break; 488 break;
481 case '?': 489 case '?':
482 if (!optional) 490 if (proto_version != p9_proto_2000u)
483 return 0; 491 return 0;
484 break; 492 break;
485 default: 493 default:
@@ -494,32 +502,32 @@ p9pdu_vwritef(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap)
494 return errcode; 502 return errcode;
495} 503}
496 504
497int p9pdu_readf(struct p9_fcall *pdu, int optional, const char *fmt, ...) 505int p9pdu_readf(struct p9_fcall *pdu, int proto_version, const char *fmt, ...)
498{ 506{
499 va_list ap; 507 va_list ap;
500 int ret; 508 int ret;
501 509
502 va_start(ap, fmt); 510 va_start(ap, fmt);
503 ret = p9pdu_vreadf(pdu, optional, fmt, ap); 511 ret = p9pdu_vreadf(pdu, proto_version, fmt, ap);
504 va_end(ap); 512 va_end(ap);
505 513
506 return ret; 514 return ret;
507} 515}
508 516
509static int 517static int
510p9pdu_writef(struct p9_fcall *pdu, int optional, const char *fmt, ...) 518p9pdu_writef(struct p9_fcall *pdu, int proto_version, const char *fmt, ...)
511{ 519{
512 va_list ap; 520 va_list ap;
513 int ret; 521 int ret;
514 522
515 va_start(ap, fmt); 523 va_start(ap, fmt);
516 ret = p9pdu_vwritef(pdu, optional, fmt, ap); 524 ret = p9pdu_vwritef(pdu, proto_version, fmt, ap);
517 va_end(ap); 525 va_end(ap);
518 526
519 return ret; 527 return ret;
520} 528}
521 529
522int p9stat_read(char *buf, int len, struct p9_wstat *st, int dotu) 530int p9stat_read(char *buf, int len, struct p9_wstat *st, int proto_version)
523{ 531{
524 struct p9_fcall fake_pdu; 532 struct p9_fcall fake_pdu;
525 int ret; 533 int ret;
@@ -529,7 +537,7 @@ int p9stat_read(char *buf, int len, struct p9_wstat *st, int dotu)
529 fake_pdu.sdata = buf; 537 fake_pdu.sdata = buf;
530 fake_pdu.offset = 0; 538 fake_pdu.offset = 0;
531 539
532 ret = p9pdu_readf(&fake_pdu, dotu, "S", st); 540 ret = p9pdu_readf(&fake_pdu, proto_version, "S", st);
533 if (ret) { 541 if (ret) {
534 P9_DPRINTK(P9_DEBUG_9P, "<<< p9stat_read failed: %d\n", ret); 542 P9_DPRINTK(P9_DEBUG_9P, "<<< p9stat_read failed: %d\n", ret);
535 p9pdu_dump(1, &fake_pdu); 543 p9pdu_dump(1, &fake_pdu);
diff --git a/net/9p/protocol.h b/net/9p/protocol.h
index ccde462e7ac5..2431c0f38d56 100644
--- a/net/9p/protocol.h
+++ b/net/9p/protocol.h
@@ -25,9 +25,9 @@
25 * 25 *
26 */ 26 */
27 27
28int 28int p9pdu_vwritef(struct p9_fcall *pdu, int proto_version, const char *fmt,
29p9pdu_vwritef(struct p9_fcall *pdu, int optional, const char *fmt, va_list ap); 29 va_list ap);
30int p9pdu_readf(struct p9_fcall *pdu, int optional, const char *fmt, ...); 30int p9pdu_readf(struct p9_fcall *pdu, int proto_version, const char *fmt, ...);
31int p9pdu_prepare(struct p9_fcall *pdu, int16_t tag, int8_t type); 31int p9pdu_prepare(struct p9_fcall *pdu, int16_t tag, int8_t type);
32int p9pdu_finalize(struct p9_fcall *pdu); 32int p9pdu_finalize(struct p9_fcall *pdu);
33void p9pdu_dump(int, struct p9_fcall *); 33void p9pdu_dump(int, struct p9_fcall *);
diff --git a/net/9p/trans_virtio.c b/net/9p/trans_virtio.c
index cb50f4ae5eef..afde1a89fbb3 100644
--- a/net/9p/trans_virtio.c
+++ b/net/9p/trans_virtio.c
@@ -49,8 +49,6 @@
49 49
50/* a single mutex to manage channel initialization and attachment */ 50/* a single mutex to manage channel initialization and attachment */
51static DEFINE_MUTEX(virtio_9p_lock); 51static DEFINE_MUTEX(virtio_9p_lock);
52/* global which tracks highest initialized channel */
53static int chan_index;
54 52
55/** 53/**
56 * struct virtio_chan - per-instance transport information 54 * struct virtio_chan - per-instance transport information
@@ -68,8 +66,7 @@ static int chan_index;
68 * 66 *
69 */ 67 */
70 68
71static struct virtio_chan { 69struct virtio_chan {
72 bool initialized;
73 bool inuse; 70 bool inuse;
74 71
75 spinlock_t lock; 72 spinlock_t lock;
@@ -80,7 +77,17 @@ static struct virtio_chan {
80 77
81 /* Scatterlist: can be too big for stack. */ 78 /* Scatterlist: can be too big for stack. */
82 struct scatterlist sg[VIRTQUEUE_NUM]; 79 struct scatterlist sg[VIRTQUEUE_NUM];
83} channels[MAX_9P_CHAN]; 80
81 int tag_len;
82 /*
83 * tag name to identify a mount Non-null terminated
84 */
85 char *tag;
86
87 struct list_head chan_list;
88};
89
90static struct list_head virtio_chan_list;
84 91
85/* How many bytes left in this page. */ 92/* How many bytes left in this page. */
86static unsigned int rest_of_page(void *data) 93static unsigned int rest_of_page(void *data)
@@ -213,30 +220,38 @@ p9_virtio_request(struct p9_client *client, struct p9_req_t *req)
213 return 0; 220 return 0;
214} 221}
215 222
223static ssize_t p9_mount_tag_show(struct device *dev,
224 struct device_attribute *attr, char *buf)
225{
226 struct virtio_chan *chan;
227 struct virtio_device *vdev;
228
229 vdev = dev_to_virtio(dev);
230 chan = vdev->priv;
231
232 return snprintf(buf, chan->tag_len + 1, "%s", chan->tag);
233}
234
235static DEVICE_ATTR(mount_tag, 0444, p9_mount_tag_show, NULL);
236
216/** 237/**
217 * p9_virtio_probe - probe for existence of 9P virtio channels 238 * p9_virtio_probe - probe for existence of 9P virtio channels
218 * @vdev: virtio device to probe 239 * @vdev: virtio device to probe
219 * 240 *
220 * This probes for existing virtio channels. At present only 241 * This probes for existing virtio channels.
221 * a single channel is in use, so in the future more work may need
222 * to be done here.
223 * 242 *
224 */ 243 */
225 244
226static int p9_virtio_probe(struct virtio_device *vdev) 245static int p9_virtio_probe(struct virtio_device *vdev)
227{ 246{
247 __u16 tag_len;
248 char *tag;
228 int err; 249 int err;
229 struct virtio_chan *chan; 250 struct virtio_chan *chan;
230 int index;
231 251
232 mutex_lock(&virtio_9p_lock); 252 chan = kmalloc(sizeof(struct virtio_chan), GFP_KERNEL);
233 index = chan_index++; 253 if (!chan) {
234 chan = &channels[index]; 254 printk(KERN_ERR "9p: Failed to allocate virtio 9P channel\n");
235 mutex_unlock(&virtio_9p_lock);
236
237 if (chan_index > MAX_9P_CHAN) {
238 printk(KERN_ERR "9p: virtio: Maximum channels exceeded\n");
239 BUG();
240 err = -ENOMEM; 255 err = -ENOMEM;
241 goto fail; 256 goto fail;
242 } 257 }
@@ -255,15 +270,37 @@ static int p9_virtio_probe(struct virtio_device *vdev)
255 sg_init_table(chan->sg, VIRTQUEUE_NUM); 270 sg_init_table(chan->sg, VIRTQUEUE_NUM);
256 271
257 chan->inuse = false; 272 chan->inuse = false;
258 chan->initialized = true; 273 if (virtio_has_feature(vdev, VIRTIO_9P_MOUNT_TAG)) {
274 vdev->config->get(vdev,
275 offsetof(struct virtio_9p_config, tag_len),
276 &tag_len, sizeof(tag_len));
277 } else {
278 err = -EINVAL;
279 goto out_free_vq;
280 }
281 tag = kmalloc(tag_len, GFP_KERNEL);
282 if (!tag) {
283 err = -ENOMEM;
284 goto out_free_vq;
285 }
286 vdev->config->get(vdev, offsetof(struct virtio_9p_config, tag),
287 tag, tag_len);
288 chan->tag = tag;
289 chan->tag_len = tag_len;
290 err = sysfs_create_file(&(vdev->dev.kobj), &dev_attr_mount_tag.attr);
291 if (err) {
292 kfree(tag);
293 goto out_free_vq;
294 }
295 mutex_lock(&virtio_9p_lock);
296 list_add_tail(&chan->chan_list, &virtio_chan_list);
297 mutex_unlock(&virtio_9p_lock);
259 return 0; 298 return 0;
260 299
261out_free_vq: 300out_free_vq:
262 vdev->config->del_vqs(vdev); 301 vdev->config->del_vqs(vdev);
302 kfree(chan);
263fail: 303fail:
264 mutex_lock(&virtio_9p_lock);
265 chan_index--;
266 mutex_unlock(&virtio_9p_lock);
267 return err; 304 return err;
268} 305}
269 306
@@ -280,35 +317,31 @@ fail:
280 * We use a simple reference count mechanism to ensure that only a single 317 * We use a simple reference count mechanism to ensure that only a single
281 * mount has a channel open at a time. 318 * mount has a channel open at a time.
282 * 319 *
283 * Bugs: doesn't allow identification of a specific channel
284 * to allocate, channels are allocated sequentially. This was
285 * a pragmatic decision to get things rolling, but ideally some
286 * way of identifying the channel to attach to would be nice
287 * if we are going to support multiple channels.
288 *
289 */ 320 */
290 321
291static int 322static int
292p9_virtio_create(struct p9_client *client, const char *devname, char *args) 323p9_virtio_create(struct p9_client *client, const char *devname, char *args)
293{ 324{
294 struct virtio_chan *chan = channels; 325 struct virtio_chan *chan;
295 int index = 0; 326 int ret = -ENOENT;
327 int found = 0;
296 328
297 mutex_lock(&virtio_9p_lock); 329 mutex_lock(&virtio_9p_lock);
298 while (index < MAX_9P_CHAN) { 330 list_for_each_entry(chan, &virtio_chan_list, chan_list) {
299 if (chan->initialized && !chan->inuse) { 331 if (!strncmp(devname, chan->tag, chan->tag_len)) {
300 chan->inuse = true; 332 if (!chan->inuse) {
301 break; 333 chan->inuse = true;
302 } else { 334 found = 1;
303 index++; 335 break;
304 chan = &channels[index]; 336 }
337 ret = -EBUSY;
305 } 338 }
306 } 339 }
307 mutex_unlock(&virtio_9p_lock); 340 mutex_unlock(&virtio_9p_lock);
308 341
309 if (index >= MAX_9P_CHAN) { 342 if (!found) {
310 printk(KERN_ERR "9p: no channels available\n"); 343 printk(KERN_ERR "9p: no channels available\n");
311 return -ENODEV; 344 return ret;
312 } 345 }
313 346
314 client->trans = (void *)chan; 347 client->trans = (void *)chan;
@@ -329,11 +362,15 @@ static void p9_virtio_remove(struct virtio_device *vdev)
329 struct virtio_chan *chan = vdev->priv; 362 struct virtio_chan *chan = vdev->priv;
330 363
331 BUG_ON(chan->inuse); 364 BUG_ON(chan->inuse);
365 vdev->config->del_vqs(vdev);
366
367 mutex_lock(&virtio_9p_lock);
368 list_del(&chan->chan_list);
369 mutex_unlock(&virtio_9p_lock);
370 sysfs_remove_file(&(vdev->dev.kobj), &dev_attr_mount_tag.attr);
371 kfree(chan->tag);
372 kfree(chan);
332 373
333 if (chan->initialized) {
334 vdev->config->del_vqs(vdev);
335 chan->initialized = false;
336 }
337} 374}
338 375
339static struct virtio_device_id id_table[] = { 376static struct virtio_device_id id_table[] = {
@@ -341,13 +378,19 @@ static struct virtio_device_id id_table[] = {
341 { 0 }, 378 { 0 },
342}; 379};
343 380
381static unsigned int features[] = {
382 VIRTIO_9P_MOUNT_TAG,
383};
384
344/* The standard "struct lguest_driver": */ 385/* The standard "struct lguest_driver": */
345static struct virtio_driver p9_virtio_drv = { 386static struct virtio_driver p9_virtio_drv = {
346 .driver.name = KBUILD_MODNAME, 387 .feature_table = features,
347 .driver.owner = THIS_MODULE, 388 .feature_table_size = ARRAY_SIZE(features),
348 .id_table = id_table, 389 .driver.name = KBUILD_MODNAME,
349 .probe = p9_virtio_probe, 390 .driver.owner = THIS_MODULE,
350 .remove = p9_virtio_remove, 391 .id_table = id_table,
392 .probe = p9_virtio_probe,
393 .remove = p9_virtio_remove,
351}; 394};
352 395
353static struct p9_trans_module p9_virtio_trans = { 396static struct p9_trans_module p9_virtio_trans = {
@@ -364,10 +407,7 @@ static struct p9_trans_module p9_virtio_trans = {
364/* The standard init function */ 407/* The standard init function */
365static int __init p9_virtio_init(void) 408static int __init p9_virtio_init(void)
366{ 409{
367 int count; 410 INIT_LIST_HEAD(&virtio_chan_list);
368
369 for (count = 0; count < MAX_9P_CHAN; count++)
370 channels[count].initialized = false;
371 411
372 v9fs_register_trans(&p9_virtio_trans); 412 v9fs_register_trans(&p9_virtio_trans);
373 return register_virtio_driver(&p9_virtio_drv); 413 return register_virtio_driver(&p9_virtio_drv);
diff --git a/net/Kconfig b/net/Kconfig
index 041c35edb763..68514644ce91 100644
--- a/net/Kconfig
+++ b/net/Kconfig
@@ -203,6 +203,11 @@ source "net/ieee802154/Kconfig"
203source "net/sched/Kconfig" 203source "net/sched/Kconfig"
204source "net/dcb/Kconfig" 204source "net/dcb/Kconfig"
205 205
206config RPS
207 boolean
208 depends on SMP && SYSFS
209 default y
210
206menu "Network testing" 211menu "Network testing"
207 212
208config NET_PKTGEN 213config NET_PKTGEN
diff --git a/net/atm/proc.c b/net/atm/proc.c
index 7a96b2376bd7..f188a399c679 100644
--- a/net/atm/proc.c
+++ b/net/atm/proc.c
@@ -406,7 +406,6 @@ EXPORT_SYMBOL(atm_proc_root);
406 406
407int atm_proc_dev_register(struct atm_dev *dev) 407int atm_proc_dev_register(struct atm_dev *dev)
408{ 408{
409 int digits, num;
410 int error; 409 int error;
411 410
412 /* No proc info */ 411 /* No proc info */
@@ -414,16 +413,9 @@ int atm_proc_dev_register(struct atm_dev *dev)
414 return 0; 413 return 0;
415 414
416 error = -ENOMEM; 415 error = -ENOMEM;
417 digits = 0; 416 dev->proc_name = kasprintf(GFP_KERNEL, "%s:%d", dev->type, dev->number);
418 for (num = dev->number; num; num /= 10)
419 digits++;
420 if (!digits)
421 digits++;
422
423 dev->proc_name = kmalloc(strlen(dev->type) + digits + 2, GFP_KERNEL);
424 if (!dev->proc_name) 417 if (!dev->proc_name)
425 goto err_out; 418 goto err_out;
426 sprintf(dev->proc_name, "%s:%d", dev->type, dev->number);
427 419
428 dev->proc_entry = proc_create_data(dev->proc_name, 0, atm_proc_root, 420 dev->proc_entry = proc_create_data(dev->proc_name, 0, atm_proc_root,
429 &proc_atm_dev_ops, dev); 421 &proc_atm_dev_ops, dev);
diff --git a/net/bluetooth/bnep/netdev.c b/net/bluetooth/bnep/netdev.c
index b6234b73c4cf..326ab453edb7 100644
--- a/net/bluetooth/bnep/netdev.c
+++ b/net/bluetooth/bnep/netdev.c
@@ -87,7 +87,7 @@ static void bnep_net_set_mc_list(struct net_device *dev)
87 memcpy(__skb_put(skb, ETH_ALEN), dev->broadcast, ETH_ALEN); 87 memcpy(__skb_put(skb, ETH_ALEN), dev->broadcast, ETH_ALEN);
88 r->len = htons(ETH_ALEN * 2); 88 r->len = htons(ETH_ALEN * 2);
89 } else { 89 } else {
90 struct dev_mc_list *dmi = dev->mc_list; 90 struct dev_mc_list *dmi;
91 int i, len = skb->len; 91 int i, len = skb->len;
92 92
93 if (dev->flags & IFF_BROADCAST) { 93 if (dev->flags & IFF_BROADCAST) {
@@ -97,12 +97,12 @@ static void bnep_net_set_mc_list(struct net_device *dev)
97 97
98 /* FIXME: We should group addresses here. */ 98 /* FIXME: We should group addresses here. */
99 99
100 for (i = 0; 100 i = 0;
101 i < netdev_mc_count(dev) && i < BNEP_MAX_MULTICAST_FILTERS; 101 netdev_for_each_mc_addr(dmi, dev) {
102 i++) { 102 if (i == BNEP_MAX_MULTICAST_FILTERS)
103 break;
103 memcpy(__skb_put(skb, ETH_ALEN), dmi->dmi_addr, ETH_ALEN); 104 memcpy(__skb_put(skb, ETH_ALEN), dmi->dmi_addr, ETH_ALEN);
104 memcpy(__skb_put(skb, ETH_ALEN), dmi->dmi_addr, ETH_ALEN); 105 memcpy(__skb_put(skb, ETH_ALEN), dmi->dmi_addr, ETH_ALEN);
105 dmi = dmi->next;
106 } 106 }
107 r->len = htons(skb->len - len); 107 r->len = htons(skb->len - len);
108 } 108 }
diff --git a/net/bluetooth/hci_sysfs.c b/net/bluetooth/hci_sysfs.c
index 1a79a6c7e30e..cafb55b0cea5 100644
--- a/net/bluetooth/hci_sysfs.c
+++ b/net/bluetooth/hci_sysfs.c
@@ -3,6 +3,7 @@
3#include <linux/kernel.h> 3#include <linux/kernel.h>
4#include <linux/init.h> 4#include <linux/init.h>
5#include <linux/debugfs.h> 5#include <linux/debugfs.h>
6#include <linux/seq_file.h>
6 7
7#include <net/bluetooth/bluetooth.h> 8#include <net/bluetooth/bluetooth.h>
8#include <net/bluetooth/hci_core.h> 9#include <net/bluetooth/hci_core.h>
@@ -405,20 +406,11 @@ static struct device_type bt_host = {
405 .release = bt_host_release, 406 .release = bt_host_release,
406}; 407};
407 408
408static int inquiry_cache_open(struct inode *inode, struct file *file) 409static int inquiry_cache_show(struct seq_file *f, void *p)
409{
410 file->private_data = inode->i_private;
411 return 0;
412}
413
414static ssize_t inquiry_cache_read(struct file *file, char __user *userbuf,
415 size_t count, loff_t *ppos)
416{ 410{
417 struct hci_dev *hdev = file->private_data; 411 struct hci_dev *hdev = f->private;
418 struct inquiry_cache *cache = &hdev->inq_cache; 412 struct inquiry_cache *cache = &hdev->inq_cache;
419 struct inquiry_entry *e; 413 struct inquiry_entry *e;
420 char buf[4096];
421 int n = 0;
422 414
423 hci_dev_lock_bh(hdev); 415 hci_dev_lock_bh(hdev);
424 416
@@ -426,23 +418,30 @@ static ssize_t inquiry_cache_read(struct file *file, char __user *userbuf,
426 struct inquiry_data *data = &e->data; 418 struct inquiry_data *data = &e->data;
427 bdaddr_t bdaddr; 419 bdaddr_t bdaddr;
428 baswap(&bdaddr, &data->bdaddr); 420 baswap(&bdaddr, &data->bdaddr);
429 n += sprintf(buf + n, "%s %d %d %d 0x%.2x%.2x%.2x 0x%.4x %d %d %u\n", 421 seq_printf(f, "%s %d %d %d 0x%.2x%.2x%.2x 0x%.4x %d %d %u\n",
430 batostr(&bdaddr), 422 batostr(&bdaddr),
431 data->pscan_rep_mode, data->pscan_period_mode, 423 data->pscan_rep_mode, data->pscan_period_mode,
432 data->pscan_mode, data->dev_class[2], 424 data->pscan_mode, data->dev_class[2],
433 data->dev_class[1], data->dev_class[0], 425 data->dev_class[1], data->dev_class[0],
434 __le16_to_cpu(data->clock_offset), 426 __le16_to_cpu(data->clock_offset),
435 data->rssi, data->ssp_mode, e->timestamp); 427 data->rssi, data->ssp_mode, e->timestamp);
436 } 428 }
437 429
438 hci_dev_unlock_bh(hdev); 430 hci_dev_unlock_bh(hdev);
439 431
440 return simple_read_from_buffer(userbuf, count, ppos, buf, n); 432 return 0;
433}
434
435static int inquiry_cache_open(struct inode *inode, struct file *file)
436{
437 return single_open(file, inquiry_cache_show, inode->i_private);
441} 438}
442 439
443static const struct file_operations inquiry_cache_fops = { 440static const struct file_operations inquiry_cache_fops = {
444 .open = inquiry_cache_open, 441 .open = inquiry_cache_open,
445 .read = inquiry_cache_read, 442 .read = seq_read,
443 .llseek = seq_lseek,
444 .release = single_release,
446}; 445};
447 446
448int hci_register_sysfs(struct hci_dev *hdev) 447int hci_register_sysfs(struct hci_dev *hdev)
diff --git a/net/bluetooth/l2cap.c b/net/bluetooth/l2cap.c
index 400efa26ddba..4db7ae2fe07d 100644
--- a/net/bluetooth/l2cap.c
+++ b/net/bluetooth/l2cap.c
@@ -3937,7 +3937,9 @@ drop:
3937 return 0; 3937 return 0;
3938} 3938}
3939 3939
3940static ssize_t l2cap_sysfs_show(struct class *dev, char *buf) 3940static ssize_t l2cap_sysfs_show(struct class *dev,
3941 struct class_attribute *attr,
3942 char *buf)
3941{ 3943{
3942 struct sock *sk; 3944 struct sock *sk;
3943 struct hlist_node *node; 3945 struct hlist_node *node;
diff --git a/net/bluetooth/rfcomm/core.c b/net/bluetooth/rfcomm/core.c
index 89f4a59eb82b..db8a68e1a5ba 100644
--- a/net/bluetooth/rfcomm/core.c
+++ b/net/bluetooth/rfcomm/core.c
@@ -2098,7 +2098,9 @@ static struct hci_cb rfcomm_cb = {
2098 .security_cfm = rfcomm_security_cfm 2098 .security_cfm = rfcomm_security_cfm
2099}; 2099};
2100 2100
2101static ssize_t rfcomm_dlc_sysfs_show(struct class *dev, char *buf) 2101static ssize_t rfcomm_dlc_sysfs_show(struct class *dev,
2102 struct class_attribute *attr,
2103 char *buf)
2102{ 2104{
2103 struct rfcomm_session *s; 2105 struct rfcomm_session *s;
2104 struct list_head *pp, *p; 2106 struct list_head *pp, *p;
diff --git a/net/bluetooth/rfcomm/sock.c b/net/bluetooth/rfcomm/sock.c
index 4b5968dda673..ca87d6ac6a20 100644
--- a/net/bluetooth/rfcomm/sock.c
+++ b/net/bluetooth/rfcomm/sock.c
@@ -1061,7 +1061,9 @@ done:
1061 return result; 1061 return result;
1062} 1062}
1063 1063
1064static ssize_t rfcomm_sock_sysfs_show(struct class *dev, char *buf) 1064static ssize_t rfcomm_sock_sysfs_show(struct class *dev,
1065 struct class_attribute *attr,
1066 char *buf)
1065{ 1067{
1066 struct sock *sk; 1068 struct sock *sk;
1067 struct hlist_node *node; 1069 struct hlist_node *node;
diff --git a/net/bluetooth/sco.c b/net/bluetooth/sco.c
index dd8f6ec57dce..f93b939539bc 100644
--- a/net/bluetooth/sco.c
+++ b/net/bluetooth/sco.c
@@ -953,7 +953,9 @@ drop:
953 return 0; 953 return 0;
954} 954}
955 955
956static ssize_t sco_sysfs_show(struct class *dev, char *buf) 956static ssize_t sco_sysfs_show(struct class *dev,
957 struct class_attribute *attr,
958 char *buf)
957{ 959{
958 struct sock *sk; 960 struct sock *sk;
959 struct hlist_node *node; 961 struct hlist_node *node;
diff --git a/net/bridge/Kconfig b/net/bridge/Kconfig
index 19a6b9629c51..d115d5cea5b6 100644
--- a/net/bridge/Kconfig
+++ b/net/bridge/Kconfig
@@ -35,6 +35,7 @@ config BRIDGE
35config BRIDGE_IGMP_SNOOPING 35config BRIDGE_IGMP_SNOOPING
36 bool "IGMP snooping" 36 bool "IGMP snooping"
37 depends on BRIDGE 37 depends on BRIDGE
38 depends on INET
38 default y 39 default y
39 ---help--- 40 ---help---
40 If you say Y here, then the Ethernet bridge will be able selectively 41 If you say Y here, then the Ethernet bridge will be able selectively
diff --git a/net/bridge/br_device.c b/net/bridge/br_device.c
index eb7062d2e9e5..5b8a6e73b02f 100644
--- a/net/bridge/br_device.c
+++ b/net/bridge/br_device.c
@@ -26,11 +26,12 @@ netdev_tx_t br_dev_xmit(struct sk_buff *skb, struct net_device *dev)
26 const unsigned char *dest = skb->data; 26 const unsigned char *dest = skb->data;
27 struct net_bridge_fdb_entry *dst; 27 struct net_bridge_fdb_entry *dst;
28 struct net_bridge_mdb_entry *mdst; 28 struct net_bridge_mdb_entry *mdst;
29 struct br_cpu_netstats *brstats = this_cpu_ptr(br->stats);
29 30
30 BR_INPUT_SKB_CB(skb)->brdev = dev; 31 brstats->tx_packets++;
32 brstats->tx_bytes += skb->len;
31 33
32 dev->stats.tx_packets++; 34 BR_INPUT_SKB_CB(skb)->brdev = dev;
33 dev->stats.tx_bytes += skb->len;
34 35
35 skb_reset_mac_header(skb); 36 skb_reset_mac_header(skb);
36 skb_pull(skb, ETH_HLEN); 37 skb_pull(skb, ETH_HLEN);
@@ -40,7 +41,7 @@ netdev_tx_t br_dev_xmit(struct sk_buff *skb, struct net_device *dev)
40 goto out; 41 goto out;
41 42
42 mdst = br_mdb_get(br, skb); 43 mdst = br_mdb_get(br, skb);
43 if (mdst || BR_INPUT_SKB_CB(skb)->mrouters_only) 44 if (mdst || BR_INPUT_SKB_CB_MROUTERS_ONLY(skb))
44 br_multicast_deliver(mdst, skb); 45 br_multicast_deliver(mdst, skb);
45 else 46 else
46 br_flood_deliver(br, skb); 47 br_flood_deliver(br, skb);
@@ -81,6 +82,31 @@ static int br_dev_stop(struct net_device *dev)
81 return 0; 82 return 0;
82} 83}
83 84
85static struct net_device_stats *br_get_stats(struct net_device *dev)
86{
87 struct net_bridge *br = netdev_priv(dev);
88 struct net_device_stats *stats = &dev->stats;
89 struct br_cpu_netstats sum = { 0 };
90 unsigned int cpu;
91
92 for_each_possible_cpu(cpu) {
93 const struct br_cpu_netstats *bstats
94 = per_cpu_ptr(br->stats, cpu);
95
96 sum.tx_bytes += bstats->tx_bytes;
97 sum.tx_packets += bstats->tx_packets;
98 sum.rx_bytes += bstats->rx_bytes;
99 sum.rx_packets += bstats->rx_packets;
100 }
101
102 stats->tx_bytes = sum.tx_bytes;
103 stats->tx_packets = sum.tx_packets;
104 stats->rx_bytes = sum.rx_bytes;
105 stats->rx_packets = sum.rx_packets;
106
107 return stats;
108}
109
84static int br_change_mtu(struct net_device *dev, int new_mtu) 110static int br_change_mtu(struct net_device *dev, int new_mtu)
85{ 111{
86 struct net_bridge *br = netdev_priv(dev); 112 struct net_bridge *br = netdev_priv(dev);
@@ -180,19 +206,28 @@ static const struct net_device_ops br_netdev_ops = {
180 .ndo_open = br_dev_open, 206 .ndo_open = br_dev_open,
181 .ndo_stop = br_dev_stop, 207 .ndo_stop = br_dev_stop,
182 .ndo_start_xmit = br_dev_xmit, 208 .ndo_start_xmit = br_dev_xmit,
209 .ndo_get_stats = br_get_stats,
183 .ndo_set_mac_address = br_set_mac_address, 210 .ndo_set_mac_address = br_set_mac_address,
184 .ndo_set_multicast_list = br_dev_set_multicast_list, 211 .ndo_set_multicast_list = br_dev_set_multicast_list,
185 .ndo_change_mtu = br_change_mtu, 212 .ndo_change_mtu = br_change_mtu,
186 .ndo_do_ioctl = br_dev_ioctl, 213 .ndo_do_ioctl = br_dev_ioctl,
187}; 214};
188 215
216static void br_dev_free(struct net_device *dev)
217{
218 struct net_bridge *br = netdev_priv(dev);
219
220 free_percpu(br->stats);
221 free_netdev(dev);
222}
223
189void br_dev_setup(struct net_device *dev) 224void br_dev_setup(struct net_device *dev)
190{ 225{
191 random_ether_addr(dev->dev_addr); 226 random_ether_addr(dev->dev_addr);
192 ether_setup(dev); 227 ether_setup(dev);
193 228
194 dev->netdev_ops = &br_netdev_ops; 229 dev->netdev_ops = &br_netdev_ops;
195 dev->destructor = free_netdev; 230 dev->destructor = br_dev_free;
196 SET_ETHTOOL_OPS(dev, &br_ethtool_ops); 231 SET_ETHTOOL_OPS(dev, &br_ethtool_ops);
197 dev->tx_queue_len = 0; 232 dev->tx_queue_len = 0;
198 dev->priv_flags = IFF_EBRIDGE; 233 dev->priv_flags = IFF_EBRIDGE;
diff --git a/net/bridge/br_forward.c b/net/bridge/br_forward.c
index d61e6f741125..8dbec83e50ca 100644
--- a/net/bridge/br_forward.c
+++ b/net/bridge/br_forward.c
@@ -19,6 +19,11 @@
19#include <linux/netfilter_bridge.h> 19#include <linux/netfilter_bridge.h>
20#include "br_private.h" 20#include "br_private.h"
21 21
22static int deliver_clone(const struct net_bridge_port *prev,
23 struct sk_buff *skb,
24 void (*__packet_hook)(const struct net_bridge_port *p,
25 struct sk_buff *skb));
26
22/* Don't forward packets to originating port or forwarding diasabled */ 27/* Don't forward packets to originating port or forwarding diasabled */
23static inline int should_deliver(const struct net_bridge_port *p, 28static inline int should_deliver(const struct net_bridge_port *p,
24 const struct sk_buff *skb) 29 const struct sk_buff *skb)
@@ -94,17 +99,22 @@ void br_deliver(const struct net_bridge_port *to, struct sk_buff *skb)
94} 99}
95 100
96/* called with rcu_read_lock */ 101/* called with rcu_read_lock */
97void br_forward(const struct net_bridge_port *to, struct sk_buff *skb) 102void br_forward(const struct net_bridge_port *to, struct sk_buff *skb, struct sk_buff *skb0)
98{ 103{
99 if (should_deliver(to, skb)) { 104 if (should_deliver(to, skb)) {
100 __br_forward(to, skb); 105 if (skb0)
106 deliver_clone(to, skb, __br_forward);
107 else
108 __br_forward(to, skb);
101 return; 109 return;
102 } 110 }
103 111
104 kfree_skb(skb); 112 if (!skb0)
113 kfree_skb(skb);
105} 114}
106 115
107static int deliver_clone(struct net_bridge_port *prev, struct sk_buff *skb, 116static int deliver_clone(const struct net_bridge_port *prev,
117 struct sk_buff *skb,
108 void (*__packet_hook)(const struct net_bridge_port *p, 118 void (*__packet_hook)(const struct net_bridge_port *p,
109 struct sk_buff *skb)) 119 struct sk_buff *skb))
110{ 120{
diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c
index b6a3872f5681..b7cdd2e98050 100644
--- a/net/bridge/br_if.c
+++ b/net/bridge/br_if.c
@@ -185,6 +185,12 @@ static struct net_device *new_bridge_dev(struct net *net, const char *name)
185 br = netdev_priv(dev); 185 br = netdev_priv(dev);
186 br->dev = dev; 186 br->dev = dev;
187 187
188 br->stats = alloc_percpu(struct br_cpu_netstats);
189 if (!br->stats) {
190 free_netdev(dev);
191 return NULL;
192 }
193
188 spin_lock_init(&br->lock); 194 spin_lock_init(&br->lock);
189 INIT_LIST_HEAD(&br->port_list); 195 INIT_LIST_HEAD(&br->port_list);
190 spin_lock_init(&br->hash_lock); 196 spin_lock_init(&br->hash_lock);
diff --git a/net/bridge/br_input.c b/net/bridge/br_input.c
index 53b39851d87d..333dfb7c5886 100644
--- a/net/bridge/br_input.c
+++ b/net/bridge/br_input.c
@@ -23,9 +23,11 @@ const u8 br_group_address[ETH_ALEN] = { 0x01, 0x80, 0xc2, 0x00, 0x00, 0x00 };
23static int br_pass_frame_up(struct sk_buff *skb) 23static int br_pass_frame_up(struct sk_buff *skb)
24{ 24{
25 struct net_device *indev, *brdev = BR_INPUT_SKB_CB(skb)->brdev; 25 struct net_device *indev, *brdev = BR_INPUT_SKB_CB(skb)->brdev;
26 struct net_bridge *br = netdev_priv(brdev);
27 struct br_cpu_netstats *brstats = this_cpu_ptr(br->stats);
26 28
27 brdev->stats.rx_packets++; 29 brstats->rx_packets++;
28 brdev->stats.rx_bytes += skb->len; 30 brstats->rx_bytes += skb->len;
29 31
30 indev = skb->dev; 32 indev = skb->dev;
31 skb->dev = brdev; 33 skb->dev = brdev;
@@ -70,7 +72,7 @@ int br_handle_frame_finish(struct sk_buff *skb)
70 72
71 if (is_multicast_ether_addr(dest)) { 73 if (is_multicast_ether_addr(dest)) {
72 mdst = br_mdb_get(br, skb); 74 mdst = br_mdb_get(br, skb);
73 if (mdst || BR_INPUT_SKB_CB(skb)->mrouters_only) { 75 if (mdst || BR_INPUT_SKB_CB_MROUTERS_ONLY(skb)) {
74 if ((mdst && !hlist_unhashed(&mdst->mglist)) || 76 if ((mdst && !hlist_unhashed(&mdst->mglist)) ||
75 br_multicast_is_router(br)) 77 br_multicast_is_router(br))
76 skb2 = skb; 78 skb2 = skb;
@@ -90,7 +92,7 @@ int br_handle_frame_finish(struct sk_buff *skb)
90 92
91 if (skb) { 93 if (skb) {
92 if (dst) 94 if (dst)
93 br_forward(dst->dst, skb); 95 br_forward(dst->dst, skb, skb2);
94 else 96 else
95 br_flood_forward(br, skb, skb2); 97 br_flood_forward(br, skb, skb2);
96 } 98 }
diff --git a/net/bridge/br_multicast.c b/net/bridge/br_multicast.c
index 2559fb539836..9f0c4f065604 100644
--- a/net/bridge/br_multicast.c
+++ b/net/bridge/br_multicast.c
@@ -38,7 +38,7 @@ static struct net_bridge_mdb_entry *__br_mdb_ip_get(
38 struct net_bridge_mdb_entry *mp; 38 struct net_bridge_mdb_entry *mp;
39 struct hlist_node *p; 39 struct hlist_node *p;
40 40
41 hlist_for_each_entry(mp, p, &mdb->mhash[hash], hlist[mdb->ver]) { 41 hlist_for_each_entry_rcu(mp, p, &mdb->mhash[hash], hlist[mdb->ver]) {
42 if (dst == mp->addr) 42 if (dst == mp->addr)
43 return mp; 43 return mp;
44 } 44 }
@@ -49,22 +49,23 @@ static struct net_bridge_mdb_entry *__br_mdb_ip_get(
49static struct net_bridge_mdb_entry *br_mdb_ip_get( 49static struct net_bridge_mdb_entry *br_mdb_ip_get(
50 struct net_bridge_mdb_htable *mdb, __be32 dst) 50 struct net_bridge_mdb_htable *mdb, __be32 dst)
51{ 51{
52 if (!mdb)
53 return NULL;
54
52 return __br_mdb_ip_get(mdb, dst, br_ip_hash(mdb, dst)); 55 return __br_mdb_ip_get(mdb, dst, br_ip_hash(mdb, dst));
53} 56}
54 57
55struct net_bridge_mdb_entry *br_mdb_get(struct net_bridge *br, 58struct net_bridge_mdb_entry *br_mdb_get(struct net_bridge *br,
56 struct sk_buff *skb) 59 struct sk_buff *skb)
57{ 60{
58 struct net_bridge_mdb_htable *mdb = br->mdb; 61 if (br->multicast_disabled)
59
60 if (!mdb || br->multicast_disabled)
61 return NULL; 62 return NULL;
62 63
63 switch (skb->protocol) { 64 switch (skb->protocol) {
64 case htons(ETH_P_IP): 65 case htons(ETH_P_IP):
65 if (BR_INPUT_SKB_CB(skb)->igmp) 66 if (BR_INPUT_SKB_CB(skb)->igmp)
66 break; 67 break;
67 return br_mdb_ip_get(mdb, ip_hdr(skb)->daddr); 68 return br_mdb_ip_get(br->mdb, ip_hdr(skb)->daddr);
68 } 69 }
69 70
70 return NULL; 71 return NULL;
@@ -627,8 +628,8 @@ static void br_multicast_port_query_expired(unsigned long data)
627 struct net_bridge *br = port->br; 628 struct net_bridge *br = port->br;
628 629
629 spin_lock(&br->multicast_lock); 630 spin_lock(&br->multicast_lock);
630 if (port && (port->state == BR_STATE_DISABLED || 631 if (port->state == BR_STATE_DISABLED ||
631 port->state == BR_STATE_BLOCKING)) 632 port->state == BR_STATE_BLOCKING)
632 goto out; 633 goto out;
633 634
634 if (port->multicast_startup_queries_sent < 635 if (port->multicast_startup_queries_sent <
@@ -823,6 +824,7 @@ static int br_multicast_query(struct net_bridge *br,
823 unsigned long max_delay; 824 unsigned long max_delay;
824 unsigned long now = jiffies; 825 unsigned long now = jiffies;
825 __be32 group; 826 __be32 group;
827 int err = 0;
826 828
827 spin_lock(&br->multicast_lock); 829 spin_lock(&br->multicast_lock);
828 if (!netif_running(br->dev) || 830 if (!netif_running(br->dev) ||
@@ -841,15 +843,17 @@ static int br_multicast_query(struct net_bridge *br,
841 group = 0; 843 group = 0;
842 } 844 }
843 } else { 845 } else {
844 if (!pskb_may_pull(skb, sizeof(struct igmpv3_query))) 846 if (!pskb_may_pull(skb, sizeof(struct igmpv3_query))) {
845 return -EINVAL; 847 err = -EINVAL;
848 goto out;
849 }
846 850
847 ih3 = igmpv3_query_hdr(skb); 851 ih3 = igmpv3_query_hdr(skb);
848 if (ih3->nsrcs) 852 if (ih3->nsrcs)
849 return 0; 853 goto out;
850 854
851 max_delay = ih3->code ? 1 : 855 max_delay = ih3->code ?
852 IGMPV3_MRC(ih3->code) * (HZ / IGMP_TIMER_SCALE); 856 IGMPV3_MRC(ih3->code) * (HZ / IGMP_TIMER_SCALE) : 1;
853 } 857 }
854 858
855 if (!group) 859 if (!group)
@@ -876,7 +880,7 @@ static int br_multicast_query(struct net_bridge *br,
876 880
877out: 881out:
878 spin_unlock(&br->multicast_lock); 882 spin_unlock(&br->multicast_lock);
879 return 0; 883 return err;
880} 884}
881 885
882static void br_multicast_leave_group(struct net_bridge *br, 886static void br_multicast_leave_group(struct net_bridge *br,
@@ -987,7 +991,7 @@ static int br_multicast_ipv4_rcv(struct net_bridge *br,
987 991
988 err = pskb_trim_rcsum(skb2, len); 992 err = pskb_trim_rcsum(skb2, len);
989 if (err) 993 if (err)
990 return err; 994 goto err_out;
991 } 995 }
992 996
993 len -= ip_hdrlen(skb2); 997 len -= ip_hdrlen(skb2);
@@ -999,8 +1003,6 @@ static int br_multicast_ipv4_rcv(struct net_bridge *br,
999 if (!pskb_may_pull(skb2, sizeof(*ih))) 1003 if (!pskb_may_pull(skb2, sizeof(*ih)))
1000 goto out; 1004 goto out;
1001 1005
1002 iph = ip_hdr(skb2);
1003
1004 switch (skb2->ip_summed) { 1006 switch (skb2->ip_summed) {
1005 case CHECKSUM_COMPLETE: 1007 case CHECKSUM_COMPLETE:
1006 if (!csum_fold(skb2->csum)) 1008 if (!csum_fold(skb2->csum))
@@ -1009,7 +1011,7 @@ static int br_multicast_ipv4_rcv(struct net_bridge *br,
1009 case CHECKSUM_NONE: 1011 case CHECKSUM_NONE:
1010 skb2->csum = 0; 1012 skb2->csum = 0;
1011 if (skb_checksum_complete(skb2)) 1013 if (skb_checksum_complete(skb2))
1012 return -EINVAL; 1014 goto out;
1013 } 1015 }
1014 1016
1015 err = 0; 1017 err = 0;
@@ -1036,6 +1038,7 @@ static int br_multicast_ipv4_rcv(struct net_bridge *br,
1036 1038
1037out: 1039out:
1038 __skb_push(skb2, offset); 1040 __skb_push(skb2, offset);
1041err_out:
1039 if (skb2 != skb) 1042 if (skb2 != skb)
1040 kfree_skb(skb2); 1043 kfree_skb(skb2);
1041 return err; 1044 return err;
@@ -1135,7 +1138,7 @@ void br_multicast_stop(struct net_bridge *br)
1135 1138
1136 if (mdb->old) { 1139 if (mdb->old) {
1137 spin_unlock_bh(&br->multicast_lock); 1140 spin_unlock_bh(&br->multicast_lock);
1138 synchronize_rcu_bh(); 1141 rcu_barrier_bh();
1139 spin_lock_bh(&br->multicast_lock); 1142 spin_lock_bh(&br->multicast_lock);
1140 WARN_ON(mdb->old); 1143 WARN_ON(mdb->old);
1141 } 1144 }
diff --git a/net/bridge/br_notify.c b/net/bridge/br_notify.c
index 763a3ec292e5..1413b72acc7f 100644
--- a/net/bridge/br_notify.c
+++ b/net/bridge/br_notify.c
@@ -82,6 +82,10 @@ static int br_device_event(struct notifier_block *unused, unsigned long event, v
82 case NETDEV_UNREGISTER: 82 case NETDEV_UNREGISTER:
83 br_del_if(br, dev); 83 br_del_if(br, dev);
84 break; 84 break;
85
86 case NETDEV_PRE_TYPE_CHANGE:
87 /* Forbid underlaying device to change its type. */
88 return NOTIFY_BAD;
85 } 89 }
86 90
87 /* Events that may cause spanning tree to refresh */ 91 /* Events that may cause spanning tree to refresh */
diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index 1cf2cef78584..791d4ab0fd4d 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -135,6 +135,14 @@ struct net_bridge
135 spinlock_t lock; 135 spinlock_t lock;
136 struct list_head port_list; 136 struct list_head port_list;
137 struct net_device *dev; 137 struct net_device *dev;
138
139 struct br_cpu_netstats __percpu {
140 unsigned long rx_packets;
141 unsigned long rx_bytes;
142 unsigned long tx_packets;
143 unsigned long tx_bytes;
144 } *stats;
145
138 spinlock_t hash_lock; 146 spinlock_t hash_lock;
139 struct hlist_head hash[BR_HASH_SIZE]; 147 struct hlist_head hash[BR_HASH_SIZE];
140 unsigned long feature_mask; 148 unsigned long feature_mask;
@@ -206,12 +214,20 @@ struct net_bridge
206 214
207struct br_input_skb_cb { 215struct br_input_skb_cb {
208 struct net_device *brdev; 216 struct net_device *brdev;
217#ifdef CONFIG_BRIDGE_IGMP_SNOOPING
209 int igmp; 218 int igmp;
210 int mrouters_only; 219 int mrouters_only;
220#endif
211}; 221};
212 222
213#define BR_INPUT_SKB_CB(__skb) ((struct br_input_skb_cb *)(__skb)->cb) 223#define BR_INPUT_SKB_CB(__skb) ((struct br_input_skb_cb *)(__skb)->cb)
214 224
225#ifdef CONFIG_BRIDGE_IGMP_SNOOPING
226# define BR_INPUT_SKB_CB_MROUTERS_ONLY(__skb) (BR_INPUT_SKB_CB(__skb)->mrouters_only)
227#else
228# define BR_INPUT_SKB_CB_MROUTERS_ONLY(__skb) (0)
229#endif
230
215extern struct notifier_block br_device_notifier; 231extern struct notifier_block br_device_notifier;
216extern const u8 br_group_address[ETH_ALEN]; 232extern const u8 br_group_address[ETH_ALEN];
217 233
@@ -252,7 +268,7 @@ extern void br_deliver(const struct net_bridge_port *to,
252 struct sk_buff *skb); 268 struct sk_buff *skb);
253extern int br_dev_queue_push_xmit(struct sk_buff *skb); 269extern int br_dev_queue_push_xmit(struct sk_buff *skb);
254extern void br_forward(const struct net_bridge_port *to, 270extern void br_forward(const struct net_bridge_port *to,
255 struct sk_buff *skb); 271 struct sk_buff *skb, struct sk_buff *skb0);
256extern int br_forward_finish(struct sk_buff *skb); 272extern int br_forward_finish(struct sk_buff *skb);
257extern void br_flood_deliver(struct net_bridge *br, struct sk_buff *skb); 273extern void br_flood_deliver(struct net_bridge *br, struct sk_buff *skb);
258extern void br_flood_forward(struct net_bridge *br, struct sk_buff *skb, 274extern void br_flood_forward(struct net_bridge *br, struct sk_buff *skb,
@@ -423,7 +439,7 @@ extern void br_ifinfo_notify(int event, struct net_bridge_port *port);
423 439
424#ifdef CONFIG_SYSFS 440#ifdef CONFIG_SYSFS
425/* br_sysfs_if.c */ 441/* br_sysfs_if.c */
426extern struct sysfs_ops brport_sysfs_ops; 442extern const struct sysfs_ops brport_sysfs_ops;
427extern int br_sysfs_addif(struct net_bridge_port *p); 443extern int br_sysfs_addif(struct net_bridge_port *p);
428 444
429/* br_sysfs_br.c */ 445/* br_sysfs_br.c */
diff --git a/net/bridge/br_sysfs_if.c b/net/bridge/br_sysfs_if.c
index 696596cd3384..0b9916489d6b 100644
--- a/net/bridge/br_sysfs_if.c
+++ b/net/bridge/br_sysfs_if.c
@@ -238,7 +238,7 @@ static ssize_t brport_store(struct kobject * kobj,
238 return ret; 238 return ret;
239} 239}
240 240
241struct sysfs_ops brport_sysfs_ops = { 241const struct sysfs_ops brport_sysfs_ops = {
242 .show = brport_show, 242 .show = brport_show,
243 .store = brport_store, 243 .store = brport_store,
244}; 244};
diff --git a/net/core/dev.c b/net/core/dev.c
index bcc490cc9452..887aa84fcd46 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -772,14 +772,17 @@ EXPORT_SYMBOL(__dev_getfirstbyhwtype);
772 772
773struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type) 773struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type)
774{ 774{
775 struct net_device *dev; 775 struct net_device *dev, *ret = NULL;
776 776
777 rtnl_lock(); 777 rcu_read_lock();
778 dev = __dev_getfirstbyhwtype(net, type); 778 for_each_netdev_rcu(net, dev)
779 if (dev) 779 if (dev->type == type) {
780 dev_hold(dev); 780 dev_hold(dev);
781 rtnl_unlock(); 781 ret = dev;
782 return dev; 782 break;
783 }
784 rcu_read_unlock();
785 return ret;
783} 786}
784EXPORT_SYMBOL(dev_getfirstbyhwtype); 787EXPORT_SYMBOL(dev_getfirstbyhwtype);
785 788
@@ -1084,9 +1087,9 @@ void netdev_state_change(struct net_device *dev)
1084} 1087}
1085EXPORT_SYMBOL(netdev_state_change); 1088EXPORT_SYMBOL(netdev_state_change);
1086 1089
1087void netdev_bonding_change(struct net_device *dev, unsigned long event) 1090int netdev_bonding_change(struct net_device *dev, unsigned long event)
1088{ 1091{
1089 call_netdevice_notifiers(event, dev); 1092 return call_netdevice_notifiers(event, dev);
1090} 1093}
1091EXPORT_SYMBOL(netdev_bonding_change); 1094EXPORT_SYMBOL(netdev_bonding_change);
1092 1095
@@ -1931,7 +1934,7 @@ out_kfree_skb:
1931 return rc; 1934 return rc;
1932} 1935}
1933 1936
1934static u32 skb_tx_hashrnd; 1937static u32 hashrnd __read_mostly;
1935 1938
1936u16 skb_tx_hash(const struct net_device *dev, const struct sk_buff *skb) 1939u16 skb_tx_hash(const struct net_device *dev, const struct sk_buff *skb)
1937{ 1940{
@@ -1949,7 +1952,7 @@ u16 skb_tx_hash(const struct net_device *dev, const struct sk_buff *skb)
1949 else 1952 else
1950 hash = skb->protocol; 1953 hash = skb->protocol;
1951 1954
1952 hash = jhash_1word(hash, skb_tx_hashrnd); 1955 hash = jhash_1word(hash, hashrnd);
1953 1956
1954 return (u16) (((u64) hash * dev->real_num_tx_queues) >> 32); 1957 return (u16) (((u64) hash * dev->real_num_tx_queues) >> 32);
1955} 1958}
@@ -1959,10 +1962,9 @@ static inline u16 dev_cap_txqueue(struct net_device *dev, u16 queue_index)
1959{ 1962{
1960 if (unlikely(queue_index >= dev->real_num_tx_queues)) { 1963 if (unlikely(queue_index >= dev->real_num_tx_queues)) {
1961 if (net_ratelimit()) { 1964 if (net_ratelimit()) {
1962 WARN(1, "%s selects TX queue %d, but " 1965 netdev_warn(dev, "selects TX queue %d, but "
1963 "real number of TX queues is %d\n", 1966 "real number of TX queues is %d\n",
1964 dev->name, queue_index, 1967 queue_index, dev->real_num_tx_queues);
1965 dev->real_num_tx_queues);
1966 } 1968 }
1967 return 0; 1969 return 0;
1968 } 1970 }
@@ -2175,6 +2177,178 @@ int weight_p __read_mostly = 64; /* old backlog weight */
2175 2177
2176DEFINE_PER_CPU(struct netif_rx_stats, netdev_rx_stat) = { 0, }; 2178DEFINE_PER_CPU(struct netif_rx_stats, netdev_rx_stat) = { 0, };
2177 2179
2180#ifdef CONFIG_RPS
2181/*
2182 * get_rps_cpu is called from netif_receive_skb and returns the target
2183 * CPU from the RPS map of the receiving queue for a given skb.
2184 */
2185static int get_rps_cpu(struct net_device *dev, struct sk_buff *skb)
2186{
2187 struct ipv6hdr *ip6;
2188 struct iphdr *ip;
2189 struct netdev_rx_queue *rxqueue;
2190 struct rps_map *map;
2191 int cpu = -1;
2192 u8 ip_proto;
2193 u32 addr1, addr2, ports, ihl;
2194
2195 rcu_read_lock();
2196
2197 if (skb_rx_queue_recorded(skb)) {
2198 u16 index = skb_get_rx_queue(skb);
2199 if (unlikely(index >= dev->num_rx_queues)) {
2200 if (net_ratelimit()) {
2201 netdev_warn(dev, "received packet on queue "
2202 "%u, but number of RX queues is %u\n",
2203 index, dev->num_rx_queues);
2204 }
2205 goto done;
2206 }
2207 rxqueue = dev->_rx + index;
2208 } else
2209 rxqueue = dev->_rx;
2210
2211 if (!rxqueue->rps_map)
2212 goto done;
2213
2214 if (skb->rxhash)
2215 goto got_hash; /* Skip hash computation on packet header */
2216
2217 switch (skb->protocol) {
2218 case __constant_htons(ETH_P_IP):
2219 if (!pskb_may_pull(skb, sizeof(*ip)))
2220 goto done;
2221
2222 ip = (struct iphdr *) skb->data;
2223 ip_proto = ip->protocol;
2224 addr1 = ip->saddr;
2225 addr2 = ip->daddr;
2226 ihl = ip->ihl;
2227 break;
2228 case __constant_htons(ETH_P_IPV6):
2229 if (!pskb_may_pull(skb, sizeof(*ip6)))
2230 goto done;
2231
2232 ip6 = (struct ipv6hdr *) skb->data;
2233 ip_proto = ip6->nexthdr;
2234 addr1 = ip6->saddr.s6_addr32[3];
2235 addr2 = ip6->daddr.s6_addr32[3];
2236 ihl = (40 >> 2);
2237 break;
2238 default:
2239 goto done;
2240 }
2241 ports = 0;
2242 switch (ip_proto) {
2243 case IPPROTO_TCP:
2244 case IPPROTO_UDP:
2245 case IPPROTO_DCCP:
2246 case IPPROTO_ESP:
2247 case IPPROTO_AH:
2248 case IPPROTO_SCTP:
2249 case IPPROTO_UDPLITE:
2250 if (pskb_may_pull(skb, (ihl * 4) + 4))
2251 ports = *((u32 *) (skb->data + (ihl * 4)));
2252 break;
2253
2254 default:
2255 break;
2256 }
2257
2258 skb->rxhash = jhash_3words(addr1, addr2, ports, hashrnd);
2259 if (!skb->rxhash)
2260 skb->rxhash = 1;
2261
2262got_hash:
2263 map = rcu_dereference(rxqueue->rps_map);
2264 if (map) {
2265 u16 tcpu = map->cpus[((u64) skb->rxhash * map->len) >> 32];
2266
2267 if (cpu_online(tcpu)) {
2268 cpu = tcpu;
2269 goto done;
2270 }
2271 }
2272
2273done:
2274 rcu_read_unlock();
2275 return cpu;
2276}
2277
2278/*
2279 * This structure holds the per-CPU mask of CPUs for which IPIs are scheduled
2280 * to be sent to kick remote softirq processing. There are two masks since
2281 * the sending of IPIs must be done with interrupts enabled. The select field
2282 * indicates the current mask that enqueue_backlog uses to schedule IPIs.
2283 * select is flipped before net_rps_action is called while still under lock,
2284 * net_rps_action then uses the non-selected mask to send the IPIs and clears
2285 * it without conflicting with enqueue_backlog operation.
2286 */
2287struct rps_remote_softirq_cpus {
2288 cpumask_t mask[2];
2289 int select;
2290};
2291static DEFINE_PER_CPU(struct rps_remote_softirq_cpus, rps_remote_softirq_cpus);
2292
2293/* Called from hardirq (IPI) context */
2294static void trigger_softirq(void *data)
2295{
2296 struct softnet_data *queue = data;
2297 __napi_schedule(&queue->backlog);
2298 __get_cpu_var(netdev_rx_stat).received_rps++;
2299}
2300#endif /* CONFIG_SMP */
2301
2302/*
2303 * enqueue_to_backlog is called to queue an skb to a per CPU backlog
2304 * queue (may be a remote CPU queue).
2305 */
2306static int enqueue_to_backlog(struct sk_buff *skb, int cpu)
2307{
2308 struct softnet_data *queue;
2309 unsigned long flags;
2310
2311 queue = &per_cpu(softnet_data, cpu);
2312
2313 local_irq_save(flags);
2314 __get_cpu_var(netdev_rx_stat).total++;
2315
2316 spin_lock(&queue->input_pkt_queue.lock);
2317 if (queue->input_pkt_queue.qlen <= netdev_max_backlog) {
2318 if (queue->input_pkt_queue.qlen) {
2319enqueue:
2320 __skb_queue_tail(&queue->input_pkt_queue, skb);
2321 spin_unlock_irqrestore(&queue->input_pkt_queue.lock,
2322 flags);
2323 return NET_RX_SUCCESS;
2324 }
2325
2326 /* Schedule NAPI for backlog device */
2327 if (napi_schedule_prep(&queue->backlog)) {
2328#ifdef CONFIG_RPS
2329 if (cpu != smp_processor_id()) {
2330 struct rps_remote_softirq_cpus *rcpus =
2331 &__get_cpu_var(rps_remote_softirq_cpus);
2332
2333 cpu_set(cpu, rcpus->mask[rcpus->select]);
2334 __raise_softirq_irqoff(NET_RX_SOFTIRQ);
2335 } else
2336 __napi_schedule(&queue->backlog);
2337#else
2338 __napi_schedule(&queue->backlog);
2339#endif
2340 }
2341 goto enqueue;
2342 }
2343
2344 spin_unlock(&queue->input_pkt_queue.lock);
2345
2346 __get_cpu_var(netdev_rx_stat).dropped++;
2347 local_irq_restore(flags);
2348
2349 kfree_skb(skb);
2350 return NET_RX_DROP;
2351}
2178 2352
2179/** 2353/**
2180 * netif_rx - post buffer to the network code 2354 * netif_rx - post buffer to the network code
@@ -2193,8 +2367,7 @@ DEFINE_PER_CPU(struct netif_rx_stats, netdev_rx_stat) = { 0, };
2193 2367
2194int netif_rx(struct sk_buff *skb) 2368int netif_rx(struct sk_buff *skb)
2195{ 2369{
2196 struct softnet_data *queue; 2370 int cpu;
2197 unsigned long flags;
2198 2371
2199 /* if netpoll wants it, pretend we never saw it */ 2372 /* if netpoll wants it, pretend we never saw it */
2200 if (netpoll_rx(skb)) 2373 if (netpoll_rx(skb))
@@ -2203,31 +2376,15 @@ int netif_rx(struct sk_buff *skb)
2203 if (!skb->tstamp.tv64) 2376 if (!skb->tstamp.tv64)
2204 net_timestamp(skb); 2377 net_timestamp(skb);
2205 2378
2206 /* 2379#ifdef CONFIG_RPS
2207 * The code is rearranged so that the path is the most 2380 cpu = get_rps_cpu(skb->dev, skb);
2208 * short when CPU is congested, but is still operating. 2381 if (cpu < 0)
2209 */ 2382 cpu = smp_processor_id();
2210 local_irq_save(flags); 2383#else
2211 queue = &__get_cpu_var(softnet_data); 2384 cpu = smp_processor_id();
2212 2385#endif
2213 __get_cpu_var(netdev_rx_stat).total++;
2214 if (queue->input_pkt_queue.qlen <= netdev_max_backlog) {
2215 if (queue->input_pkt_queue.qlen) {
2216enqueue:
2217 __skb_queue_tail(&queue->input_pkt_queue, skb);
2218 local_irq_restore(flags);
2219 return NET_RX_SUCCESS;
2220 }
2221
2222 napi_schedule(&queue->backlog);
2223 goto enqueue;
2224 }
2225
2226 __get_cpu_var(netdev_rx_stat).dropped++;
2227 local_irq_restore(flags);
2228 2386
2229 kfree_skb(skb); 2387 return enqueue_to_backlog(skb, cpu);
2230 return NET_RX_DROP;
2231} 2388}
2232EXPORT_SYMBOL(netif_rx); 2389EXPORT_SYMBOL(netif_rx);
2233 2390
@@ -2464,25 +2621,11 @@ void netif_nit_deliver(struct sk_buff *skb)
2464 rcu_read_unlock(); 2621 rcu_read_unlock();
2465} 2622}
2466 2623
2467/** 2624static int __netif_receive_skb(struct sk_buff *skb)
2468 * netif_receive_skb - process receive buffer from network
2469 * @skb: buffer to process
2470 *
2471 * netif_receive_skb() is the main receive data processing function.
2472 * It always succeeds. The buffer may be dropped during processing
2473 * for congestion control or by the protocol layers.
2474 *
2475 * This function may only be called from softirq context and interrupts
2476 * should be enabled.
2477 *
2478 * Return values (usually ignored):
2479 * NET_RX_SUCCESS: no congestion
2480 * NET_RX_DROP: packet was dropped
2481 */
2482int netif_receive_skb(struct sk_buff *skb)
2483{ 2625{
2484 struct packet_type *ptype, *pt_prev; 2626 struct packet_type *ptype, *pt_prev;
2485 struct net_device *orig_dev; 2627 struct net_device *orig_dev;
2628 struct net_device *master;
2486 struct net_device *null_or_orig; 2629 struct net_device *null_or_orig;
2487 struct net_device *null_or_bond; 2630 struct net_device *null_or_bond;
2488 int ret = NET_RX_DROP; 2631 int ret = NET_RX_DROP;
@@ -2503,11 +2646,12 @@ int netif_receive_skb(struct sk_buff *skb)
2503 2646
2504 null_or_orig = NULL; 2647 null_or_orig = NULL;
2505 orig_dev = skb->dev; 2648 orig_dev = skb->dev;
2506 if (orig_dev->master) { 2649 master = ACCESS_ONCE(orig_dev->master);
2507 if (skb_bond_should_drop(skb)) 2650 if (master) {
2651 if (skb_bond_should_drop(skb, master))
2508 null_or_orig = orig_dev; /* deliver only exact match */ 2652 null_or_orig = orig_dev; /* deliver only exact match */
2509 else 2653 else
2510 skb->dev = orig_dev->master; 2654 skb->dev = master;
2511 } 2655 }
2512 2656
2513 __get_cpu_var(netdev_rx_stat).total++; 2657 __get_cpu_var(netdev_rx_stat).total++;
@@ -2588,20 +2732,53 @@ out:
2588 rcu_read_unlock(); 2732 rcu_read_unlock();
2589 return ret; 2733 return ret;
2590} 2734}
2735
2736/**
2737 * netif_receive_skb - process receive buffer from network
2738 * @skb: buffer to process
2739 *
2740 * netif_receive_skb() is the main receive data processing function.
2741 * It always succeeds. The buffer may be dropped during processing
2742 * for congestion control or by the protocol layers.
2743 *
2744 * This function may only be called from softirq context and interrupts
2745 * should be enabled.
2746 *
2747 * Return values (usually ignored):
2748 * NET_RX_SUCCESS: no congestion
2749 * NET_RX_DROP: packet was dropped
2750 */
2751int netif_receive_skb(struct sk_buff *skb)
2752{
2753#ifdef CONFIG_RPS
2754 int cpu;
2755
2756 cpu = get_rps_cpu(skb->dev, skb);
2757
2758 if (cpu < 0)
2759 return __netif_receive_skb(skb);
2760 else
2761 return enqueue_to_backlog(skb, cpu);
2762#else
2763 return __netif_receive_skb(skb);
2764#endif
2765}
2591EXPORT_SYMBOL(netif_receive_skb); 2766EXPORT_SYMBOL(netif_receive_skb);
2592 2767
2593/* Network device is going away, flush any packets still pending */ 2768/* Network device is going away, flush any packets still pending */
2594static void flush_backlog(void *arg) 2769static void flush_backlog(struct net_device *dev, int cpu)
2595{ 2770{
2596 struct net_device *dev = arg; 2771 struct softnet_data *queue = &per_cpu(softnet_data, cpu);
2597 struct softnet_data *queue = &__get_cpu_var(softnet_data);
2598 struct sk_buff *skb, *tmp; 2772 struct sk_buff *skb, *tmp;
2773 unsigned long flags;
2599 2774
2775 spin_lock_irqsave(&queue->input_pkt_queue.lock, flags);
2600 skb_queue_walk_safe(&queue->input_pkt_queue, skb, tmp) 2776 skb_queue_walk_safe(&queue->input_pkt_queue, skb, tmp)
2601 if (skb->dev == dev) { 2777 if (skb->dev == dev) {
2602 __skb_unlink(skb, &queue->input_pkt_queue); 2778 __skb_unlink(skb, &queue->input_pkt_queue);
2603 kfree_skb(skb); 2779 kfree_skb(skb);
2604 } 2780 }
2781 spin_unlock_irqrestore(&queue->input_pkt_queue.lock, flags);
2605} 2782}
2606 2783
2607static int napi_gro_complete(struct sk_buff *skb) 2784static int napi_gro_complete(struct sk_buff *skb)
@@ -2914,16 +3091,16 @@ static int process_backlog(struct napi_struct *napi, int quota)
2914 do { 3091 do {
2915 struct sk_buff *skb; 3092 struct sk_buff *skb;
2916 3093
2917 local_irq_disable(); 3094 spin_lock_irq(&queue->input_pkt_queue.lock);
2918 skb = __skb_dequeue(&queue->input_pkt_queue); 3095 skb = __skb_dequeue(&queue->input_pkt_queue);
2919 if (!skb) { 3096 if (!skb) {
2920 __napi_complete(napi); 3097 __napi_complete(napi);
2921 local_irq_enable(); 3098 spin_unlock_irq(&queue->input_pkt_queue.lock);
2922 break; 3099 break;
2923 } 3100 }
2924 local_irq_enable(); 3101 spin_unlock_irq(&queue->input_pkt_queue.lock);
2925 3102
2926 netif_receive_skb(skb); 3103 __netif_receive_skb(skb);
2927 } while (++work < quota && jiffies == start_time); 3104 } while (++work < quota && jiffies == start_time);
2928 3105
2929 return work; 3106 return work;
@@ -3012,6 +3189,24 @@ void netif_napi_del(struct napi_struct *napi)
3012} 3189}
3013EXPORT_SYMBOL(netif_napi_del); 3190EXPORT_SYMBOL(netif_napi_del);
3014 3191
3192#ifdef CONFIG_RPS
3193/*
3194 * net_rps_action sends any pending IPI's for rps. This is only called from
3195 * softirq and interrupts must be enabled.
3196 */
3197static void net_rps_action(cpumask_t *mask)
3198{
3199 int cpu;
3200
3201 /* Send pending IPI's to kick RPS processing on remote cpus. */
3202 for_each_cpu_mask_nr(cpu, *mask) {
3203 struct softnet_data *queue = &per_cpu(softnet_data, cpu);
3204 if (cpu_online(cpu))
3205 __smp_call_function_single(cpu, &queue->csd, 0);
3206 }
3207 cpus_clear(*mask);
3208}
3209#endif
3015 3210
3016static void net_rx_action(struct softirq_action *h) 3211static void net_rx_action(struct softirq_action *h)
3017{ 3212{
@@ -3019,6 +3214,10 @@ static void net_rx_action(struct softirq_action *h)
3019 unsigned long time_limit = jiffies + 2; 3214 unsigned long time_limit = jiffies + 2;
3020 int budget = netdev_budget; 3215 int budget = netdev_budget;
3021 void *have; 3216 void *have;
3217#ifdef CONFIG_RPS
3218 int select;
3219 struct rps_remote_softirq_cpus *rcpus;
3220#endif
3022 3221
3023 local_irq_disable(); 3222 local_irq_disable();
3024 3223
@@ -3081,7 +3280,17 @@ static void net_rx_action(struct softirq_action *h)
3081 netpoll_poll_unlock(have); 3280 netpoll_poll_unlock(have);
3082 } 3281 }
3083out: 3282out:
3283#ifdef CONFIG_RPS
3284 rcpus = &__get_cpu_var(rps_remote_softirq_cpus);
3285 select = rcpus->select;
3286 rcpus->select ^= 1;
3287
3288 local_irq_enable();
3289
3290 net_rps_action(&rcpus->mask[select]);
3291#else
3084 local_irq_enable(); 3292 local_irq_enable();
3293#endif
3085 3294
3086#ifdef CONFIG_NET_DMA 3295#ifdef CONFIG_NET_DMA
3087 /* 3296 /*
@@ -3327,10 +3536,10 @@ static int softnet_seq_show(struct seq_file *seq, void *v)
3327{ 3536{
3328 struct netif_rx_stats *s = v; 3537 struct netif_rx_stats *s = v;
3329 3538
3330 seq_printf(seq, "%08x %08x %08x %08x %08x %08x %08x %08x %08x\n", 3539 seq_printf(seq, "%08x %08x %08x %08x %08x %08x %08x %08x %08x %08x\n",
3331 s->total, s->dropped, s->time_squeeze, 0, 3540 s->total, s->dropped, s->time_squeeze, 0,
3332 0, 0, 0, 0, /* was fastroute */ 3541 0, 0, 0, 0, /* was fastroute */
3333 s->cpu_collision); 3542 s->cpu_collision, s->received_rps);
3334 return 0; 3543 return 0;
3335} 3544}
3336 3545
@@ -3553,11 +3762,10 @@ int netdev_set_master(struct net_device *slave, struct net_device *master)
3553 3762
3554 slave->master = master; 3763 slave->master = master;
3555 3764
3556 synchronize_net(); 3765 if (old) {
3557 3766 synchronize_net();
3558 if (old)
3559 dev_put(old); 3767 dev_put(old);
3560 3768 }
3561 if (master) 3769 if (master)
3562 slave->flags |= IFF_SLAVE; 3770 slave->flags |= IFF_SLAVE;
3563 else 3771 else
@@ -4253,12 +4461,13 @@ void dev_unicast_unsync(struct net_device *to, struct net_device *from)
4253} 4461}
4254EXPORT_SYMBOL(dev_unicast_unsync); 4462EXPORT_SYMBOL(dev_unicast_unsync);
4255 4463
4256static void dev_unicast_flush(struct net_device *dev) 4464void dev_unicast_flush(struct net_device *dev)
4257{ 4465{
4258 netif_addr_lock_bh(dev); 4466 netif_addr_lock_bh(dev);
4259 __hw_addr_flush(&dev->uc); 4467 __hw_addr_flush(&dev->uc);
4260 netif_addr_unlock_bh(dev); 4468 netif_addr_unlock_bh(dev);
4261} 4469}
4470EXPORT_SYMBOL(dev_unicast_flush);
4262 4471
4263static void dev_unicast_init(struct net_device *dev) 4472static void dev_unicast_init(struct net_device *dev)
4264{ 4473{
@@ -4280,7 +4489,7 @@ static void __dev_addr_discard(struct dev_addr_list **list)
4280 } 4489 }
4281} 4490}
4282 4491
4283static void dev_addr_discard(struct net_device *dev) 4492void dev_addr_discard(struct net_device *dev)
4284{ 4493{
4285 netif_addr_lock_bh(dev); 4494 netif_addr_lock_bh(dev);
4286 4495
@@ -4289,6 +4498,7 @@ static void dev_addr_discard(struct net_device *dev)
4289 4498
4290 netif_addr_unlock_bh(dev); 4499 netif_addr_unlock_bh(dev);
4291} 4500}
4501EXPORT_SYMBOL(dev_addr_discard);
4292 4502
4293/** 4503/**
4294 * dev_get_flags - get flags reported to userspace 4504 * dev_get_flags - get flags reported to userspace
@@ -5067,6 +5277,24 @@ int register_netdevice(struct net_device *dev)
5067 5277
5068 dev->iflink = -1; 5278 dev->iflink = -1;
5069 5279
5280#ifdef CONFIG_RPS
5281 if (!dev->num_rx_queues) {
5282 /*
5283 * Allocate a single RX queue if driver never called
5284 * alloc_netdev_mq
5285 */
5286
5287 dev->_rx = kzalloc(sizeof(struct netdev_rx_queue), GFP_KERNEL);
5288 if (!dev->_rx) {
5289 ret = -ENOMEM;
5290 goto out;
5291 }
5292
5293 dev->_rx->first = dev->_rx;
5294 atomic_set(&dev->_rx->count, 1);
5295 dev->num_rx_queues = 1;
5296 }
5297#endif
5070 /* Init, if this function is available */ 5298 /* Init, if this function is available */
5071 if (dev->netdev_ops->ndo_init) { 5299 if (dev->netdev_ops->ndo_init) {
5072 ret = dev->netdev_ops->ndo_init(dev); 5300 ret = dev->netdev_ops->ndo_init(dev);
@@ -5320,6 +5548,7 @@ void netdev_run_todo(void)
5320 while (!list_empty(&list)) { 5548 while (!list_empty(&list)) {
5321 struct net_device *dev 5549 struct net_device *dev
5322 = list_first_entry(&list, struct net_device, todo_list); 5550 = list_first_entry(&list, struct net_device, todo_list);
5551 int i;
5323 list_del(&dev->todo_list); 5552 list_del(&dev->todo_list);
5324 5553
5325 if (unlikely(dev->reg_state != NETREG_UNREGISTERING)) { 5554 if (unlikely(dev->reg_state != NETREG_UNREGISTERING)) {
@@ -5331,7 +5560,8 @@ void netdev_run_todo(void)
5331 5560
5332 dev->reg_state = NETREG_UNREGISTERED; 5561 dev->reg_state = NETREG_UNREGISTERED;
5333 5562
5334 on_each_cpu(flush_backlog, dev, 1); 5563 for_each_online_cpu(i)
5564 flush_backlog(dev, i);
5335 5565
5336 netdev_wait_allrefs(dev); 5566 netdev_wait_allrefs(dev);
5337 5567
@@ -5427,6 +5657,10 @@ struct net_device *alloc_netdev_mq(int sizeof_priv, const char *name,
5427 struct net_device *dev; 5657 struct net_device *dev;
5428 size_t alloc_size; 5658 size_t alloc_size;
5429 struct net_device *p; 5659 struct net_device *p;
5660#ifdef CONFIG_RPS
5661 struct netdev_rx_queue *rx;
5662 int i;
5663#endif
5430 5664
5431 BUG_ON(strlen(name) >= sizeof(dev->name)); 5665 BUG_ON(strlen(name) >= sizeof(dev->name));
5432 5666
@@ -5452,11 +5686,29 @@ struct net_device *alloc_netdev_mq(int sizeof_priv, const char *name,
5452 goto free_p; 5686 goto free_p;
5453 } 5687 }
5454 5688
5689#ifdef CONFIG_RPS
5690 rx = kcalloc(queue_count, sizeof(struct netdev_rx_queue), GFP_KERNEL);
5691 if (!rx) {
5692 printk(KERN_ERR "alloc_netdev: Unable to allocate "
5693 "rx queues.\n");
5694 goto free_tx;
5695 }
5696
5697 atomic_set(&rx->count, queue_count);
5698
5699 /*
5700 * Set a pointer to first element in the array which holds the
5701 * reference count.
5702 */
5703 for (i = 0; i < queue_count; i++)
5704 rx[i].first = rx;
5705#endif
5706
5455 dev = PTR_ALIGN(p, NETDEV_ALIGN); 5707 dev = PTR_ALIGN(p, NETDEV_ALIGN);
5456 dev->padded = (char *)dev - (char *)p; 5708 dev->padded = (char *)dev - (char *)p;
5457 5709
5458 if (dev_addr_init(dev)) 5710 if (dev_addr_init(dev))
5459 goto free_tx; 5711 goto free_rx;
5460 5712
5461 dev_unicast_init(dev); 5713 dev_unicast_init(dev);
5462 5714
@@ -5466,6 +5718,11 @@ struct net_device *alloc_netdev_mq(int sizeof_priv, const char *name,
5466 dev->num_tx_queues = queue_count; 5718 dev->num_tx_queues = queue_count;
5467 dev->real_num_tx_queues = queue_count; 5719 dev->real_num_tx_queues = queue_count;
5468 5720
5721#ifdef CONFIG_RPS
5722 dev->_rx = rx;
5723 dev->num_rx_queues = queue_count;
5724#endif
5725
5469 dev->gso_max_size = GSO_MAX_SIZE; 5726 dev->gso_max_size = GSO_MAX_SIZE;
5470 5727
5471 netdev_init_queues(dev); 5728 netdev_init_queues(dev);
@@ -5480,9 +5737,12 @@ struct net_device *alloc_netdev_mq(int sizeof_priv, const char *name,
5480 strcpy(dev->name, name); 5737 strcpy(dev->name, name);
5481 return dev; 5738 return dev;
5482 5739
5740free_rx:
5741#ifdef CONFIG_RPS
5742 kfree(rx);
5483free_tx: 5743free_tx:
5744#endif
5484 kfree(tx); 5745 kfree(tx);
5485
5486free_p: 5746free_p:
5487 kfree(p); 5747 kfree(p);
5488 return NULL; 5748 return NULL;
@@ -5985,6 +6245,12 @@ static int __init net_dev_init(void)
5985 queue->completion_queue = NULL; 6245 queue->completion_queue = NULL;
5986 INIT_LIST_HEAD(&queue->poll_list); 6246 INIT_LIST_HEAD(&queue->poll_list);
5987 6247
6248#ifdef CONFIG_RPS
6249 queue->csd.func = trigger_softirq;
6250 queue->csd.info = queue;
6251 queue->csd.flags = 0;
6252#endif
6253
5988 queue->backlog.poll = process_backlog; 6254 queue->backlog.poll = process_backlog;
5989 queue->backlog.weight = weight_p; 6255 queue->backlog.weight = weight_p;
5990 queue->backlog.gro_list = NULL; 6256 queue->backlog.gro_list = NULL;
@@ -6023,7 +6289,7 @@ subsys_initcall(net_dev_init);
6023 6289
6024static int __init initialize_hashrnd(void) 6290static int __init initialize_hashrnd(void)
6025{ 6291{
6026 get_random_bytes(&skb_tx_hashrnd, sizeof(skb_tx_hashrnd)); 6292 get_random_bytes(&hashrnd, sizeof(hashrnd));
6027 return 0; 6293 return 0;
6028} 6294}
6029 6295
diff --git a/net/core/dev_mcast.c b/net/core/dev_mcast.c
index fd91569e2394..3dc295beb483 100644
--- a/net/core/dev_mcast.c
+++ b/net/core/dev_mcast.c
@@ -97,8 +97,9 @@ int dev_mc_add(struct net_device *dev, void *addr, int alen, int glbl)
97 97
98 netif_addr_lock_bh(dev); 98 netif_addr_lock_bh(dev);
99 if (alen != dev->addr_len) 99 if (alen != dev->addr_len)
100 return -EINVAL; 100 err = -EINVAL;
101 err = __dev_addr_add(&dev->mc_list, &dev->mc_count, addr, alen, glbl); 101 else
102 err = __dev_addr_add(&dev->mc_list, &dev->mc_count, addr, alen, glbl);
102 if (!err) 103 if (!err)
103 __dev_set_rx_mode(dev); 104 __dev_set_rx_mode(dev);
104 netif_addr_unlock_bh(dev); 105 netif_addr_unlock_bh(dev);
diff --git a/net/core/ethtool.c b/net/core/ethtool.c
index 0f2f82185ec4..f4cb6b6299d9 100644
--- a/net/core/ethtool.c
+++ b/net/core/ethtool.c
@@ -17,6 +17,7 @@
17#include <linux/errno.h> 17#include <linux/errno.h>
18#include <linux/ethtool.h> 18#include <linux/ethtool.h>
19#include <linux/netdevice.h> 19#include <linux/netdevice.h>
20#include <linux/bitops.h>
20#include <asm/uaccess.h> 21#include <asm/uaccess.h>
21 22
22/* 23/*
@@ -199,10 +200,7 @@ static int ethtool_set_settings(struct net_device *dev, void __user *useraddr)
199 return dev->ethtool_ops->set_settings(dev, &cmd); 200 return dev->ethtool_ops->set_settings(dev, &cmd);
200} 201}
201 202
202/* 203static noinline_for_stack int ethtool_get_drvinfo(struct net_device *dev, void __user *useraddr)
203 * noinline attribute so that gcc doesnt use too much stack in dev_ethtool()
204 */
205static noinline int ethtool_get_drvinfo(struct net_device *dev, void __user *useraddr)
206{ 204{
207 struct ethtool_drvinfo info; 205 struct ethtool_drvinfo info;
208 const struct ethtool_ops *ops = dev->ethtool_ops; 206 const struct ethtool_ops *ops = dev->ethtool_ops;
@@ -214,6 +212,10 @@ static noinline int ethtool_get_drvinfo(struct net_device *dev, void __user *use
214 info.cmd = ETHTOOL_GDRVINFO; 212 info.cmd = ETHTOOL_GDRVINFO;
215 ops->get_drvinfo(dev, &info); 213 ops->get_drvinfo(dev, &info);
216 214
215 /*
216 * this method of obtaining string set info is deprecated;
217 * Use ETHTOOL_GSSET_INFO instead.
218 */
217 if (ops->get_sset_count) { 219 if (ops->get_sset_count) {
218 int rc; 220 int rc;
219 221
@@ -237,10 +239,67 @@ static noinline int ethtool_get_drvinfo(struct net_device *dev, void __user *use
237 return 0; 239 return 0;
238} 240}
239 241
240/* 242static noinline_for_stack int ethtool_get_sset_info(struct net_device *dev,
241 * noinline attribute so that gcc doesnt use too much stack in dev_ethtool() 243 void __user *useraddr)
242 */ 244{
243static noinline int ethtool_set_rxnfc(struct net_device *dev, void __user *useraddr) 245 struct ethtool_sset_info info;
246 const struct ethtool_ops *ops = dev->ethtool_ops;
247 u64 sset_mask;
248 int i, idx = 0, n_bits = 0, ret, rc;
249 u32 *info_buf = NULL;
250
251 if (!ops->get_sset_count)
252 return -EOPNOTSUPP;
253
254 if (copy_from_user(&info, useraddr, sizeof(info)))
255 return -EFAULT;
256
257 /* store copy of mask, because we zero struct later on */
258 sset_mask = info.sset_mask;
259 if (!sset_mask)
260 return 0;
261
262 /* calculate size of return buffer */
263 n_bits = hweight64(sset_mask);
264
265 memset(&info, 0, sizeof(info));
266 info.cmd = ETHTOOL_GSSET_INFO;
267
268 info_buf = kzalloc(n_bits * sizeof(u32), GFP_USER);
269 if (!info_buf)
270 return -ENOMEM;
271
272 /*
273 * fill return buffer based on input bitmask and successful
274 * get_sset_count return
275 */
276 for (i = 0; i < 64; i++) {
277 if (!(sset_mask & (1ULL << i)))
278 continue;
279
280 rc = ops->get_sset_count(dev, i);
281 if (rc >= 0) {
282 info.sset_mask |= (1ULL << i);
283 info_buf[idx++] = rc;
284 }
285 }
286
287 ret = -EFAULT;
288 if (copy_to_user(useraddr, &info, sizeof(info)))
289 goto out;
290
291 useraddr += offsetof(struct ethtool_sset_info, data);
292 if (copy_to_user(useraddr, info_buf, idx * sizeof(u32)))
293 goto out;
294
295 ret = 0;
296
297out:
298 kfree(info_buf);
299 return ret;
300}
301
302static noinline_for_stack int ethtool_set_rxnfc(struct net_device *dev, void __user *useraddr)
244{ 303{
245 struct ethtool_rxnfc cmd; 304 struct ethtool_rxnfc cmd;
246 305
@@ -253,10 +312,7 @@ static noinline int ethtool_set_rxnfc(struct net_device *dev, void __user *usera
253 return dev->ethtool_ops->set_rxnfc(dev, &cmd); 312 return dev->ethtool_ops->set_rxnfc(dev, &cmd);
254} 313}
255 314
256/* 315static noinline_for_stack int ethtool_get_rxnfc(struct net_device *dev, void __user *useraddr)
257 * noinline attribute so that gcc doesnt use too much stack in dev_ethtool()
258 */
259static noinline int ethtool_get_rxnfc(struct net_device *dev, void __user *useraddr)
260{ 316{
261 struct ethtool_rxnfc info; 317 struct ethtool_rxnfc info;
262 const struct ethtool_ops *ops = dev->ethtool_ops; 318 const struct ethtool_ops *ops = dev->ethtool_ops;
@@ -328,10 +384,7 @@ static void __rx_ntuple_filter_add(struct ethtool_rx_ntuple_list *list,
328 list->count++; 384 list->count++;
329} 385}
330 386
331/* 387static noinline_for_stack int ethtool_set_rx_ntuple(struct net_device *dev, void __user *useraddr)
332 * noinline attribute so that gcc doesnt use too much stack in dev_ethtool()
333 */
334static noinline int ethtool_set_rx_ntuple(struct net_device *dev, void __user *useraddr)
335{ 388{
336 struct ethtool_rx_ntuple cmd; 389 struct ethtool_rx_ntuple cmd;
337 const struct ethtool_ops *ops = dev->ethtool_ops; 390 const struct ethtool_ops *ops = dev->ethtool_ops;
@@ -799,10 +852,7 @@ static int ethtool_set_eeprom(struct net_device *dev, void __user *useraddr)
799 return ret; 852 return ret;
800} 853}
801 854
802/* 855static noinline_for_stack int ethtool_get_coalesce(struct net_device *dev, void __user *useraddr)
803 * noinline attribute so that gcc doesnt use too much stack in dev_ethtool()
804 */
805static noinline int ethtool_get_coalesce(struct net_device *dev, void __user *useraddr)
806{ 856{
807 struct ethtool_coalesce coalesce = { .cmd = ETHTOOL_GCOALESCE }; 857 struct ethtool_coalesce coalesce = { .cmd = ETHTOOL_GCOALESCE };
808 858
@@ -816,10 +866,7 @@ static noinline int ethtool_get_coalesce(struct net_device *dev, void __user *us
816 return 0; 866 return 0;
817} 867}
818 868
819/* 869static noinline_for_stack int ethtool_set_coalesce(struct net_device *dev, void __user *useraddr)
820 * noinline attribute so that gcc doesnt use too much stack in dev_ethtool()
821 */
822static noinline int ethtool_set_coalesce(struct net_device *dev, void __user *useraddr)
823{ 870{
824 struct ethtool_coalesce coalesce; 871 struct ethtool_coalesce coalesce;
825 872
@@ -1229,10 +1276,7 @@ static int ethtool_set_value(struct net_device *dev, char __user *useraddr,
1229 return actor(dev, edata.data); 1276 return actor(dev, edata.data);
1230} 1277}
1231 1278
1232/* 1279static noinline_for_stack int ethtool_flash_device(struct net_device *dev, char __user *useraddr)
1233 * noinline attribute so that gcc doesnt use too much stack in dev_ethtool()
1234 */
1235static noinline int ethtool_flash_device(struct net_device *dev, char __user *useraddr)
1236{ 1280{
1237 struct ethtool_flash efl; 1281 struct ethtool_flash efl;
1238 1282
@@ -1471,6 +1515,9 @@ int dev_ethtool(struct net *net, struct ifreq *ifr)
1471 case ETHTOOL_GRXNTUPLE: 1515 case ETHTOOL_GRXNTUPLE:
1472 rc = ethtool_get_rx_ntuple(dev, useraddr); 1516 rc = ethtool_get_rx_ntuple(dev, useraddr);
1473 break; 1517 break;
1518 case ETHTOOL_GSSET_INFO:
1519 rc = ethtool_get_sset_info(dev, useraddr);
1520 break;
1474 default: 1521 default:
1475 rc = -EOPNOTSUPP; 1522 rc = -EOPNOTSUPP;
1476 } 1523 }
diff --git a/net/core/fib_rules.c b/net/core/fib_rules.c
index 9a24377146bf..2ff34894357a 100644
--- a/net/core/fib_rules.c
+++ b/net/core/fib_rules.c
@@ -108,7 +108,7 @@ fib_rules_register(struct fib_rules_ops *tmpl, struct net *net)
108 struct fib_rules_ops *ops; 108 struct fib_rules_ops *ops;
109 int err; 109 int err;
110 110
111 ops = kmemdup(tmpl, sizeof (*ops), GFP_KERNEL); 111 ops = kmemdup(tmpl, sizeof(*ops), GFP_KERNEL);
112 if (ops == NULL) 112 if (ops == NULL)
113 return ERR_PTR(-ENOMEM); 113 return ERR_PTR(-ENOMEM);
114 114
@@ -123,7 +123,6 @@ fib_rules_register(struct fib_rules_ops *tmpl, struct net *net)
123 123
124 return ops; 124 return ops;
125} 125}
126
127EXPORT_SYMBOL_GPL(fib_rules_register); 126EXPORT_SYMBOL_GPL(fib_rules_register);
128 127
129void fib_rules_cleanup_ops(struct fib_rules_ops *ops) 128void fib_rules_cleanup_ops(struct fib_rules_ops *ops)
@@ -157,7 +156,6 @@ void fib_rules_unregister(struct fib_rules_ops *ops)
157 156
158 call_rcu(&ops->rcu, fib_rules_put_rcu); 157 call_rcu(&ops->rcu, fib_rules_put_rcu);
159} 158}
160
161EXPORT_SYMBOL_GPL(fib_rules_unregister); 159EXPORT_SYMBOL_GPL(fib_rules_unregister);
162 160
163static int fib_rule_match(struct fib_rule *rule, struct fib_rules_ops *ops, 161static int fib_rule_match(struct fib_rule *rule, struct fib_rules_ops *ops,
@@ -220,7 +218,6 @@ out:
220 218
221 return err; 219 return err;
222} 220}
223
224EXPORT_SYMBOL_GPL(fib_rules_lookup); 221EXPORT_SYMBOL_GPL(fib_rules_lookup);
225 222
226static int validate_rulemsg(struct fib_rule_hdr *frh, struct nlattr **tb, 223static int validate_rulemsg(struct fib_rule_hdr *frh, struct nlattr **tb,
@@ -613,7 +610,7 @@ static int fib_nl_dumprule(struct sk_buff *skb, struct netlink_callback *cb)
613 break; 610 break;
614 611
615 cb->args[1] = 0; 612 cb->args[1] = 0;
616 skip: 613skip:
617 idx++; 614 idx++;
618 } 615 }
619 rcu_read_unlock(); 616 rcu_read_unlock();
@@ -685,7 +682,6 @@ static int fib_rules_event(struct notifier_block *this, unsigned long event,
685 struct fib_rules_ops *ops; 682 struct fib_rules_ops *ops;
686 683
687 ASSERT_RTNL(); 684 ASSERT_RTNL();
688 rcu_read_lock();
689 685
690 switch (event) { 686 switch (event) {
691 case NETDEV_REGISTER: 687 case NETDEV_REGISTER:
@@ -699,8 +695,6 @@ static int fib_rules_event(struct notifier_block *this, unsigned long event,
699 break; 695 break;
700 } 696 }
701 697
702 rcu_read_unlock();
703
704 return NOTIFY_DONE; 698 return NOTIFY_DONE;
705} 699}
706 700
diff --git a/net/core/neighbour.c b/net/core/neighbour.c
index d102f6d9abdc..6cee6434da67 100644
--- a/net/core/neighbour.c
+++ b/net/core/neighbour.c
@@ -771,6 +771,8 @@ static __inline__ int neigh_max_probes(struct neighbour *n)
771} 771}
772 772
773static void neigh_invalidate(struct neighbour *neigh) 773static void neigh_invalidate(struct neighbour *neigh)
774 __releases(neigh->lock)
775 __acquires(neigh->lock)
774{ 776{
775 struct sk_buff *skb; 777 struct sk_buff *skb;
776 778
diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c
index 099c753c4213..1e7fdd6029a2 100644
--- a/net/core/net-sysfs.c
+++ b/net/core/net-sysfs.c
@@ -466,6 +466,217 @@ static struct attribute_group wireless_group = {
466}; 466};
467#endif 467#endif
468 468
469#ifdef CONFIG_RPS
470/*
471 * RX queue sysfs structures and functions.
472 */
473struct rx_queue_attribute {
474 struct attribute attr;
475 ssize_t (*show)(struct netdev_rx_queue *queue,
476 struct rx_queue_attribute *attr, char *buf);
477 ssize_t (*store)(struct netdev_rx_queue *queue,
478 struct rx_queue_attribute *attr, const char *buf, size_t len);
479};
480#define to_rx_queue_attr(_attr) container_of(_attr, \
481 struct rx_queue_attribute, attr)
482
483#define to_rx_queue(obj) container_of(obj, struct netdev_rx_queue, kobj)
484
485static ssize_t rx_queue_attr_show(struct kobject *kobj, struct attribute *attr,
486 char *buf)
487{
488 struct rx_queue_attribute *attribute = to_rx_queue_attr(attr);
489 struct netdev_rx_queue *queue = to_rx_queue(kobj);
490
491 if (!attribute->show)
492 return -EIO;
493
494 return attribute->show(queue, attribute, buf);
495}
496
497static ssize_t rx_queue_attr_store(struct kobject *kobj, struct attribute *attr,
498 const char *buf, size_t count)
499{
500 struct rx_queue_attribute *attribute = to_rx_queue_attr(attr);
501 struct netdev_rx_queue *queue = to_rx_queue(kobj);
502
503 if (!attribute->store)
504 return -EIO;
505
506 return attribute->store(queue, attribute, buf, count);
507}
508
509static struct sysfs_ops rx_queue_sysfs_ops = {
510 .show = rx_queue_attr_show,
511 .store = rx_queue_attr_store,
512};
513
514static ssize_t show_rps_map(struct netdev_rx_queue *queue,
515 struct rx_queue_attribute *attribute, char *buf)
516{
517 struct rps_map *map;
518 cpumask_var_t mask;
519 size_t len = 0;
520 int i;
521
522 if (!zalloc_cpumask_var(&mask, GFP_KERNEL))
523 return -ENOMEM;
524
525 rcu_read_lock();
526 map = rcu_dereference(queue->rps_map);
527 if (map)
528 for (i = 0; i < map->len; i++)
529 cpumask_set_cpu(map->cpus[i], mask);
530
531 len += cpumask_scnprintf(buf + len, PAGE_SIZE, mask);
532 if (PAGE_SIZE - len < 3) {
533 rcu_read_unlock();
534 free_cpumask_var(mask);
535 return -EINVAL;
536 }
537 rcu_read_unlock();
538
539 free_cpumask_var(mask);
540 len += sprintf(buf + len, "\n");
541 return len;
542}
543
544static void rps_map_release(struct rcu_head *rcu)
545{
546 struct rps_map *map = container_of(rcu, struct rps_map, rcu);
547
548 kfree(map);
549}
550
551ssize_t store_rps_map(struct netdev_rx_queue *queue,
552 struct rx_queue_attribute *attribute,
553 const char *buf, size_t len)
554{
555 struct rps_map *old_map, *map;
556 cpumask_var_t mask;
557 int err, cpu, i;
558 static DEFINE_SPINLOCK(rps_map_lock);
559
560 if (!capable(CAP_NET_ADMIN))
561 return -EPERM;
562
563 if (!alloc_cpumask_var(&mask, GFP_KERNEL))
564 return -ENOMEM;
565
566 err = bitmap_parse(buf, len, cpumask_bits(mask), nr_cpumask_bits);
567 if (err) {
568 free_cpumask_var(mask);
569 return err;
570 }
571
572 map = kzalloc(max_t(unsigned,
573 RPS_MAP_SIZE(cpumask_weight(mask)), L1_CACHE_BYTES),
574 GFP_KERNEL);
575 if (!map) {
576 free_cpumask_var(mask);
577 return -ENOMEM;
578 }
579
580 i = 0;
581 for_each_cpu_and(cpu, mask, cpu_online_mask)
582 map->cpus[i++] = cpu;
583
584 if (i)
585 map->len = i;
586 else {
587 kfree(map);
588 map = NULL;
589 }
590
591 spin_lock(&rps_map_lock);
592 old_map = queue->rps_map;
593 rcu_assign_pointer(queue->rps_map, map);
594 spin_unlock(&rps_map_lock);
595
596 if (old_map)
597 call_rcu(&old_map->rcu, rps_map_release);
598
599 free_cpumask_var(mask);
600 return len;
601}
602
603static struct rx_queue_attribute rps_cpus_attribute =
604 __ATTR(rps_cpus, S_IRUGO | S_IWUSR, show_rps_map, store_rps_map);
605
606static struct attribute *rx_queue_default_attrs[] = {
607 &rps_cpus_attribute.attr,
608 NULL
609};
610
611static void rx_queue_release(struct kobject *kobj)
612{
613 struct netdev_rx_queue *queue = to_rx_queue(kobj);
614 struct rps_map *map = queue->rps_map;
615 struct netdev_rx_queue *first = queue->first;
616
617 if (map)
618 call_rcu(&map->rcu, rps_map_release);
619
620 if (atomic_dec_and_test(&first->count))
621 kfree(first);
622}
623
624static struct kobj_type rx_queue_ktype = {
625 .sysfs_ops = &rx_queue_sysfs_ops,
626 .release = rx_queue_release,
627 .default_attrs = rx_queue_default_attrs,
628};
629
630static int rx_queue_add_kobject(struct net_device *net, int index)
631{
632 struct netdev_rx_queue *queue = net->_rx + index;
633 struct kobject *kobj = &queue->kobj;
634 int error = 0;
635
636 kobj->kset = net->queues_kset;
637 error = kobject_init_and_add(kobj, &rx_queue_ktype, NULL,
638 "rx-%u", index);
639 if (error) {
640 kobject_put(kobj);
641 return error;
642 }
643
644 kobject_uevent(kobj, KOBJ_ADD);
645
646 return error;
647}
648
649static int rx_queue_register_kobjects(struct net_device *net)
650{
651 int i;
652 int error = 0;
653
654 net->queues_kset = kset_create_and_add("queues",
655 NULL, &net->dev.kobj);
656 if (!net->queues_kset)
657 return -ENOMEM;
658 for (i = 0; i < net->num_rx_queues; i++) {
659 error = rx_queue_add_kobject(net, i);
660 if (error)
661 break;
662 }
663
664 if (error)
665 while (--i >= 0)
666 kobject_put(&net->_rx[i].kobj);
667
668 return error;
669}
670
671static void rx_queue_remove_kobjects(struct net_device *net)
672{
673 int i;
674
675 for (i = 0; i < net->num_rx_queues; i++)
676 kobject_put(&net->_rx[i].kobj);
677 kset_unregister(net->queues_kset);
678}
679#endif /* CONFIG_RPS */
469#endif /* CONFIG_SYSFS */ 680#endif /* CONFIG_SYSFS */
470 681
471#ifdef CONFIG_HOTPLUG 682#ifdef CONFIG_HOTPLUG
@@ -529,6 +740,10 @@ void netdev_unregister_kobject(struct net_device * net)
529 if (!net_eq(dev_net(net), &init_net)) 740 if (!net_eq(dev_net(net), &init_net))
530 return; 741 return;
531 742
743#ifdef CONFIG_RPS
744 rx_queue_remove_kobjects(net);
745#endif
746
532 device_del(dev); 747 device_del(dev);
533} 748}
534 749
@@ -537,6 +752,7 @@ int netdev_register_kobject(struct net_device *net)
537{ 752{
538 struct device *dev = &(net->dev); 753 struct device *dev = &(net->dev);
539 const struct attribute_group **groups = net->sysfs_groups; 754 const struct attribute_group **groups = net->sysfs_groups;
755 int error = 0;
540 756
541 dev->class = &net_class; 757 dev->class = &net_class;
542 dev->platform_data = net; 758 dev->platform_data = net;
@@ -563,7 +779,19 @@ int netdev_register_kobject(struct net_device *net)
563 if (!net_eq(dev_net(net), &init_net)) 779 if (!net_eq(dev_net(net), &init_net))
564 return 0; 780 return 0;
565 781
566 return device_add(dev); 782 error = device_add(dev);
783 if (error)
784 return error;
785
786#ifdef CONFIG_RPS
787 error = rx_queue_register_kobjects(net);
788 if (error) {
789 device_del(dev);
790 return error;
791 }
792#endif
793
794 return error;
567} 795}
568 796
569int netdev_class_create_file(struct class_attribute *class_attr) 797int netdev_class_create_file(struct class_attribute *class_attr)
diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index 7aa697253765..d4ec38fa64e6 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -735,7 +735,7 @@ int netpoll_setup(struct netpoll *np)
735 npinfo = kmalloc(sizeof(*npinfo), GFP_KERNEL); 735 npinfo = kmalloc(sizeof(*npinfo), GFP_KERNEL);
736 if (!npinfo) { 736 if (!npinfo) {
737 err = -ENOMEM; 737 err = -ENOMEM;
738 goto release; 738 goto put;
739 } 739 }
740 740
741 npinfo->rx_flags = 0; 741 npinfo->rx_flags = 0;
@@ -845,7 +845,7 @@ int netpoll_setup(struct netpoll *np)
845 845
846 kfree(npinfo); 846 kfree(npinfo);
847 } 847 }
848 848put:
849 dev_put(ndev); 849 dev_put(ndev);
850 return err; 850 return err;
851} 851}
diff --git a/net/core/pktgen.c b/net/core/pktgen.c
index 43923811bd6a..2ad68da418df 100644
--- a/net/core/pktgen.c
+++ b/net/core/pktgen.c
@@ -169,7 +169,7 @@
169#include <asm/dma.h> 169#include <asm/dma.h>
170#include <asm/div64.h> /* do_div */ 170#include <asm/div64.h> /* do_div */
171 171
172#define VERSION "2.72" 172#define VERSION "2.73"
173#define IP_NAME_SZ 32 173#define IP_NAME_SZ 32
174#define MAX_MPLS_LABELS 16 /* This is the max label stack depth */ 174#define MAX_MPLS_LABELS 16 /* This is the max label stack depth */
175#define MPLS_STACK_BOTTOM htonl(0x00000100) 175#define MPLS_STACK_BOTTOM htonl(0x00000100)
@@ -190,6 +190,7 @@
190#define F_IPSEC_ON (1<<12) /* ipsec on for flows */ 190#define F_IPSEC_ON (1<<12) /* ipsec on for flows */
191#define F_QUEUE_MAP_RND (1<<13) /* queue map Random */ 191#define F_QUEUE_MAP_RND (1<<13) /* queue map Random */
192#define F_QUEUE_MAP_CPU (1<<14) /* queue map mirrors smp_processor_id() */ 192#define F_QUEUE_MAP_CPU (1<<14) /* queue map mirrors smp_processor_id() */
193#define F_NODE (1<<15) /* Node memory alloc*/
193 194
194/* Thread control flag bits */ 195/* Thread control flag bits */
195#define T_STOP (1<<0) /* Stop run */ 196#define T_STOP (1<<0) /* Stop run */
@@ -372,6 +373,7 @@ struct pktgen_dev {
372 373
373 u16 queue_map_min; 374 u16 queue_map_min;
374 u16 queue_map_max; 375 u16 queue_map_max;
376 int node; /* Memory node */
375 377
376#ifdef CONFIG_XFRM 378#ifdef CONFIG_XFRM
377 __u8 ipsmode; /* IPSEC mode (config) */ 379 __u8 ipsmode; /* IPSEC mode (config) */
@@ -607,6 +609,9 @@ static int pktgen_if_show(struct seq_file *seq, void *v)
607 if (pkt_dev->traffic_class) 609 if (pkt_dev->traffic_class)
608 seq_printf(seq, " traffic_class: 0x%02x\n", pkt_dev->traffic_class); 610 seq_printf(seq, " traffic_class: 0x%02x\n", pkt_dev->traffic_class);
609 611
612 if (pkt_dev->node >= 0)
613 seq_printf(seq, " node: %d\n", pkt_dev->node);
614
610 seq_printf(seq, " Flags: "); 615 seq_printf(seq, " Flags: ");
611 616
612 if (pkt_dev->flags & F_IPV6) 617 if (pkt_dev->flags & F_IPV6)
@@ -660,6 +665,9 @@ static int pktgen_if_show(struct seq_file *seq, void *v)
660 if (pkt_dev->flags & F_SVID_RND) 665 if (pkt_dev->flags & F_SVID_RND)
661 seq_printf(seq, "SVID_RND "); 666 seq_printf(seq, "SVID_RND ");
662 667
668 if (pkt_dev->flags & F_NODE)
669 seq_printf(seq, "NODE_ALLOC ");
670
663 seq_puts(seq, "\n"); 671 seq_puts(seq, "\n");
664 672
665 /* not really stopped, more like last-running-at */ 673 /* not really stopped, more like last-running-at */
@@ -1074,6 +1082,21 @@ static ssize_t pktgen_if_write(struct file *file,
1074 pkt_dev->dst_mac_count); 1082 pkt_dev->dst_mac_count);
1075 return count; 1083 return count;
1076 } 1084 }
1085 if (!strcmp(name, "node")) {
1086 len = num_arg(&user_buffer[i], 10, &value);
1087 if (len < 0)
1088 return len;
1089
1090 i += len;
1091
1092 if (node_possible(value)) {
1093 pkt_dev->node = value;
1094 sprintf(pg_result, "OK: node=%d", pkt_dev->node);
1095 }
1096 else
1097 sprintf(pg_result, "ERROR: node not possible");
1098 return count;
1099 }
1077 if (!strcmp(name, "flag")) { 1100 if (!strcmp(name, "flag")) {
1078 char f[32]; 1101 char f[32];
1079 memset(f, 0, 32); 1102 memset(f, 0, 32);
@@ -1166,12 +1189,18 @@ static ssize_t pktgen_if_write(struct file *file,
1166 else if (strcmp(f, "!IPV6") == 0) 1189 else if (strcmp(f, "!IPV6") == 0)
1167 pkt_dev->flags &= ~F_IPV6; 1190 pkt_dev->flags &= ~F_IPV6;
1168 1191
1192 else if (strcmp(f, "NODE_ALLOC") == 0)
1193 pkt_dev->flags |= F_NODE;
1194
1195 else if (strcmp(f, "!NODE_ALLOC") == 0)
1196 pkt_dev->flags &= ~F_NODE;
1197
1169 else { 1198 else {
1170 sprintf(pg_result, 1199 sprintf(pg_result,
1171 "Flag -:%s:- unknown\nAvailable flags, (prepend ! to un-set flag):\n%s", 1200 "Flag -:%s:- unknown\nAvailable flags, (prepend ! to un-set flag):\n%s",
1172 f, 1201 f,
1173 "IPSRC_RND, IPDST_RND, UDPSRC_RND, UDPDST_RND, " 1202 "IPSRC_RND, IPDST_RND, UDPSRC_RND, UDPDST_RND, "
1174 "MACSRC_RND, MACDST_RND, TXSIZE_RND, IPV6, MPLS_RND, VID_RND, SVID_RND, FLOW_SEQ, IPSEC\n"); 1203 "MACSRC_RND, MACDST_RND, TXSIZE_RND, IPV6, MPLS_RND, VID_RND, SVID_RND, FLOW_SEQ, IPSEC, NODE_ALLOC\n");
1175 return count; 1204 return count;
1176 } 1205 }
1177 sprintf(pg_result, "OK: flags=0x%x", pkt_dev->flags); 1206 sprintf(pg_result, "OK: flags=0x%x", pkt_dev->flags);
@@ -2572,9 +2601,27 @@ static struct sk_buff *fill_packet_ipv4(struct net_device *odev,
2572 mod_cur_headers(pkt_dev); 2601 mod_cur_headers(pkt_dev);
2573 2602
2574 datalen = (odev->hard_header_len + 16) & ~0xf; 2603 datalen = (odev->hard_header_len + 16) & ~0xf;
2575 skb = __netdev_alloc_skb(odev, 2604
2576 pkt_dev->cur_pkt_size + 64 2605 if (pkt_dev->flags & F_NODE) {
2577 + datalen + pkt_dev->pkt_overhead, GFP_NOWAIT); 2606 int node;
2607
2608 if (pkt_dev->node >= 0)
2609 node = pkt_dev->node;
2610 else
2611 node = numa_node_id();
2612
2613 skb = __alloc_skb(NET_SKB_PAD + pkt_dev->cur_pkt_size + 64
2614 + datalen + pkt_dev->pkt_overhead, GFP_NOWAIT, 0, node);
2615 if (likely(skb)) {
2616 skb_reserve(skb, NET_SKB_PAD);
2617 skb->dev = odev;
2618 }
2619 }
2620 else
2621 skb = __netdev_alloc_skb(odev,
2622 pkt_dev->cur_pkt_size + 64
2623 + datalen + pkt_dev->pkt_overhead, GFP_NOWAIT);
2624
2578 if (!skb) { 2625 if (!skb) {
2579 sprintf(pkt_dev->result, "No memory"); 2626 sprintf(pkt_dev->result, "No memory");
2580 return NULL; 2627 return NULL;
@@ -3674,6 +3721,7 @@ static int pktgen_add_device(struct pktgen_thread *t, const char *ifname)
3674 pkt_dev->svlan_p = 0; 3721 pkt_dev->svlan_p = 0;
3675 pkt_dev->svlan_cfi = 0; 3722 pkt_dev->svlan_cfi = 0;
3676 pkt_dev->svlan_id = 0xffff; 3723 pkt_dev->svlan_id = 0xffff;
3724 pkt_dev->node = -1;
3677 3725
3678 err = pktgen_setup_dev(pkt_dev, ifname); 3726 err = pktgen_setup_dev(pkt_dev, ifname);
3679 if (err) 3727 if (err)
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index 4568120d8533..bf919b6acea2 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -600,7 +600,41 @@ static void copy_rtnl_link_stats(struct rtnl_link_stats *a,
600 600
601 a->rx_compressed = b->rx_compressed; 601 a->rx_compressed = b->rx_compressed;
602 a->tx_compressed = b->tx_compressed; 602 a->tx_compressed = b->tx_compressed;
603}; 603}
604
605static void copy_rtnl_link_stats64(void *v, const struct net_device_stats *b)
606{
607 struct rtnl_link_stats64 a;
608
609 a.rx_packets = b->rx_packets;
610 a.tx_packets = b->tx_packets;
611 a.rx_bytes = b->rx_bytes;
612 a.tx_bytes = b->tx_bytes;
613 a.rx_errors = b->rx_errors;
614 a.tx_errors = b->tx_errors;
615 a.rx_dropped = b->rx_dropped;
616 a.tx_dropped = b->tx_dropped;
617
618 a.multicast = b->multicast;
619 a.collisions = b->collisions;
620
621 a.rx_length_errors = b->rx_length_errors;
622 a.rx_over_errors = b->rx_over_errors;
623 a.rx_crc_errors = b->rx_crc_errors;
624 a.rx_frame_errors = b->rx_frame_errors;
625 a.rx_fifo_errors = b->rx_fifo_errors;
626 a.rx_missed_errors = b->rx_missed_errors;
627
628 a.tx_aborted_errors = b->tx_aborted_errors;
629 a.tx_carrier_errors = b->tx_carrier_errors;
630 a.tx_fifo_errors = b->tx_fifo_errors;
631 a.tx_heartbeat_errors = b->tx_heartbeat_errors;
632 a.tx_window_errors = b->tx_window_errors;
633
634 a.rx_compressed = b->rx_compressed;
635 a.tx_compressed = b->tx_compressed;
636 memcpy(v, &a, sizeof(a));
637}
604 638
605static inline int rtnl_vfinfo_size(const struct net_device *dev) 639static inline int rtnl_vfinfo_size(const struct net_device *dev)
606{ 640{
@@ -619,6 +653,7 @@ static inline size_t if_nlmsg_size(const struct net_device *dev)
619 + nla_total_size(IFNAMSIZ) /* IFLA_QDISC */ 653 + nla_total_size(IFNAMSIZ) /* IFLA_QDISC */
620 + nla_total_size(sizeof(struct rtnl_link_ifmap)) 654 + nla_total_size(sizeof(struct rtnl_link_ifmap))
621 + nla_total_size(sizeof(struct rtnl_link_stats)) 655 + nla_total_size(sizeof(struct rtnl_link_stats))
656 + nla_total_size(sizeof(struct rtnl_link_stats64))
622 + nla_total_size(MAX_ADDR_LEN) /* IFLA_ADDRESS */ 657 + nla_total_size(MAX_ADDR_LEN) /* IFLA_ADDRESS */
623 + nla_total_size(MAX_ADDR_LEN) /* IFLA_BROADCAST */ 658 + nla_total_size(MAX_ADDR_LEN) /* IFLA_BROADCAST */
624 + nla_total_size(4) /* IFLA_TXQLEN */ 659 + nla_total_size(4) /* IFLA_TXQLEN */
@@ -698,6 +733,12 @@ static int rtnl_fill_ifinfo(struct sk_buff *skb, struct net_device *dev,
698 stats = dev_get_stats(dev); 733 stats = dev_get_stats(dev);
699 copy_rtnl_link_stats(nla_data(attr), stats); 734 copy_rtnl_link_stats(nla_data(attr), stats);
700 735
736 attr = nla_reserve(skb, IFLA_STATS64,
737 sizeof(struct rtnl_link_stats64));
738 if (attr == NULL)
739 goto nla_put_failure;
740 copy_rtnl_link_stats64(nla_data(attr), stats);
741
701 if (dev->netdev_ops->ndo_get_vf_config && dev->dev.parent) { 742 if (dev->netdev_ops->ndo_get_vf_config && dev->dev.parent) {
702 int i; 743 int i;
703 struct ifla_vf_info ivi; 744 struct ifla_vf_info ivi;
@@ -1473,6 +1514,7 @@ static int rtnetlink_event(struct notifier_block *this, unsigned long event, voi
1473 case NETDEV_POST_INIT: 1514 case NETDEV_POST_INIT:
1474 case NETDEV_REGISTER: 1515 case NETDEV_REGISTER:
1475 case NETDEV_CHANGE: 1516 case NETDEV_CHANGE:
1517 case NETDEV_PRE_TYPE_CHANGE:
1476 case NETDEV_GOING_DOWN: 1518 case NETDEV_GOING_DOWN:
1477 case NETDEV_UNREGISTER: 1519 case NETDEV_UNREGISTER:
1478 case NETDEV_UNREGISTER_BATCH: 1520 case NETDEV_UNREGISTER_BATCH:
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 93c4e060c91e..bdea0efdf8cb 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -534,6 +534,7 @@ static void __copy_skb_header(struct sk_buff *new, const struct sk_buff *old)
534 new->network_header = old->network_header; 534 new->network_header = old->network_header;
535 new->mac_header = old->mac_header; 535 new->mac_header = old->mac_header;
536 skb_dst_set(new, dst_clone(skb_dst(old))); 536 skb_dst_set(new, dst_clone(skb_dst(old)));
537 new->rxhash = old->rxhash;
537#ifdef CONFIG_XFRM 538#ifdef CONFIG_XFRM
538 new->sp = secpath_get(old->sp); 539 new->sp = secpath_get(old->sp);
539#endif 540#endif
@@ -581,6 +582,7 @@ static struct sk_buff *__skb_clone(struct sk_buff *n, struct sk_buff *skb)
581 C(len); 582 C(len);
582 C(data_len); 583 C(data_len);
583 C(mac_len); 584 C(mac_len);
585 C(rxhash);
584 n->hdr_len = skb->nohdr ? skb_headroom(skb) : skb->hdr_len; 586 n->hdr_len = skb->nohdr ? skb_headroom(skb) : skb->hdr_len;
585 n->cloned = 1; 587 n->cloned = 1;
586 n->nohdr = 0; 588 n->nohdr = 0;
diff --git a/net/core/sock.c b/net/core/sock.c
index fcd397a762ff..c5812bbc2cc9 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -340,8 +340,12 @@ int sk_receive_skb(struct sock *sk, struct sk_buff *skb, const int nested)
340 rc = sk_backlog_rcv(sk, skb); 340 rc = sk_backlog_rcv(sk, skb);
341 341
342 mutex_release(&sk->sk_lock.dep_map, 1, _RET_IP_); 342 mutex_release(&sk->sk_lock.dep_map, 1, _RET_IP_);
343 } else 343 } else if (sk_add_backlog(sk, skb)) {
344 sk_add_backlog(sk, skb); 344 bh_unlock_sock(sk);
345 atomic_inc(&sk->sk_drops);
346 goto discard_and_relse;
347 }
348
345 bh_unlock_sock(sk); 349 bh_unlock_sock(sk);
346out: 350out:
347 sock_put(sk); 351 sock_put(sk);
@@ -1139,6 +1143,7 @@ struct sock *sk_clone(const struct sock *sk, const gfp_t priority)
1139 sock_lock_init(newsk); 1143 sock_lock_init(newsk);
1140 bh_lock_sock(newsk); 1144 bh_lock_sock(newsk);
1141 newsk->sk_backlog.head = newsk->sk_backlog.tail = NULL; 1145 newsk->sk_backlog.head = newsk->sk_backlog.tail = NULL;
1146 newsk->sk_backlog.len = 0;
1142 1147
1143 atomic_set(&newsk->sk_rmem_alloc, 0); 1148 atomic_set(&newsk->sk_rmem_alloc, 0);
1144 /* 1149 /*
@@ -1542,6 +1547,12 @@ static void __release_sock(struct sock *sk)
1542 1547
1543 bh_lock_sock(sk); 1548 bh_lock_sock(sk);
1544 } while ((skb = sk->sk_backlog.head) != NULL); 1549 } while ((skb = sk->sk_backlog.head) != NULL);
1550
1551 /*
1552 * Doing the zeroing here guarantee we can not loop forever
1553 * while a wild producer attempts to flood us.
1554 */
1555 sk->sk_backlog.len = 0;
1545} 1556}
1546 1557
1547/** 1558/**
@@ -1874,6 +1885,7 @@ void sock_init_data(struct socket *sock, struct sock *sk)
1874 sk->sk_allocation = GFP_KERNEL; 1885 sk->sk_allocation = GFP_KERNEL;
1875 sk->sk_rcvbuf = sysctl_rmem_default; 1886 sk->sk_rcvbuf = sysctl_rmem_default;
1876 sk->sk_sndbuf = sysctl_wmem_default; 1887 sk->sk_sndbuf = sysctl_wmem_default;
1888 sk->sk_backlog.limit = sk->sk_rcvbuf << 1;
1877 sk->sk_state = TCP_CLOSE; 1889 sk->sk_state = TCP_CLOSE;
1878 sk_set_socket(sk, sock); 1890 sk_set_socket(sk, sock);
1879 1891
@@ -2276,7 +2288,8 @@ out_free_request_sock_slab:
2276 prot->rsk_prot->slab = NULL; 2288 prot->rsk_prot->slab = NULL;
2277 } 2289 }
2278out_free_request_sock_slab_name: 2290out_free_request_sock_slab_name:
2279 kfree(prot->rsk_prot->slab_name); 2291 if (prot->rsk_prot)
2292 kfree(prot->rsk_prot->slab_name);
2280out_free_sock_slab: 2293out_free_sock_slab:
2281 kmem_cache_destroy(prot->slab); 2294 kmem_cache_destroy(prot->slab);
2282 prot->slab = NULL; 2295 prot->slab = NULL;
diff --git a/net/dccp/ccids/ccid3.c b/net/dccp/ccids/ccid3.c
index bcd7632299f5..d3235899c7e3 100644
--- a/net/dccp/ccids/ccid3.c
+++ b/net/dccp/ccids/ccid3.c
@@ -208,7 +208,7 @@ static void ccid3_hc_tx_no_feedback_timer(unsigned long data)
208 goto restart_timer; 208 goto restart_timer;
209 } 209 }
210 210
211 ccid3_pr_debug("%s(%p, state=%s) - entry \n", dccp_role(sk), sk, 211 ccid3_pr_debug("%s(%p, state=%s) - entry\n", dccp_role(sk), sk,
212 ccid3_tx_state_name(hc->tx_state)); 212 ccid3_tx_state_name(hc->tx_state));
213 213
214 if (hc->tx_state == TFRC_SSTATE_FBACK) 214 if (hc->tx_state == TFRC_SSTATE_FBACK)
diff --git a/net/dccp/dccp.h b/net/dccp/dccp.h
index 5ef32c2f0d6a..53f8e12d0c10 100644
--- a/net/dccp/dccp.h
+++ b/net/dccp/dccp.h
@@ -189,7 +189,7 @@ enum {
189#define DCCP_MIB_MAX __DCCP_MIB_MAX 189#define DCCP_MIB_MAX __DCCP_MIB_MAX
190struct dccp_mib { 190struct dccp_mib {
191 unsigned long mibs[DCCP_MIB_MAX]; 191 unsigned long mibs[DCCP_MIB_MAX];
192} __SNMP_MIB_ALIGN__; 192};
193 193
194DECLARE_SNMP_STAT(struct dccp_mib, dccp_statistics); 194DECLARE_SNMP_STAT(struct dccp_mib, dccp_statistics);
195#define DCCP_INC_STATS(field) SNMP_INC_STATS(dccp_statistics, field) 195#define DCCP_INC_STATS(field) SNMP_INC_STATS(dccp_statistics, field)
diff --git a/net/dccp/input.c b/net/dccp/input.c
index 7648f316310f..5daa4bdfdb10 100644
--- a/net/dccp/input.c
+++ b/net/dccp/input.c
@@ -414,7 +414,7 @@ static int dccp_rcv_request_sent_state_process(struct sock *sk,
414 if (!between48(DCCP_SKB_CB(skb)->dccpd_ack_seq, 414 if (!between48(DCCP_SKB_CB(skb)->dccpd_ack_seq,
415 dp->dccps_awl, dp->dccps_awh)) { 415 dp->dccps_awl, dp->dccps_awh)) {
416 dccp_pr_debug("invalid ackno: S.AWL=%llu, " 416 dccp_pr_debug("invalid ackno: S.AWL=%llu, "
417 "P.ackno=%llu, S.AWH=%llu \n", 417 "P.ackno=%llu, S.AWH=%llu\n",
418 (unsigned long long)dp->dccps_awl, 418 (unsigned long long)dp->dccps_awl,
419 (unsigned long long)DCCP_SKB_CB(skb)->dccpd_ack_seq, 419 (unsigned long long)DCCP_SKB_CB(skb)->dccpd_ack_seq,
420 (unsigned long long)dp->dccps_awh); 420 (unsigned long long)dp->dccps_awh);
diff --git a/net/dccp/ipv4.c b/net/dccp/ipv4.c
index b195c4feaa0a..4071eaf2b361 100644
--- a/net/dccp/ipv4.c
+++ b/net/dccp/ipv4.c
@@ -998,11 +998,11 @@ static struct inet_protosw dccp_v4_protosw = {
998 998
999static int __net_init dccp_v4_init_net(struct net *net) 999static int __net_init dccp_v4_init_net(struct net *net)
1000{ 1000{
1001 int err; 1001 if (dccp_hashinfo.bhash == NULL)
1002 return -ESOCKTNOSUPPORT;
1002 1003
1003 err = inet_ctl_sock_create(&net->dccp.v4_ctl_sk, PF_INET, 1004 return inet_ctl_sock_create(&net->dccp.v4_ctl_sk, PF_INET,
1004 SOCK_DCCP, IPPROTO_DCCP, net); 1005 SOCK_DCCP, IPPROTO_DCCP, net);
1005 return err;
1006} 1006}
1007 1007
1008static void __net_exit dccp_v4_exit_net(struct net *net) 1008static void __net_exit dccp_v4_exit_net(struct net *net)
diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c
index 1aec6349e858..af3394df63b7 100644
--- a/net/dccp/ipv6.c
+++ b/net/dccp/ipv6.c
@@ -1191,11 +1191,11 @@ static struct inet_protosw dccp_v6_protosw = {
1191 1191
1192static int __net_init dccp_v6_init_net(struct net *net) 1192static int __net_init dccp_v6_init_net(struct net *net)
1193{ 1193{
1194 int err; 1194 if (dccp_hashinfo.bhash == NULL)
1195 return -ESOCKTNOSUPPORT;
1195 1196
1196 err = inet_ctl_sock_create(&net->dccp.v6_ctl_sk, PF_INET6, 1197 return inet_ctl_sock_create(&net->dccp.v6_ctl_sk, PF_INET6,
1197 SOCK_DCCP, IPPROTO_DCCP, net); 1198 SOCK_DCCP, IPPROTO_DCCP, net);
1198 return err;
1199} 1199}
1200 1200
1201static void __net_exit dccp_v6_exit_net(struct net *net) 1201static void __net_exit dccp_v6_exit_net(struct net *net)
diff --git a/net/dccp/minisocks.c b/net/dccp/minisocks.c
index af226a063141..0d508c359fa9 100644
--- a/net/dccp/minisocks.c
+++ b/net/dccp/minisocks.c
@@ -254,7 +254,7 @@ int dccp_child_process(struct sock *parent, struct sock *child,
254 * in main socket hash table and lock on listening 254 * in main socket hash table and lock on listening
255 * socket does not protect us more. 255 * socket does not protect us more.
256 */ 256 */
257 sk_add_backlog(child, skb); 257 __sk_add_backlog(child, skb);
258 } 258 }
259 259
260 bh_unlock_sock(child); 260 bh_unlock_sock(child);
diff --git a/net/dccp/proto.c b/net/dccp/proto.c
index 0ef7061920c0..aa4cef374fd0 100644
--- a/net/dccp/proto.c
+++ b/net/dccp/proto.c
@@ -1036,7 +1036,7 @@ static int __init dccp_init(void)
1036 FIELD_SIZEOF(struct sk_buff, cb)); 1036 FIELD_SIZEOF(struct sk_buff, cb));
1037 rc = percpu_counter_init(&dccp_orphan_count, 0); 1037 rc = percpu_counter_init(&dccp_orphan_count, 0);
1038 if (rc) 1038 if (rc)
1039 goto out; 1039 goto out_fail;
1040 rc = -ENOBUFS; 1040 rc = -ENOBUFS;
1041 inet_hashinfo_init(&dccp_hashinfo); 1041 inet_hashinfo_init(&dccp_hashinfo);
1042 dccp_hashinfo.bind_bucket_cachep = 1042 dccp_hashinfo.bind_bucket_cachep =
@@ -1125,8 +1125,9 @@ static int __init dccp_init(void)
1125 goto out_sysctl_exit; 1125 goto out_sysctl_exit;
1126 1126
1127 dccp_timestamping_init(); 1127 dccp_timestamping_init();
1128out: 1128
1129 return rc; 1129 return 0;
1130
1130out_sysctl_exit: 1131out_sysctl_exit:
1131 dccp_sysctl_exit(); 1132 dccp_sysctl_exit();
1132out_ackvec_exit: 1133out_ackvec_exit:
@@ -1135,18 +1136,19 @@ out_free_dccp_mib:
1135 dccp_mib_exit(); 1136 dccp_mib_exit();
1136out_free_dccp_bhash: 1137out_free_dccp_bhash:
1137 free_pages((unsigned long)dccp_hashinfo.bhash, bhash_order); 1138 free_pages((unsigned long)dccp_hashinfo.bhash, bhash_order);
1138 dccp_hashinfo.bhash = NULL;
1139out_free_dccp_locks: 1139out_free_dccp_locks:
1140 inet_ehash_locks_free(&dccp_hashinfo); 1140 inet_ehash_locks_free(&dccp_hashinfo);
1141out_free_dccp_ehash: 1141out_free_dccp_ehash:
1142 free_pages((unsigned long)dccp_hashinfo.ehash, ehash_order); 1142 free_pages((unsigned long)dccp_hashinfo.ehash, ehash_order);
1143 dccp_hashinfo.ehash = NULL;
1144out_free_bind_bucket_cachep: 1143out_free_bind_bucket_cachep:
1145 kmem_cache_destroy(dccp_hashinfo.bind_bucket_cachep); 1144 kmem_cache_destroy(dccp_hashinfo.bind_bucket_cachep);
1146 dccp_hashinfo.bind_bucket_cachep = NULL;
1147out_free_percpu: 1145out_free_percpu:
1148 percpu_counter_destroy(&dccp_orphan_count); 1146 percpu_counter_destroy(&dccp_orphan_count);
1149 goto out; 1147out_fail:
1148 dccp_hashinfo.bhash = NULL;
1149 dccp_hashinfo.ehash = NULL;
1150 dccp_hashinfo.bind_bucket_cachep = NULL;
1151 return rc;
1150} 1152}
1151 1153
1152static void __exit dccp_fini(void) 1154static void __exit dccp_fini(void)
diff --git a/net/ipv4/Kconfig b/net/ipv4/Kconfig
index 0c94a1ac2946..c9a1c68767ff 100644
--- a/net/ipv4/Kconfig
+++ b/net/ipv4/Kconfig
@@ -587,9 +587,15 @@ choice
587 config DEFAULT_HTCP 587 config DEFAULT_HTCP
588 bool "Htcp" if TCP_CONG_HTCP=y 588 bool "Htcp" if TCP_CONG_HTCP=y
589 589
590 config DEFAULT_HYBLA
591 bool "Hybla" if TCP_CONG_HYBLA=y
592
590 config DEFAULT_VEGAS 593 config DEFAULT_VEGAS
591 bool "Vegas" if TCP_CONG_VEGAS=y 594 bool "Vegas" if TCP_CONG_VEGAS=y
592 595
596 config DEFAULT_VENO
597 bool "Veno" if TCP_CONG_VENO=y
598
593 config DEFAULT_WESTWOOD 599 config DEFAULT_WESTWOOD
594 bool "Westwood" if TCP_CONG_WESTWOOD=y 600 bool "Westwood" if TCP_CONG_WESTWOOD=y
595 601
@@ -610,8 +616,10 @@ config DEFAULT_TCP_CONG
610 default "bic" if DEFAULT_BIC 616 default "bic" if DEFAULT_BIC
611 default "cubic" if DEFAULT_CUBIC 617 default "cubic" if DEFAULT_CUBIC
612 default "htcp" if DEFAULT_HTCP 618 default "htcp" if DEFAULT_HTCP
619 default "hybla" if DEFAULT_HYBLA
613 default "vegas" if DEFAULT_VEGAS 620 default "vegas" if DEFAULT_VEGAS
614 default "westwood" if DEFAULT_WESTWOOD 621 default "westwood" if DEFAULT_WESTWOOD
622 default "veno" if DEFAULT_VENO
615 default "reno" if DEFAULT_RENO 623 default "reno" if DEFAULT_RENO
616 default "cubic" 624 default "cubic"
617 625
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 33b7dffa7732..55e11906a73a 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -1401,10 +1401,10 @@ EXPORT_SYMBOL_GPL(snmp_fold_field);
1401int snmp_mib_init(void __percpu *ptr[2], size_t mibsize) 1401int snmp_mib_init(void __percpu *ptr[2], size_t mibsize)
1402{ 1402{
1403 BUG_ON(ptr == NULL); 1403 BUG_ON(ptr == NULL);
1404 ptr[0] = __alloc_percpu(mibsize, __alignof__(unsigned long long)); 1404 ptr[0] = __alloc_percpu(mibsize, __alignof__(unsigned long));
1405 if (!ptr[0]) 1405 if (!ptr[0])
1406 goto err0; 1406 goto err0;
1407 ptr[1] = __alloc_percpu(mibsize, __alignof__(unsigned long long)); 1407 ptr[1] = __alloc_percpu(mibsize, __alignof__(unsigned long));
1408 if (!ptr[1]) 1408 if (!ptr[1])
1409 goto err1; 1409 goto err1;
1410 return 0; 1410 return 0;
diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 51ca946e3392..c75320ef95c2 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -1095,10 +1095,10 @@ static int inetdev_event(struct notifier_block *this, unsigned long event,
1095 case NETDEV_DOWN: 1095 case NETDEV_DOWN:
1096 ip_mc_down(in_dev); 1096 ip_mc_down(in_dev);
1097 break; 1097 break;
1098 case NETDEV_BONDING_OLDTYPE: 1098 case NETDEV_PRE_TYPE_CHANGE:
1099 ip_mc_unmap(in_dev); 1099 ip_mc_unmap(in_dev);
1100 break; 1100 break;
1101 case NETDEV_BONDING_NEWTYPE: 1101 case NETDEV_POST_TYPE_CHANGE:
1102 ip_mc_remap(in_dev); 1102 ip_mc_remap(in_dev);
1103 break; 1103 break;
1104 case NETDEV_CHANGEMTU: 1104 case NETDEV_CHANGEMTU:
diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index c0c5274d0271..f47c9f76754b 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -1144,12 +1144,9 @@ static int ipgre_header(struct sk_buff *skb, struct net_device *dev,
1144 1144
1145 if (saddr) 1145 if (saddr)
1146 memcpy(&iph->saddr, saddr, 4); 1146 memcpy(&iph->saddr, saddr, 4);
1147 1147 if (daddr)
1148 if (daddr) {
1149 memcpy(&iph->daddr, daddr, 4); 1148 memcpy(&iph->daddr, daddr, 4);
1150 return t->hlen; 1149 if (iph->daddr)
1151 }
1152 if (iph->daddr && !ipv4_is_multicast(iph->daddr))
1153 return t->hlen; 1150 return t->hlen;
1154 1151
1155 return -t->hlen; 1152 return -t->hlen;
diff --git a/net/ipv4/ipconfig.c b/net/ipv4/ipconfig.c
index 10a6a604bf32..bf12d2a7a0c7 100644
--- a/net/ipv4/ipconfig.c
+++ b/net/ipv4/ipconfig.c
@@ -187,6 +187,16 @@ struct ic_device {
187static struct ic_device *ic_first_dev __initdata = NULL;/* List of open device */ 187static struct ic_device *ic_first_dev __initdata = NULL;/* List of open device */
188static struct net_device *ic_dev __initdata = NULL; /* Selected device */ 188static struct net_device *ic_dev __initdata = NULL; /* Selected device */
189 189
190static bool __init ic_device_match(struct net_device *dev)
191{
192 if (user_dev_name[0] ? !strcmp(dev->name, user_dev_name) :
193 (!(dev->flags & IFF_LOOPBACK) &&
194 (dev->flags & (IFF_POINTOPOINT|IFF_BROADCAST)) &&
195 strncmp(dev->name, "dummy", 5)))
196 return true;
197 return false;
198}
199
190static int __init ic_open_devs(void) 200static int __init ic_open_devs(void)
191{ 201{
192 struct ic_device *d, **last; 202 struct ic_device *d, **last;
@@ -207,10 +217,7 @@ static int __init ic_open_devs(void)
207 for_each_netdev(&init_net, dev) { 217 for_each_netdev(&init_net, dev) {
208 if (dev->flags & IFF_LOOPBACK) 218 if (dev->flags & IFF_LOOPBACK)
209 continue; 219 continue;
210 if (user_dev_name[0] ? !strcmp(dev->name, user_dev_name) : 220 if (ic_device_match(dev)) {
211 (!(dev->flags & IFF_LOOPBACK) &&
212 (dev->flags & (IFF_POINTOPOINT|IFF_BROADCAST)) &&
213 strncmp(dev->name, "dummy", 5))) {
214 int able = 0; 221 int able = 0;
215 if (dev->mtu >= 364) 222 if (dev->mtu >= 364)
216 able |= IC_BOOTP; 223 able |= IC_BOOTP;
@@ -228,7 +235,7 @@ static int __init ic_open_devs(void)
228 } 235 }
229 if (!(d = kmalloc(sizeof(struct ic_device), GFP_KERNEL))) { 236 if (!(d = kmalloc(sizeof(struct ic_device), GFP_KERNEL))) {
230 rtnl_unlock(); 237 rtnl_unlock();
231 return -1; 238 return -ENOMEM;
232 } 239 }
233 d->dev = dev; 240 d->dev = dev;
234 *last = d; 241 *last = d;
@@ -253,7 +260,7 @@ static int __init ic_open_devs(void)
253 printk(KERN_ERR "IP-Config: Device `%s' not found.\n", user_dev_name); 260 printk(KERN_ERR "IP-Config: Device `%s' not found.\n", user_dev_name);
254 else 261 else
255 printk(KERN_ERR "IP-Config: No network devices available.\n"); 262 printk(KERN_ERR "IP-Config: No network devices available.\n");
256 return -1; 263 return -ENODEV;
257 } 264 }
258 return 0; 265 return 0;
259} 266}
@@ -968,7 +975,7 @@ static int __init ic_bootp_recv(struct sk_buff *skb, struct net_device *dev, str
968 /* Is it a reply for the device we are configuring? */ 975 /* Is it a reply for the device we are configuring? */
969 if (b->xid != ic_dev_xid) { 976 if (b->xid != ic_dev_xid) {
970 if (net_ratelimit()) 977 if (net_ratelimit())
971 printk(KERN_ERR "DHCP/BOOTP: Ignoring delayed packet \n"); 978 printk(KERN_ERR "DHCP/BOOTP: Ignoring delayed packet\n");
972 goto drop_unlock; 979 goto drop_unlock;
973 } 980 }
974 981
@@ -1303,6 +1310,32 @@ __be32 __init root_nfs_parse_addr(char *name)
1303 return addr; 1310 return addr;
1304} 1311}
1305 1312
1313#define DEVICE_WAIT_MAX 12 /* 12 seconds */
1314
1315static int __init wait_for_devices(void)
1316{
1317 int i;
1318
1319 msleep(CONF_PRE_OPEN);
1320 for (i = 0; i < DEVICE_WAIT_MAX; i++) {
1321 struct net_device *dev;
1322 int found = 0;
1323
1324 rtnl_lock();
1325 for_each_netdev(&init_net, dev) {
1326 if (ic_device_match(dev)) {
1327 found = 1;
1328 break;
1329 }
1330 }
1331 rtnl_unlock();
1332 if (found)
1333 return 0;
1334 ssleep(1);
1335 }
1336 return -ENODEV;
1337}
1338
1306/* 1339/*
1307 * IP Autoconfig dispatcher. 1340 * IP Autoconfig dispatcher.
1308 */ 1341 */
@@ -1313,6 +1346,7 @@ static int __init ip_auto_config(void)
1313#ifdef IPCONFIG_DYNAMIC 1346#ifdef IPCONFIG_DYNAMIC
1314 int retries = CONF_OPEN_RETRIES; 1347 int retries = CONF_OPEN_RETRIES;
1315#endif 1348#endif
1349 int err;
1316 1350
1317#ifdef CONFIG_PROC_FS 1351#ifdef CONFIG_PROC_FS
1318 proc_net_fops_create(&init_net, "pnp", S_IRUGO, &pnp_seq_fops); 1352 proc_net_fops_create(&init_net, "pnp", S_IRUGO, &pnp_seq_fops);
@@ -1325,12 +1359,15 @@ static int __init ip_auto_config(void)
1325#ifdef IPCONFIG_DYNAMIC 1359#ifdef IPCONFIG_DYNAMIC
1326 try_try_again: 1360 try_try_again:
1327#endif 1361#endif
1328 /* Give hardware a chance to settle */ 1362 /* Wait for devices to appear */
1329 msleep(CONF_PRE_OPEN); 1363 err = wait_for_devices();
1364 if (err)
1365 return err;
1330 1366
1331 /* Setup all network devices */ 1367 /* Setup all network devices */
1332 if (ic_open_devs() < 0) 1368 err = ic_open_devs();
1333 return -1; 1369 if (err)
1370 return err;
1334 1371
1335 /* Give drivers a chance to settle */ 1372 /* Give drivers a chance to settle */
1336 ssleep(CONF_POST_OPEN); 1373 ssleep(CONF_POST_OPEN);
diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
index 8582e12e4a62..0b9d03c54dc3 100644
--- a/net/ipv4/ipmr.c
+++ b/net/ipv4/ipmr.c
@@ -802,6 +802,9 @@ static int ipmr_mfc_add(struct net *net, struct mfcctl *mfc, int mrtsock)
802 int line; 802 int line;
803 struct mfc_cache *uc, *c, **cp; 803 struct mfc_cache *uc, *c, **cp;
804 804
805 if (mfc->mfcc_parent >= MAXVIFS)
806 return -ENFILE;
807
805 line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); 808 line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr);
806 809
807 for (cp = &net->ipv4.mfc_cache_array[line]; 810 for (cp = &net->ipv4.mfc_cache_array[line];
diff --git a/net/ipv4/proc.c b/net/ipv4/proc.c
index 242ed2307370..3dc9914c1dce 100644
--- a/net/ipv4/proc.c
+++ b/net/ipv4/proc.c
@@ -249,6 +249,9 @@ static const struct snmp_mib snmp4_net_list[] = {
249 SNMP_MIB_ITEM("TCPSackShifted", LINUX_MIB_SACKSHIFTED), 249 SNMP_MIB_ITEM("TCPSackShifted", LINUX_MIB_SACKSHIFTED),
250 SNMP_MIB_ITEM("TCPSackMerged", LINUX_MIB_SACKMERGED), 250 SNMP_MIB_ITEM("TCPSackMerged", LINUX_MIB_SACKMERGED),
251 SNMP_MIB_ITEM("TCPSackShiftFallback", LINUX_MIB_SACKSHIFTFALLBACK), 251 SNMP_MIB_ITEM("TCPSackShiftFallback", LINUX_MIB_SACKSHIFTFALLBACK),
252 SNMP_MIB_ITEM("TCPBacklogDrop", LINUX_MIB_TCPBACKLOGDROP),
253 SNMP_MIB_ITEM("TCPMinTTLDrop", LINUX_MIB_TCPMINTTLDROP),
254 SNMP_MIB_ITEM("TCPDeferAcceptDrop", LINUX_MIB_TCPDEFERACCEPTDROP),
252 SNMP_MIB_SENTINEL 255 SNMP_MIB_SENTINEL
253}; 256};
254 257
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index b2ba5581d2ae..32d396196df8 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -146,7 +146,6 @@ static struct dst_entry *ipv4_negative_advice(struct dst_entry *dst);
146static void ipv4_link_failure(struct sk_buff *skb); 146static void ipv4_link_failure(struct sk_buff *skb);
147static void ip_rt_update_pmtu(struct dst_entry *dst, u32 mtu); 147static void ip_rt_update_pmtu(struct dst_entry *dst, u32 mtu);
148static int rt_garbage_collect(struct dst_ops *ops); 148static int rt_garbage_collect(struct dst_ops *ops);
149static void rt_emergency_hash_rebuild(struct net *net);
150 149
151 150
152static struct dst_ops ipv4_dst_ops = { 151static struct dst_ops ipv4_dst_ops = {
@@ -780,11 +779,30 @@ static void rt_do_flush(int process_context)
780#define FRACT_BITS 3 779#define FRACT_BITS 3
781#define ONE (1UL << FRACT_BITS) 780#define ONE (1UL << FRACT_BITS)
782 781
782/*
783 * Given a hash chain and an item in this hash chain,
784 * find if a previous entry has the same hash_inputs
785 * (but differs on tos, mark or oif)
786 * Returns 0 if an alias is found.
787 * Returns ONE if rth has no alias before itself.
788 */
789static int has_noalias(const struct rtable *head, const struct rtable *rth)
790{
791 const struct rtable *aux = head;
792
793 while (aux != rth) {
794 if (compare_hash_inputs(&aux->fl, &rth->fl))
795 return 0;
796 aux = aux->u.dst.rt_next;
797 }
798 return ONE;
799}
800
783static void rt_check_expire(void) 801static void rt_check_expire(void)
784{ 802{
785 static unsigned int rover; 803 static unsigned int rover;
786 unsigned int i = rover, goal; 804 unsigned int i = rover, goal;
787 struct rtable *rth, *aux, **rthp; 805 struct rtable *rth, **rthp;
788 unsigned long samples = 0; 806 unsigned long samples = 0;
789 unsigned long sum = 0, sum2 = 0; 807 unsigned long sum = 0, sum2 = 0;
790 unsigned long delta; 808 unsigned long delta;
@@ -835,15 +853,7 @@ nofree:
835 * attributes don't unfairly skew 853 * attributes don't unfairly skew
836 * the length computation 854 * the length computation
837 */ 855 */
838 for (aux = rt_hash_table[i].chain;;) { 856 length += has_noalias(rt_hash_table[i].chain, rth);
839 if (aux == rth) {
840 length += ONE;
841 break;
842 }
843 if (compare_hash_inputs(&aux->fl, &rth->fl))
844 break;
845 aux = aux->u.dst.rt_next;
846 }
847 continue; 857 continue;
848 } 858 }
849 } else if (!rt_may_expire(rth, tmo, ip_rt_gc_timeout)) 859 } else if (!rt_may_expire(rth, tmo, ip_rt_gc_timeout))
@@ -922,10 +932,8 @@ static void rt_secret_rebuild_oneshot(struct net *net)
922{ 932{
923 del_timer_sync(&net->ipv4.rt_secret_timer); 933 del_timer_sync(&net->ipv4.rt_secret_timer);
924 rt_cache_invalidate(net); 934 rt_cache_invalidate(net);
925 if (ip_rt_secret_interval) { 935 if (ip_rt_secret_interval)
926 net->ipv4.rt_secret_timer.expires += ip_rt_secret_interval; 936 mod_timer(&net->ipv4.rt_secret_timer, jiffies + ip_rt_secret_interval);
927 add_timer(&net->ipv4.rt_secret_timer);
928 }
929} 937}
930 938
931static void rt_emergency_hash_rebuild(struct net *net) 939static void rt_emergency_hash_rebuild(struct net *net)
@@ -1073,6 +1081,21 @@ work_done:
1073out: return 0; 1081out: return 0;
1074} 1082}
1075 1083
1084/*
1085 * Returns number of entries in a hash chain that have different hash_inputs
1086 */
1087static int slow_chain_length(const struct rtable *head)
1088{
1089 int length = 0;
1090 const struct rtable *rth = head;
1091
1092 while (rth) {
1093 length += has_noalias(head, rth);
1094 rth = rth->u.dst.rt_next;
1095 }
1096 return length >> FRACT_BITS;
1097}
1098
1076static int rt_intern_hash(unsigned hash, struct rtable *rt, 1099static int rt_intern_hash(unsigned hash, struct rtable *rt,
1077 struct rtable **rp, struct sk_buff *skb) 1100 struct rtable **rp, struct sk_buff *skb)
1078{ 1101{
@@ -1185,7 +1208,8 @@ restart:
1185 rt_free(cand); 1208 rt_free(cand);
1186 } 1209 }
1187 } else { 1210 } else {
1188 if (chain_length > rt_chain_length_max) { 1211 if (chain_length > rt_chain_length_max &&
1212 slow_chain_length(rt_hash_table[hash].chain) > rt_chain_length_max) {
1189 struct net *net = dev_net(rt->u.dst.dev); 1213 struct net *net = dev_net(rt->u.dst.dev);
1190 int num = ++net->ipv4.current_rt_cache_rebuild_count; 1214 int num = ++net->ipv4.current_rt_cache_rebuild_count;
1191 if (!rt_caching(dev_net(rt->u.dst.dev))) { 1215 if (!rt_caching(dev_net(rt->u.dst.dev))) {
@@ -1417,7 +1441,7 @@ void ip_rt_redirect(__be32 old_gw, __be32 daddr, __be32 new_gw,
1417 dev_hold(rt->u.dst.dev); 1441 dev_hold(rt->u.dst.dev);
1418 if (rt->idev) 1442 if (rt->idev)
1419 in_dev_hold(rt->idev); 1443 in_dev_hold(rt->idev);
1420 rt->u.dst.obsolete = 0; 1444 rt->u.dst.obsolete = -1;
1421 rt->u.dst.lastuse = jiffies; 1445 rt->u.dst.lastuse = jiffies;
1422 rt->u.dst.path = &rt->u.dst; 1446 rt->u.dst.path = &rt->u.dst;
1423 rt->u.dst.neighbour = NULL; 1447 rt->u.dst.neighbour = NULL;
@@ -1482,7 +1506,7 @@ static struct dst_entry *ipv4_negative_advice(struct dst_entry *dst)
1482 struct dst_entry *ret = dst; 1506 struct dst_entry *ret = dst;
1483 1507
1484 if (rt) { 1508 if (rt) {
1485 if (dst->obsolete) { 1509 if (dst->obsolete > 0) {
1486 ip_rt_put(rt); 1510 ip_rt_put(rt);
1487 ret = NULL; 1511 ret = NULL;
1488 } else if ((rt->rt_flags & RTCF_REDIRECTED) || 1512 } else if ((rt->rt_flags & RTCF_REDIRECTED) ||
@@ -1702,7 +1726,9 @@ static void ip_rt_update_pmtu(struct dst_entry *dst, u32 mtu)
1702 1726
1703static struct dst_entry *ipv4_dst_check(struct dst_entry *dst, u32 cookie) 1727static struct dst_entry *ipv4_dst_check(struct dst_entry *dst, u32 cookie)
1704{ 1728{
1705 return NULL; 1729 if (rt_is_expired((struct rtable *)dst))
1730 return NULL;
1731 return dst;
1706} 1732}
1707 1733
1708static void ipv4_dst_destroy(struct dst_entry *dst) 1734static void ipv4_dst_destroy(struct dst_entry *dst)
@@ -1864,7 +1890,8 @@ static int ip_route_input_mc(struct sk_buff *skb, __be32 daddr, __be32 saddr,
1864 if (!rth) 1890 if (!rth)
1865 goto e_nobufs; 1891 goto e_nobufs;
1866 1892
1867 rth->u.dst.output= ip_rt_bug; 1893 rth->u.dst.output = ip_rt_bug;
1894 rth->u.dst.obsolete = -1;
1868 1895
1869 atomic_set(&rth->u.dst.__refcnt, 1); 1896 atomic_set(&rth->u.dst.__refcnt, 1);
1870 rth->u.dst.flags= DST_HOST; 1897 rth->u.dst.flags= DST_HOST;
@@ -2030,6 +2057,7 @@ static int __mkroute_input(struct sk_buff *skb,
2030 rth->fl.oif = 0; 2057 rth->fl.oif = 0;
2031 rth->rt_spec_dst= spec_dst; 2058 rth->rt_spec_dst= spec_dst;
2032 2059
2060 rth->u.dst.obsolete = -1;
2033 rth->u.dst.input = ip_forward; 2061 rth->u.dst.input = ip_forward;
2034 rth->u.dst.output = ip_output; 2062 rth->u.dst.output = ip_output;
2035 rth->rt_genid = rt_genid(dev_net(rth->u.dst.dev)); 2063 rth->rt_genid = rt_genid(dev_net(rth->u.dst.dev));
@@ -2194,6 +2222,7 @@ local_input:
2194 goto e_nobufs; 2222 goto e_nobufs;
2195 2223
2196 rth->u.dst.output= ip_rt_bug; 2224 rth->u.dst.output= ip_rt_bug;
2225 rth->u.dst.obsolete = -1;
2197 rth->rt_genid = rt_genid(net); 2226 rth->rt_genid = rt_genid(net);
2198 2227
2199 atomic_set(&rth->u.dst.__refcnt, 1); 2228 atomic_set(&rth->u.dst.__refcnt, 1);
@@ -2420,6 +2449,7 @@ static int __mkroute_output(struct rtable **result,
2420 rth->rt_spec_dst= fl->fl4_src; 2449 rth->rt_spec_dst= fl->fl4_src;
2421 2450
2422 rth->u.dst.output=ip_output; 2451 rth->u.dst.output=ip_output;
2452 rth->u.dst.obsolete = -1;
2423 rth->rt_genid = rt_genid(dev_net(dev_out)); 2453 rth->rt_genid = rt_genid(dev_net(dev_out));
2424 2454
2425 RT_CACHE_STAT_INC(out_slow_tot); 2455 RT_CACHE_STAT_INC(out_slow_tot);
@@ -3077,22 +3107,20 @@ static void rt_secret_reschedule(int old)
3077 rtnl_lock(); 3107 rtnl_lock();
3078 for_each_net(net) { 3108 for_each_net(net) {
3079 int deleted = del_timer_sync(&net->ipv4.rt_secret_timer); 3109 int deleted = del_timer_sync(&net->ipv4.rt_secret_timer);
3110 long time;
3080 3111
3081 if (!new) 3112 if (!new)
3082 continue; 3113 continue;
3083 3114
3084 if (deleted) { 3115 if (deleted) {
3085 long time = net->ipv4.rt_secret_timer.expires - jiffies; 3116 time = net->ipv4.rt_secret_timer.expires - jiffies;
3086 3117
3087 if (time <= 0 || (time += diff) <= 0) 3118 if (time <= 0 || (time += diff) <= 0)
3088 time = 0; 3119 time = 0;
3089
3090 net->ipv4.rt_secret_timer.expires = time;
3091 } else 3120 } else
3092 net->ipv4.rt_secret_timer.expires = new; 3121 time = new;
3093 3122
3094 net->ipv4.rt_secret_timer.expires += jiffies; 3123 mod_timer(&net->ipv4.rt_secret_timer, jiffies + time);
3095 add_timer(&net->ipv4.rt_secret_timer);
3096 } 3124 }
3097 rtnl_unlock(); 3125 rtnl_unlock();
3098} 3126}
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 5901010fad55..6afb6d8662b2 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -429,7 +429,7 @@ unsigned int tcp_poll(struct file *file, struct socket *sock, poll_table *wait)
429 if (tp->urg_seq == tp->copied_seq && 429 if (tp->urg_seq == tp->copied_seq &&
430 !sock_flag(sk, SOCK_URGINLINE) && 430 !sock_flag(sk, SOCK_URGINLINE) &&
431 tp->urg_data) 431 tp->urg_data)
432 target--; 432 target++;
433 433
434 /* Potential race condition. If read of tp below will 434 /* Potential race condition. If read of tp below will
435 * escape above sk->sk_state, we can be illegally awaken 435 * escape above sk->sk_state, we can be illegally awaken
@@ -1254,6 +1254,39 @@ static void tcp_prequeue_process(struct sock *sk)
1254 tp->ucopy.memory = 0; 1254 tp->ucopy.memory = 0;
1255} 1255}
1256 1256
1257#ifdef CONFIG_NET_DMA
1258static void tcp_service_net_dma(struct sock *sk, bool wait)
1259{
1260 dma_cookie_t done, used;
1261 dma_cookie_t last_issued;
1262 struct tcp_sock *tp = tcp_sk(sk);
1263
1264 if (!tp->ucopy.dma_chan)
1265 return;
1266
1267 last_issued = tp->ucopy.dma_cookie;
1268 dma_async_memcpy_issue_pending(tp->ucopy.dma_chan);
1269
1270 do {
1271 if (dma_async_memcpy_complete(tp->ucopy.dma_chan,
1272 last_issued, &done,
1273 &used) == DMA_SUCCESS) {
1274 /* Safe to free early-copied skbs now */
1275 __skb_queue_purge(&sk->sk_async_wait_queue);
1276 break;
1277 } else {
1278 struct sk_buff *skb;
1279 while ((skb = skb_peek(&sk->sk_async_wait_queue)) &&
1280 (dma_async_is_complete(skb->dma_cookie, done,
1281 used) == DMA_SUCCESS)) {
1282 __skb_dequeue(&sk->sk_async_wait_queue);
1283 kfree_skb(skb);
1284 }
1285 }
1286 } while (wait);
1287}
1288#endif
1289
1257static inline struct sk_buff *tcp_recv_skb(struct sock *sk, u32 seq, u32 *off) 1290static inline struct sk_buff *tcp_recv_skb(struct sock *sk, u32 seq, u32 *off)
1258{ 1291{
1259 struct sk_buff *skb; 1292 struct sk_buff *skb;
@@ -1546,6 +1579,10 @@ int tcp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
1546 /* __ Set realtime policy in scheduler __ */ 1579 /* __ Set realtime policy in scheduler __ */
1547 } 1580 }
1548 1581
1582#ifdef CONFIG_NET_DMA
1583 if (tp->ucopy.dma_chan)
1584 dma_async_memcpy_issue_pending(tp->ucopy.dma_chan);
1585#endif
1549 if (copied >= target) { 1586 if (copied >= target) {
1550 /* Do not sleep, just process backlog. */ 1587 /* Do not sleep, just process backlog. */
1551 release_sock(sk); 1588 release_sock(sk);
@@ -1554,6 +1591,7 @@ int tcp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
1554 sk_wait_data(sk, &timeo); 1591 sk_wait_data(sk, &timeo);
1555 1592
1556#ifdef CONFIG_NET_DMA 1593#ifdef CONFIG_NET_DMA
1594 tcp_service_net_dma(sk, false); /* Don't block */
1557 tp->ucopy.wakeup = 0; 1595 tp->ucopy.wakeup = 0;
1558#endif 1596#endif
1559 1597
@@ -1633,6 +1671,9 @@ do_prequeue:
1633 copied = -EFAULT; 1671 copied = -EFAULT;
1634 break; 1672 break;
1635 } 1673 }
1674
1675 dma_async_memcpy_issue_pending(tp->ucopy.dma_chan);
1676
1636 if ((offset + used) == skb->len) 1677 if ((offset + used) == skb->len)
1637 copied_early = 1; 1678 copied_early = 1;
1638 1679
@@ -1702,27 +1743,9 @@ skip_copy:
1702 } 1743 }
1703 1744
1704#ifdef CONFIG_NET_DMA 1745#ifdef CONFIG_NET_DMA
1705 if (tp->ucopy.dma_chan) { 1746 tcp_service_net_dma(sk, true); /* Wait for queue to drain */
1706 dma_cookie_t done, used; 1747 tp->ucopy.dma_chan = NULL;
1707
1708 dma_async_memcpy_issue_pending(tp->ucopy.dma_chan);
1709
1710 while (dma_async_memcpy_complete(tp->ucopy.dma_chan,
1711 tp->ucopy.dma_cookie, &done,
1712 &used) == DMA_IN_PROGRESS) {
1713 /* do partial cleanup of sk_async_wait_queue */
1714 while ((skb = skb_peek(&sk->sk_async_wait_queue)) &&
1715 (dma_async_is_complete(skb->dma_cookie, done,
1716 used) == DMA_SUCCESS)) {
1717 __skb_dequeue(&sk->sk_async_wait_queue);
1718 kfree_skb(skb);
1719 }
1720 }
1721 1748
1722 /* Safe to free early-copied skbs now */
1723 __skb_queue_purge(&sk->sk_async_wait_queue);
1724 tp->ucopy.dma_chan = NULL;
1725 }
1726 if (tp->ucopy.pinned_list) { 1749 if (tp->ucopy.pinned_list) {
1727 dma_unpin_iovec_pages(tp->ucopy.pinned_list); 1750 dma_unpin_iovec_pages(tp->ucopy.pinned_list);
1728 tp->ucopy.pinned_list = NULL; 1751 tp->ucopy.pinned_list = NULL;
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 788851ca8c5d..7b31476a4063 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -2511,6 +2511,9 @@ static void tcp_mark_head_lost(struct sock *sk, int packets)
2511 int err; 2511 int err;
2512 unsigned int mss; 2512 unsigned int mss;
2513 2513
2514 if (packets == 0)
2515 return;
2516
2514 WARN_ON(packets > tp->packets_out); 2517 WARN_ON(packets > tp->packets_out);
2515 if (tp->lost_skb_hint) { 2518 if (tp->lost_skb_hint) {
2516 skb = tp->lost_skb_hint; 2519 skb = tp->lost_skb_hint;
@@ -4315,7 +4318,7 @@ static void tcp_ofo_queue(struct sock *sk)
4315 } 4318 }
4316 4319
4317 if (!after(TCP_SKB_CB(skb)->end_seq, tp->rcv_nxt)) { 4320 if (!after(TCP_SKB_CB(skb)->end_seq, tp->rcv_nxt)) {
4318 SOCK_DEBUG(sk, "ofo packet was already received \n"); 4321 SOCK_DEBUG(sk, "ofo packet was already received\n");
4319 __skb_unlink(skb, &tp->out_of_order_queue); 4322 __skb_unlink(skb, &tp->out_of_order_queue);
4320 __kfree_skb(skb); 4323 __kfree_skb(skb);
4321 continue; 4324 continue;
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index c3588b4fd979..f4df5f931f36 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -370,6 +370,11 @@ void tcp_v4_err(struct sk_buff *icmp_skb, u32 info)
370 if (sk->sk_state == TCP_CLOSE) 370 if (sk->sk_state == TCP_CLOSE)
371 goto out; 371 goto out;
372 372
373 if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
374 NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
375 goto out;
376 }
377
373 icsk = inet_csk(sk); 378 icsk = inet_csk(sk);
374 tp = tcp_sk(sk); 379 tp = tcp_sk(sk);
375 seq = ntohl(th->seq); 380 seq = ntohl(th->seq);
@@ -1651,13 +1656,15 @@ int tcp_v4_rcv(struct sk_buff *skb)
1651 if (!sk) 1656 if (!sk)
1652 goto no_tcp_socket; 1657 goto no_tcp_socket;
1653 1658
1654 if (iph->ttl < inet_sk(sk)->min_ttl)
1655 goto discard_and_relse;
1656
1657process: 1659process:
1658 if (sk->sk_state == TCP_TIME_WAIT) 1660 if (sk->sk_state == TCP_TIME_WAIT)
1659 goto do_time_wait; 1661 goto do_time_wait;
1660 1662
1663 if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
1664 NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
1665 goto discard_and_relse;
1666 }
1667
1661 if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) 1668 if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
1662 goto discard_and_relse; 1669 goto discard_and_relse;
1663 nf_reset(skb); 1670 nf_reset(skb);
@@ -1682,8 +1689,11 @@ process:
1682 if (!tcp_prequeue(sk, skb)) 1689 if (!tcp_prequeue(sk, skb))
1683 ret = tcp_v4_do_rcv(sk, skb); 1690 ret = tcp_v4_do_rcv(sk, skb);
1684 } 1691 }
1685 } else 1692 } else if (unlikely(sk_add_backlog(sk, skb))) {
1686 sk_add_backlog(sk, skb); 1693 bh_unlock_sock(sk);
1694 NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
1695 goto discard_and_relse;
1696 }
1687 bh_unlock_sock(sk); 1697 bh_unlock_sock(sk);
1688 1698
1689 sock_put(sk); 1699 sock_put(sk);
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index f206ee5dda80..32f96278a24a 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -671,6 +671,7 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb,
671 if (req->retrans < inet_csk(sk)->icsk_accept_queue.rskq_defer_accept && 671 if (req->retrans < inet_csk(sk)->icsk_accept_queue.rskq_defer_accept &&
672 TCP_SKB_CB(skb)->end_seq == tcp_rsk(req)->rcv_isn + 1) { 672 TCP_SKB_CB(skb)->end_seq == tcp_rsk(req)->rcv_isn + 1) {
673 inet_rsk(req)->acked = 1; 673 inet_rsk(req)->acked = 1;
674 NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPDEFERACCEPTDROP);
674 return NULL; 675 return NULL;
675 } 676 }
676 677
@@ -728,7 +729,7 @@ int tcp_child_process(struct sock *parent, struct sock *child,
728 * in main socket hash table and lock on listening 729 * in main socket hash table and lock on listening
729 * socket does not protect us more. 730 * socket does not protect us more.
730 */ 731 */
731 sk_add_backlog(child, skb); 732 __sk_add_backlog(child, skb);
732 } 733 }
733 734
734 bh_unlock_sock(child); 735 bh_unlock_sock(child);
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 4a1605d3f909..f181b78f2385 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -2395,13 +2395,17 @@ struct sk_buff *tcp_make_synack(struct sock *sk, struct dst_entry *dst,
2395 struct tcp_extend_values *xvp = tcp_xv(rvp); 2395 struct tcp_extend_values *xvp = tcp_xv(rvp);
2396 struct inet_request_sock *ireq = inet_rsk(req); 2396 struct inet_request_sock *ireq = inet_rsk(req);
2397 struct tcp_sock *tp = tcp_sk(sk); 2397 struct tcp_sock *tp = tcp_sk(sk);
2398 const struct tcp_cookie_values *cvp = tp->cookie_values;
2398 struct tcphdr *th; 2399 struct tcphdr *th;
2399 struct sk_buff *skb; 2400 struct sk_buff *skb;
2400 struct tcp_md5sig_key *md5; 2401 struct tcp_md5sig_key *md5;
2401 int tcp_header_size; 2402 int tcp_header_size;
2402 int mss; 2403 int mss;
2404 int s_data_desired = 0;
2403 2405
2404 skb = sock_wmalloc(sk, MAX_TCP_HEADER + 15, 1, GFP_ATOMIC); 2406 if (cvp != NULL && cvp->s_data_constant && cvp->s_data_desired)
2407 s_data_desired = cvp->s_data_desired;
2408 skb = sock_wmalloc(sk, MAX_TCP_HEADER + 15 + s_data_desired, 1, GFP_ATOMIC);
2405 if (skb == NULL) 2409 if (skb == NULL)
2406 return NULL; 2410 return NULL;
2407 2411
@@ -2457,16 +2461,12 @@ struct sk_buff *tcp_make_synack(struct sock *sk, struct dst_entry *dst,
2457 TCPCB_FLAG_SYN | TCPCB_FLAG_ACK); 2461 TCPCB_FLAG_SYN | TCPCB_FLAG_ACK);
2458 2462
2459 if (OPTION_COOKIE_EXTENSION & opts.options) { 2463 if (OPTION_COOKIE_EXTENSION & opts.options) {
2460 const struct tcp_cookie_values *cvp = tp->cookie_values; 2464 if (s_data_desired) {
2461 2465 u8 *buf = skb_put(skb, s_data_desired);
2462 if (cvp != NULL &&
2463 cvp->s_data_constant &&
2464 cvp->s_data_desired > 0) {
2465 u8 *buf = skb_put(skb, cvp->s_data_desired);
2466 2466
2467 /* copy data directly from the listening socket. */ 2467 /* copy data directly from the listening socket. */
2468 memcpy(buf, cvp->s_data_payload, cvp->s_data_desired); 2468 memcpy(buf, cvp->s_data_payload, s_data_desired);
2469 TCP_SKB_CB(skb)->end_seq += cvp->s_data_desired; 2469 TCP_SKB_CB(skb)->end_seq += s_data_desired;
2470 } 2470 }
2471 2471
2472 if (opts.hash_size > 0) { 2472 if (opts.hash_size > 0) {
diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
index a17629b8912e..b2e6bbccaee1 100644
--- a/net/ipv4/tcp_timer.c
+++ b/net/ipv4/tcp_timer.c
@@ -134,7 +134,7 @@ static void tcp_mtu_probing(struct inet_connection_sock *icsk, struct sock *sk)
134} 134}
135 135
136/* This function calculates a "timeout" which is equivalent to the timeout of a 136/* This function calculates a "timeout" which is equivalent to the timeout of a
137 * TCP connection after "boundary" unsucessful, exponentially backed-off 137 * TCP connection after "boundary" unsuccessful, exponentially backed-off
138 * retransmissions with an initial RTO of TCP_RTO_MIN. 138 * retransmissions with an initial RTO of TCP_RTO_MIN.
139 */ 139 */
140static bool retransmits_timed_out(struct sock *sk, 140static bool retransmits_timed_out(struct sock *sk,
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index 608a5446d05b..7af756d0f931 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -1371,8 +1371,10 @@ int udp_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
1371 bh_lock_sock(sk); 1371 bh_lock_sock(sk);
1372 if (!sock_owned_by_user(sk)) 1372 if (!sock_owned_by_user(sk))
1373 rc = __udp_queue_rcv_skb(sk, skb); 1373 rc = __udp_queue_rcv_skb(sk, skb);
1374 else 1374 else if (sk_add_backlog(sk, skb)) {
1375 sk_add_backlog(sk, skb); 1375 bh_unlock_sock(sk);
1376 goto drop;
1377 }
1376 bh_unlock_sock(sk); 1378 bh_unlock_sock(sk);
1377 1379
1378 return rc; 1380 return rc;
diff --git a/net/ipv4/xfrm4_policy.c b/net/ipv4/xfrm4_policy.c
index 67107d63c1cd..e4a1483fba77 100644
--- a/net/ipv4/xfrm4_policy.c
+++ b/net/ipv4/xfrm4_policy.c
@@ -91,11 +91,12 @@ static int xfrm4_init_path(struct xfrm_dst *path, struct dst_entry *dst,
91 return 0; 91 return 0;
92} 92}
93 93
94static int xfrm4_fill_dst(struct xfrm_dst *xdst, struct net_device *dev) 94static int xfrm4_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
95 struct flowi *fl)
95{ 96{
96 struct rtable *rt = (struct rtable *)xdst->route; 97 struct rtable *rt = (struct rtable *)xdst->route;
97 98
98 xdst->u.rt.fl = rt->fl; 99 xdst->u.rt.fl = *fl;
99 100
100 xdst->u.dst.dev = dev; 101 xdst->u.dst.dev = dev;
101 dev_hold(dev); 102 dev_hold(dev);
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 88fd8c5877ee..21b4c9e1a682 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -81,7 +81,7 @@
81#include <linux/random.h> 81#include <linux/random.h>
82#endif 82#endif
83 83
84#include <asm/uaccess.h> 84#include <linux/uaccess.h>
85#include <asm/unaligned.h> 85#include <asm/unaligned.h>
86 86
87#include <linux/proc_fs.h> 87#include <linux/proc_fs.h>
@@ -97,7 +97,11 @@
97#endif 97#endif
98 98
99#define INFINITY_LIFE_TIME 0xFFFFFFFF 99#define INFINITY_LIFE_TIME 0xFFFFFFFF
100#define TIME_DELTA(a,b) ((unsigned long)((long)(a) - (long)(b))) 100#define TIME_DELTA(a, b) ((unsigned long)((long)(a) - (long)(b)))
101
102#define ADDRCONF_TIMER_FUZZ_MINUS (HZ > 50 ? HZ/50 : 1)
103#define ADDRCONF_TIMER_FUZZ (HZ / 4)
104#define ADDRCONF_TIMER_FUZZ_MAX (HZ)
101 105
102#ifdef CONFIG_SYSCTL 106#ifdef CONFIG_SYSCTL
103static void addrconf_sysctl_register(struct inet6_dev *idev); 107static void addrconf_sysctl_register(struct inet6_dev *idev);
@@ -126,8 +130,8 @@ static int ipv6_count_addresses(struct inet6_dev *idev);
126/* 130/*
127 * Configured unicast address hash table 131 * Configured unicast address hash table
128 */ 132 */
129static struct inet6_ifaddr *inet6_addr_lst[IN6_ADDR_HSIZE]; 133static struct hlist_head inet6_addr_lst[IN6_ADDR_HSIZE];
130static DEFINE_RWLOCK(addrconf_hash_lock); 134static DEFINE_SPINLOCK(addrconf_hash_lock);
131 135
132static void addrconf_verify(unsigned long); 136static void addrconf_verify(unsigned long);
133 137
@@ -137,8 +141,8 @@ static DEFINE_SPINLOCK(addrconf_verify_lock);
137static void addrconf_join_anycast(struct inet6_ifaddr *ifp); 141static void addrconf_join_anycast(struct inet6_ifaddr *ifp);
138static void addrconf_leave_anycast(struct inet6_ifaddr *ifp); 142static void addrconf_leave_anycast(struct inet6_ifaddr *ifp);
139 143
140static void addrconf_bonding_change(struct net_device *dev, 144static void addrconf_type_change(struct net_device *dev,
141 unsigned long event); 145 unsigned long event);
142static int addrconf_ifdown(struct net_device *dev, int how); 146static int addrconf_ifdown(struct net_device *dev, int how);
143 147
144static void addrconf_dad_start(struct inet6_ifaddr *ifp, u32 flags); 148static void addrconf_dad_start(struct inet6_ifaddr *ifp, u32 flags);
@@ -151,8 +155,8 @@ static void ipv6_ifa_notify(int event, struct inet6_ifaddr *ifa);
151 155
152static void inet6_prefix_notify(int event, struct inet6_dev *idev, 156static void inet6_prefix_notify(int event, struct inet6_dev *idev,
153 struct prefix_info *pinfo); 157 struct prefix_info *pinfo);
154static int ipv6_chk_same_addr(struct net *net, const struct in6_addr *addr, 158static bool ipv6_chk_same_addr(struct net *net, const struct in6_addr *addr,
155 struct net_device *dev); 159 struct net_device *dev);
156 160
157static ATOMIC_NOTIFIER_HEAD(inet6addr_chain); 161static ATOMIC_NOTIFIER_HEAD(inet6addr_chain);
158 162
@@ -249,8 +253,7 @@ static void addrconf_del_timer(struct inet6_ifaddr *ifp)
249 __in6_ifa_put(ifp); 253 __in6_ifa_put(ifp);
250} 254}
251 255
252enum addrconf_timer_t 256enum addrconf_timer_t {
253{
254 AC_NONE, 257 AC_NONE,
255 AC_DAD, 258 AC_DAD,
256 AC_RS, 259 AC_RS,
@@ -270,7 +273,8 @@ static void addrconf_mod_timer(struct inet6_ifaddr *ifp,
270 case AC_RS: 273 case AC_RS:
271 ifp->timer.function = addrconf_rs_timer; 274 ifp->timer.function = addrconf_rs_timer;
272 break; 275 break;
273 default:; 276 default:
277 break;
274 } 278 }
275 ifp->timer.expires = jiffies + when; 279 ifp->timer.expires = jiffies + when;
276 add_timer(&ifp->timer); 280 add_timer(&ifp->timer);
@@ -317,7 +321,7 @@ void in6_dev_finish_destroy(struct inet6_dev *idev)
317{ 321{
318 struct net_device *dev = idev->dev; 322 struct net_device *dev = idev->dev;
319 323
320 WARN_ON(idev->addr_list != NULL); 324 WARN_ON(!list_empty(&idev->addr_list));
321 WARN_ON(idev->mc_list != NULL); 325 WARN_ON(idev->mc_list != NULL);
322 326
323#ifdef NET_REFCNT_DEBUG 327#ifdef NET_REFCNT_DEBUG
@@ -325,7 +329,7 @@ void in6_dev_finish_destroy(struct inet6_dev *idev)
325#endif 329#endif
326 dev_put(dev); 330 dev_put(dev);
327 if (!idev->dead) { 331 if (!idev->dead) {
328 printk("Freeing alive inet6 device %p\n", idev); 332 pr_warning("Freeing alive inet6 device %p\n", idev);
329 return; 333 return;
330 } 334 }
331 snmp6_free_dev(idev); 335 snmp6_free_dev(idev);
@@ -350,6 +354,8 @@ static struct inet6_dev * ipv6_add_dev(struct net_device *dev)
350 354
351 rwlock_init(&ndev->lock); 355 rwlock_init(&ndev->lock);
352 ndev->dev = dev; 356 ndev->dev = dev;
357 INIT_LIST_HEAD(&ndev->addr_list);
358
353 memcpy(&ndev->cnf, dev_net(dev)->ipv6.devconf_dflt, sizeof(ndev->cnf)); 359 memcpy(&ndev->cnf, dev_net(dev)->ipv6.devconf_dflt, sizeof(ndev->cnf));
354 ndev->cnf.mtu6 = dev->mtu; 360 ndev->cnf.mtu6 = dev->mtu;
355 ndev->cnf.sysctl = NULL; 361 ndev->cnf.sysctl = NULL;
@@ -401,6 +407,7 @@ static struct inet6_dev * ipv6_add_dev(struct net_device *dev)
401#endif 407#endif
402 408
403#ifdef CONFIG_IPV6_PRIVACY 409#ifdef CONFIG_IPV6_PRIVACY
410 INIT_LIST_HEAD(&ndev->tempaddr_list);
404 setup_timer(&ndev->regen_timer, ipv6_regen_rndid, (unsigned long)ndev); 411 setup_timer(&ndev->regen_timer, ipv6_regen_rndid, (unsigned long)ndev);
405 if ((dev->flags&IFF_LOOPBACK) || 412 if ((dev->flags&IFF_LOOPBACK) ||
406 dev->type == ARPHRD_TUNNEL || 413 dev->type == ARPHRD_TUNNEL ||
@@ -438,8 +445,10 @@ static struct inet6_dev * ipv6_find_idev(struct net_device *dev)
438 445
439 ASSERT_RTNL(); 446 ASSERT_RTNL();
440 447
441 if ((idev = __in6_dev_get(dev)) == NULL) { 448 idev = __in6_dev_get(dev);
442 if ((idev = ipv6_add_dev(dev)) == NULL) 449 if (!idev) {
450 idev = ipv6_add_dev(dev);
451 if (!idev)
443 return NULL; 452 return NULL;
444 } 453 }
445 454
@@ -465,7 +474,8 @@ static void dev_forward_change(struct inet6_dev *idev)
465 else 474 else
466 ipv6_dev_mc_dec(dev, &in6addr_linklocal_allrouters); 475 ipv6_dev_mc_dec(dev, &in6addr_linklocal_allrouters);
467 } 476 }
468 for (ifa=idev->addr_list; ifa; ifa=ifa->if_next) { 477
478 list_for_each_entry(ifa, &idev->addr_list, if_list) {
469 if (ifa->flags&IFA_F_TENTATIVE) 479 if (ifa->flags&IFA_F_TENTATIVE)
470 continue; 480 continue;
471 if (idev->cnf.forwarding) 481 if (idev->cnf.forwarding)
@@ -522,12 +532,16 @@ static int addrconf_fixup_forwarding(struct ctl_table *table, int *p, int old)
522} 532}
523#endif 533#endif
524 534
525/* Nobody refers to this ifaddr, destroy it */ 535static void inet6_ifa_finish_destroy_rcu(struct rcu_head *head)
536{
537 struct inet6_ifaddr *ifp = container_of(head, struct inet6_ifaddr, rcu);
538 kfree(ifp);
539}
526 540
541/* Nobody refers to this ifaddr, destroy it */
527void inet6_ifa_finish_destroy(struct inet6_ifaddr *ifp) 542void inet6_ifa_finish_destroy(struct inet6_ifaddr *ifp)
528{ 543{
529 WARN_ON(ifp->if_next != NULL); 544 WARN_ON(!hlist_unhashed(&ifp->addr_lst));
530 WARN_ON(ifp->lst_next != NULL);
531 545
532#ifdef NET_REFCNT_DEBUG 546#ifdef NET_REFCNT_DEBUG
533 printk(KERN_DEBUG "inet6_ifa_finish_destroy\n"); 547 printk(KERN_DEBUG "inet6_ifa_finish_destroy\n");
@@ -536,54 +550,45 @@ void inet6_ifa_finish_destroy(struct inet6_ifaddr *ifp)
536 in6_dev_put(ifp->idev); 550 in6_dev_put(ifp->idev);
537 551
538 if (del_timer(&ifp->timer)) 552 if (del_timer(&ifp->timer))
539 printk("Timer is still running, when freeing ifa=%p\n", ifp); 553 pr_notice("Timer is still running, when freeing ifa=%p\n", ifp);
540 554
541 if (!ifp->dead) { 555 if (!ifp->dead) {
542 printk("Freeing alive inet6 address %p\n", ifp); 556 pr_warning("Freeing alive inet6 address %p\n", ifp);
543 return; 557 return;
544 } 558 }
545 dst_release(&ifp->rt->u.dst); 559 dst_release(&ifp->rt->u.dst);
546 560
547 kfree(ifp); 561 call_rcu(&ifp->rcu, inet6_ifa_finish_destroy_rcu);
548} 562}
549 563
550static void 564static void
551ipv6_link_dev_addr(struct inet6_dev *idev, struct inet6_ifaddr *ifp) 565ipv6_link_dev_addr(struct inet6_dev *idev, struct inet6_ifaddr *ifp)
552{ 566{
553 struct inet6_ifaddr *ifa, **ifap; 567 struct list_head *p;
554 int ifp_scope = ipv6_addr_src_scope(&ifp->addr); 568 int ifp_scope = ipv6_addr_src_scope(&ifp->addr);
555 569
556 /* 570 /*
557 * Each device address list is sorted in order of scope - 571 * Each device address list is sorted in order of scope -
558 * global before linklocal. 572 * global before linklocal.
559 */ 573 */
560 for (ifap = &idev->addr_list; (ifa = *ifap) != NULL; 574 list_for_each(p, &idev->addr_list) {
561 ifap = &ifa->if_next) { 575 struct inet6_ifaddr *ifa
576 = list_entry(p, struct inet6_ifaddr, if_list);
562 if (ifp_scope >= ipv6_addr_src_scope(&ifa->addr)) 577 if (ifp_scope >= ipv6_addr_src_scope(&ifa->addr))
563 break; 578 break;
564 } 579 }
565 580
566 ifp->if_next = *ifap; 581 list_add_tail(&ifp->if_list, p);
567 *ifap = ifp;
568} 582}
569 583
570/* 584static u32 ipv6_addr_hash(const struct in6_addr *addr)
571 * Hash function taken from net_alias.c
572 */
573static u8 ipv6_addr_hash(const struct in6_addr *addr)
574{ 585{
575 __u32 word;
576
577 /* 586 /*
578 * We perform the hash function over the last 64 bits of the address 587 * We perform the hash function over the last 64 bits of the address
579 * This will include the IEEE address token on links that support it. 588 * This will include the IEEE address token on links that support it.
580 */ 589 */
581 590 return jhash_2words(addr->s6_addr32[2], addr->s6_addr32[3], 0)
582 word = (__force u32)(addr->s6_addr32[2] ^ addr->s6_addr32[3]); 591 & (IN6_ADDR_HSIZE - 1);
583 word ^= (word >> 16);
584 word ^= (word >> 8);
585
586 return ((word ^ (word >> 4)) & 0x0f);
587} 592}
588 593
589/* On success it returns ifp with increased reference count */ 594/* On success it returns ifp with increased reference count */
@@ -594,7 +599,7 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, int pfxlen,
594{ 599{
595 struct inet6_ifaddr *ifa = NULL; 600 struct inet6_ifaddr *ifa = NULL;
596 struct rt6_info *rt; 601 struct rt6_info *rt;
597 int hash; 602 unsigned int hash;
598 int err = 0; 603 int err = 0;
599 int addr_type = ipv6_addr_type(addr); 604 int addr_type = ipv6_addr_type(addr);
600 605
@@ -615,7 +620,7 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, int pfxlen,
615 goto out2; 620 goto out2;
616 } 621 }
617 622
618 write_lock(&addrconf_hash_lock); 623 spin_lock(&addrconf_hash_lock);
619 624
620 /* Ignore adding duplicate addresses on an interface */ 625 /* Ignore adding duplicate addresses on an interface */
621 if (ipv6_chk_same_addr(dev_net(idev->dev), addr, idev->dev)) { 626 if (ipv6_chk_same_addr(dev_net(idev->dev), addr, idev->dev)) {
@@ -642,6 +647,7 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, int pfxlen,
642 647
643 spin_lock_init(&ifa->lock); 648 spin_lock_init(&ifa->lock);
644 init_timer(&ifa->timer); 649 init_timer(&ifa->timer);
650 INIT_HLIST_NODE(&ifa->addr_lst);
645 ifa->timer.data = (unsigned long) ifa; 651 ifa->timer.data = (unsigned long) ifa;
646 ifa->scope = scope; 652 ifa->scope = scope;
647 ifa->prefix_len = pfxlen; 653 ifa->prefix_len = pfxlen;
@@ -668,10 +674,9 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, int pfxlen,
668 /* Add to big hash table */ 674 /* Add to big hash table */
669 hash = ipv6_addr_hash(addr); 675 hash = ipv6_addr_hash(addr);
670 676
671 ifa->lst_next = inet6_addr_lst[hash]; 677 hlist_add_head_rcu(&ifa->addr_lst, &inet6_addr_lst[hash]);
672 inet6_addr_lst[hash] = ifa;
673 in6_ifa_hold(ifa); 678 in6_ifa_hold(ifa);
674 write_unlock(&addrconf_hash_lock); 679 spin_unlock(&addrconf_hash_lock);
675 680
676 write_lock(&idev->lock); 681 write_lock(&idev->lock);
677 /* Add to inet6_dev unicast addr list. */ 682 /* Add to inet6_dev unicast addr list. */
@@ -679,8 +684,7 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, int pfxlen,
679 684
680#ifdef CONFIG_IPV6_PRIVACY 685#ifdef CONFIG_IPV6_PRIVACY
681 if (ifa->flags&IFA_F_TEMPORARY) { 686 if (ifa->flags&IFA_F_TEMPORARY) {
682 ifa->tmp_next = idev->tempaddr_list; 687 list_add(&ifa->tmp_list, &idev->tempaddr_list);
683 idev->tempaddr_list = ifa;
684 in6_ifa_hold(ifa); 688 in6_ifa_hold(ifa);
685 } 689 }
686#endif 690#endif
@@ -699,7 +703,7 @@ out2:
699 703
700 return ifa; 704 return ifa;
701out: 705out:
702 write_unlock(&addrconf_hash_lock); 706 spin_unlock(&addrconf_hash_lock);
703 goto out2; 707 goto out2;
704} 708}
705 709
@@ -707,7 +711,7 @@ out:
707 711
708static void ipv6_del_addr(struct inet6_ifaddr *ifp) 712static void ipv6_del_addr(struct inet6_ifaddr *ifp)
709{ 713{
710 struct inet6_ifaddr *ifa, **ifap; 714 struct inet6_ifaddr *ifa, *ifn;
711 struct inet6_dev *idev = ifp->idev; 715 struct inet6_dev *idev = ifp->idev;
712 int hash; 716 int hash;
713 int deleted = 0, onlink = 0; 717 int deleted = 0, onlink = 0;
@@ -717,42 +721,28 @@ static void ipv6_del_addr(struct inet6_ifaddr *ifp)
717 721
718 ifp->dead = 1; 722 ifp->dead = 1;
719 723
720 write_lock_bh(&addrconf_hash_lock); 724 spin_lock_bh(&addrconf_hash_lock);
721 for (ifap = &inet6_addr_lst[hash]; (ifa=*ifap) != NULL; 725 hlist_del_init_rcu(&ifp->addr_lst);
722 ifap = &ifa->lst_next) { 726 __in6_ifa_put(ifp);
723 if (ifa == ifp) { 727 spin_unlock_bh(&addrconf_hash_lock);
724 *ifap = ifa->lst_next;
725 __in6_ifa_put(ifp);
726 ifa->lst_next = NULL;
727 break;
728 }
729 }
730 write_unlock_bh(&addrconf_hash_lock);
731 728
732 write_lock_bh(&idev->lock); 729 write_lock_bh(&idev->lock);
733#ifdef CONFIG_IPV6_PRIVACY 730#ifdef CONFIG_IPV6_PRIVACY
734 if (ifp->flags&IFA_F_TEMPORARY) { 731 if (ifp->flags&IFA_F_TEMPORARY) {
735 for (ifap = &idev->tempaddr_list; (ifa=*ifap) != NULL; 732 list_del(&ifp->tmp_list);
736 ifap = &ifa->tmp_next) { 733 if (ifp->ifpub) {
737 if (ifa == ifp) { 734 in6_ifa_put(ifp->ifpub);
738 *ifap = ifa->tmp_next; 735 ifp->ifpub = NULL;
739 if (ifp->ifpub) {
740 in6_ifa_put(ifp->ifpub);
741 ifp->ifpub = NULL;
742 }
743 __in6_ifa_put(ifp);
744 ifa->tmp_next = NULL;
745 break;
746 }
747 } 736 }
737 __in6_ifa_put(ifp);
748 } 738 }
749#endif 739#endif
750 740
751 for (ifap = &idev->addr_list; (ifa=*ifap) != NULL;) { 741 list_for_each_entry_safe(ifa, ifn, &idev->addr_list, if_list) {
752 if (ifa == ifp) { 742 if (ifa == ifp) {
753 *ifap = ifa->if_next; 743 list_del_init(&ifp->if_list);
754 __in6_ifa_put(ifp); 744 __in6_ifa_put(ifp);
755 ifa->if_next = NULL; 745
756 if (!(ifp->flags & IFA_F_PERMANENT) || onlink > 0) 746 if (!(ifp->flags & IFA_F_PERMANENT) || onlink > 0)
757 break; 747 break;
758 deleted = 1; 748 deleted = 1;
@@ -785,7 +775,6 @@ static void ipv6_del_addr(struct inet6_ifaddr *ifp)
785 } 775 }
786 } 776 }
787 } 777 }
788 ifap = &ifa->if_next;
789 } 778 }
790 write_unlock_bh(&idev->lock); 779 write_unlock_bh(&idev->lock);
791 780
@@ -1164,7 +1153,7 @@ int ipv6_dev_get_saddr(struct net *net, struct net_device *dst_dev,
1164 continue; 1153 continue;
1165 1154
1166 read_lock_bh(&idev->lock); 1155 read_lock_bh(&idev->lock);
1167 for (score->ifa = idev->addr_list; score->ifa; score->ifa = score->ifa->if_next) { 1156 list_for_each_entry(score->ifa, &idev->addr_list, if_list) {
1168 int i; 1157 int i;
1169 1158
1170 /* 1159 /*
@@ -1242,7 +1231,6 @@ try_nextdev:
1242 in6_ifa_put(hiscore->ifa); 1231 in6_ifa_put(hiscore->ifa);
1243 return 0; 1232 return 0;
1244} 1233}
1245
1246EXPORT_SYMBOL(ipv6_dev_get_saddr); 1234EXPORT_SYMBOL(ipv6_dev_get_saddr);
1247 1235
1248int ipv6_get_lladdr(struct net_device *dev, struct in6_addr *addr, 1236int ipv6_get_lladdr(struct net_device *dev, struct in6_addr *addr,
@@ -1252,12 +1240,14 @@ int ipv6_get_lladdr(struct net_device *dev, struct in6_addr *addr,
1252 int err = -EADDRNOTAVAIL; 1240 int err = -EADDRNOTAVAIL;
1253 1241
1254 rcu_read_lock(); 1242 rcu_read_lock();
1255 if ((idev = __in6_dev_get(dev)) != NULL) { 1243 idev = __in6_dev_get(dev);
1244 if (idev) {
1256 struct inet6_ifaddr *ifp; 1245 struct inet6_ifaddr *ifp;
1257 1246
1258 read_lock_bh(&idev->lock); 1247 read_lock_bh(&idev->lock);
1259 for (ifp=idev->addr_list; ifp; ifp=ifp->if_next) { 1248 list_for_each_entry(ifp, &idev->addr_list, if_list) {
1260 if (ifp->scope == IFA_LINK && !(ifp->flags & banned_flags)) { 1249 if (ifp->scope == IFA_LINK &&
1250 !(ifp->flags & banned_flags)) {
1261 ipv6_addr_copy(addr, &ifp->addr); 1251 ipv6_addr_copy(addr, &ifp->addr);
1262 err = 0; 1252 err = 0;
1263 break; 1253 break;
@@ -1275,7 +1265,7 @@ static int ipv6_count_addresses(struct inet6_dev *idev)
1275 struct inet6_ifaddr *ifp; 1265 struct inet6_ifaddr *ifp;
1276 1266
1277 read_lock_bh(&idev->lock); 1267 read_lock_bh(&idev->lock);
1278 for (ifp=idev->addr_list; ifp; ifp=ifp->if_next) 1268 list_for_each_entry(ifp, &idev->addr_list, if_list)
1279 cnt++; 1269 cnt++;
1280 read_unlock_bh(&idev->lock); 1270 read_unlock_bh(&idev->lock);
1281 return cnt; 1271 return cnt;
@@ -1284,11 +1274,12 @@ static int ipv6_count_addresses(struct inet6_dev *idev)
1284int ipv6_chk_addr(struct net *net, struct in6_addr *addr, 1274int ipv6_chk_addr(struct net *net, struct in6_addr *addr,
1285 struct net_device *dev, int strict) 1275 struct net_device *dev, int strict)
1286{ 1276{
1287 struct inet6_ifaddr * ifp; 1277 struct inet6_ifaddr *ifp = NULL;
1288 u8 hash = ipv6_addr_hash(addr); 1278 struct hlist_node *node;
1279 unsigned int hash = ipv6_addr_hash(addr);
1289 1280
1290 read_lock_bh(&addrconf_hash_lock); 1281 rcu_read_lock_bh();
1291 for(ifp = inet6_addr_lst[hash]; ifp; ifp=ifp->lst_next) { 1282 hlist_for_each_entry_rcu(ifp, node, &inet6_addr_lst[hash], addr_lst) {
1292 if (!net_eq(dev_net(ifp->idev->dev), net)) 1283 if (!net_eq(dev_net(ifp->idev->dev), net))
1293 continue; 1284 continue;
1294 if (ipv6_addr_equal(&ifp->addr, addr) && 1285 if (ipv6_addr_equal(&ifp->addr, addr) &&
@@ -1298,27 +1289,28 @@ int ipv6_chk_addr(struct net *net, struct in6_addr *addr,
1298 break; 1289 break;
1299 } 1290 }
1300 } 1291 }
1301 read_unlock_bh(&addrconf_hash_lock); 1292 rcu_read_unlock_bh();
1293
1302 return ifp != NULL; 1294 return ifp != NULL;
1303} 1295}
1304EXPORT_SYMBOL(ipv6_chk_addr); 1296EXPORT_SYMBOL(ipv6_chk_addr);
1305 1297
1306static 1298static bool ipv6_chk_same_addr(struct net *net, const struct in6_addr *addr,
1307int ipv6_chk_same_addr(struct net *net, const struct in6_addr *addr, 1299 struct net_device *dev)
1308 struct net_device *dev)
1309{ 1300{
1310 struct inet6_ifaddr * ifp; 1301 unsigned int hash = ipv6_addr_hash(addr);
1311 u8 hash = ipv6_addr_hash(addr); 1302 struct inet6_ifaddr *ifp;
1303 struct hlist_node *node;
1312 1304
1313 for(ifp = inet6_addr_lst[hash]; ifp; ifp=ifp->lst_next) { 1305 hlist_for_each_entry(ifp, node, &inet6_addr_lst[hash], addr_lst) {
1314 if (!net_eq(dev_net(ifp->idev->dev), net)) 1306 if (!net_eq(dev_net(ifp->idev->dev), net))
1315 continue; 1307 continue;
1316 if (ipv6_addr_equal(&ifp->addr, addr)) { 1308 if (ipv6_addr_equal(&ifp->addr, addr)) {
1317 if (dev == NULL || ifp->idev->dev == dev) 1309 if (dev == NULL || ifp->idev->dev == dev)
1318 break; 1310 return true;
1319 } 1311 }
1320 } 1312 }
1321 return ifp != NULL; 1313 return false;
1322} 1314}
1323 1315
1324int ipv6_chk_prefix(struct in6_addr *addr, struct net_device *dev) 1316int ipv6_chk_prefix(struct in6_addr *addr, struct net_device *dev)
@@ -1332,7 +1324,7 @@ int ipv6_chk_prefix(struct in6_addr *addr, struct net_device *dev)
1332 idev = __in6_dev_get(dev); 1324 idev = __in6_dev_get(dev);
1333 if (idev) { 1325 if (idev) {
1334 read_lock_bh(&idev->lock); 1326 read_lock_bh(&idev->lock);
1335 for (ifa = idev->addr_list; ifa; ifa = ifa->if_next) { 1327 list_for_each_entry(ifa, &idev->addr_list, if_list) {
1336 onlink = ipv6_prefix_equal(addr, &ifa->addr, 1328 onlink = ipv6_prefix_equal(addr, &ifa->addr,
1337 ifa->prefix_len); 1329 ifa->prefix_len);
1338 if (onlink) 1330 if (onlink)
@@ -1349,24 +1341,26 @@ EXPORT_SYMBOL(ipv6_chk_prefix);
1349struct inet6_ifaddr *ipv6_get_ifaddr(struct net *net, const struct in6_addr *addr, 1341struct inet6_ifaddr *ipv6_get_ifaddr(struct net *net, const struct in6_addr *addr,
1350 struct net_device *dev, int strict) 1342 struct net_device *dev, int strict)
1351{ 1343{
1352 struct inet6_ifaddr * ifp; 1344 struct inet6_ifaddr *ifp, *result = NULL;
1353 u8 hash = ipv6_addr_hash(addr); 1345 unsigned int hash = ipv6_addr_hash(addr);
1346 struct hlist_node *node;
1354 1347
1355 read_lock_bh(&addrconf_hash_lock); 1348 rcu_read_lock_bh();
1356 for(ifp = inet6_addr_lst[hash]; ifp; ifp=ifp->lst_next) { 1349 hlist_for_each_entry_rcu(ifp, node, &inet6_addr_lst[hash], addr_lst) {
1357 if (!net_eq(dev_net(ifp->idev->dev), net)) 1350 if (!net_eq(dev_net(ifp->idev->dev), net))
1358 continue; 1351 continue;
1359 if (ipv6_addr_equal(&ifp->addr, addr)) { 1352 if (ipv6_addr_equal(&ifp->addr, addr)) {
1360 if (dev == NULL || ifp->idev->dev == dev || 1353 if (dev == NULL || ifp->idev->dev == dev ||
1361 !(ifp->scope&(IFA_LINK|IFA_HOST) || strict)) { 1354 !(ifp->scope&(IFA_LINK|IFA_HOST) || strict)) {
1355 result = ifp;
1362 in6_ifa_hold(ifp); 1356 in6_ifa_hold(ifp);
1363 break; 1357 break;
1364 } 1358 }
1365 } 1359 }
1366 } 1360 }
1367 read_unlock_bh(&addrconf_hash_lock); 1361 rcu_read_unlock_bh();
1368 1362
1369 return ifp; 1363 return result;
1370} 1364}
1371 1365
1372/* Gets referenced address, destroys ifaddr */ 1366/* Gets referenced address, destroys ifaddr */
@@ -1380,6 +1374,8 @@ static void addrconf_dad_stop(struct inet6_ifaddr *ifp, int dad_failed)
1380 if (dad_failed) 1374 if (dad_failed)
1381 ifp->flags |= IFA_F_DADFAILED; 1375 ifp->flags |= IFA_F_DADFAILED;
1382 spin_unlock_bh(&ifp->lock); 1376 spin_unlock_bh(&ifp->lock);
1377 if (dad_failed)
1378 ipv6_ifa_notify(0, ifp);
1383 in6_ifa_put(ifp); 1379 in6_ifa_put(ifp);
1384#ifdef CONFIG_IPV6_PRIVACY 1380#ifdef CONFIG_IPV6_PRIVACY
1385 } else if (ifp->flags&IFA_F_TEMPORARY) { 1381 } else if (ifp->flags&IFA_F_TEMPORARY) {
@@ -1567,7 +1563,7 @@ static int ipv6_inherit_eui64(u8 *eui, struct inet6_dev *idev)
1567 struct inet6_ifaddr *ifp; 1563 struct inet6_ifaddr *ifp;
1568 1564
1569 read_lock_bh(&idev->lock); 1565 read_lock_bh(&idev->lock);
1570 for (ifp=idev->addr_list; ifp; ifp=ifp->if_next) { 1566 list_for_each_entry(ifp, &idev->addr_list, if_list) {
1571 if (ifp->scope == IFA_LINK && !(ifp->flags&IFA_F_TENTATIVE)) { 1567 if (ifp->scope == IFA_LINK && !(ifp->flags&IFA_F_TENTATIVE)) {
1572 memcpy(eui, ifp->addr.s6_addr+8, 8); 1568 memcpy(eui, ifp->addr.s6_addr+8, 8);
1573 err = 0; 1569 err = 0;
@@ -1735,7 +1731,8 @@ static struct inet6_dev *addrconf_add_dev(struct net_device *dev)
1735 1731
1736 ASSERT_RTNL(); 1732 ASSERT_RTNL();
1737 1733
1738 if ((idev = ipv6_find_idev(dev)) == NULL) 1734 idev = ipv6_find_idev(dev);
1735 if (!idev)
1739 return NULL; 1736 return NULL;
1740 1737
1741 /* Add default multicast route */ 1738 /* Add default multicast route */
@@ -1968,7 +1965,7 @@ ok:
1968#ifdef CONFIG_IPV6_PRIVACY 1965#ifdef CONFIG_IPV6_PRIVACY
1969 read_lock_bh(&in6_dev->lock); 1966 read_lock_bh(&in6_dev->lock);
1970 /* update all temporary addresses in the list */ 1967 /* update all temporary addresses in the list */
1971 for (ift=in6_dev->tempaddr_list; ift; ift=ift->tmp_next) { 1968 list_for_each_entry(ift, &in6_dev->tempaddr_list, tmp_list) {
1972 /* 1969 /*
1973 * When adjusting the lifetimes of an existing 1970 * When adjusting the lifetimes of an existing
1974 * temporary address, only lower the lifetimes. 1971 * temporary address, only lower the lifetimes.
@@ -2171,7 +2168,7 @@ static int inet6_addr_del(struct net *net, int ifindex, struct in6_addr *pfx,
2171 return -ENXIO; 2168 return -ENXIO;
2172 2169
2173 read_lock_bh(&idev->lock); 2170 read_lock_bh(&idev->lock);
2174 for (ifp = idev->addr_list; ifp; ifp=ifp->if_next) { 2171 list_for_each_entry(ifp, &idev->addr_list, if_list) {
2175 if (ifp->prefix_len == plen && 2172 if (ifp->prefix_len == plen &&
2176 ipv6_addr_equal(pfx, &ifp->addr)) { 2173 ipv6_addr_equal(pfx, &ifp->addr)) {
2177 in6_ifa_hold(ifp); 2174 in6_ifa_hold(ifp);
@@ -2182,7 +2179,7 @@ static int inet6_addr_del(struct net *net, int ifindex, struct in6_addr *pfx,
2182 /* If the last address is deleted administratively, 2179 /* If the last address is deleted administratively,
2183 disable IPv6 on this interface. 2180 disable IPv6 on this interface.
2184 */ 2181 */
2185 if (idev->addr_list == NULL) 2182 if (list_empty(&idev->addr_list))
2186 addrconf_ifdown(idev->dev, 1); 2183 addrconf_ifdown(idev->dev, 1);
2187 return 0; 2184 return 0;
2188 } 2185 }
@@ -2443,7 +2440,8 @@ static void addrconf_ip6_tnl_config(struct net_device *dev)
2443 2440
2444 ASSERT_RTNL(); 2441 ASSERT_RTNL();
2445 2442
2446 if ((idev = addrconf_add_dev(dev)) == NULL) { 2443 idev = addrconf_add_dev(dev);
2444 if (!idev) {
2447 printk(KERN_DEBUG "init ip6-ip6: add_dev failed\n"); 2445 printk(KERN_DEBUG "init ip6-ip6: add_dev failed\n");
2448 return; 2446 return;
2449 } 2447 }
@@ -2458,7 +2456,7 @@ static int addrconf_notify(struct notifier_block *this, unsigned long event,
2458 int run_pending = 0; 2456 int run_pending = 0;
2459 int err; 2457 int err;
2460 2458
2461 switch(event) { 2459 switch (event) {
2462 case NETDEV_REGISTER: 2460 case NETDEV_REGISTER:
2463 if (!idev && dev->mtu >= IPV6_MIN_MTU) { 2461 if (!idev && dev->mtu >= IPV6_MIN_MTU) {
2464 idev = ipv6_add_dev(dev); 2462 idev = ipv6_add_dev(dev);
@@ -2466,6 +2464,7 @@ static int addrconf_notify(struct notifier_block *this, unsigned long event,
2466 return notifier_from_errno(-ENOMEM); 2464 return notifier_from_errno(-ENOMEM);
2467 } 2465 }
2468 break; 2466 break;
2467
2469 case NETDEV_UP: 2468 case NETDEV_UP:
2470 case NETDEV_CHANGE: 2469 case NETDEV_CHANGE:
2471 if (dev->flags & IFF_SLAVE) 2470 if (dev->flags & IFF_SLAVE)
@@ -2495,10 +2494,9 @@ static int addrconf_notify(struct notifier_block *this, unsigned long event,
2495 } 2494 }
2496 2495
2497 if (idev) { 2496 if (idev) {
2498 if (idev->if_flags & IF_READY) { 2497 if (idev->if_flags & IF_READY)
2499 /* device is already configured. */ 2498 /* device is already configured. */
2500 break; 2499 break;
2501 }
2502 idev->if_flags |= IF_READY; 2500 idev->if_flags |= IF_READY;
2503 } 2501 }
2504 2502
@@ -2510,7 +2508,7 @@ static int addrconf_notify(struct notifier_block *this, unsigned long event,
2510 run_pending = 1; 2508 run_pending = 1;
2511 } 2509 }
2512 2510
2513 switch(dev->type) { 2511 switch (dev->type) {
2514#if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE) 2512#if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
2515 case ARPHRD_SIT: 2513 case ARPHRD_SIT:
2516 addrconf_sit_config(dev); 2514 addrconf_sit_config(dev);
@@ -2527,25 +2525,30 @@ static int addrconf_notify(struct notifier_block *this, unsigned long event,
2527 addrconf_dev_config(dev); 2525 addrconf_dev_config(dev);
2528 break; 2526 break;
2529 } 2527 }
2528
2530 if (idev) { 2529 if (idev) {
2531 if (run_pending) 2530 if (run_pending)
2532 addrconf_dad_run(idev); 2531 addrconf_dad_run(idev);
2533 2532
2534 /* If the MTU changed during the interface down, when the 2533 /*
2535 interface up, the changed MTU must be reflected in the 2534 * If the MTU changed during the interface down,
2536 idev as well as routers. 2535 * when the interface up, the changed MTU must be
2536 * reflected in the idev as well as routers.
2537 */ 2537 */
2538 if (idev->cnf.mtu6 != dev->mtu && dev->mtu >= IPV6_MIN_MTU) { 2538 if (idev->cnf.mtu6 != dev->mtu &&
2539 dev->mtu >= IPV6_MIN_MTU) {
2539 rt6_mtu_change(dev, dev->mtu); 2540 rt6_mtu_change(dev, dev->mtu);
2540 idev->cnf.mtu6 = dev->mtu; 2541 idev->cnf.mtu6 = dev->mtu;
2541 } 2542 }
2542 idev->tstamp = jiffies; 2543 idev->tstamp = jiffies;
2543 inet6_ifinfo_notify(RTM_NEWLINK, idev); 2544 inet6_ifinfo_notify(RTM_NEWLINK, idev);
2544 /* If the changed mtu during down is lower than IPV6_MIN_MTU 2545
2545 stop IPv6 on this interface. 2546 /*
2547 * If the changed mtu during down is lower than
2548 * IPV6_MIN_MTU stop IPv6 on this interface.
2546 */ 2549 */
2547 if (dev->mtu < IPV6_MIN_MTU) 2550 if (dev->mtu < IPV6_MIN_MTU)
2548 addrconf_ifdown(dev, event != NETDEV_DOWN); 2551 addrconf_ifdown(dev, 1);
2549 } 2552 }
2550 break; 2553 break;
2551 2554
@@ -2562,7 +2565,10 @@ static int addrconf_notify(struct notifier_block *this, unsigned long event,
2562 break; 2565 break;
2563 } 2566 }
2564 2567
2565 /* MTU falled under IPV6_MIN_MTU. Stop IPv6 on this interface. */ 2568 /*
2569 * MTU falled under IPV6_MIN_MTU.
2570 * Stop IPv6 on this interface.
2571 */
2566 2572
2567 case NETDEV_DOWN: 2573 case NETDEV_DOWN:
2568 case NETDEV_UNREGISTER: 2574 case NETDEV_UNREGISTER:
@@ -2582,9 +2588,10 @@ static int addrconf_notify(struct notifier_block *this, unsigned long event,
2582 return notifier_from_errno(err); 2588 return notifier_from_errno(err);
2583 } 2589 }
2584 break; 2590 break;
2585 case NETDEV_BONDING_OLDTYPE: 2591
2586 case NETDEV_BONDING_NEWTYPE: 2592 case NETDEV_PRE_TYPE_CHANGE:
2587 addrconf_bonding_change(dev, event); 2593 case NETDEV_POST_TYPE_CHANGE:
2594 addrconf_type_change(dev, event);
2588 break; 2595 break;
2589 } 2596 }
2590 2597
@@ -2596,28 +2603,27 @@ static int addrconf_notify(struct notifier_block *this, unsigned long event,
2596 */ 2603 */
2597static struct notifier_block ipv6_dev_notf = { 2604static struct notifier_block ipv6_dev_notf = {
2598 .notifier_call = addrconf_notify, 2605 .notifier_call = addrconf_notify,
2599 .priority = 0
2600}; 2606};
2601 2607
2602static void addrconf_bonding_change(struct net_device *dev, unsigned long event) 2608static void addrconf_type_change(struct net_device *dev, unsigned long event)
2603{ 2609{
2604 struct inet6_dev *idev; 2610 struct inet6_dev *idev;
2605 ASSERT_RTNL(); 2611 ASSERT_RTNL();
2606 2612
2607 idev = __in6_dev_get(dev); 2613 idev = __in6_dev_get(dev);
2608 2614
2609 if (event == NETDEV_BONDING_NEWTYPE) 2615 if (event == NETDEV_POST_TYPE_CHANGE)
2610 ipv6_mc_remap(idev); 2616 ipv6_mc_remap(idev);
2611 else if (event == NETDEV_BONDING_OLDTYPE) 2617 else if (event == NETDEV_PRE_TYPE_CHANGE)
2612 ipv6_mc_unmap(idev); 2618 ipv6_mc_unmap(idev);
2613} 2619}
2614 2620
2615static int addrconf_ifdown(struct net_device *dev, int how) 2621static int addrconf_ifdown(struct net_device *dev, int how)
2616{ 2622{
2617 struct inet6_dev *idev;
2618 struct inet6_ifaddr *ifa, **bifa;
2619 struct net *net = dev_net(dev); 2623 struct net *net = dev_net(dev);
2620 int i; 2624 struct inet6_dev *idev;
2625 struct inet6_ifaddr *ifa;
2626 LIST_HEAD(keep_list);
2621 2627
2622 ASSERT_RTNL(); 2628 ASSERT_RTNL();
2623 2629
@@ -2628,8 +2634,9 @@ static int addrconf_ifdown(struct net_device *dev, int how)
2628 if (idev == NULL) 2634 if (idev == NULL)
2629 return -ENODEV; 2635 return -ENODEV;
2630 2636
2631 /* Step 1: remove reference to ipv6 device from parent device. 2637 /*
2632 Do not dev_put! 2638 * Step 1: remove reference to ipv6 device from parent device.
2639 * Do not dev_put!
2633 */ 2640 */
2634 if (how) { 2641 if (how) {
2635 idev->dead = 1; 2642 idev->dead = 1;
@@ -2642,40 +2649,21 @@ static int addrconf_ifdown(struct net_device *dev, int how)
2642 2649
2643 } 2650 }
2644 2651
2645 /* Step 2: clear hash table */
2646 for (i=0; i<IN6_ADDR_HSIZE; i++) {
2647 bifa = &inet6_addr_lst[i];
2648
2649 write_lock_bh(&addrconf_hash_lock);
2650 while ((ifa = *bifa) != NULL) {
2651 if (ifa->idev == idev &&
2652 (how || !(ifa->flags&IFA_F_PERMANENT))) {
2653 *bifa = ifa->lst_next;
2654 ifa->lst_next = NULL;
2655 addrconf_del_timer(ifa);
2656 in6_ifa_put(ifa);
2657 continue;
2658 }
2659 bifa = &ifa->lst_next;
2660 }
2661 write_unlock_bh(&addrconf_hash_lock);
2662 }
2663
2664 write_lock_bh(&idev->lock); 2652 write_lock_bh(&idev->lock);
2665 2653
2666 /* Step 3: clear flags for stateless addrconf */ 2654 /* Step 2: clear flags for stateless addrconf */
2667 if (!how) 2655 if (!how)
2668 idev->if_flags &= ~(IF_RS_SENT|IF_RA_RCVD|IF_READY); 2656 idev->if_flags &= ~(IF_RS_SENT|IF_RA_RCVD|IF_READY);
2669 2657
2670 /* Step 4: clear address list */
2671#ifdef CONFIG_IPV6_PRIVACY 2658#ifdef CONFIG_IPV6_PRIVACY
2672 if (how && del_timer(&idev->regen_timer)) 2659 if (how && del_timer(&idev->regen_timer))
2673 in6_dev_put(idev); 2660 in6_dev_put(idev);
2674 2661
2675 /* clear tempaddr list */ 2662 /* Step 3: clear tempaddr list */
2676 while ((ifa = idev->tempaddr_list) != NULL) { 2663 while (!list_empty(&idev->tempaddr_list)) {
2677 idev->tempaddr_list = ifa->tmp_next; 2664 ifa = list_first_entry(&idev->tempaddr_list,
2678 ifa->tmp_next = NULL; 2665 struct inet6_ifaddr, tmp_list);
2666 list_del(&ifa->tmp_list);
2679 ifa->dead = 1; 2667 ifa->dead = 1;
2680 write_unlock_bh(&idev->lock); 2668 write_unlock_bh(&idev->lock);
2681 spin_lock_bh(&ifa->lock); 2669 spin_lock_bh(&ifa->lock);
@@ -2689,35 +2677,56 @@ static int addrconf_ifdown(struct net_device *dev, int how)
2689 write_lock_bh(&idev->lock); 2677 write_lock_bh(&idev->lock);
2690 } 2678 }
2691#endif 2679#endif
2692 bifa = &idev->addr_list;
2693 while ((ifa = *bifa) != NULL) {
2694 if (how == 0 && (ifa->flags&IFA_F_PERMANENT)) {
2695 /* Retain permanent address on admin down */
2696 bifa = &ifa->if_next;
2697
2698 /* Restart DAD if needed when link comes back up */
2699 if ( !((dev->flags&(IFF_NOARP|IFF_LOOPBACK)) ||
2700 idev->cnf.accept_dad <= 0 ||
2701 (ifa->flags & IFA_F_NODAD)))
2702 ifa->flags |= IFA_F_TENTATIVE;
2703 } else {
2704 *bifa = ifa->if_next;
2705 ifa->if_next = NULL;
2706 2680
2707 ifa->dead = 1; 2681 while (!list_empty(&idev->addr_list)) {
2708 write_unlock_bh(&idev->lock); 2682 ifa = list_first_entry(&idev->addr_list,
2683 struct inet6_ifaddr, if_list);
2684 addrconf_del_timer(ifa);
2685
2686 /* If just doing link down, and address is permanent
2687 and not link-local, then retain it. */
2688 if (!how &&
2689 (ifa->flags&IFA_F_PERMANENT) &&
2690 !(ipv6_addr_type(&ifa->addr) & IPV6_ADDR_LINKLOCAL)) {
2691 list_move_tail(&ifa->if_list, &keep_list);
2692
2693 /* If not doing DAD on this address, just keep it. */
2694 if ((dev->flags&(IFF_NOARP|IFF_LOOPBACK)) ||
2695 idev->cnf.accept_dad <= 0 ||
2696 (ifa->flags & IFA_F_NODAD))
2697 continue;
2709 2698
2710 __ipv6_ifa_notify(RTM_DELADDR, ifa); 2699 /* If it was tentative already, no need to notify */
2711 atomic_notifier_call_chain(&inet6addr_chain, NETDEV_DOWN, ifa); 2700 if (ifa->flags & IFA_F_TENTATIVE)
2712 in6_ifa_put(ifa); 2701 continue;
2713 2702
2714 write_lock_bh(&idev->lock); 2703 /* Flag it for later restoration when link comes up */
2704 ifa->flags |= IFA_F_TENTATIVE;
2705 in6_ifa_hold(ifa);
2706 } else {
2707 list_del(&ifa->if_list);
2708 ifa->dead = 1;
2715 } 2709 }
2710 write_unlock_bh(&idev->lock);
2711
2712 /* clear hash table */
2713 spin_lock_bh(&addrconf_hash_lock);
2714 hlist_del_init_rcu(&ifa->addr_lst);
2715 __in6_ifa_put(ifa);
2716 spin_unlock_bh(&addrconf_hash_lock);
2717
2718 __ipv6_ifa_notify(RTM_DELADDR, ifa);
2719 atomic_notifier_call_chain(&inet6addr_chain, NETDEV_DOWN, ifa);
2720 in6_ifa_put(ifa);
2721
2722 write_lock_bh(&idev->lock);
2716 } 2723 }
2724
2725 list_splice(&keep_list, &idev->addr_list);
2726
2717 write_unlock_bh(&idev->lock); 2727 write_unlock_bh(&idev->lock);
2718 2728
2719 /* Step 5: Discard multicast list */ 2729 /* Step 5: Discard multicast list */
2720
2721 if (how) 2730 if (how)
2722 ipv6_mc_destroy_dev(idev); 2731 ipv6_mc_destroy_dev(idev);
2723 else 2732 else
@@ -2725,8 +2734,7 @@ static int addrconf_ifdown(struct net_device *dev, int how)
2725 2734
2726 idev->tstamp = jiffies; 2735 idev->tstamp = jiffies;
2727 2736
2728 /* Shot the device (if unregistered) */ 2737 /* Last: Shot the device (if unregistered) */
2729
2730 if (how) { 2738 if (how) {
2731 addrconf_sysctl_unregister(idev); 2739 addrconf_sysctl_unregister(idev);
2732 neigh_parms_release(&nd_tbl, idev->nd_parms); 2740 neigh_parms_release(&nd_tbl, idev->nd_parms);
@@ -2739,28 +2747,29 @@ static int addrconf_ifdown(struct net_device *dev, int how)
2739static void addrconf_rs_timer(unsigned long data) 2747static void addrconf_rs_timer(unsigned long data)
2740{ 2748{
2741 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *) data; 2749 struct inet6_ifaddr *ifp = (struct inet6_ifaddr *) data;
2750 struct inet6_dev *idev = ifp->idev;
2742 2751
2743 if (ifp->idev->cnf.forwarding) 2752 read_lock(&idev->lock);
2753 if (idev->dead || !(idev->if_flags & IF_READY))
2744 goto out; 2754 goto out;
2745 2755
2746 if (ifp->idev->if_flags & IF_RA_RCVD) { 2756 if (idev->cnf.forwarding)
2747 /* 2757 goto out;
2748 * Announcement received after solicitation 2758
2749 * was sent 2759 /* Announcement received after solicitation was sent */
2750 */ 2760 if (idev->if_flags & IF_RA_RCVD)
2751 goto out; 2761 goto out;
2752 }
2753 2762
2754 spin_lock(&ifp->lock); 2763 spin_lock(&ifp->lock);
2755 if (ifp->probes++ < ifp->idev->cnf.rtr_solicits) { 2764 if (ifp->probes++ < idev->cnf.rtr_solicits) {
2756 /* The wait after the last probe can be shorter */ 2765 /* The wait after the last probe can be shorter */
2757 addrconf_mod_timer(ifp, AC_RS, 2766 addrconf_mod_timer(ifp, AC_RS,
2758 (ifp->probes == ifp->idev->cnf.rtr_solicits) ? 2767 (ifp->probes == idev->cnf.rtr_solicits) ?
2759 ifp->idev->cnf.rtr_solicit_delay : 2768 idev->cnf.rtr_solicit_delay :
2760 ifp->idev->cnf.rtr_solicit_interval); 2769 idev->cnf.rtr_solicit_interval);
2761 spin_unlock(&ifp->lock); 2770 spin_unlock(&ifp->lock);
2762 2771
2763 ndisc_send_rs(ifp->idev->dev, &ifp->addr, &in6addr_linklocal_allrouters); 2772 ndisc_send_rs(idev->dev, &ifp->addr, &in6addr_linklocal_allrouters);
2764 } else { 2773 } else {
2765 spin_unlock(&ifp->lock); 2774 spin_unlock(&ifp->lock);
2766 /* 2775 /*
@@ -2768,10 +2777,11 @@ static void addrconf_rs_timer(unsigned long data)
2768 * assumption any longer. 2777 * assumption any longer.
2769 */ 2778 */
2770 printk(KERN_DEBUG "%s: no IPv6 routers present\n", 2779 printk(KERN_DEBUG "%s: no IPv6 routers present\n",
2771 ifp->idev->dev->name); 2780 idev->dev->name);
2772 } 2781 }
2773 2782
2774out: 2783out:
2784 read_unlock(&idev->lock);
2775 in6_ifa_put(ifp); 2785 in6_ifa_put(ifp);
2776} 2786}
2777 2787
@@ -2835,7 +2845,7 @@ static void addrconf_dad_start(struct inet6_ifaddr *ifp, u32 flags)
2835 * Optimistic nodes can start receiving 2845 * Optimistic nodes can start receiving
2836 * Frames right away 2846 * Frames right away
2837 */ 2847 */
2838 if(ifp->flags & IFA_F_OPTIMISTIC) 2848 if (ifp->flags & IFA_F_OPTIMISTIC)
2839 ip6_ins_rt(ifp->rt); 2849 ip6_ins_rt(ifp->rt);
2840 2850
2841 addrconf_dad_kick(ifp); 2851 addrconf_dad_kick(ifp);
@@ -2850,9 +2860,9 @@ static void addrconf_dad_timer(unsigned long data)
2850 struct inet6_dev *idev = ifp->idev; 2860 struct inet6_dev *idev = ifp->idev;
2851 struct in6_addr mcaddr; 2861 struct in6_addr mcaddr;
2852 2862
2853 read_lock_bh(&idev->lock); 2863 read_lock(&idev->lock);
2854 if (idev->dead) { 2864 if (idev->dead || !(idev->if_flags & IF_READY)) {
2855 read_unlock_bh(&idev->lock); 2865 read_unlock(&idev->lock);
2856 goto out; 2866 goto out;
2857 } 2867 }
2858 2868
@@ -2864,7 +2874,7 @@ static void addrconf_dad_timer(unsigned long data)
2864 2874
2865 ifp->flags &= ~(IFA_F_TENTATIVE|IFA_F_OPTIMISTIC|IFA_F_DADFAILED); 2875 ifp->flags &= ~(IFA_F_TENTATIVE|IFA_F_OPTIMISTIC|IFA_F_DADFAILED);
2866 spin_unlock(&ifp->lock); 2876 spin_unlock(&ifp->lock);
2867 read_unlock_bh(&idev->lock); 2877 read_unlock(&idev->lock);
2868 2878
2869 addrconf_dad_completed(ifp); 2879 addrconf_dad_completed(ifp);
2870 2880
@@ -2874,7 +2884,7 @@ static void addrconf_dad_timer(unsigned long data)
2874 ifp->probes--; 2884 ifp->probes--;
2875 addrconf_mod_timer(ifp, AC_DAD, ifp->idev->nd_parms->retrans_time); 2885 addrconf_mod_timer(ifp, AC_DAD, ifp->idev->nd_parms->retrans_time);
2876 spin_unlock(&ifp->lock); 2886 spin_unlock(&ifp->lock);
2877 read_unlock_bh(&idev->lock); 2887 read_unlock(&idev->lock);
2878 2888
2879 /* send a neighbour solicitation for our addr */ 2889 /* send a neighbour solicitation for our addr */
2880 addrconf_addr_solict_mult(&ifp->addr, &mcaddr); 2890 addrconf_addr_solict_mult(&ifp->addr, &mcaddr);
@@ -2885,7 +2895,7 @@ out:
2885 2895
2886static void addrconf_dad_completed(struct inet6_ifaddr *ifp) 2896static void addrconf_dad_completed(struct inet6_ifaddr *ifp)
2887{ 2897{
2888 struct net_device * dev = ifp->idev->dev; 2898 struct net_device *dev = ifp->idev->dev;
2889 2899
2890 /* 2900 /*
2891 * Configure the address for reception. Now it is valid. 2901 * Configure the address for reception. Now it is valid.
@@ -2916,11 +2926,12 @@ static void addrconf_dad_completed(struct inet6_ifaddr *ifp)
2916 } 2926 }
2917} 2927}
2918 2928
2919static void addrconf_dad_run(struct inet6_dev *idev) { 2929static void addrconf_dad_run(struct inet6_dev *idev)
2930{
2920 struct inet6_ifaddr *ifp; 2931 struct inet6_ifaddr *ifp;
2921 2932
2922 read_lock_bh(&idev->lock); 2933 read_lock_bh(&idev->lock);
2923 for (ifp = idev->addr_list; ifp; ifp = ifp->if_next) { 2934 list_for_each_entry(ifp, &idev->addr_list, if_list) {
2924 spin_lock(&ifp->lock); 2935 spin_lock(&ifp->lock);
2925 if (!(ifp->flags & IFA_F_TENTATIVE)) { 2936 if (!(ifp->flags & IFA_F_TENTATIVE)) {
2926 spin_unlock(&ifp->lock); 2937 spin_unlock(&ifp->lock);
@@ -2945,36 +2956,35 @@ static struct inet6_ifaddr *if6_get_first(struct seq_file *seq)
2945 struct net *net = seq_file_net(seq); 2956 struct net *net = seq_file_net(seq);
2946 2957
2947 for (state->bucket = 0; state->bucket < IN6_ADDR_HSIZE; ++state->bucket) { 2958 for (state->bucket = 0; state->bucket < IN6_ADDR_HSIZE; ++state->bucket) {
2948 ifa = inet6_addr_lst[state->bucket]; 2959 struct hlist_node *n;
2949 2960 hlist_for_each_entry_rcu(ifa, n, &inet6_addr_lst[state->bucket],
2950 while (ifa && !net_eq(dev_net(ifa->idev->dev), net)) 2961 addr_lst)
2951 ifa = ifa->lst_next; 2962 if (net_eq(dev_net(ifa->idev->dev), net))
2952 if (ifa) 2963 return ifa;
2953 break;
2954 } 2964 }
2955 return ifa; 2965 return NULL;
2956} 2966}
2957 2967
2958static struct inet6_ifaddr *if6_get_next(struct seq_file *seq, struct inet6_ifaddr *ifa) 2968static struct inet6_ifaddr *if6_get_next(struct seq_file *seq,
2969 struct inet6_ifaddr *ifa)
2959{ 2970{
2960 struct if6_iter_state *state = seq->private; 2971 struct if6_iter_state *state = seq->private;
2961 struct net *net = seq_file_net(seq); 2972 struct net *net = seq_file_net(seq);
2973 struct hlist_node *n = &ifa->addr_lst;
2962 2974
2963 ifa = ifa->lst_next; 2975 hlist_for_each_entry_continue_rcu(ifa, n, addr_lst)
2964try_again: 2976 if (net_eq(dev_net(ifa->idev->dev), net))
2965 if (ifa) { 2977 return ifa;
2966 if (!net_eq(dev_net(ifa->idev->dev), net)) {
2967 ifa = ifa->lst_next;
2968 goto try_again;
2969 }
2970 }
2971 2978
2972 if (!ifa && ++state->bucket < IN6_ADDR_HSIZE) { 2979 while (++state->bucket < IN6_ADDR_HSIZE) {
2973 ifa = inet6_addr_lst[state->bucket]; 2980 hlist_for_each_entry(ifa, n,
2974 goto try_again; 2981 &inet6_addr_lst[state->bucket], addr_lst) {
2982 if (net_eq(dev_net(ifa->idev->dev), net))
2983 return ifa;
2984 }
2975 } 2985 }
2976 2986
2977 return ifa; 2987 return NULL;
2978} 2988}
2979 2989
2980static struct inet6_ifaddr *if6_get_idx(struct seq_file *seq, loff_t pos) 2990static struct inet6_ifaddr *if6_get_idx(struct seq_file *seq, loff_t pos)
@@ -2982,15 +2992,15 @@ static struct inet6_ifaddr *if6_get_idx(struct seq_file *seq, loff_t pos)
2982 struct inet6_ifaddr *ifa = if6_get_first(seq); 2992 struct inet6_ifaddr *ifa = if6_get_first(seq);
2983 2993
2984 if (ifa) 2994 if (ifa)
2985 while(pos && (ifa = if6_get_next(seq, ifa)) != NULL) 2995 while (pos && (ifa = if6_get_next(seq, ifa)) != NULL)
2986 --pos; 2996 --pos;
2987 return pos ? NULL : ifa; 2997 return pos ? NULL : ifa;
2988} 2998}
2989 2999
2990static void *if6_seq_start(struct seq_file *seq, loff_t *pos) 3000static void *if6_seq_start(struct seq_file *seq, loff_t *pos)
2991 __acquires(addrconf_hash_lock) 3001 __acquires(rcu)
2992{ 3002{
2993 read_lock_bh(&addrconf_hash_lock); 3003 rcu_read_lock_bh();
2994 return if6_get_idx(seq, *pos); 3004 return if6_get_idx(seq, *pos);
2995} 3005}
2996 3006
@@ -3004,9 +3014,9 @@ static void *if6_seq_next(struct seq_file *seq, void *v, loff_t *pos)
3004} 3014}
3005 3015
3006static void if6_seq_stop(struct seq_file *seq, void *v) 3016static void if6_seq_stop(struct seq_file *seq, void *v)
3007 __releases(addrconf_hash_lock) 3017 __releases(rcu)
3008{ 3018{
3009 read_unlock_bh(&addrconf_hash_lock); 3019 rcu_read_unlock_bh();
3010} 3020}
3011 3021
3012static int if6_seq_show(struct seq_file *seq, void *v) 3022static int if6_seq_show(struct seq_file *seq, void *v)
@@ -3076,10 +3086,12 @@ void if6_proc_exit(void)
3076int ipv6_chk_home_addr(struct net *net, struct in6_addr *addr) 3086int ipv6_chk_home_addr(struct net *net, struct in6_addr *addr)
3077{ 3087{
3078 int ret = 0; 3088 int ret = 0;
3079 struct inet6_ifaddr * ifp; 3089 struct inet6_ifaddr *ifp = NULL;
3080 u8 hash = ipv6_addr_hash(addr); 3090 struct hlist_node *n;
3081 read_lock_bh(&addrconf_hash_lock); 3091 unsigned int hash = ipv6_addr_hash(addr);
3082 for (ifp = inet6_addr_lst[hash]; ifp; ifp = ifp->lst_next) { 3092
3093 rcu_read_lock_bh();
3094 hlist_for_each_entry_rcu(ifp, n, &inet6_addr_lst[hash], addr_lst) {
3083 if (!net_eq(dev_net(ifp->idev->dev), net)) 3095 if (!net_eq(dev_net(ifp->idev->dev), net))
3084 continue; 3096 continue;
3085 if (ipv6_addr_equal(&ifp->addr, addr) && 3097 if (ipv6_addr_equal(&ifp->addr, addr) &&
@@ -3088,7 +3100,7 @@ int ipv6_chk_home_addr(struct net *net, struct in6_addr *addr)
3088 break; 3100 break;
3089 } 3101 }
3090 } 3102 }
3091 read_unlock_bh(&addrconf_hash_lock); 3103 rcu_read_unlock_bh();
3092 return ret; 3104 return ret;
3093} 3105}
3094#endif 3106#endif
@@ -3099,43 +3111,35 @@ int ipv6_chk_home_addr(struct net *net, struct in6_addr *addr)
3099 3111
3100static void addrconf_verify(unsigned long foo) 3112static void addrconf_verify(unsigned long foo)
3101{ 3113{
3114 unsigned long now, next, next_sec, next_sched;
3102 struct inet6_ifaddr *ifp; 3115 struct inet6_ifaddr *ifp;
3103 unsigned long now, next; 3116 struct hlist_node *node;
3104 int i; 3117 int i;
3105 3118
3106 spin_lock_bh(&addrconf_verify_lock); 3119 rcu_read_lock_bh();
3120 spin_lock(&addrconf_verify_lock);
3107 now = jiffies; 3121 now = jiffies;
3108 next = now + ADDR_CHECK_FREQUENCY; 3122 next = round_jiffies_up(now + ADDR_CHECK_FREQUENCY);
3109 3123
3110 del_timer(&addr_chk_timer); 3124 del_timer(&addr_chk_timer);
3111 3125
3112 for (i=0; i < IN6_ADDR_HSIZE; i++) { 3126 for (i = 0; i < IN6_ADDR_HSIZE; i++) {
3113
3114restart: 3127restart:
3115 read_lock(&addrconf_hash_lock); 3128 hlist_for_each_entry_rcu(ifp, node,
3116 for (ifp=inet6_addr_lst[i]; ifp; ifp=ifp->lst_next) { 3129 &inet6_addr_lst[i], addr_lst) {
3117 unsigned long age; 3130 unsigned long age;
3118#ifdef CONFIG_IPV6_PRIVACY
3119 unsigned long regen_advance;
3120#endif
3121 3131
3122 if (ifp->flags & IFA_F_PERMANENT) 3132 if (ifp->flags & IFA_F_PERMANENT)
3123 continue; 3133 continue;
3124 3134
3125 spin_lock(&ifp->lock); 3135 spin_lock(&ifp->lock);
3126 age = (now - ifp->tstamp) / HZ; 3136 /* We try to batch several events at once. */
3127 3137 age = (now - ifp->tstamp + ADDRCONF_TIMER_FUZZ_MINUS) / HZ;
3128#ifdef CONFIG_IPV6_PRIVACY
3129 regen_advance = ifp->idev->cnf.regen_max_retry *
3130 ifp->idev->cnf.dad_transmits *
3131 ifp->idev->nd_parms->retrans_time / HZ;
3132#endif
3133 3138
3134 if (ifp->valid_lft != INFINITY_LIFE_TIME && 3139 if (ifp->valid_lft != INFINITY_LIFE_TIME &&
3135 age >= ifp->valid_lft) { 3140 age >= ifp->valid_lft) {
3136 spin_unlock(&ifp->lock); 3141 spin_unlock(&ifp->lock);
3137 in6_ifa_hold(ifp); 3142 in6_ifa_hold(ifp);
3138 read_unlock(&addrconf_hash_lock);
3139 ipv6_del_addr(ifp); 3143 ipv6_del_addr(ifp);
3140 goto restart; 3144 goto restart;
3141 } else if (ifp->prefered_lft == INFINITY_LIFE_TIME) { 3145 } else if (ifp->prefered_lft == INFINITY_LIFE_TIME) {
@@ -3157,7 +3161,6 @@ restart:
3157 3161
3158 if (deprecate) { 3162 if (deprecate) {
3159 in6_ifa_hold(ifp); 3163 in6_ifa_hold(ifp);
3160 read_unlock(&addrconf_hash_lock);
3161 3164
3162 ipv6_ifa_notify(0, ifp); 3165 ipv6_ifa_notify(0, ifp);
3163 in6_ifa_put(ifp); 3166 in6_ifa_put(ifp);
@@ -3166,6 +3169,10 @@ restart:
3166#ifdef CONFIG_IPV6_PRIVACY 3169#ifdef CONFIG_IPV6_PRIVACY
3167 } else if ((ifp->flags&IFA_F_TEMPORARY) && 3170 } else if ((ifp->flags&IFA_F_TEMPORARY) &&
3168 !(ifp->flags&IFA_F_TENTATIVE)) { 3171 !(ifp->flags&IFA_F_TENTATIVE)) {
3172 unsigned long regen_advance = ifp->idev->cnf.regen_max_retry *
3173 ifp->idev->cnf.dad_transmits *
3174 ifp->idev->nd_parms->retrans_time / HZ;
3175
3169 if (age >= ifp->prefered_lft - regen_advance) { 3176 if (age >= ifp->prefered_lft - regen_advance) {
3170 struct inet6_ifaddr *ifpub = ifp->ifpub; 3177 struct inet6_ifaddr *ifpub = ifp->ifpub;
3171 if (time_before(ifp->tstamp + ifp->prefered_lft * HZ, next)) 3178 if (time_before(ifp->tstamp + ifp->prefered_lft * HZ, next))
@@ -3175,7 +3182,7 @@ restart:
3175 in6_ifa_hold(ifp); 3182 in6_ifa_hold(ifp);
3176 in6_ifa_hold(ifpub); 3183 in6_ifa_hold(ifpub);
3177 spin_unlock(&ifp->lock); 3184 spin_unlock(&ifp->lock);
3178 read_unlock(&addrconf_hash_lock); 3185
3179 spin_lock(&ifpub->lock); 3186 spin_lock(&ifpub->lock);
3180 ifpub->regen_count = 0; 3187 ifpub->regen_count = 0;
3181 spin_unlock(&ifpub->lock); 3188 spin_unlock(&ifpub->lock);
@@ -3195,12 +3202,26 @@ restart:
3195 spin_unlock(&ifp->lock); 3202 spin_unlock(&ifp->lock);
3196 } 3203 }
3197 } 3204 }
3198 read_unlock(&addrconf_hash_lock);
3199 } 3205 }
3200 3206
3201 addr_chk_timer.expires = time_before(next, jiffies + HZ) ? jiffies + HZ : next; 3207 next_sec = round_jiffies_up(next);
3208 next_sched = next;
3209
3210 /* If rounded timeout is accurate enough, accept it. */
3211 if (time_before(next_sec, next + ADDRCONF_TIMER_FUZZ))
3212 next_sched = next_sec;
3213
3214 /* And minimum interval is ADDRCONF_TIMER_FUZZ_MAX. */
3215 if (time_before(next_sched, jiffies + ADDRCONF_TIMER_FUZZ_MAX))
3216 next_sched = jiffies + ADDRCONF_TIMER_FUZZ_MAX;
3217
3218 ADBG((KERN_DEBUG "now = %lu, schedule = %lu, rounded schedule = %lu => %lu\n",
3219 now, next, next_sec, next_sched));
3220
3221 addr_chk_timer.expires = next_sched;
3202 add_timer(&addr_chk_timer); 3222 add_timer(&addr_chk_timer);
3203 spin_unlock_bh(&addrconf_verify_lock); 3223 spin_unlock(&addrconf_verify_lock);
3224 rcu_read_unlock_bh();
3204} 3225}
3205 3226
3206static struct in6_addr *extract_addr(struct nlattr *addr, struct nlattr *local) 3227static struct in6_addr *extract_addr(struct nlattr *addr, struct nlattr *local)
@@ -3490,8 +3511,7 @@ static int inet6_fill_ifacaddr(struct sk_buff *skb, struct ifacaddr6 *ifaca,
3490 return nlmsg_end(skb, nlh); 3511 return nlmsg_end(skb, nlh);
3491} 3512}
3492 3513
3493enum addr_type_t 3514enum addr_type_t {
3494{
3495 UNICAST_ADDR, 3515 UNICAST_ADDR,
3496 MULTICAST_ADDR, 3516 MULTICAST_ADDR,
3497 ANYCAST_ADDR, 3517 ANYCAST_ADDR,
@@ -3502,7 +3522,6 @@ static int in6_dump_addrs(struct inet6_dev *idev, struct sk_buff *skb,
3502 struct netlink_callback *cb, enum addr_type_t type, 3522 struct netlink_callback *cb, enum addr_type_t type,
3503 int s_ip_idx, int *p_ip_idx) 3523 int s_ip_idx, int *p_ip_idx)
3504{ 3524{
3505 struct inet6_ifaddr *ifa;
3506 struct ifmcaddr6 *ifmca; 3525 struct ifmcaddr6 *ifmca;
3507 struct ifacaddr6 *ifaca; 3526 struct ifacaddr6 *ifaca;
3508 int err = 1; 3527 int err = 1;
@@ -3510,11 +3529,12 @@ static int in6_dump_addrs(struct inet6_dev *idev, struct sk_buff *skb,
3510 3529
3511 read_lock_bh(&idev->lock); 3530 read_lock_bh(&idev->lock);
3512 switch (type) { 3531 switch (type) {
3513 case UNICAST_ADDR: 3532 case UNICAST_ADDR: {
3533 struct inet6_ifaddr *ifa;
3534
3514 /* unicast address incl. temp addr */ 3535 /* unicast address incl. temp addr */
3515 for (ifa = idev->addr_list; ifa; 3536 list_for_each_entry(ifa, &idev->addr_list, if_list) {
3516 ifa = ifa->if_next, ip_idx++) { 3537 if (++ip_idx < s_ip_idx)
3517 if (ip_idx < s_ip_idx)
3518 continue; 3538 continue;
3519 err = inet6_fill_ifaddr(skb, ifa, 3539 err = inet6_fill_ifaddr(skb, ifa,
3520 NETLINK_CB(cb->skb).pid, 3540 NETLINK_CB(cb->skb).pid,
@@ -3525,6 +3545,7 @@ static int in6_dump_addrs(struct inet6_dev *idev, struct sk_buff *skb,
3525 break; 3545 break;
3526 } 3546 }
3527 break; 3547 break;
3548 }
3528 case MULTICAST_ADDR: 3549 case MULTICAST_ADDR:
3529 /* multicast address */ 3550 /* multicast address */
3530 for (ifmca = idev->mc_list; ifmca; 3551 for (ifmca = idev->mc_list; ifmca;
@@ -3589,7 +3610,8 @@ static int inet6_dump_addr(struct sk_buff *skb, struct netlink_callback *cb,
3589 if (idx > s_idx) 3610 if (idx > s_idx)
3590 s_ip_idx = 0; 3611 s_ip_idx = 0;
3591 ip_idx = 0; 3612 ip_idx = 0;
3592 if ((idev = __in6_dev_get(dev)) == NULL) 3613 idev = __in6_dev_get(dev);
3614 if (!idev)
3593 goto cont; 3615 goto cont;
3594 3616
3595 if (in6_dump_addrs(idev, skb, cb, type, 3617 if (in6_dump_addrs(idev, skb, cb, type,
@@ -3656,12 +3678,14 @@ static int inet6_rtm_getaddr(struct sk_buff *in_skb, struct nlmsghdr* nlh,
3656 if (ifm->ifa_index) 3678 if (ifm->ifa_index)
3657 dev = __dev_get_by_index(net, ifm->ifa_index); 3679 dev = __dev_get_by_index(net, ifm->ifa_index);
3658 3680
3659 if ((ifa = ipv6_get_ifaddr(net, addr, dev, 1)) == NULL) { 3681 ifa = ipv6_get_ifaddr(net, addr, dev, 1);
3682 if (!ifa) {
3660 err = -EADDRNOTAVAIL; 3683 err = -EADDRNOTAVAIL;
3661 goto errout; 3684 goto errout;
3662 } 3685 }
3663 3686
3664 if ((skb = nlmsg_new(inet6_ifaddr_msgsize(), GFP_KERNEL)) == NULL) { 3687 skb = nlmsg_new(inet6_ifaddr_msgsize(), GFP_KERNEL);
3688 if (!skb) {
3665 err = -ENOBUFS; 3689 err = -ENOBUFS;
3666 goto errout_ifa; 3690 goto errout_ifa;
3667 } 3691 }
@@ -3786,7 +3810,7 @@ static inline void __snmp6_fill_stats(u64 *stats, void __percpu **mib,
3786static void snmp6_fill_stats(u64 *stats, struct inet6_dev *idev, int attrtype, 3810static void snmp6_fill_stats(u64 *stats, struct inet6_dev *idev, int attrtype,
3787 int bytes) 3811 int bytes)
3788{ 3812{
3789 switch(attrtype) { 3813 switch (attrtype) {
3790 case IFLA_INET6_STATS: 3814 case IFLA_INET6_STATS:
3791 __snmp6_fill_stats(stats, (void __percpu **)idev->stats.ipv6, IPSTATS_MIB_MAX, bytes); 3815 __snmp6_fill_stats(stats, (void __percpu **)idev->stats.ipv6, IPSTATS_MIB_MAX, bytes);
3792 break; 3816 break;
@@ -4138,211 +4162,211 @@ static struct addrconf_sysctl_table
4138 .sysctl_header = NULL, 4162 .sysctl_header = NULL,
4139 .addrconf_vars = { 4163 .addrconf_vars = {
4140 { 4164 {
4141 .procname = "forwarding", 4165 .procname = "forwarding",
4142 .data = &ipv6_devconf.forwarding, 4166 .data = &ipv6_devconf.forwarding,
4143 .maxlen = sizeof(int), 4167 .maxlen = sizeof(int),
4144 .mode = 0644, 4168 .mode = 0644,
4145 .proc_handler = addrconf_sysctl_forward, 4169 .proc_handler = addrconf_sysctl_forward,
4146 }, 4170 },
4147 { 4171 {
4148 .procname = "hop_limit", 4172 .procname = "hop_limit",
4149 .data = &ipv6_devconf.hop_limit, 4173 .data = &ipv6_devconf.hop_limit,
4150 .maxlen = sizeof(int), 4174 .maxlen = sizeof(int),
4151 .mode = 0644, 4175 .mode = 0644,
4152 .proc_handler = proc_dointvec, 4176 .proc_handler = proc_dointvec,
4153 }, 4177 },
4154 { 4178 {
4155 .procname = "mtu", 4179 .procname = "mtu",
4156 .data = &ipv6_devconf.mtu6, 4180 .data = &ipv6_devconf.mtu6,
4157 .maxlen = sizeof(int), 4181 .maxlen = sizeof(int),
4158 .mode = 0644, 4182 .mode = 0644,
4159 .proc_handler = proc_dointvec, 4183 .proc_handler = proc_dointvec,
4160 }, 4184 },
4161 { 4185 {
4162 .procname = "accept_ra", 4186 .procname = "accept_ra",
4163 .data = &ipv6_devconf.accept_ra, 4187 .data = &ipv6_devconf.accept_ra,
4164 .maxlen = sizeof(int), 4188 .maxlen = sizeof(int),
4165 .mode = 0644, 4189 .mode = 0644,
4166 .proc_handler = proc_dointvec, 4190 .proc_handler = proc_dointvec,
4167 }, 4191 },
4168 { 4192 {
4169 .procname = "accept_redirects", 4193 .procname = "accept_redirects",
4170 .data = &ipv6_devconf.accept_redirects, 4194 .data = &ipv6_devconf.accept_redirects,
4171 .maxlen = sizeof(int), 4195 .maxlen = sizeof(int),
4172 .mode = 0644, 4196 .mode = 0644,
4173 .proc_handler = proc_dointvec, 4197 .proc_handler = proc_dointvec,
4174 }, 4198 },
4175 { 4199 {
4176 .procname = "autoconf", 4200 .procname = "autoconf",
4177 .data = &ipv6_devconf.autoconf, 4201 .data = &ipv6_devconf.autoconf,
4178 .maxlen = sizeof(int), 4202 .maxlen = sizeof(int),
4179 .mode = 0644, 4203 .mode = 0644,
4180 .proc_handler = proc_dointvec, 4204 .proc_handler = proc_dointvec,
4181 }, 4205 },
4182 { 4206 {
4183 .procname = "dad_transmits", 4207 .procname = "dad_transmits",
4184 .data = &ipv6_devconf.dad_transmits, 4208 .data = &ipv6_devconf.dad_transmits,
4185 .maxlen = sizeof(int), 4209 .maxlen = sizeof(int),
4186 .mode = 0644, 4210 .mode = 0644,
4187 .proc_handler = proc_dointvec, 4211 .proc_handler = proc_dointvec,
4188 }, 4212 },
4189 { 4213 {
4190 .procname = "router_solicitations", 4214 .procname = "router_solicitations",
4191 .data = &ipv6_devconf.rtr_solicits, 4215 .data = &ipv6_devconf.rtr_solicits,
4192 .maxlen = sizeof(int), 4216 .maxlen = sizeof(int),
4193 .mode = 0644, 4217 .mode = 0644,
4194 .proc_handler = proc_dointvec, 4218 .proc_handler = proc_dointvec,
4195 }, 4219 },
4196 { 4220 {
4197 .procname = "router_solicitation_interval", 4221 .procname = "router_solicitation_interval",
4198 .data = &ipv6_devconf.rtr_solicit_interval, 4222 .data = &ipv6_devconf.rtr_solicit_interval,
4199 .maxlen = sizeof(int), 4223 .maxlen = sizeof(int),
4200 .mode = 0644, 4224 .mode = 0644,
4201 .proc_handler = proc_dointvec_jiffies, 4225 .proc_handler = proc_dointvec_jiffies,
4202 }, 4226 },
4203 { 4227 {
4204 .procname = "router_solicitation_delay", 4228 .procname = "router_solicitation_delay",
4205 .data = &ipv6_devconf.rtr_solicit_delay, 4229 .data = &ipv6_devconf.rtr_solicit_delay,
4206 .maxlen = sizeof(int), 4230 .maxlen = sizeof(int),
4207 .mode = 0644, 4231 .mode = 0644,
4208 .proc_handler = proc_dointvec_jiffies, 4232 .proc_handler = proc_dointvec_jiffies,
4209 }, 4233 },
4210 { 4234 {
4211 .procname = "force_mld_version", 4235 .procname = "force_mld_version",
4212 .data = &ipv6_devconf.force_mld_version, 4236 .data = &ipv6_devconf.force_mld_version,
4213 .maxlen = sizeof(int), 4237 .maxlen = sizeof(int),
4214 .mode = 0644, 4238 .mode = 0644,
4215 .proc_handler = proc_dointvec, 4239 .proc_handler = proc_dointvec,
4216 }, 4240 },
4217#ifdef CONFIG_IPV6_PRIVACY 4241#ifdef CONFIG_IPV6_PRIVACY
4218 { 4242 {
4219 .procname = "use_tempaddr", 4243 .procname = "use_tempaddr",
4220 .data = &ipv6_devconf.use_tempaddr, 4244 .data = &ipv6_devconf.use_tempaddr,
4221 .maxlen = sizeof(int), 4245 .maxlen = sizeof(int),
4222 .mode = 0644, 4246 .mode = 0644,
4223 .proc_handler = proc_dointvec, 4247 .proc_handler = proc_dointvec,
4224 }, 4248 },
4225 { 4249 {
4226 .procname = "temp_valid_lft", 4250 .procname = "temp_valid_lft",
4227 .data = &ipv6_devconf.temp_valid_lft, 4251 .data = &ipv6_devconf.temp_valid_lft,
4228 .maxlen = sizeof(int), 4252 .maxlen = sizeof(int),
4229 .mode = 0644, 4253 .mode = 0644,
4230 .proc_handler = proc_dointvec, 4254 .proc_handler = proc_dointvec,
4231 }, 4255 },
4232 { 4256 {
4233 .procname = "temp_prefered_lft", 4257 .procname = "temp_prefered_lft",
4234 .data = &ipv6_devconf.temp_prefered_lft, 4258 .data = &ipv6_devconf.temp_prefered_lft,
4235 .maxlen = sizeof(int), 4259 .maxlen = sizeof(int),
4236 .mode = 0644, 4260 .mode = 0644,
4237 .proc_handler = proc_dointvec, 4261 .proc_handler = proc_dointvec,
4238 }, 4262 },
4239 { 4263 {
4240 .procname = "regen_max_retry", 4264 .procname = "regen_max_retry",
4241 .data = &ipv6_devconf.regen_max_retry, 4265 .data = &ipv6_devconf.regen_max_retry,
4242 .maxlen = sizeof(int), 4266 .maxlen = sizeof(int),
4243 .mode = 0644, 4267 .mode = 0644,
4244 .proc_handler = proc_dointvec, 4268 .proc_handler = proc_dointvec,
4245 }, 4269 },
4246 { 4270 {
4247 .procname = "max_desync_factor", 4271 .procname = "max_desync_factor",
4248 .data = &ipv6_devconf.max_desync_factor, 4272 .data = &ipv6_devconf.max_desync_factor,
4249 .maxlen = sizeof(int), 4273 .maxlen = sizeof(int),
4250 .mode = 0644, 4274 .mode = 0644,
4251 .proc_handler = proc_dointvec, 4275 .proc_handler = proc_dointvec,
4252 }, 4276 },
4253#endif 4277#endif
4254 { 4278 {
4255 .procname = "max_addresses", 4279 .procname = "max_addresses",
4256 .data = &ipv6_devconf.max_addresses, 4280 .data = &ipv6_devconf.max_addresses,
4257 .maxlen = sizeof(int), 4281 .maxlen = sizeof(int),
4258 .mode = 0644, 4282 .mode = 0644,
4259 .proc_handler = proc_dointvec, 4283 .proc_handler = proc_dointvec,
4260 }, 4284 },
4261 { 4285 {
4262 .procname = "accept_ra_defrtr", 4286 .procname = "accept_ra_defrtr",
4263 .data = &ipv6_devconf.accept_ra_defrtr, 4287 .data = &ipv6_devconf.accept_ra_defrtr,
4264 .maxlen = sizeof(int), 4288 .maxlen = sizeof(int),
4265 .mode = 0644, 4289 .mode = 0644,
4266 .proc_handler = proc_dointvec, 4290 .proc_handler = proc_dointvec,
4267 }, 4291 },
4268 { 4292 {
4269 .procname = "accept_ra_pinfo", 4293 .procname = "accept_ra_pinfo",
4270 .data = &ipv6_devconf.accept_ra_pinfo, 4294 .data = &ipv6_devconf.accept_ra_pinfo,
4271 .maxlen = sizeof(int), 4295 .maxlen = sizeof(int),
4272 .mode = 0644, 4296 .mode = 0644,
4273 .proc_handler = proc_dointvec, 4297 .proc_handler = proc_dointvec,
4274 }, 4298 },
4275#ifdef CONFIG_IPV6_ROUTER_PREF 4299#ifdef CONFIG_IPV6_ROUTER_PREF
4276 { 4300 {
4277 .procname = "accept_ra_rtr_pref", 4301 .procname = "accept_ra_rtr_pref",
4278 .data = &ipv6_devconf.accept_ra_rtr_pref, 4302 .data = &ipv6_devconf.accept_ra_rtr_pref,
4279 .maxlen = sizeof(int), 4303 .maxlen = sizeof(int),
4280 .mode = 0644, 4304 .mode = 0644,
4281 .proc_handler = proc_dointvec, 4305 .proc_handler = proc_dointvec,
4282 }, 4306 },
4283 { 4307 {
4284 .procname = "router_probe_interval", 4308 .procname = "router_probe_interval",
4285 .data = &ipv6_devconf.rtr_probe_interval, 4309 .data = &ipv6_devconf.rtr_probe_interval,
4286 .maxlen = sizeof(int), 4310 .maxlen = sizeof(int),
4287 .mode = 0644, 4311 .mode = 0644,
4288 .proc_handler = proc_dointvec_jiffies, 4312 .proc_handler = proc_dointvec_jiffies,
4289 }, 4313 },
4290#ifdef CONFIG_IPV6_ROUTE_INFO 4314#ifdef CONFIG_IPV6_ROUTE_INFO
4291 { 4315 {
4292 .procname = "accept_ra_rt_info_max_plen", 4316 .procname = "accept_ra_rt_info_max_plen",
4293 .data = &ipv6_devconf.accept_ra_rt_info_max_plen, 4317 .data = &ipv6_devconf.accept_ra_rt_info_max_plen,
4294 .maxlen = sizeof(int), 4318 .maxlen = sizeof(int),
4295 .mode = 0644, 4319 .mode = 0644,
4296 .proc_handler = proc_dointvec, 4320 .proc_handler = proc_dointvec,
4297 }, 4321 },
4298#endif 4322#endif
4299#endif 4323#endif
4300 { 4324 {
4301 .procname = "proxy_ndp", 4325 .procname = "proxy_ndp",
4302 .data = &ipv6_devconf.proxy_ndp, 4326 .data = &ipv6_devconf.proxy_ndp,
4303 .maxlen = sizeof(int), 4327 .maxlen = sizeof(int),
4304 .mode = 0644, 4328 .mode = 0644,
4305 .proc_handler = proc_dointvec, 4329 .proc_handler = proc_dointvec,
4306 }, 4330 },
4307 { 4331 {
4308 .procname = "accept_source_route", 4332 .procname = "accept_source_route",
4309 .data = &ipv6_devconf.accept_source_route, 4333 .data = &ipv6_devconf.accept_source_route,
4310 .maxlen = sizeof(int), 4334 .maxlen = sizeof(int),
4311 .mode = 0644, 4335 .mode = 0644,
4312 .proc_handler = proc_dointvec, 4336 .proc_handler = proc_dointvec,
4313 }, 4337 },
4314#ifdef CONFIG_IPV6_OPTIMISTIC_DAD 4338#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
4315 { 4339 {
4316 .procname = "optimistic_dad", 4340 .procname = "optimistic_dad",
4317 .data = &ipv6_devconf.optimistic_dad, 4341 .data = &ipv6_devconf.optimistic_dad,
4318 .maxlen = sizeof(int), 4342 .maxlen = sizeof(int),
4319 .mode = 0644, 4343 .mode = 0644,
4320 .proc_handler = proc_dointvec, 4344 .proc_handler = proc_dointvec,
4321 4345
4322 }, 4346 },
4323#endif 4347#endif
4324#ifdef CONFIG_IPV6_MROUTE 4348#ifdef CONFIG_IPV6_MROUTE
4325 { 4349 {
4326 .procname = "mc_forwarding", 4350 .procname = "mc_forwarding",
4327 .data = &ipv6_devconf.mc_forwarding, 4351 .data = &ipv6_devconf.mc_forwarding,
4328 .maxlen = sizeof(int), 4352 .maxlen = sizeof(int),
4329 .mode = 0444, 4353 .mode = 0444,
4330 .proc_handler = proc_dointvec, 4354 .proc_handler = proc_dointvec,
4331 }, 4355 },
4332#endif 4356#endif
4333 { 4357 {
4334 .procname = "disable_ipv6", 4358 .procname = "disable_ipv6",
4335 .data = &ipv6_devconf.disable_ipv6, 4359 .data = &ipv6_devconf.disable_ipv6,
4336 .maxlen = sizeof(int), 4360 .maxlen = sizeof(int),
4337 .mode = 0644, 4361 .mode = 0644,
4338 .proc_handler = addrconf_sysctl_disable, 4362 .proc_handler = addrconf_sysctl_disable,
4339 }, 4363 },
4340 { 4364 {
4341 .procname = "accept_dad", 4365 .procname = "accept_dad",
4342 .data = &ipv6_devconf.accept_dad, 4366 .data = &ipv6_devconf.accept_dad,
4343 .maxlen = sizeof(int), 4367 .maxlen = sizeof(int),
4344 .mode = 0644, 4368 .mode = 0644,
4345 .proc_handler = proc_dointvec, 4369 .proc_handler = proc_dointvec,
4346 }, 4370 },
4347 { 4371 {
4348 .procname = "force_tllao", 4372 .procname = "force_tllao",
@@ -4378,8 +4402,8 @@ static int __addrconf_sysctl_register(struct net *net, char *dev_name,
4378 if (t == NULL) 4402 if (t == NULL)
4379 goto out; 4403 goto out;
4380 4404
4381 for (i=0; t->addrconf_vars[i].data; i++) { 4405 for (i = 0; t->addrconf_vars[i].data; i++) {
4382 t->addrconf_vars[i].data += (char*)p - (char*)&ipv6_devconf; 4406 t->addrconf_vars[i].data += (char *)p - (char *)&ipv6_devconf;
4383 t->addrconf_vars[i].extra1 = idev; /* embedded; no ref */ 4407 t->addrconf_vars[i].extra1 = idev; /* embedded; no ref */
4384 t->addrconf_vars[i].extra2 = net; 4408 t->addrconf_vars[i].extra2 = net;
4385 } 4409 }
@@ -4516,14 +4540,12 @@ int register_inet6addr_notifier(struct notifier_block *nb)
4516{ 4540{
4517 return atomic_notifier_chain_register(&inet6addr_chain, nb); 4541 return atomic_notifier_chain_register(&inet6addr_chain, nb);
4518} 4542}
4519
4520EXPORT_SYMBOL(register_inet6addr_notifier); 4543EXPORT_SYMBOL(register_inet6addr_notifier);
4521 4544
4522int unregister_inet6addr_notifier(struct notifier_block *nb) 4545int unregister_inet6addr_notifier(struct notifier_block *nb)
4523{ 4546{
4524 return atomic_notifier_chain_unregister(&inet6addr_chain,nb); 4547 return atomic_notifier_chain_unregister(&inet6addr_chain, nb);
4525} 4548}
4526
4527EXPORT_SYMBOL(unregister_inet6addr_notifier); 4549EXPORT_SYMBOL(unregister_inet6addr_notifier);
4528 4550
4529/* 4551/*
@@ -4532,11 +4554,12 @@ EXPORT_SYMBOL(unregister_inet6addr_notifier);
4532 4554
4533int __init addrconf_init(void) 4555int __init addrconf_init(void)
4534{ 4556{
4535 int err; 4557 int i, err;
4536 4558
4537 if ((err = ipv6_addr_label_init()) < 0) { 4559 err = ipv6_addr_label_init();
4538 printk(KERN_CRIT "IPv6 Addrconf: cannot initialize default policy table: %d.\n", 4560 if (err < 0) {
4539 err); 4561 printk(KERN_CRIT "IPv6 Addrconf:"
4562 " cannot initialize default policy table: %d.\n", err);
4540 return err; 4563 return err;
4541 } 4564 }
4542 4565
@@ -4567,6 +4590,9 @@ int __init addrconf_init(void)
4567 if (err) 4590 if (err)
4568 goto errlo; 4591 goto errlo;
4569 4592
4593 for (i = 0; i < IN6_ADDR_HSIZE; i++)
4594 INIT_HLIST_HEAD(&inet6_addr_lst[i]);
4595
4570 register_netdevice_notifier(&ipv6_dev_notf); 4596 register_netdevice_notifier(&ipv6_dev_notf);
4571 4597
4572 addrconf_verify(0); 4598 addrconf_verify(0);
@@ -4595,7 +4621,6 @@ errlo:
4595 4621
4596void addrconf_cleanup(void) 4622void addrconf_cleanup(void)
4597{ 4623{
4598 struct inet6_ifaddr *ifa;
4599 struct net_device *dev; 4624 struct net_device *dev;
4600 int i; 4625 int i;
4601 4626
@@ -4615,20 +4640,10 @@ void addrconf_cleanup(void)
4615 /* 4640 /*
4616 * Check hash table. 4641 * Check hash table.
4617 */ 4642 */
4618 write_lock_bh(&addrconf_hash_lock); 4643 spin_lock_bh(&addrconf_hash_lock);
4619 for (i=0; i < IN6_ADDR_HSIZE; i++) { 4644 for (i = 0; i < IN6_ADDR_HSIZE; i++)
4620 for (ifa=inet6_addr_lst[i]; ifa; ) { 4645 WARN_ON(!hlist_empty(&inet6_addr_lst[i]));
4621 struct inet6_ifaddr *bifa; 4646 spin_unlock_bh(&addrconf_hash_lock);
4622
4623 bifa = ifa;
4624 ifa = ifa->lst_next;
4625 printk(KERN_DEBUG "bug: IPv6 address leakage detected: ifa=%p\n", bifa);
4626 /* Do not free it; something is wrong.
4627 Now we can investigate it with debugger.
4628 */
4629 }
4630 }
4631 write_unlock_bh(&addrconf_hash_lock);
4632 4647
4633 del_timer(&addr_chk_timer); 4648 del_timer(&addr_chk_timer);
4634 rtnl_unlock(); 4649 rtnl_unlock();
diff --git a/net/ipv6/fib6_rules.c b/net/ipv6/fib6_rules.c
index 551882b9dfd6..5e463c43fcc2 100644
--- a/net/ipv6/fib6_rules.c
+++ b/net/ipv6/fib6_rules.c
@@ -84,18 +84,11 @@ static int fib6_rule_action(struct fib_rule *rule, struct flowi *flp,
84 if ((rule->flags & FIB_RULE_FIND_SADDR) && 84 if ((rule->flags & FIB_RULE_FIND_SADDR) &&
85 r->src.plen && !(flags & RT6_LOOKUP_F_HAS_SADDR)) { 85 r->src.plen && !(flags & RT6_LOOKUP_F_HAS_SADDR)) {
86 struct in6_addr saddr; 86 struct in6_addr saddr;
87 unsigned int srcprefs = 0;
88
89 if (flags & RT6_LOOKUP_F_SRCPREF_TMP)
90 srcprefs |= IPV6_PREFER_SRC_TMP;
91 if (flags & RT6_LOOKUP_F_SRCPREF_PUBLIC)
92 srcprefs |= IPV6_PREFER_SRC_PUBLIC;
93 if (flags & RT6_LOOKUP_F_SRCPREF_COA)
94 srcprefs |= IPV6_PREFER_SRC_COA;
95 87
96 if (ipv6_dev_get_saddr(net, 88 if (ipv6_dev_get_saddr(net,
97 ip6_dst_idev(&rt->u.dst)->dev, 89 ip6_dst_idev(&rt->u.dst)->dev,
98 &flp->fl6_dst, srcprefs, 90 &flp->fl6_dst,
91 rt6_flags2srcprefs(flags),
99 &saddr)) 92 &saddr))
100 goto again; 93 goto again;
101 if (!ipv6_prefix_equal(&saddr, &r->src.addr, 94 if (!ipv6_prefix_equal(&saddr, &r->src.addr,
diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index 52e0f74fdfe0..23e4ac0cc30e 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -1113,6 +1113,9 @@ static int ip6mr_mfc_add(struct net *net, struct mf6cctl *mfc, int mrtsock)
1113 unsigned char ttls[MAXMIFS]; 1113 unsigned char ttls[MAXMIFS];
1114 int i; 1114 int i;
1115 1115
1116 if (mfc->mf6cc_parent >= MAXMIFS)
1117 return -ENFILE;
1118
1116 memset(ttls, 255, MAXMIFS); 1119 memset(ttls, 255, MAXMIFS);
1117 for (i = 0; i < MAXMIFS; i++) { 1120 for (i = 0; i < MAXMIFS; i++) {
1118 if (IF_ISSET(i, &mfc->mf6cc_ifset)) 1121 if (IF_ISSET(i, &mfc->mf6cc_ifset))
diff --git a/net/ipv6/netfilter/ip6t_hbh.c b/net/ipv6/netfilter/ip6t_hbh.c
index cbe8dec9744b..e60677519e40 100644
--- a/net/ipv6/netfilter/ip6t_hbh.c
+++ b/net/ipv6/netfilter/ip6t_hbh.c
@@ -141,11 +141,11 @@ hbh_mt6(const struct sk_buff *skb, const struct xt_match_param *par)
141 } 141 }
142 142
143 /* Step to the next */ 143 /* Step to the next */
144 pr_debug("len%04X \n", optlen); 144 pr_debug("len%04X\n", optlen);
145 145
146 if ((ptr > skb->len - optlen || hdrlen < optlen) && 146 if ((ptr > skb->len - optlen || hdrlen < optlen) &&
147 temp < optinfo->optsnr - 1) { 147 temp < optinfo->optsnr - 1) {
148 pr_debug("new pointer is too large! \n"); 148 pr_debug("new pointer is too large!\n");
149 break; 149 break;
150 } 150 }
151 ptr += optlen; 151 ptr += optlen;
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index b08879e97f22..7fcb0e5d1213 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -819,15 +819,8 @@ struct dst_entry * ip6_route_output(struct net *net, struct sock *sk,
819 819
820 if (!ipv6_addr_any(&fl->fl6_src)) 820 if (!ipv6_addr_any(&fl->fl6_src))
821 flags |= RT6_LOOKUP_F_HAS_SADDR; 821 flags |= RT6_LOOKUP_F_HAS_SADDR;
822 else if (sk) { 822 else if (sk)
823 unsigned int prefs = inet6_sk(sk)->srcprefs; 823 flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
824 if (prefs & IPV6_PREFER_SRC_TMP)
825 flags |= RT6_LOOKUP_F_SRCPREF_TMP;
826 if (prefs & IPV6_PREFER_SRC_PUBLIC)
827 flags |= RT6_LOOKUP_F_SRCPREF_PUBLIC;
828 if (prefs & IPV6_PREFER_SRC_COA)
829 flags |= RT6_LOOKUP_F_SRCPREF_COA;
830 }
831 824
832 return fib6_rule_lookup(net, fl, flags, ip6_pol_route_output); 825 return fib6_rule_lookup(net, fl, flags, ip6_pol_route_output);
833} 826}
@@ -886,7 +879,7 @@ static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)
886 879
887 rt = (struct rt6_info *) dst; 880 rt = (struct rt6_info *) dst;
888 881
889 if (rt && rt->rt6i_node && (rt->rt6i_node->fn_sernum == cookie)) 882 if (rt->rt6i_node && (rt->rt6i_node->fn_sernum == cookie))
890 return dst; 883 return dst;
891 884
892 return NULL; 885 return NULL;
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 6963a6b6763e..9b6dbba80d31 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -1740,8 +1740,11 @@ process:
1740 if (!tcp_prequeue(sk, skb)) 1740 if (!tcp_prequeue(sk, skb))
1741 ret = tcp_v6_do_rcv(sk, skb); 1741 ret = tcp_v6_do_rcv(sk, skb);
1742 } 1742 }
1743 } else 1743 } else if (unlikely(sk_add_backlog(sk, skb))) {
1744 sk_add_backlog(sk, skb); 1744 bh_unlock_sock(sk);
1745 NET_INC_STATS_BH(net, LINUX_MIB_TCPBACKLOGDROP);
1746 goto discard_and_relse;
1747 }
1745 bh_unlock_sock(sk); 1748 bh_unlock_sock(sk);
1746 1749
1747 sock_put(sk); 1750 sock_put(sk);
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index 52b8347ae3b2..3c0c9c755c92 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -583,16 +583,20 @@ static void flush_stack(struct sock **stack, unsigned int count,
583 bh_lock_sock(sk); 583 bh_lock_sock(sk);
584 if (!sock_owned_by_user(sk)) 584 if (!sock_owned_by_user(sk))
585 udpv6_queue_rcv_skb(sk, skb1); 585 udpv6_queue_rcv_skb(sk, skb1);
586 else 586 else if (sk_add_backlog(sk, skb1)) {
587 sk_add_backlog(sk, skb1); 587 kfree_skb(skb1);
588 bh_unlock_sock(sk);
589 goto drop;
590 }
588 bh_unlock_sock(sk); 591 bh_unlock_sock(sk);
589 } else { 592 continue;
590 atomic_inc(&sk->sk_drops);
591 UDP6_INC_STATS_BH(sock_net(sk),
592 UDP_MIB_RCVBUFERRORS, IS_UDPLITE(sk));
593 UDP6_INC_STATS_BH(sock_net(sk),
594 UDP_MIB_INERRORS, IS_UDPLITE(sk));
595 } 593 }
594drop:
595 atomic_inc(&sk->sk_drops);
596 UDP6_INC_STATS_BH(sock_net(sk),
597 UDP_MIB_RCVBUFERRORS, IS_UDPLITE(sk));
598 UDP6_INC_STATS_BH(sock_net(sk),
599 UDP_MIB_INERRORS, IS_UDPLITE(sk));
596 } 600 }
597} 601}
598/* 602/*
@@ -754,8 +758,12 @@ int __udp6_lib_rcv(struct sk_buff *skb, struct udp_table *udptable,
754 bh_lock_sock(sk); 758 bh_lock_sock(sk);
755 if (!sock_owned_by_user(sk)) 759 if (!sock_owned_by_user(sk))
756 udpv6_queue_rcv_skb(sk, skb); 760 udpv6_queue_rcv_skb(sk, skb);
757 else 761 else if (sk_add_backlog(sk, skb)) {
758 sk_add_backlog(sk, skb); 762 atomic_inc(&sk->sk_drops);
763 bh_unlock_sock(sk);
764 sock_put(sk);
765 goto discard;
766 }
759 bh_unlock_sock(sk); 767 bh_unlock_sock(sk);
760 sock_put(sk); 768 sock_put(sk);
761 return 0; 769 return 0;
diff --git a/net/ipv6/xfrm6_policy.c b/net/ipv6/xfrm6_policy.c
index dbdc696f5fc5..ae181651c75a 100644
--- a/net/ipv6/xfrm6_policy.c
+++ b/net/ipv6/xfrm6_policy.c
@@ -116,7 +116,8 @@ static int xfrm6_init_path(struct xfrm_dst *path, struct dst_entry *dst,
116 return 0; 116 return 0;
117} 117}
118 118
119static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev) 119static int xfrm6_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
120 struct flowi *fl)
120{ 121{
121 struct rt6_info *rt = (struct rt6_info*)xdst->route; 122 struct rt6_info *rt = (struct rt6_info*)xdst->route;
122 123
diff --git a/net/irda/ircomm/ircomm_param.c b/net/irda/ircomm/ircomm_param.c
index d57aefd9fe77..0804532d970f 100644
--- a/net/irda/ircomm/ircomm_param.c
+++ b/net/irda/ircomm/ircomm_param.c
@@ -474,7 +474,7 @@ static int ircomm_param_dce(void *instance, irda_param_t *param, int get)
474 /* Check if any of the settings have changed */ 474 /* Check if any of the settings have changed */
475 if (dce & 0x0f) { 475 if (dce & 0x0f) {
476 if (dce & IRCOMM_DELTA_CTS) { 476 if (dce & IRCOMM_DELTA_CTS) {
477 IRDA_DEBUG(2, "%s(), CTS \n", __func__ ); 477 IRDA_DEBUG(2, "%s(), CTS\n", __func__ );
478 } 478 }
479 } 479 }
480 480
diff --git a/net/llc/llc_c_ac.c b/net/llc/llc_c_ac.c
index 019c780512e8..86d6985b9d49 100644
--- a/net/llc/llc_c_ac.c
+++ b/net/llc/llc_c_ac.c
@@ -1437,7 +1437,7 @@ static void llc_process_tmr_ev(struct sock *sk, struct sk_buff *skb)
1437 llc_conn_state_process(sk, skb); 1437 llc_conn_state_process(sk, skb);
1438 else { 1438 else {
1439 llc_set_backlog_type(skb, LLC_EVENT); 1439 llc_set_backlog_type(skb, LLC_EVENT);
1440 sk_add_backlog(sk, skb); 1440 __sk_add_backlog(sk, skb);
1441 } 1441 }
1442 } 1442 }
1443} 1443}
diff --git a/net/llc/llc_conn.c b/net/llc/llc_conn.c
index a8dde9b010da..a12144da7974 100644
--- a/net/llc/llc_conn.c
+++ b/net/llc/llc_conn.c
@@ -827,7 +827,8 @@ void llc_conn_handler(struct llc_sap *sap, struct sk_buff *skb)
827 else { 827 else {
828 dprintk("%s: adding to backlog...\n", __func__); 828 dprintk("%s: adding to backlog...\n", __func__);
829 llc_set_backlog_type(skb, LLC_PACKET); 829 llc_set_backlog_type(skb, LLC_PACKET);
830 sk_add_backlog(sk, skb); 830 if (sk_add_backlog(sk, skb))
831 goto drop_unlock;
831 } 832 }
832out: 833out:
833 bh_unlock_sock(sk); 834 bh_unlock_sock(sk);
diff --git a/net/llc/llc_core.c b/net/llc/llc_core.c
index 78167e81dfeb..2bb0ddff8c0f 100644
--- a/net/llc/llc_core.c
+++ b/net/llc/llc_core.c
@@ -144,12 +144,6 @@ static struct packet_type llc_tr_packet_type __read_mostly = {
144 144
145static int __init llc_init(void) 145static int __init llc_init(void)
146{ 146{
147 struct net_device *dev;
148
149 dev = first_net_device(&init_net);
150 if (dev != NULL)
151 dev = next_net_device(dev);
152
153 dev_add_pack(&llc_packet_type); 147 dev_add_pack(&llc_packet_type);
154 dev_add_pack(&llc_tr_packet_type); 148 dev_add_pack(&llc_tr_packet_type);
155 return 0; 149 return 0;
diff --git a/net/mac80211/debugfs_netdev.c b/net/mac80211/debugfs_netdev.c
index 9affe2cd185f..b4ddb2f83914 100644
--- a/net/mac80211/debugfs_netdev.c
+++ b/net/mac80211/debugfs_netdev.c
@@ -48,20 +48,24 @@ static ssize_t ieee80211_if_write(
48 ssize_t (*write)(struct ieee80211_sub_if_data *, const char *, int)) 48 ssize_t (*write)(struct ieee80211_sub_if_data *, const char *, int))
49{ 49{
50 u8 *buf; 50 u8 *buf;
51 ssize_t ret = -ENODEV; 51 ssize_t ret;
52 52
53 buf = kzalloc(count, GFP_KERNEL); 53 buf = kmalloc(count, GFP_KERNEL);
54 if (!buf) 54 if (!buf)
55 return -ENOMEM; 55 return -ENOMEM;
56 56
57 ret = -EFAULT;
57 if (copy_from_user(buf, userbuf, count)) 58 if (copy_from_user(buf, userbuf, count))
58 return -EFAULT; 59 goto freebuf;
59 60
61 ret = -ENODEV;
60 rtnl_lock(); 62 rtnl_lock();
61 if (sdata->dev->reg_state == NETREG_REGISTERED) 63 if (sdata->dev->reg_state == NETREG_REGISTERED)
62 ret = (*write)(sdata, buf, count); 64 ret = (*write)(sdata, buf, count);
63 rtnl_unlock(); 65 rtnl_unlock();
64 66
67freebuf:
68 kfree(buf);
65 return ret; 69 return ret;
66} 70}
67 71
diff --git a/net/mac80211/mesh_plink.c b/net/mac80211/mesh_plink.c
index bc4e20e57ff5..1a29c4a8139e 100644
--- a/net/mac80211/mesh_plink.c
+++ b/net/mac80211/mesh_plink.c
@@ -744,7 +744,7 @@ void mesh_rx_plink_frame(struct ieee80211_sub_if_data *sdata, struct ieee80211_m
744 break; 744 break;
745 default: 745 default:
746 /* should not get here, PLINK_BLOCKED is dealt with at the 746 /* should not get here, PLINK_BLOCKED is dealt with at the
747 * beggining of the function 747 * beginning of the function
748 */ 748 */
749 spin_unlock_bh(&sta->lock); 749 spin_unlock_bh(&sta->lock);
750 break; 750 break;
diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
index 65eafda5738a..34e0650e9ef8 100644
--- a/net/mac80211/mlme.c
+++ b/net/mac80211/mlme.c
@@ -436,10 +436,12 @@ static void ieee80211_enable_ps(struct ieee80211_local *local,
436 if (local->hw.flags & IEEE80211_HW_PS_NULLFUNC_STACK) 436 if (local->hw.flags & IEEE80211_HW_PS_NULLFUNC_STACK)
437 ieee80211_send_nullfunc(local, sdata, 1); 437 ieee80211_send_nullfunc(local, sdata, 1);
438 438
439 if (!(local->hw.flags & IEEE80211_HW_REPORTS_TX_ACK_STATUS)) { 439 if ((local->hw.flags & IEEE80211_HW_PS_NULLFUNC_STACK) &&
440 conf->flags |= IEEE80211_CONF_PS; 440 (local->hw.flags & IEEE80211_HW_REPORTS_TX_ACK_STATUS))
441 ieee80211_hw_config(local, IEEE80211_CONF_CHANGE_PS); 441 return;
442 } 442
443 conf->flags |= IEEE80211_CONF_PS;
444 ieee80211_hw_config(local, IEEE80211_CONF_CHANGE_PS);
443 } 445 }
444} 446}
445 447
@@ -558,7 +560,8 @@ void ieee80211_dynamic_ps_enable_work(struct work_struct *work)
558 (!(ifmgd->flags & IEEE80211_STA_NULLFUNC_ACKED))) 560 (!(ifmgd->flags & IEEE80211_STA_NULLFUNC_ACKED)))
559 ieee80211_send_nullfunc(local, sdata, 1); 561 ieee80211_send_nullfunc(local, sdata, 1);
560 562
561 if (!(local->hw.flags & IEEE80211_HW_REPORTS_TX_ACK_STATUS) || 563 if (!((local->hw.flags & IEEE80211_HW_REPORTS_TX_ACK_STATUS) &&
564 (local->hw.flags & IEEE80211_HW_PS_NULLFUNC_STACK)) ||
562 (ifmgd->flags & IEEE80211_STA_NULLFUNC_ACKED)) { 565 (ifmgd->flags & IEEE80211_STA_NULLFUNC_ACKED)) {
563 ifmgd->flags &= ~IEEE80211_STA_NULLFUNC_ACKED; 566 ifmgd->flags &= ~IEEE80211_STA_NULLFUNC_ACKED;
564 local->hw.conf.flags |= IEEE80211_CONF_PS; 567 local->hw.conf.flags |= IEEE80211_CONF_PS;
diff --git a/net/mac80211/sta_info.c b/net/mac80211/sta_info.c
index 211c475f73c6..56422d894351 100644
--- a/net/mac80211/sta_info.c
+++ b/net/mac80211/sta_info.c
@@ -434,6 +434,7 @@ int sta_info_insert_rcu(struct sta_info *sta) __acquires(RCU)
434 /* check if STA exists already */ 434 /* check if STA exists already */
435 if (sta_info_get_bss(sdata, sta->sta.addr)) { 435 if (sta_info_get_bss(sdata, sta->sta.addr)) {
436 spin_unlock_irqrestore(&local->sta_lock, flags); 436 spin_unlock_irqrestore(&local->sta_lock, flags);
437 mutex_unlock(&local->sta_mtx);
437 rcu_read_lock(); 438 rcu_read_lock();
438 err = -EEXIST; 439 err = -EEXIST;
439 goto out_free; 440 goto out_free;
diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c
index 2b2af631d2b8..569410a85953 100644
--- a/net/netfilter/nf_conntrack_netlink.c
+++ b/net/netfilter/nf_conntrack_netlink.c
@@ -582,7 +582,9 @@ nla_put_failure:
582nlmsg_failure: 582nlmsg_failure:
583 kfree_skb(skb); 583 kfree_skb(skb);
584errout: 584errout:
585 nfnetlink_set_err(net, 0, group, -ENOBUFS); 585 if (nfnetlink_set_err(net, 0, group, -ENOBUFS) > 0)
586 return -ENOBUFS;
587
586 return 0; 588 return 0;
587} 589}
588#endif /* CONFIG_NF_CONNTRACK_EVENTS */ 590#endif /* CONFIG_NF_CONNTRACK_EVENTS */
diff --git a/net/netfilter/nf_conntrack_sip.c b/net/netfilter/nf_conntrack_sip.c
index 8dd75d90efc0..c6cd1b84eddd 100644
--- a/net/netfilter/nf_conntrack_sip.c
+++ b/net/netfilter/nf_conntrack_sip.c
@@ -284,7 +284,7 @@ EXPORT_SYMBOL_GPL(ct_sip_parse_request);
284 * tabs, spaces and continuation lines, which are treated as a single whitespace 284 * tabs, spaces and continuation lines, which are treated as a single whitespace
285 * character. 285 * character.
286 * 286 *
287 * Some headers may appear multiple times. A comma seperated list of values is 287 * Some headers may appear multiple times. A comma separated list of values is
288 * equivalent to multiple headers. 288 * equivalent to multiple headers.
289 */ 289 */
290static const struct sip_header ct_sip_hdrs[] = { 290static const struct sip_header ct_sip_hdrs[] = {
@@ -421,7 +421,7 @@ int ct_sip_get_header(const struct nf_conn *ct, const char *dptr,
421} 421}
422EXPORT_SYMBOL_GPL(ct_sip_get_header); 422EXPORT_SYMBOL_GPL(ct_sip_get_header);
423 423
424/* Get next header field in a list of comma seperated values */ 424/* Get next header field in a list of comma separated values */
425static int ct_sip_next_header(const struct nf_conn *ct, const char *dptr, 425static int ct_sip_next_header(const struct nf_conn *ct, const char *dptr,
426 unsigned int dataoff, unsigned int datalen, 426 unsigned int dataoff, unsigned int datalen,
427 enum sip_header_types type, 427 enum sip_header_types type,
diff --git a/net/netfilter/nfnetlink.c b/net/netfilter/nfnetlink.c
index 8eb0cc23ada3..6afa3d52ea5f 100644
--- a/net/netfilter/nfnetlink.c
+++ b/net/netfilter/nfnetlink.c
@@ -113,9 +113,9 @@ int nfnetlink_send(struct sk_buff *skb, struct net *net, u32 pid,
113} 113}
114EXPORT_SYMBOL_GPL(nfnetlink_send); 114EXPORT_SYMBOL_GPL(nfnetlink_send);
115 115
116void nfnetlink_set_err(struct net *net, u32 pid, u32 group, int error) 116int nfnetlink_set_err(struct net *net, u32 pid, u32 group, int error)
117{ 117{
118 netlink_set_err(net->nfnl, pid, group, error); 118 return netlink_set_err(net->nfnl, pid, group, error);
119} 119}
120EXPORT_SYMBOL_GPL(nfnetlink_set_err); 120EXPORT_SYMBOL_GPL(nfnetlink_set_err);
121 121
diff --git a/net/netfilter/xt_hashlimit.c b/net/netfilter/xt_hashlimit.c
index d952806b6469..9e9c48963942 100644
--- a/net/netfilter/xt_hashlimit.c
+++ b/net/netfilter/xt_hashlimit.c
@@ -1,6 +1,6 @@
1/* 1/*
2 * xt_hashlimit - Netfilter module to limit the number of packets per time 2 * xt_hashlimit - Netfilter module to limit the number of packets per time
3 * seperately for each hashbucket (sourceip/sourceport/dstip/dstport) 3 * separately for each hashbucket (sourceip/sourceport/dstip/dstport)
4 * 4 *
5 * (C) 2003-2004 by Harald Welte <laforge@netfilter.org> 5 * (C) 2003-2004 by Harald Welte <laforge@netfilter.org>
6 * Copyright © CC Computer Consultants GmbH, 2007 - 2008 6 * Copyright © CC Computer Consultants GmbH, 2007 - 2008
diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index 320d0423a240..274d977166b7 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c
@@ -545,7 +545,7 @@ static int netlink_autobind(struct socket *sock)
545 struct hlist_head *head; 545 struct hlist_head *head;
546 struct sock *osk; 546 struct sock *osk;
547 struct hlist_node *node; 547 struct hlist_node *node;
548 s32 pid = current->tgid; 548 s32 pid = task_tgid_vnr(current);
549 int err; 549 int err;
550 static s32 rover = -4097; 550 static s32 rover = -4097;
551 551
@@ -1093,6 +1093,7 @@ static inline int do_one_set_err(struct sock *sk,
1093 struct netlink_set_err_data *p) 1093 struct netlink_set_err_data *p)
1094{ 1094{
1095 struct netlink_sock *nlk = nlk_sk(sk); 1095 struct netlink_sock *nlk = nlk_sk(sk);
1096 int ret = 0;
1096 1097
1097 if (sk == p->exclude_sk) 1098 if (sk == p->exclude_sk)
1098 goto out; 1099 goto out;
@@ -1104,10 +1105,15 @@ static inline int do_one_set_err(struct sock *sk,
1104 !test_bit(p->group - 1, nlk->groups)) 1105 !test_bit(p->group - 1, nlk->groups))
1105 goto out; 1106 goto out;
1106 1107
1108 if (p->code == ENOBUFS && nlk->flags & NETLINK_RECV_NO_ENOBUFS) {
1109 ret = 1;
1110 goto out;
1111 }
1112
1107 sk->sk_err = p->code; 1113 sk->sk_err = p->code;
1108 sk->sk_error_report(sk); 1114 sk->sk_error_report(sk);
1109out: 1115out:
1110 return 0; 1116 return ret;
1111} 1117}
1112 1118
1113/** 1119/**
@@ -1116,12 +1122,16 @@ out:
1116 * @pid: the PID of a process that we want to skip (if any) 1122 * @pid: the PID of a process that we want to skip (if any)
1117 * @groups: the broadcast group that will notice the error 1123 * @groups: the broadcast group that will notice the error
1118 * @code: error code, must be negative (as usual in kernelspace) 1124 * @code: error code, must be negative (as usual in kernelspace)
1125 *
1126 * This function returns the number of broadcast listeners that have set the
1127 * NETLINK_RECV_NO_ENOBUFS socket option.
1119 */ 1128 */
1120void netlink_set_err(struct sock *ssk, u32 pid, u32 group, int code) 1129int netlink_set_err(struct sock *ssk, u32 pid, u32 group, int code)
1121{ 1130{
1122 struct netlink_set_err_data info; 1131 struct netlink_set_err_data info;
1123 struct hlist_node *node; 1132 struct hlist_node *node;
1124 struct sock *sk; 1133 struct sock *sk;
1134 int ret = 0;
1125 1135
1126 info.exclude_sk = ssk; 1136 info.exclude_sk = ssk;
1127 info.pid = pid; 1137 info.pid = pid;
@@ -1132,9 +1142,10 @@ void netlink_set_err(struct sock *ssk, u32 pid, u32 group, int code)
1132 read_lock(&nl_table_lock); 1142 read_lock(&nl_table_lock);
1133 1143
1134 sk_for_each_bound(sk, node, &nl_table[ssk->sk_protocol].mc_list) 1144 sk_for_each_bound(sk, node, &nl_table[ssk->sk_protocol].mc_list)
1135 do_one_set_err(sk, &info); 1145 ret += do_one_set_err(sk, &info);
1136 1146
1137 read_unlock(&nl_table_lock); 1147 read_unlock(&nl_table_lock);
1148 return ret;
1138} 1149}
1139EXPORT_SYMBOL(netlink_set_err); 1150EXPORT_SYMBOL(netlink_set_err);
1140 1151
diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 031a5e6fb4aa..1612d417d10c 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -1688,6 +1688,8 @@ static int packet_dev_mc(struct net_device *dev, struct packet_mclist *i,
1688{ 1688{
1689 switch (i->type) { 1689 switch (i->type) {
1690 case PACKET_MR_MULTICAST: 1690 case PACKET_MR_MULTICAST:
1691 if (i->alen != dev->addr_len)
1692 return -EINVAL;
1691 if (what > 0) 1693 if (what > 0)
1692 return dev_mc_add(dev, i->addr, i->alen, 0); 1694 return dev_mc_add(dev, i->addr, i->alen, 0);
1693 else 1695 else
@@ -1700,6 +1702,8 @@ static int packet_dev_mc(struct net_device *dev, struct packet_mclist *i,
1700 return dev_set_allmulti(dev, what); 1702 return dev_set_allmulti(dev, what);
1701 break; 1703 break;
1702 case PACKET_MR_UNICAST: 1704 case PACKET_MR_UNICAST:
1705 if (i->alen != dev->addr_len)
1706 return -EINVAL;
1703 if (what > 0) 1707 if (what > 0)
1704 return dev_unicast_add(dev, i->addr); 1708 return dev_unicast_add(dev, i->addr);
1705 else 1709 else
@@ -1734,7 +1738,7 @@ static int packet_mc_add(struct sock *sk, struct packet_mreq_max *mreq)
1734 goto done; 1738 goto done;
1735 1739
1736 err = -EINVAL; 1740 err = -EINVAL;
1737 if (mreq->mr_alen != dev->addr_len) 1741 if (mreq->mr_alen > dev->addr_len)
1738 goto done; 1742 goto done;
1739 1743
1740 err = -ENOBUFS; 1744 err = -ENOBUFS;
diff --git a/net/phonet/pn_dev.c b/net/phonet/pn_dev.c
index c597cc53a6fb..5c6ae0c701c0 100644
--- a/net/phonet/pn_dev.c
+++ b/net/phonet/pn_dev.c
@@ -107,8 +107,7 @@ static void phonet_device_destroy(struct net_device *dev)
107 if (pnd) { 107 if (pnd) {
108 u8 addr; 108 u8 addr;
109 109
110 for (addr = find_first_bit(pnd->addrs, 64); addr < 64; 110 for_each_set_bit(addr, pnd->addrs, 64)
111 addr = find_next_bit(pnd->addrs, 64, 1+addr))
112 phonet_address_notify(RTM_DELADDR, dev, addr); 111 phonet_address_notify(RTM_DELADDR, dev, addr);
113 kfree(pnd); 112 kfree(pnd);
114 } 113 }
diff --git a/net/phonet/pn_netlink.c b/net/phonet/pn_netlink.c
index 2e6c7eb8e76a..fe2e7088ee07 100644
--- a/net/phonet/pn_netlink.c
+++ b/net/phonet/pn_netlink.c
@@ -141,8 +141,7 @@ static int getaddr_dumpit(struct sk_buff *skb, struct netlink_callback *cb)
141 continue; 141 continue;
142 142
143 addr_idx = 0; 143 addr_idx = 0;
144 for (addr = find_first_bit(pnd->addrs, 64); addr < 64; 144 for_each_set_bit(addr, pnd->addrs, 64) {
145 addr = find_next_bit(pnd->addrs, 64, 1+addr)) {
146 if (addr_idx++ < addr_start_idx) 145 if (addr_idx++ < addr_start_idx)
147 continue; 146 continue;
148 147
diff --git a/net/rds/af_rds.c b/net/rds/af_rds.c
index 853c52be781f..c2e45e8efa20 100644
--- a/net/rds/af_rds.c
+++ b/net/rds/af_rds.c
@@ -159,7 +159,8 @@ static unsigned int rds_poll(struct file *file, struct socket *sock,
159 159
160 poll_wait(file, sk->sk_sleep, wait); 160 poll_wait(file, sk->sk_sleep, wait);
161 161
162 poll_wait(file, &rds_poll_waitq, wait); 162 if (rs->rs_seen_congestion)
163 poll_wait(file, &rds_poll_waitq, wait);
163 164
164 read_lock_irqsave(&rs->rs_recv_lock, flags); 165 read_lock_irqsave(&rs->rs_recv_lock, flags);
165 if (!rs->rs_cong_monitor) { 166 if (!rs->rs_cong_monitor) {
@@ -181,6 +182,10 @@ static unsigned int rds_poll(struct file *file, struct socket *sock,
181 mask |= (POLLOUT | POLLWRNORM); 182 mask |= (POLLOUT | POLLWRNORM);
182 read_unlock_irqrestore(&rs->rs_recv_lock, flags); 183 read_unlock_irqrestore(&rs->rs_recv_lock, flags);
183 184
185 /* clear state any time we wake a seen-congested socket */
186 if (mask)
187 rs->rs_seen_congestion = 0;
188
184 return mask; 189 return mask;
185} 190}
186 191
@@ -446,7 +451,6 @@ static void rds_sock_inc_info(struct socket *sock, unsigned int len,
446 struct rds_info_lengths *lens) 451 struct rds_info_lengths *lens)
447{ 452{
448 struct rds_sock *rs; 453 struct rds_sock *rs;
449 struct sock *sk;
450 struct rds_incoming *inc; 454 struct rds_incoming *inc;
451 unsigned long flags; 455 unsigned long flags;
452 unsigned int total = 0; 456 unsigned int total = 0;
@@ -456,7 +460,6 @@ static void rds_sock_inc_info(struct socket *sock, unsigned int len,
456 spin_lock_irqsave(&rds_sock_lock, flags); 460 spin_lock_irqsave(&rds_sock_lock, flags);
457 461
458 list_for_each_entry(rs, &rds_sock_list, rs_item) { 462 list_for_each_entry(rs, &rds_sock_list, rs_item) {
459 sk = rds_rs_to_sk(rs);
460 read_lock(&rs->rs_recv_lock); 463 read_lock(&rs->rs_recv_lock);
461 464
462 /* XXX too lazy to maintain counts.. */ 465 /* XXX too lazy to maintain counts.. */
diff --git a/net/rds/cong.c b/net/rds/cong.c
index 6d06cac2649c..dd2711df640b 100644
--- a/net/rds/cong.c
+++ b/net/rds/cong.c
@@ -218,8 +218,6 @@ void rds_cong_queue_updates(struct rds_cong_map *map)
218 spin_lock_irqsave(&rds_cong_lock, flags); 218 spin_lock_irqsave(&rds_cong_lock, flags);
219 219
220 list_for_each_entry(conn, &map->m_conn_list, c_map_item) { 220 list_for_each_entry(conn, &map->m_conn_list, c_map_item) {
221 if (conn->c_loopback)
222 continue;
223 if (!test_and_set_bit(0, &conn->c_map_queued)) { 221 if (!test_and_set_bit(0, &conn->c_map_queued)) {
224 rds_stats_inc(s_cong_update_queued); 222 rds_stats_inc(s_cong_update_queued);
225 queue_delayed_work(rds_wq, &conn->c_send_w, 0); 223 queue_delayed_work(rds_wq, &conn->c_send_w, 0);
diff --git a/net/rds/ib_cm.c b/net/rds/ib_cm.c
index 647cb8ffc39b..e1f124bf03bb 100644
--- a/net/rds/ib_cm.c
+++ b/net/rds/ib_cm.c
@@ -203,9 +203,10 @@ static void rds_ib_qp_event_handler(struct ib_event *event, void *data)
203 rdma_notify(ic->i_cm_id, IB_EVENT_COMM_EST); 203 rdma_notify(ic->i_cm_id, IB_EVENT_COMM_EST);
204 break; 204 break;
205 default: 205 default:
206 rds_ib_conn_error(conn, "RDS/IB: Fatal QP Event %u " 206 rdsdebug("Fatal QP Event %u "
207 "- connection %pI4->%pI4, reconnecting\n", 207 "- connection %pI4->%pI4, reconnecting\n",
208 event->event, &conn->c_laddr, &conn->c_faddr); 208 event->event, &conn->c_laddr, &conn->c_faddr);
209 rds_conn_drop(conn);
209 break; 210 break;
210 } 211 }
211} 212}
diff --git a/net/rds/ib_rdma.c b/net/rds/ib_rdma.c
index 4b0da865a72c..cfb1d904ed00 100644
--- a/net/rds/ib_rdma.c
+++ b/net/rds/ib_rdma.c
@@ -234,8 +234,8 @@ void rds_ib_destroy_mr_pool(struct rds_ib_mr_pool *pool)
234{ 234{
235 flush_workqueue(rds_wq); 235 flush_workqueue(rds_wq);
236 rds_ib_flush_mr_pool(pool, 1); 236 rds_ib_flush_mr_pool(pool, 1);
237 BUG_ON(atomic_read(&pool->item_count)); 237 WARN_ON(atomic_read(&pool->item_count));
238 BUG_ON(atomic_read(&pool->free_pinned)); 238 WARN_ON(atomic_read(&pool->free_pinned));
239 kfree(pool); 239 kfree(pool);
240} 240}
241 241
@@ -440,6 +440,7 @@ static void __rds_ib_teardown_mr(struct rds_ib_mr *ibmr)
440 440
441 /* FIXME we need a way to tell a r/w MR 441 /* FIXME we need a way to tell a r/w MR
442 * from a r/o MR */ 442 * from a r/o MR */
443 BUG_ON(in_interrupt());
443 set_page_dirty(page); 444 set_page_dirty(page);
444 put_page(page); 445 put_page(page);
445 } 446 }
diff --git a/net/rds/ib_recv.c b/net/rds/ib_recv.c
index 04dc0d3f3c95..c338881eca71 100644
--- a/net/rds/ib_recv.c
+++ b/net/rds/ib_recv.c
@@ -468,8 +468,8 @@ static void rds_ib_send_ack(struct rds_ib_connection *ic, unsigned int adv_credi
468 set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags); 468 set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
469 469
470 rds_ib_stats_inc(s_ib_ack_send_failure); 470 rds_ib_stats_inc(s_ib_ack_send_failure);
471 /* Need to finesse this later. */ 471
472 BUG(); 472 rds_ib_conn_error(ic->conn, "sending ack failed\n");
473 } else 473 } else
474 rds_ib_stats_inc(s_ib_ack_sent); 474 rds_ib_stats_inc(s_ib_ack_sent);
475} 475}
diff --git a/net/rds/ib_send.c b/net/rds/ib_send.c
index a10fab6886d1..17fa80803ab0 100644
--- a/net/rds/ib_send.c
+++ b/net/rds/ib_send.c
@@ -243,8 +243,12 @@ void rds_ib_send_cq_comp_handler(struct ib_cq *cq, void *context)
243 struct rds_message *rm; 243 struct rds_message *rm;
244 244
245 rm = rds_send_get_message(conn, send->s_op); 245 rm = rds_send_get_message(conn, send->s_op);
246 if (rm) 246 if (rm) {
247 if (rm->m_rdma_op)
248 rds_ib_send_unmap_rdma(ic, rm->m_rdma_op);
247 rds_ib_send_rdma_complete(rm, wc.status); 249 rds_ib_send_rdma_complete(rm, wc.status);
250 rds_message_put(rm);
251 }
248 } 252 }
249 253
250 oldest = (oldest + 1) % ic->i_send_ring.w_nr; 254 oldest = (oldest + 1) % ic->i_send_ring.w_nr;
@@ -482,6 +486,13 @@ int rds_ib_xmit(struct rds_connection *conn, struct rds_message *rm,
482 BUG_ON(off % RDS_FRAG_SIZE); 486 BUG_ON(off % RDS_FRAG_SIZE);
483 BUG_ON(hdr_off != 0 && hdr_off != sizeof(struct rds_header)); 487 BUG_ON(hdr_off != 0 && hdr_off != sizeof(struct rds_header));
484 488
489 /* Do not send cong updates to IB loopback */
490 if (conn->c_loopback
491 && rm->m_inc.i_hdr.h_flags & RDS_FLAG_CONG_BITMAP) {
492 rds_cong_map_updated(conn->c_fcong, ~(u64) 0);
493 return sizeof(struct rds_header) + RDS_CONG_MAP_BYTES;
494 }
495
485 /* FIXME we may overallocate here */ 496 /* FIXME we may overallocate here */
486 if (be32_to_cpu(rm->m_inc.i_hdr.h_len) == 0) 497 if (be32_to_cpu(rm->m_inc.i_hdr.h_len) == 0)
487 i = 1; 498 i = 1;
@@ -574,8 +585,7 @@ int rds_ib_xmit(struct rds_connection *conn, struct rds_message *rm,
574 rds_ib_send_grab_credits(ic, 0, &posted, 1, RDS_MAX_ADV_CREDIT - adv_credits); 585 rds_ib_send_grab_credits(ic, 0, &posted, 1, RDS_MAX_ADV_CREDIT - adv_credits);
575 adv_credits += posted; 586 adv_credits += posted;
576 BUG_ON(adv_credits > 255); 587 BUG_ON(adv_credits > 255);
577 } else if (ic->i_rm != rm) 588 }
578 BUG();
579 589
580 send = &ic->i_sends[pos]; 590 send = &ic->i_sends[pos];
581 first = send; 591 first = send;
@@ -714,8 +724,8 @@ add_header:
714 ic->i_rm = prev->s_rm; 724 ic->i_rm = prev->s_rm;
715 prev->s_rm = NULL; 725 prev->s_rm = NULL;
716 } 726 }
717 /* Finesse this later */ 727
718 BUG(); 728 rds_ib_conn_error(ic->conn, "ib_post_send failed\n");
719 goto out; 729 goto out;
720 } 730 }
721 731
diff --git a/net/rds/iw_cm.c b/net/rds/iw_cm.c
index 394cf6b4d0aa..6bc638fd252c 100644
--- a/net/rds/iw_cm.c
+++ b/net/rds/iw_cm.c
@@ -156,9 +156,11 @@ static void rds_iw_qp_event_handler(struct ib_event *event, void *data)
156 case IB_EVENT_QP_REQ_ERR: 156 case IB_EVENT_QP_REQ_ERR:
157 case IB_EVENT_QP_FATAL: 157 case IB_EVENT_QP_FATAL:
158 default: 158 default:
159 rds_iw_conn_error(conn, "RDS/IW: Fatal QP Event %u - connection %pI4->%pI4...reconnecting\n", 159 rdsdebug("Fatal QP Event %u "
160 "- connection %pI4->%pI4, reconnecting\n",
160 event->event, &conn->c_laddr, 161 event->event, &conn->c_laddr,
161 &conn->c_faddr); 162 &conn->c_faddr);
163 rds_conn_drop(conn);
162 break; 164 break;
163 } 165 }
164} 166}
diff --git a/net/rds/iw_recv.c b/net/rds/iw_recv.c
index 54af7d6b92da..337e4e5025e2 100644
--- a/net/rds/iw_recv.c
+++ b/net/rds/iw_recv.c
@@ -468,8 +468,8 @@ static void rds_iw_send_ack(struct rds_iw_connection *ic, unsigned int adv_credi
468 set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags); 468 set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
469 469
470 rds_iw_stats_inc(s_iw_ack_send_failure); 470 rds_iw_stats_inc(s_iw_ack_send_failure);
471 /* Need to finesse this later. */ 471
472 BUG(); 472 rds_iw_conn_error(ic->conn, "sending ack failed\n");
473 } else 473 } else
474 rds_iw_stats_inc(s_iw_ack_sent); 474 rds_iw_stats_inc(s_iw_ack_sent);
475} 475}
diff --git a/net/rds/iw_send.c b/net/rds/iw_send.c
index 1379e9d66a78..52182ff7519e 100644
--- a/net/rds/iw_send.c
+++ b/net/rds/iw_send.c
@@ -616,8 +616,7 @@ int rds_iw_xmit(struct rds_connection *conn, struct rds_message *rm,
616 rds_iw_send_grab_credits(ic, 0, &posted, 1, RDS_MAX_ADV_CREDIT - adv_credits); 616 rds_iw_send_grab_credits(ic, 0, &posted, 1, RDS_MAX_ADV_CREDIT - adv_credits);
617 adv_credits += posted; 617 adv_credits += posted;
618 BUG_ON(adv_credits > 255); 618 BUG_ON(adv_credits > 255);
619 } else if (ic->i_rm != rm) 619 }
620 BUG();
621 620
622 send = &ic->i_sends[pos]; 621 send = &ic->i_sends[pos];
623 first = send; 622 first = send;
diff --git a/net/rds/loop.c b/net/rds/loop.c
index 4a61997f554d..93a45f1ce61f 100644
--- a/net/rds/loop.c
+++ b/net/rds/loop.c
@@ -80,16 +80,9 @@ static int rds_loop_xmit_cong_map(struct rds_connection *conn,
80 struct rds_cong_map *map, 80 struct rds_cong_map *map,
81 unsigned long offset) 81 unsigned long offset)
82{ 82{
83 unsigned long i;
84
85 BUG_ON(offset); 83 BUG_ON(offset);
86 BUG_ON(map != conn->c_lcong); 84 BUG_ON(map != conn->c_lcong);
87 85
88 for (i = 0; i < RDS_CONG_MAP_PAGES; i++) {
89 memcpy((void *)conn->c_fcong->m_page_addrs[i],
90 (void *)map->m_page_addrs[i], PAGE_SIZE);
91 }
92
93 rds_cong_map_updated(conn->c_fcong, ~(u64) 0); 86 rds_cong_map_updated(conn->c_fcong, ~(u64) 0);
94 87
95 return sizeof(struct rds_header) + RDS_CONG_MAP_BYTES; 88 return sizeof(struct rds_header) + RDS_CONG_MAP_BYTES;
diff --git a/net/rds/rdma.c b/net/rds/rdma.c
index 4c64daa1f5d5..61b359d9dffd 100644
--- a/net/rds/rdma.c
+++ b/net/rds/rdma.c
@@ -438,8 +438,10 @@ void rds_rdma_free_op(struct rds_rdma_op *ro)
438 /* Mark page dirty if it was possibly modified, which 438 /* Mark page dirty if it was possibly modified, which
439 * is the case for a RDMA_READ which copies from remote 439 * is the case for a RDMA_READ which copies from remote
440 * to local memory */ 440 * to local memory */
441 if (!ro->r_write) 441 if (!ro->r_write) {
442 BUG_ON(in_interrupt());
442 set_page_dirty(page); 443 set_page_dirty(page);
444 }
443 put_page(page); 445 put_page(page);
444 } 446 }
445 447
diff --git a/net/rds/rdma_transport.c b/net/rds/rdma_transport.c
index 9ece910ea394..5ea82fc47c3e 100644
--- a/net/rds/rdma_transport.c
+++ b/net/rds/rdma_transport.c
@@ -101,7 +101,7 @@ int rds_rdma_cm_event_handler(struct rdma_cm_id *cm_id,
101 break; 101 break;
102 102
103 case RDMA_CM_EVENT_DISCONNECTED: 103 case RDMA_CM_EVENT_DISCONNECTED:
104 printk(KERN_WARNING "RDS/RDMA: DISCONNECT event - dropping connection " 104 rdsdebug("DISCONNECT event - dropping connection "
105 "%pI4->%pI4\n", &conn->c_laddr, 105 "%pI4->%pI4\n", &conn->c_laddr,
106 &conn->c_faddr); 106 &conn->c_faddr);
107 rds_conn_drop(conn); 107 rds_conn_drop(conn);
@@ -109,8 +109,7 @@ int rds_rdma_cm_event_handler(struct rdma_cm_id *cm_id,
109 109
110 default: 110 default:
111 /* things like device disconnect? */ 111 /* things like device disconnect? */
112 printk(KERN_ERR "unknown event %u\n", event->event); 112 printk(KERN_ERR "RDS: unknown event %u!\n", event->event);
113 BUG();
114 break; 113 break;
115 } 114 }
116 115
diff --git a/net/rds/rds.h b/net/rds/rds.h
index 85d6f897ecc7..4bec6e2ed495 100644
--- a/net/rds/rds.h
+++ b/net/rds/rds.h
@@ -388,6 +388,8 @@ struct rds_sock {
388 388
389 /* flag indicating we were congested or not */ 389 /* flag indicating we were congested or not */
390 int rs_congested; 390 int rs_congested;
391 /* seen congestion (ENOBUFS) when sending? */
392 int rs_seen_congestion;
391 393
392 /* rs_lock protects all these adjacent members before the newline */ 394 /* rs_lock protects all these adjacent members before the newline */
393 spinlock_t rs_lock; 395 spinlock_t rs_lock;
diff --git a/net/rds/send.c b/net/rds/send.c
index b2fccfc20769..4629a0b63bbd 100644
--- a/net/rds/send.c
+++ b/net/rds/send.c
@@ -507,12 +507,13 @@ EXPORT_SYMBOL_GPL(rds_send_get_message);
507 */ 507 */
508void rds_send_remove_from_sock(struct list_head *messages, int status) 508void rds_send_remove_from_sock(struct list_head *messages, int status)
509{ 509{
510 unsigned long flags = 0; /* silence gcc :P */ 510 unsigned long flags;
511 struct rds_sock *rs = NULL; 511 struct rds_sock *rs = NULL;
512 struct rds_message *rm; 512 struct rds_message *rm;
513 513
514 local_irq_save(flags);
515 while (!list_empty(messages)) { 514 while (!list_empty(messages)) {
515 int was_on_sock = 0;
516
516 rm = list_entry(messages->next, struct rds_message, 517 rm = list_entry(messages->next, struct rds_message,
517 m_conn_item); 518 m_conn_item);
518 list_del_init(&rm->m_conn_item); 519 list_del_init(&rm->m_conn_item);
@@ -527,20 +528,19 @@ void rds_send_remove_from_sock(struct list_head *messages, int status)
527 * while we're messing with it. It does not prevent the 528 * while we're messing with it. It does not prevent the
528 * message from being removed from the socket, though. 529 * message from being removed from the socket, though.
529 */ 530 */
530 spin_lock(&rm->m_rs_lock); 531 spin_lock_irqsave(&rm->m_rs_lock, flags);
531 if (!test_bit(RDS_MSG_ON_SOCK, &rm->m_flags)) 532 if (!test_bit(RDS_MSG_ON_SOCK, &rm->m_flags))
532 goto unlock_and_drop; 533 goto unlock_and_drop;
533 534
534 if (rs != rm->m_rs) { 535 if (rs != rm->m_rs) {
535 if (rs) { 536 if (rs) {
536 spin_unlock(&rs->rs_lock);
537 rds_wake_sk_sleep(rs); 537 rds_wake_sk_sleep(rs);
538 sock_put(rds_rs_to_sk(rs)); 538 sock_put(rds_rs_to_sk(rs));
539 } 539 }
540 rs = rm->m_rs; 540 rs = rm->m_rs;
541 spin_lock(&rs->rs_lock);
542 sock_hold(rds_rs_to_sk(rs)); 541 sock_hold(rds_rs_to_sk(rs));
543 } 542 }
543 spin_lock(&rs->rs_lock);
544 544
545 if (test_and_clear_bit(RDS_MSG_ON_SOCK, &rm->m_flags)) { 545 if (test_and_clear_bit(RDS_MSG_ON_SOCK, &rm->m_flags)) {
546 struct rds_rdma_op *ro = rm->m_rdma_op; 546 struct rds_rdma_op *ro = rm->m_rdma_op;
@@ -557,21 +557,22 @@ void rds_send_remove_from_sock(struct list_head *messages, int status)
557 notifier->n_status = status; 557 notifier->n_status = status;
558 rm->m_rdma_op->r_notifier = NULL; 558 rm->m_rdma_op->r_notifier = NULL;
559 } 559 }
560 rds_message_put(rm); 560 was_on_sock = 1;
561 rm->m_rs = NULL; 561 rm->m_rs = NULL;
562 } 562 }
563 spin_unlock(&rs->rs_lock);
563 564
564unlock_and_drop: 565unlock_and_drop:
565 spin_unlock(&rm->m_rs_lock); 566 spin_unlock_irqrestore(&rm->m_rs_lock, flags);
566 rds_message_put(rm); 567 rds_message_put(rm);
568 if (was_on_sock)
569 rds_message_put(rm);
567 } 570 }
568 571
569 if (rs) { 572 if (rs) {
570 spin_unlock(&rs->rs_lock);
571 rds_wake_sk_sleep(rs); 573 rds_wake_sk_sleep(rs);
572 sock_put(rds_rs_to_sk(rs)); 574 sock_put(rds_rs_to_sk(rs));
573 } 575 }
574 local_irq_restore(flags);
575} 576}
576 577
577/* 578/*
@@ -633,9 +634,6 @@ void rds_send_drop_to(struct rds_sock *rs, struct sockaddr_in *dest)
633 list_move(&rm->m_sock_item, &list); 634 list_move(&rm->m_sock_item, &list);
634 rds_send_sndbuf_remove(rs, rm); 635 rds_send_sndbuf_remove(rs, rm);
635 clear_bit(RDS_MSG_ON_SOCK, &rm->m_flags); 636 clear_bit(RDS_MSG_ON_SOCK, &rm->m_flags);
636
637 /* If this is a RDMA operation, notify the app. */
638 __rds_rdma_send_complete(rs, rm, RDS_RDMA_CANCELED);
639 } 637 }
640 638
641 /* order flag updates with the rs lock */ 639 /* order flag updates with the rs lock */
@@ -644,9 +642,6 @@ void rds_send_drop_to(struct rds_sock *rs, struct sockaddr_in *dest)
644 642
645 spin_unlock_irqrestore(&rs->rs_lock, flags); 643 spin_unlock_irqrestore(&rs->rs_lock, flags);
646 644
647 if (wake)
648 rds_wake_sk_sleep(rs);
649
650 conn = NULL; 645 conn = NULL;
651 646
652 /* now remove the messages from the conn list as needed */ 647 /* now remove the messages from the conn list as needed */
@@ -654,6 +649,10 @@ void rds_send_drop_to(struct rds_sock *rs, struct sockaddr_in *dest)
654 /* We do this here rather than in the loop above, so that 649 /* We do this here rather than in the loop above, so that
655 * we don't have to nest m_rs_lock under rs->rs_lock */ 650 * we don't have to nest m_rs_lock under rs->rs_lock */
656 spin_lock_irqsave(&rm->m_rs_lock, flags2); 651 spin_lock_irqsave(&rm->m_rs_lock, flags2);
652 /* If this is a RDMA operation, notify the app. */
653 spin_lock(&rs->rs_lock);
654 __rds_rdma_send_complete(rs, rm, RDS_RDMA_CANCELED);
655 spin_unlock(&rs->rs_lock);
657 rm->m_rs = NULL; 656 rm->m_rs = NULL;
658 spin_unlock_irqrestore(&rm->m_rs_lock, flags2); 657 spin_unlock_irqrestore(&rm->m_rs_lock, flags2);
659 658
@@ -682,6 +681,9 @@ void rds_send_drop_to(struct rds_sock *rs, struct sockaddr_in *dest)
682 if (conn) 681 if (conn)
683 spin_unlock_irqrestore(&conn->c_lock, flags); 682 spin_unlock_irqrestore(&conn->c_lock, flags);
684 683
684 if (wake)
685 rds_wake_sk_sleep(rs);
686
685 while (!list_empty(&list)) { 687 while (!list_empty(&list)) {
686 rm = list_entry(list.next, struct rds_message, m_sock_item); 688 rm = list_entry(list.next, struct rds_message, m_sock_item);
687 list_del_init(&rm->m_sock_item); 689 list_del_init(&rm->m_sock_item);
@@ -815,7 +817,7 @@ int rds_sendmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *msg,
815 int ret = 0; 817 int ret = 0;
816 int queued = 0, allocated_mr = 0; 818 int queued = 0, allocated_mr = 0;
817 int nonblock = msg->msg_flags & MSG_DONTWAIT; 819 int nonblock = msg->msg_flags & MSG_DONTWAIT;
818 long timeo = sock_rcvtimeo(sk, nonblock); 820 long timeo = sock_sndtimeo(sk, nonblock);
819 821
820 /* Mirror Linux UDP mirror of BSD error message compatibility */ 822 /* Mirror Linux UDP mirror of BSD error message compatibility */
821 /* XXX: Perhaps MSG_MORE someday */ 823 /* XXX: Perhaps MSG_MORE someday */
@@ -894,8 +896,10 @@ int rds_sendmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *msg,
894 queue_delayed_work(rds_wq, &conn->c_conn_w, 0); 896 queue_delayed_work(rds_wq, &conn->c_conn_w, 0);
895 897
896 ret = rds_cong_wait(conn->c_fcong, dport, nonblock, rs); 898 ret = rds_cong_wait(conn->c_fcong, dport, nonblock, rs);
897 if (ret) 899 if (ret) {
900 rs->rs_seen_congestion = 1;
898 goto out; 901 goto out;
902 }
899 903
900 while (!rds_send_queue_rm(rs, conn, rm, rs->rs_bound_port, 904 while (!rds_send_queue_rm(rs, conn, rm, rs->rs_bound_port,
901 dport, &queued)) { 905 dport, &queued)) {
diff --git a/net/rds/tcp_recv.c b/net/rds/tcp_recv.c
index c00dafffbb5a..40bfcf887465 100644
--- a/net/rds/tcp_recv.c
+++ b/net/rds/tcp_recv.c
@@ -97,6 +97,7 @@ int rds_tcp_inc_copy_to_user(struct rds_incoming *inc, struct iovec *first_iov,
97 goto out; 97 goto out;
98 } 98 }
99 99
100 rds_stats_add(s_copy_to_user, to_copy);
100 size -= to_copy; 101 size -= to_copy;
101 ret += to_copy; 102 ret += to_copy;
102 skb_off += to_copy; 103 skb_off += to_copy;
diff --git a/net/rds/tcp_send.c b/net/rds/tcp_send.c
index 34fdcc059e54..a28b895ff0d1 100644
--- a/net/rds/tcp_send.c
+++ b/net/rds/tcp_send.c
@@ -240,7 +240,9 @@ void rds_tcp_write_space(struct sock *sk)
240 tc->t_last_seen_una = rds_tcp_snd_una(tc); 240 tc->t_last_seen_una = rds_tcp_snd_una(tc);
241 rds_send_drop_acked(conn, rds_tcp_snd_una(tc), rds_tcp_is_acked); 241 rds_send_drop_acked(conn, rds_tcp_snd_una(tc), rds_tcp_is_acked);
242 242
243 queue_delayed_work(rds_wq, &conn->c_send_w, 0); 243 if ((atomic_read(&sk->sk_wmem_alloc) << 1) <= sk->sk_sndbuf)
244 queue_delayed_work(rds_wq, &conn->c_send_w, 0);
245
244out: 246out:
245 read_unlock(&sk->sk_callback_lock); 247 read_unlock(&sk->sk_callback_lock);
246 248
diff --git a/net/rds/threads.c b/net/rds/threads.c
index 00fa10e59af8..786c20eaaf5e 100644
--- a/net/rds/threads.c
+++ b/net/rds/threads.c
@@ -259,7 +259,7 @@ void rds_threads_exit(void)
259 259
260int __init rds_threads_init(void) 260int __init rds_threads_init(void)
261{ 261{
262 rds_wq = create_singlethread_workqueue("krdsd"); 262 rds_wq = create_workqueue("krdsd");
263 if (rds_wq == NULL) 263 if (rds_wq == NULL)
264 return -ENOMEM; 264 return -ENOMEM;
265 265
diff --git a/net/sched/act_api.c b/net/sched/act_api.c
index 64f5e328cee9..7a558da99bb6 100644
--- a/net/sched/act_api.c
+++ b/net/sched/act_api.c
@@ -667,7 +667,8 @@ nlmsg_failure:
667} 667}
668 668
669static int 669static int
670act_get_notify(u32 pid, struct nlmsghdr *n, struct tc_action *a, int event) 670act_get_notify(struct net *net, u32 pid, struct nlmsghdr *n,
671 struct tc_action *a, int event)
671{ 672{
672 struct sk_buff *skb; 673 struct sk_buff *skb;
673 674
@@ -679,7 +680,7 @@ act_get_notify(u32 pid, struct nlmsghdr *n, struct tc_action *a, int event)
679 return -EINVAL; 680 return -EINVAL;
680 } 681 }
681 682
682 return rtnl_unicast(skb, &init_net, pid); 683 return rtnl_unicast(skb, net, pid);
683} 684}
684 685
685static struct tc_action * 686static struct tc_action *
@@ -749,7 +750,8 @@ static struct tc_action *create_a(int i)
749 return act; 750 return act;
750} 751}
751 752
752static int tca_action_flush(struct nlattr *nla, struct nlmsghdr *n, u32 pid) 753static int tca_action_flush(struct net *net, struct nlattr *nla,
754 struct nlmsghdr *n, u32 pid)
753{ 755{
754 struct sk_buff *skb; 756 struct sk_buff *skb;
755 unsigned char *b; 757 unsigned char *b;
@@ -808,7 +810,7 @@ static int tca_action_flush(struct nlattr *nla, struct nlmsghdr *n, u32 pid)
808 nlh->nlmsg_flags |= NLM_F_ROOT; 810 nlh->nlmsg_flags |= NLM_F_ROOT;
809 module_put(a->ops->owner); 811 module_put(a->ops->owner);
810 kfree(a); 812 kfree(a);
811 err = rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO); 813 err = rtnetlink_send(skb, net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
812 if (err > 0) 814 if (err > 0)
813 return 0; 815 return 0;
814 816
@@ -825,7 +827,8 @@ noflush_out:
825} 827}
826 828
827static int 829static int
828tca_action_gd(struct nlattr *nla, struct nlmsghdr *n, u32 pid, int event) 830tca_action_gd(struct net *net, struct nlattr *nla, struct nlmsghdr *n,
831 u32 pid, int event)
829{ 832{
830 int i, ret; 833 int i, ret;
831 struct nlattr *tb[TCA_ACT_MAX_PRIO+1]; 834 struct nlattr *tb[TCA_ACT_MAX_PRIO+1];
@@ -837,7 +840,7 @@ tca_action_gd(struct nlattr *nla, struct nlmsghdr *n, u32 pid, int event)
837 840
838 if (event == RTM_DELACTION && n->nlmsg_flags&NLM_F_ROOT) { 841 if (event == RTM_DELACTION && n->nlmsg_flags&NLM_F_ROOT) {
839 if (tb[1] != NULL) 842 if (tb[1] != NULL)
840 return tca_action_flush(tb[1], n, pid); 843 return tca_action_flush(net, tb[1], n, pid);
841 else 844 else
842 return -EINVAL; 845 return -EINVAL;
843 } 846 }
@@ -858,7 +861,7 @@ tca_action_gd(struct nlattr *nla, struct nlmsghdr *n, u32 pid, int event)
858 } 861 }
859 862
860 if (event == RTM_GETACTION) 863 if (event == RTM_GETACTION)
861 ret = act_get_notify(pid, n, head, event); 864 ret = act_get_notify(net, pid, n, head, event);
862 else { /* delete */ 865 else { /* delete */
863 struct sk_buff *skb; 866 struct sk_buff *skb;
864 867
@@ -877,7 +880,7 @@ tca_action_gd(struct nlattr *nla, struct nlmsghdr *n, u32 pid, int event)
877 880
878 /* now do the delete */ 881 /* now do the delete */
879 tcf_action_destroy(head, 0); 882 tcf_action_destroy(head, 0);
880 ret = rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, 883 ret = rtnetlink_send(skb, net, pid, RTNLGRP_TC,
881 n->nlmsg_flags&NLM_F_ECHO); 884 n->nlmsg_flags&NLM_F_ECHO);
882 if (ret > 0) 885 if (ret > 0)
883 return 0; 886 return 0;
@@ -888,8 +891,8 @@ err:
888 return ret; 891 return ret;
889} 892}
890 893
891static int tcf_add_notify(struct tc_action *a, u32 pid, u32 seq, int event, 894static int tcf_add_notify(struct net *net, struct tc_action *a,
892 u16 flags) 895 u32 pid, u32 seq, int event, u16 flags)
893{ 896{
894 struct tcamsg *t; 897 struct tcamsg *t;
895 struct nlmsghdr *nlh; 898 struct nlmsghdr *nlh;
@@ -922,7 +925,7 @@ static int tcf_add_notify(struct tc_action *a, u32 pid, u32 seq, int event,
922 nlh->nlmsg_len = skb_tail_pointer(skb) - b; 925 nlh->nlmsg_len = skb_tail_pointer(skb) - b;
923 NETLINK_CB(skb).dst_group = RTNLGRP_TC; 926 NETLINK_CB(skb).dst_group = RTNLGRP_TC;
924 927
925 err = rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, flags&NLM_F_ECHO); 928 err = rtnetlink_send(skb, net, pid, RTNLGRP_TC, flags&NLM_F_ECHO);
926 if (err > 0) 929 if (err > 0)
927 err = 0; 930 err = 0;
928 return err; 931 return err;
@@ -935,7 +938,8 @@ nlmsg_failure:
935 938
936 939
937static int 940static int
938tcf_action_add(struct nlattr *nla, struct nlmsghdr *n, u32 pid, int ovr) 941tcf_action_add(struct net *net, struct nlattr *nla, struct nlmsghdr *n,
942 u32 pid, int ovr)
939{ 943{
940 int ret = 0; 944 int ret = 0;
941 struct tc_action *act; 945 struct tc_action *act;
@@ -953,7 +957,7 @@ tcf_action_add(struct nlattr *nla, struct nlmsghdr *n, u32 pid, int ovr)
953 /* dump then free all the actions after update; inserted policy 957 /* dump then free all the actions after update; inserted policy
954 * stays intact 958 * stays intact
955 * */ 959 * */
956 ret = tcf_add_notify(act, pid, seq, RTM_NEWACTION, n->nlmsg_flags); 960 ret = tcf_add_notify(net, act, pid, seq, RTM_NEWACTION, n->nlmsg_flags);
957 for (a = act; a; a = act) { 961 for (a = act; a; a = act) {
958 act = a->next; 962 act = a->next;
959 kfree(a); 963 kfree(a);
@@ -969,9 +973,6 @@ static int tc_ctl_action(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
969 u32 pid = skb ? NETLINK_CB(skb).pid : 0; 973 u32 pid = skb ? NETLINK_CB(skb).pid : 0;
970 int ret = 0, ovr = 0; 974 int ret = 0, ovr = 0;
971 975
972 if (!net_eq(net, &init_net))
973 return -EINVAL;
974
975 ret = nlmsg_parse(n, sizeof(struct tcamsg), tca, TCA_ACT_MAX, NULL); 976 ret = nlmsg_parse(n, sizeof(struct tcamsg), tca, TCA_ACT_MAX, NULL);
976 if (ret < 0) 977 if (ret < 0)
977 return ret; 978 return ret;
@@ -994,15 +995,17 @@ static int tc_ctl_action(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
994 if (n->nlmsg_flags&NLM_F_REPLACE) 995 if (n->nlmsg_flags&NLM_F_REPLACE)
995 ovr = 1; 996 ovr = 1;
996replay: 997replay:
997 ret = tcf_action_add(tca[TCA_ACT_TAB], n, pid, ovr); 998 ret = tcf_action_add(net, tca[TCA_ACT_TAB], n, pid, ovr);
998 if (ret == -EAGAIN) 999 if (ret == -EAGAIN)
999 goto replay; 1000 goto replay;
1000 break; 1001 break;
1001 case RTM_DELACTION: 1002 case RTM_DELACTION:
1002 ret = tca_action_gd(tca[TCA_ACT_TAB], n, pid, RTM_DELACTION); 1003 ret = tca_action_gd(net, tca[TCA_ACT_TAB], n,
1004 pid, RTM_DELACTION);
1003 break; 1005 break;
1004 case RTM_GETACTION: 1006 case RTM_GETACTION:
1005 ret = tca_action_gd(tca[TCA_ACT_TAB], n, pid, RTM_GETACTION); 1007 ret = tca_action_gd(net, tca[TCA_ACT_TAB], n,
1008 pid, RTM_GETACTION);
1006 break; 1009 break;
1007 default: 1010 default:
1008 BUG(); 1011 BUG();
@@ -1042,7 +1045,6 @@ find_dump_kind(const struct nlmsghdr *n)
1042static int 1045static int
1043tc_dump_action(struct sk_buff *skb, struct netlink_callback *cb) 1046tc_dump_action(struct sk_buff *skb, struct netlink_callback *cb)
1044{ 1047{
1045 struct net *net = sock_net(skb->sk);
1046 struct nlmsghdr *nlh; 1048 struct nlmsghdr *nlh;
1047 unsigned char *b = skb_tail_pointer(skb); 1049 unsigned char *b = skb_tail_pointer(skb);
1048 struct nlattr *nest; 1050 struct nlattr *nest;
@@ -1052,9 +1054,6 @@ tc_dump_action(struct sk_buff *skb, struct netlink_callback *cb)
1052 struct tcamsg *t = (struct tcamsg *) NLMSG_DATA(cb->nlh); 1054 struct tcamsg *t = (struct tcamsg *) NLMSG_DATA(cb->nlh);
1053 struct nlattr *kind = find_dump_kind(cb->nlh); 1055 struct nlattr *kind = find_dump_kind(cb->nlh);
1054 1056
1055 if (!net_eq(net, &init_net))
1056 return 0;
1057
1058 if (kind == NULL) { 1057 if (kind == NULL) {
1059 printk("tc_dump_action: action bad kind\n"); 1058 printk("tc_dump_action: action bad kind\n");
1060 return 0; 1059 return 0;
diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index 3725d8fa29db..4a795d966172 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c
@@ -98,8 +98,9 @@ out:
98} 98}
99EXPORT_SYMBOL(unregister_tcf_proto_ops); 99EXPORT_SYMBOL(unregister_tcf_proto_ops);
100 100
101static int tfilter_notify(struct sk_buff *oskb, struct nlmsghdr *n, 101static int tfilter_notify(struct net *net, struct sk_buff *oskb,
102 struct tcf_proto *tp, unsigned long fh, int event); 102 struct nlmsghdr *n, struct tcf_proto *tp,
103 unsigned long fh, int event);
103 104
104 105
105/* Select new prio value from the range, managed by kernel. */ 106/* Select new prio value from the range, managed by kernel. */
@@ -137,9 +138,6 @@ static int tc_ctl_tfilter(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
137 int err; 138 int err;
138 int tp_created = 0; 139 int tp_created = 0;
139 140
140 if (!net_eq(net, &init_net))
141 return -EINVAL;
142
143replay: 141replay:
144 t = NLMSG_DATA(n); 142 t = NLMSG_DATA(n);
145 protocol = TC_H_MIN(t->tcm_info); 143 protocol = TC_H_MIN(t->tcm_info);
@@ -158,7 +156,7 @@ replay:
158 /* Find head of filter chain. */ 156 /* Find head of filter chain. */
159 157
160 /* Find link */ 158 /* Find link */
161 dev = __dev_get_by_index(&init_net, t->tcm_ifindex); 159 dev = __dev_get_by_index(net, t->tcm_ifindex);
162 if (dev == NULL) 160 if (dev == NULL)
163 return -ENODEV; 161 return -ENODEV;
164 162
@@ -282,7 +280,7 @@ replay:
282 *back = tp->next; 280 *back = tp->next;
283 spin_unlock_bh(root_lock); 281 spin_unlock_bh(root_lock);
284 282
285 tfilter_notify(skb, n, tp, fh, RTM_DELTFILTER); 283 tfilter_notify(net, skb, n, tp, fh, RTM_DELTFILTER);
286 tcf_destroy(tp); 284 tcf_destroy(tp);
287 err = 0; 285 err = 0;
288 goto errout; 286 goto errout;
@@ -305,10 +303,10 @@ replay:
305 case RTM_DELTFILTER: 303 case RTM_DELTFILTER:
306 err = tp->ops->delete(tp, fh); 304 err = tp->ops->delete(tp, fh);
307 if (err == 0) 305 if (err == 0)
308 tfilter_notify(skb, n, tp, fh, RTM_DELTFILTER); 306 tfilter_notify(net, skb, n, tp, fh, RTM_DELTFILTER);
309 goto errout; 307 goto errout;
310 case RTM_GETTFILTER: 308 case RTM_GETTFILTER:
311 err = tfilter_notify(skb, n, tp, fh, RTM_NEWTFILTER); 309 err = tfilter_notify(net, skb, n, tp, fh, RTM_NEWTFILTER);
312 goto errout; 310 goto errout;
313 default: 311 default:
314 err = -EINVAL; 312 err = -EINVAL;
@@ -324,7 +322,7 @@ replay:
324 *back = tp; 322 *back = tp;
325 spin_unlock_bh(root_lock); 323 spin_unlock_bh(root_lock);
326 } 324 }
327 tfilter_notify(skb, n, tp, fh, RTM_NEWTFILTER); 325 tfilter_notify(net, skb, n, tp, fh, RTM_NEWTFILTER);
328 } else { 326 } else {
329 if (tp_created) 327 if (tp_created)
330 tcf_destroy(tp); 328 tcf_destroy(tp);
@@ -370,8 +368,9 @@ nla_put_failure:
370 return -1; 368 return -1;
371} 369}
372 370
373static int tfilter_notify(struct sk_buff *oskb, struct nlmsghdr *n, 371static int tfilter_notify(struct net *net, struct sk_buff *oskb,
374 struct tcf_proto *tp, unsigned long fh, int event) 372 struct nlmsghdr *n, struct tcf_proto *tp,
373 unsigned long fh, int event)
375{ 374{
376 struct sk_buff *skb; 375 struct sk_buff *skb;
377 u32 pid = oskb ? NETLINK_CB(oskb).pid : 0; 376 u32 pid = oskb ? NETLINK_CB(oskb).pid : 0;
@@ -385,7 +384,7 @@ static int tfilter_notify(struct sk_buff *oskb, struct nlmsghdr *n,
385 return -EINVAL; 384 return -EINVAL;
386 } 385 }
387 386
388 return rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, 387 return rtnetlink_send(skb, net, pid, RTNLGRP_TC,
389 n->nlmsg_flags & NLM_F_ECHO); 388 n->nlmsg_flags & NLM_F_ECHO);
390} 389}
391 390
@@ -418,12 +417,9 @@ static int tc_dump_tfilter(struct sk_buff *skb, struct netlink_callback *cb)
418 const struct Qdisc_class_ops *cops; 417 const struct Qdisc_class_ops *cops;
419 struct tcf_dump_args arg; 418 struct tcf_dump_args arg;
420 419
421 if (!net_eq(net, &init_net))
422 return 0;
423
424 if (cb->nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*tcm))) 420 if (cb->nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*tcm)))
425 return skb->len; 421 return skb->len;
426 if ((dev = __dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL) 422 if ((dev = __dev_get_by_index(net, tcm->tcm_ifindex)) == NULL)
427 return skb->len; 423 return skb->len;
428 424
429 if (!tcm->tcm_parent) 425 if (!tcm->tcm_parent)
diff --git a/net/sched/cls_u32.c b/net/sched/cls_u32.c
index 07372f60bee3..1ef76871a57b 100644
--- a/net/sched/cls_u32.c
+++ b/net/sched/cls_u32.c
@@ -772,10 +772,10 @@ static int __init init_u32(void)
772 printk(" Performance counters on\n"); 772 printk(" Performance counters on\n");
773#endif 773#endif
774#ifdef CONFIG_NET_CLS_IND 774#ifdef CONFIG_NET_CLS_IND
775 printk(" input device check on \n"); 775 printk(" input device check on\n");
776#endif 776#endif
777#ifdef CONFIG_NET_CLS_ACT 777#ifdef CONFIG_NET_CLS_ACT
778 printk(" Actions configured \n"); 778 printk(" Actions configured\n");
779#endif 779#endif
780 return register_tcf_proto_ops(&cls_u32_ops); 780 return register_tcf_proto_ops(&cls_u32_ops);
781} 781}
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index 6cd491013b50..6d6fe16289f3 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -34,10 +34,12 @@
34#include <net/netlink.h> 34#include <net/netlink.h>
35#include <net/pkt_sched.h> 35#include <net/pkt_sched.h>
36 36
37static int qdisc_notify(struct sk_buff *oskb, struct nlmsghdr *n, u32 clid, 37static int qdisc_notify(struct net *net, struct sk_buff *oskb,
38 struct nlmsghdr *n, u32 clid,
38 struct Qdisc *old, struct Qdisc *new); 39 struct Qdisc *old, struct Qdisc *new);
39static int tclass_notify(struct sk_buff *oskb, struct nlmsghdr *n, 40static int tclass_notify(struct net *net, struct sk_buff *oskb,
40 struct Qdisc *q, unsigned long cl, int event); 41 struct nlmsghdr *n, struct Qdisc *q,
42 unsigned long cl, int event);
41 43
42/* 44/*
43 45
@@ -638,11 +640,12 @@ void qdisc_tree_decrease_qlen(struct Qdisc *sch, unsigned int n)
638} 640}
639EXPORT_SYMBOL(qdisc_tree_decrease_qlen); 641EXPORT_SYMBOL(qdisc_tree_decrease_qlen);
640 642
641static void notify_and_destroy(struct sk_buff *skb, struct nlmsghdr *n, u32 clid, 643static void notify_and_destroy(struct net *net, struct sk_buff *skb,
644 struct nlmsghdr *n, u32 clid,
642 struct Qdisc *old, struct Qdisc *new) 645 struct Qdisc *old, struct Qdisc *new)
643{ 646{
644 if (new || old) 647 if (new || old)
645 qdisc_notify(skb, n, clid, old, new); 648 qdisc_notify(net, skb, n, clid, old, new);
646 649
647 if (old) 650 if (old)
648 qdisc_destroy(old); 651 qdisc_destroy(old);
@@ -662,6 +665,7 @@ static int qdisc_graft(struct net_device *dev, struct Qdisc *parent,
662 struct Qdisc *new, struct Qdisc *old) 665 struct Qdisc *new, struct Qdisc *old)
663{ 666{
664 struct Qdisc *q = old; 667 struct Qdisc *q = old;
668 struct net *net = dev_net(dev);
665 int err = 0; 669 int err = 0;
666 670
667 if (parent == NULL) { 671 if (parent == NULL) {
@@ -698,12 +702,13 @@ static int qdisc_graft(struct net_device *dev, struct Qdisc *parent,
698 } 702 }
699 703
700 if (!ingress) { 704 if (!ingress) {
701 notify_and_destroy(skb, n, classid, dev->qdisc, new); 705 notify_and_destroy(net, skb, n, classid,
706 dev->qdisc, new);
702 if (new && !new->ops->attach) 707 if (new && !new->ops->attach)
703 atomic_inc(&new->refcnt); 708 atomic_inc(&new->refcnt);
704 dev->qdisc = new ? : &noop_qdisc; 709 dev->qdisc = new ? : &noop_qdisc;
705 } else { 710 } else {
706 notify_and_destroy(skb, n, classid, old, new); 711 notify_and_destroy(net, skb, n, classid, old, new);
707 } 712 }
708 713
709 if (dev->flags & IFF_UP) 714 if (dev->flags & IFF_UP)
@@ -721,7 +726,7 @@ static int qdisc_graft(struct net_device *dev, struct Qdisc *parent,
721 err = -ENOENT; 726 err = -ENOENT;
722 } 727 }
723 if (!err) 728 if (!err)
724 notify_and_destroy(skb, n, classid, old, new); 729 notify_and_destroy(net, skb, n, classid, old, new);
725 } 730 }
726 return err; 731 return err;
727} 732}
@@ -947,10 +952,7 @@ static int tc_get_qdisc(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
947 struct Qdisc *p = NULL; 952 struct Qdisc *p = NULL;
948 int err; 953 int err;
949 954
950 if (!net_eq(net, &init_net)) 955 if ((dev = __dev_get_by_index(net, tcm->tcm_ifindex)) == NULL)
951 return -EINVAL;
952
953 if ((dev = __dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
954 return -ENODEV; 956 return -ENODEV;
955 957
956 err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL); 958 err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL);
@@ -990,7 +992,7 @@ static int tc_get_qdisc(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
990 if ((err = qdisc_graft(dev, p, skb, n, clid, NULL, q)) != 0) 992 if ((err = qdisc_graft(dev, p, skb, n, clid, NULL, q)) != 0)
991 return err; 993 return err;
992 } else { 994 } else {
993 qdisc_notify(skb, n, clid, NULL, q); 995 qdisc_notify(net, skb, n, clid, NULL, q);
994 } 996 }
995 return 0; 997 return 0;
996} 998}
@@ -1009,16 +1011,13 @@ static int tc_modify_qdisc(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
1009 struct Qdisc *q, *p; 1011 struct Qdisc *q, *p;
1010 int err; 1012 int err;
1011 1013
1012 if (!net_eq(net, &init_net))
1013 return -EINVAL;
1014
1015replay: 1014replay:
1016 /* Reinit, just in case something touches this. */ 1015 /* Reinit, just in case something touches this. */
1017 tcm = NLMSG_DATA(n); 1016 tcm = NLMSG_DATA(n);
1018 clid = tcm->tcm_parent; 1017 clid = tcm->tcm_parent;
1019 q = p = NULL; 1018 q = p = NULL;
1020 1019
1021 if ((dev = __dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL) 1020 if ((dev = __dev_get_by_index(net, tcm->tcm_ifindex)) == NULL)
1022 return -ENODEV; 1021 return -ENODEV;
1023 1022
1024 err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL); 1023 err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL);
@@ -1105,7 +1104,7 @@ replay:
1105 return -EINVAL; 1104 return -EINVAL;
1106 err = qdisc_change(q, tca); 1105 err = qdisc_change(q, tca);
1107 if (err == 0) 1106 if (err == 0)
1108 qdisc_notify(skb, n, clid, NULL, q); 1107 qdisc_notify(net, skb, n, clid, NULL, q);
1109 return err; 1108 return err;
1110 1109
1111create_n_graft: 1110create_n_graft:
@@ -1195,8 +1194,9 @@ nla_put_failure:
1195 return -1; 1194 return -1;
1196} 1195}
1197 1196
1198static int qdisc_notify(struct sk_buff *oskb, struct nlmsghdr *n, 1197static int qdisc_notify(struct net *net, struct sk_buff *oskb,
1199 u32 clid, struct Qdisc *old, struct Qdisc *new) 1198 struct nlmsghdr *n, u32 clid,
1199 struct Qdisc *old, struct Qdisc *new)
1200{ 1200{
1201 struct sk_buff *skb; 1201 struct sk_buff *skb;
1202 u32 pid = oskb ? NETLINK_CB(oskb).pid : 0; 1202 u32 pid = oskb ? NETLINK_CB(oskb).pid : 0;
@@ -1215,7 +1215,7 @@ static int qdisc_notify(struct sk_buff *oskb, struct nlmsghdr *n,
1215 } 1215 }
1216 1216
1217 if (skb->len) 1217 if (skb->len)
1218 return rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO); 1218 return rtnetlink_send(skb, net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
1219 1219
1220err_out: 1220err_out:
1221 kfree_skb(skb); 1221 kfree_skb(skb);
@@ -1274,15 +1274,12 @@ static int tc_dump_qdisc(struct sk_buff *skb, struct netlink_callback *cb)
1274 int s_idx, s_q_idx; 1274 int s_idx, s_q_idx;
1275 struct net_device *dev; 1275 struct net_device *dev;
1276 1276
1277 if (!net_eq(net, &init_net))
1278 return 0;
1279
1280 s_idx = cb->args[0]; 1277 s_idx = cb->args[0];
1281 s_q_idx = q_idx = cb->args[1]; 1278 s_q_idx = q_idx = cb->args[1];
1282 1279
1283 rcu_read_lock(); 1280 rcu_read_lock();
1284 idx = 0; 1281 idx = 0;
1285 for_each_netdev_rcu(&init_net, dev) { 1282 for_each_netdev_rcu(net, dev) {
1286 struct netdev_queue *dev_queue; 1283 struct netdev_queue *dev_queue;
1287 1284
1288 if (idx < s_idx) 1285 if (idx < s_idx)
@@ -1334,10 +1331,7 @@ static int tc_ctl_tclass(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
1334 u32 qid = TC_H_MAJ(clid); 1331 u32 qid = TC_H_MAJ(clid);
1335 int err; 1332 int err;
1336 1333
1337 if (!net_eq(net, &init_net)) 1334 if ((dev = __dev_get_by_index(net, tcm->tcm_ifindex)) == NULL)
1338 return -EINVAL;
1339
1340 if ((dev = __dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL)
1341 return -ENODEV; 1335 return -ENODEV;
1342 1336
1343 err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL); 1337 err = nlmsg_parse(n, sizeof(*tcm), tca, TCA_MAX, NULL);
@@ -1418,10 +1412,10 @@ static int tc_ctl_tclass(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
1418 if (cops->delete) 1412 if (cops->delete)
1419 err = cops->delete(q, cl); 1413 err = cops->delete(q, cl);
1420 if (err == 0) 1414 if (err == 0)
1421 tclass_notify(skb, n, q, cl, RTM_DELTCLASS); 1415 tclass_notify(net, skb, n, q, cl, RTM_DELTCLASS);
1422 goto out; 1416 goto out;
1423 case RTM_GETTCLASS: 1417 case RTM_GETTCLASS:
1424 err = tclass_notify(skb, n, q, cl, RTM_NEWTCLASS); 1418 err = tclass_notify(net, skb, n, q, cl, RTM_NEWTCLASS);
1425 goto out; 1419 goto out;
1426 default: 1420 default:
1427 err = -EINVAL; 1421 err = -EINVAL;
@@ -1434,7 +1428,7 @@ static int tc_ctl_tclass(struct sk_buff *skb, struct nlmsghdr *n, void *arg)
1434 if (cops->change) 1428 if (cops->change)
1435 err = cops->change(q, clid, pid, tca, &new_cl); 1429 err = cops->change(q, clid, pid, tca, &new_cl);
1436 if (err == 0) 1430 if (err == 0)
1437 tclass_notify(skb, n, q, new_cl, RTM_NEWTCLASS); 1431 tclass_notify(net, skb, n, q, new_cl, RTM_NEWTCLASS);
1438 1432
1439out: 1433out:
1440 if (cl) 1434 if (cl)
@@ -1486,8 +1480,9 @@ nla_put_failure:
1486 return -1; 1480 return -1;
1487} 1481}
1488 1482
1489static int tclass_notify(struct sk_buff *oskb, struct nlmsghdr *n, 1483static int tclass_notify(struct net *net, struct sk_buff *oskb,
1490 struct Qdisc *q, unsigned long cl, int event) 1484 struct nlmsghdr *n, struct Qdisc *q,
1485 unsigned long cl, int event)
1491{ 1486{
1492 struct sk_buff *skb; 1487 struct sk_buff *skb;
1493 u32 pid = oskb ? NETLINK_CB(oskb).pid : 0; 1488 u32 pid = oskb ? NETLINK_CB(oskb).pid : 0;
@@ -1501,7 +1496,7 @@ static int tclass_notify(struct sk_buff *oskb, struct nlmsghdr *n,
1501 return -EINVAL; 1496 return -EINVAL;
1502 } 1497 }
1503 1498
1504 return rtnetlink_send(skb, &init_net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO); 1499 return rtnetlink_send(skb, net, pid, RTNLGRP_TC, n->nlmsg_flags&NLM_F_ECHO);
1505} 1500}
1506 1501
1507struct qdisc_dump_args 1502struct qdisc_dump_args
@@ -1576,12 +1571,9 @@ static int tc_dump_tclass(struct sk_buff *skb, struct netlink_callback *cb)
1576 struct net_device *dev; 1571 struct net_device *dev;
1577 int t, s_t; 1572 int t, s_t;
1578 1573
1579 if (!net_eq(net, &init_net))
1580 return 0;
1581
1582 if (cb->nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*tcm))) 1574 if (cb->nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*tcm)))
1583 return 0; 1575 return 0;
1584 if ((dev = dev_get_by_index(&init_net, tcm->tcm_ifindex)) == NULL) 1576 if ((dev = dev_get_by_index(net, tcm->tcm_ifindex)) == NULL)
1585 return 0; 1577 return 0;
1586 1578
1587 s_t = cb->args[0]; 1579 s_t = cb->args[0];
@@ -1701,15 +1693,55 @@ static const struct file_operations psched_fops = {
1701 .llseek = seq_lseek, 1693 .llseek = seq_lseek,
1702 .release = single_release, 1694 .release = single_release,
1703}; 1695};
1696
1697static int __net_init psched_net_init(struct net *net)
1698{
1699 struct proc_dir_entry *e;
1700
1701 e = proc_net_fops_create(net, "psched", 0, &psched_fops);
1702 if (e == NULL)
1703 return -ENOMEM;
1704
1705 return 0;
1706}
1707
1708static void __net_exit psched_net_exit(struct net *net)
1709{
1710 proc_net_remove(net, "psched");
1711
1712 return;
1713}
1714#else
1715static int __net_init psched_net_init(struct net *net)
1716{
1717 return 0;
1718}
1719
1720static void __net_exit psched_net_exit(struct net *net)
1721{
1722}
1704#endif 1723#endif
1705 1724
1725static struct pernet_operations psched_net_ops = {
1726 .init = psched_net_init,
1727 .exit = psched_net_exit,
1728};
1729
1706static int __init pktsched_init(void) 1730static int __init pktsched_init(void)
1707{ 1731{
1732 int err;
1733
1734 err = register_pernet_subsys(&psched_net_ops);
1735 if (err) {
1736 printk(KERN_ERR "pktsched_init: "
1737 "cannot initialize per netns operations\n");
1738 return err;
1739 }
1740
1708 register_qdisc(&pfifo_qdisc_ops); 1741 register_qdisc(&pfifo_qdisc_ops);
1709 register_qdisc(&bfifo_qdisc_ops); 1742 register_qdisc(&bfifo_qdisc_ops);
1710 register_qdisc(&pfifo_head_drop_qdisc_ops); 1743 register_qdisc(&pfifo_head_drop_qdisc_ops);
1711 register_qdisc(&mq_qdisc_ops); 1744 register_qdisc(&mq_qdisc_ops);
1712 proc_net_fops_create(&init_net, "psched", 0, &psched_fops);
1713 1745
1714 rtnl_register(PF_UNSPEC, RTM_NEWQDISC, tc_modify_qdisc, NULL); 1746 rtnl_register(PF_UNSPEC, RTM_NEWQDISC, tc_modify_qdisc, NULL);
1715 rtnl_register(PF_UNSPEC, RTM_DELQDISC, tc_get_qdisc, NULL); 1747 rtnl_register(PF_UNSPEC, RTM_DELQDISC, tc_get_qdisc, NULL);
diff --git a/net/sctp/input.c b/net/sctp/input.c
index c0c973e67add..3d74b264ea22 100644
--- a/net/sctp/input.c
+++ b/net/sctp/input.c
@@ -75,7 +75,7 @@ static struct sctp_association *__sctp_lookup_association(
75 const union sctp_addr *peer, 75 const union sctp_addr *peer,
76 struct sctp_transport **pt); 76 struct sctp_transport **pt);
77 77
78static void sctp_add_backlog(struct sock *sk, struct sk_buff *skb); 78static int sctp_add_backlog(struct sock *sk, struct sk_buff *skb);
79 79
80 80
81/* Calculate the SCTP checksum of an SCTP packet. */ 81/* Calculate the SCTP checksum of an SCTP packet. */
@@ -265,8 +265,13 @@ int sctp_rcv(struct sk_buff *skb)
265 } 265 }
266 266
267 if (sock_owned_by_user(sk)) { 267 if (sock_owned_by_user(sk)) {
268 if (sctp_add_backlog(sk, skb)) {
269 sctp_bh_unlock_sock(sk);
270 sctp_chunk_free(chunk);
271 skb = NULL; /* sctp_chunk_free already freed the skb */
272 goto discard_release;
273 }
268 SCTP_INC_STATS_BH(SCTP_MIB_IN_PKT_BACKLOG); 274 SCTP_INC_STATS_BH(SCTP_MIB_IN_PKT_BACKLOG);
269 sctp_add_backlog(sk, skb);
270 } else { 275 } else {
271 SCTP_INC_STATS_BH(SCTP_MIB_IN_PKT_SOFTIRQ); 276 SCTP_INC_STATS_BH(SCTP_MIB_IN_PKT_SOFTIRQ);
272 sctp_inq_push(&chunk->rcvr->inqueue, chunk); 277 sctp_inq_push(&chunk->rcvr->inqueue, chunk);
@@ -336,8 +341,10 @@ int sctp_backlog_rcv(struct sock *sk, struct sk_buff *skb)
336 sctp_bh_lock_sock(sk); 341 sctp_bh_lock_sock(sk);
337 342
338 if (sock_owned_by_user(sk)) { 343 if (sock_owned_by_user(sk)) {
339 sk_add_backlog(sk, skb); 344 if (sk_add_backlog(sk, skb))
340 backloged = 1; 345 sctp_chunk_free(chunk);
346 else
347 backloged = 1;
341 } else 348 } else
342 sctp_inq_push(inqueue, chunk); 349 sctp_inq_push(inqueue, chunk);
343 350
@@ -362,22 +369,27 @@ done:
362 return 0; 369 return 0;
363} 370}
364 371
365static void sctp_add_backlog(struct sock *sk, struct sk_buff *skb) 372static int sctp_add_backlog(struct sock *sk, struct sk_buff *skb)
366{ 373{
367 struct sctp_chunk *chunk = SCTP_INPUT_CB(skb)->chunk; 374 struct sctp_chunk *chunk = SCTP_INPUT_CB(skb)->chunk;
368 struct sctp_ep_common *rcvr = chunk->rcvr; 375 struct sctp_ep_common *rcvr = chunk->rcvr;
376 int ret;
369 377
370 /* Hold the assoc/ep while hanging on the backlog queue. 378 ret = sk_add_backlog(sk, skb);
371 * This way, we know structures we need will not disappear from us 379 if (!ret) {
372 */ 380 /* Hold the assoc/ep while hanging on the backlog queue.
373 if (SCTP_EP_TYPE_ASSOCIATION == rcvr->type) 381 * This way, we know structures we need will not disappear
374 sctp_association_hold(sctp_assoc(rcvr)); 382 * from us
375 else if (SCTP_EP_TYPE_SOCKET == rcvr->type) 383 */
376 sctp_endpoint_hold(sctp_ep(rcvr)); 384 if (SCTP_EP_TYPE_ASSOCIATION == rcvr->type)
377 else 385 sctp_association_hold(sctp_assoc(rcvr));
378 BUG(); 386 else if (SCTP_EP_TYPE_SOCKET == rcvr->type)
387 sctp_endpoint_hold(sctp_ep(rcvr));
388 else
389 BUG();
390 }
391 return ret;
379 392
380 sk_add_backlog(sk, skb);
381} 393}
382 394
383/* Handle icmp frag needed error. */ 395/* Handle icmp frag needed error. */
diff --git a/net/sctp/ipv6.c b/net/sctp/ipv6.c
index 1d7ac70ba39f..240dceba06e5 100644
--- a/net/sctp/ipv6.c
+++ b/net/sctp/ipv6.c
@@ -371,7 +371,7 @@ static void sctp_v6_copy_addrlist(struct list_head *addrlist,
371 } 371 }
372 372
373 read_lock_bh(&in6_dev->lock); 373 read_lock_bh(&in6_dev->lock);
374 for (ifp = in6_dev->addr_list; ifp; ifp = ifp->if_next) { 374 list_for_each_entry(ifp, &in6_dev->addr_list, if_list) {
375 /* Add the address to the local list. */ 375 /* Add the address to the local list. */
376 addr = t_new(struct sctp_sockaddr_entry, GFP_ATOMIC); 376 addr = t_new(struct sctp_sockaddr_entry, GFP_ATOMIC);
377 if (addr) { 377 if (addr) {
diff --git a/net/sctp/sm_sideeffect.c b/net/sctp/sm_sideeffect.c
index 4e4ca65cd320..500886bda9b4 100644
--- a/net/sctp/sm_sideeffect.c
+++ b/net/sctp/sm_sideeffect.c
@@ -475,7 +475,7 @@ static void sctp_do_8_2_transport_strike(struct sctp_association *asoc,
475 * used to provide an upper bound to this doubling operation. 475 * used to provide an upper bound to this doubling operation.
476 * 476 *
477 * Special Case: the first HB doesn't trigger exponential backoff. 477 * Special Case: the first HB doesn't trigger exponential backoff.
478 * The first unacknowleged HB triggers it. We do this with a flag 478 * The first unacknowledged HB triggers it. We do this with a flag
479 * that indicates that we have an outstanding HB. 479 * that indicates that we have an outstanding HB.
480 */ 480 */
481 if (!is_hb || transport->hb_sent) { 481 if (!is_hb || transport->hb_sent) {
diff --git a/net/sctp/socket.c b/net/sctp/socket.c
index f6d1e59c4151..dfc5c127efd4 100644
--- a/net/sctp/socket.c
+++ b/net/sctp/socket.c
@@ -3720,6 +3720,9 @@ SCTP_STATIC int sctp_init_sock(struct sock *sk)
3720 SCTP_DBG_OBJCNT_INC(sock); 3720 SCTP_DBG_OBJCNT_INC(sock);
3721 percpu_counter_inc(&sctp_sockets_allocated); 3721 percpu_counter_inc(&sctp_sockets_allocated);
3722 3722
3723 /* Set socket backlog limit. */
3724 sk->sk_backlog.limit = sysctl_sctp_rmem[1];
3725
3723 local_bh_disable(); 3726 local_bh_disable();
3724 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1); 3727 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
3725 local_bh_enable(); 3728 local_bh_enable();
diff --git a/net/sunrpc/addr.c b/net/sunrpc/addr.c
index 6dcdd2517819..f845d9d72f73 100644
--- a/net/sunrpc/addr.c
+++ b/net/sunrpc/addr.c
@@ -71,8 +71,9 @@ static size_t rpc_ntop6(const struct sockaddr *sap,
71 if (unlikely(len == 0)) 71 if (unlikely(len == 0))
72 return len; 72 return len;
73 73
74 if (!(ipv6_addr_type(&sin6->sin6_addr) & IPV6_ADDR_LINKLOCAL) && 74 if (!(ipv6_addr_type(&sin6->sin6_addr) & IPV6_ADDR_LINKLOCAL))
75 !(ipv6_addr_type(&sin6->sin6_addr) & IPV6_ADDR_SITELOCAL)) 75 return len;
76 if (sin6->sin6_scope_id == 0)
76 return len; 77 return len;
77 78
78 rc = snprintf(scopebuf, sizeof(scopebuf), "%c%u", 79 rc = snprintf(scopebuf, sizeof(scopebuf), "%c%u",
@@ -165,8 +166,7 @@ static int rpc_parse_scope_id(const char *buf, const size_t buflen,
165 if (*delim != IPV6_SCOPE_DELIMITER) 166 if (*delim != IPV6_SCOPE_DELIMITER)
166 return 0; 167 return 0;
167 168
168 if (!(ipv6_addr_type(&sin6->sin6_addr) & IPV6_ADDR_LINKLOCAL) && 169 if (!(ipv6_addr_type(&sin6->sin6_addr) & IPV6_ADDR_LINKLOCAL))
169 !(ipv6_addr_type(&sin6->sin6_addr) & IPV6_ADDR_SITELOCAL))
170 return 0; 170 return 0;
171 171
172 len = (buf + buflen) - delim - 1; 172 len = (buf + buflen) - delim - 1;
diff --git a/net/sunrpc/auth_gss/auth_gss.c b/net/sunrpc/auth_gss/auth_gss.c
index f7a7f8380e38..0cfccc2a0297 100644
--- a/net/sunrpc/auth_gss/auth_gss.c
+++ b/net/sunrpc/auth_gss/auth_gss.c
@@ -206,8 +206,14 @@ gss_fill_context(const void *p, const void *end, struct gss_cl_ctx *ctx, struct
206 ctx->gc_win = window_size; 206 ctx->gc_win = window_size;
207 /* gssd signals an error by passing ctx->gc_win = 0: */ 207 /* gssd signals an error by passing ctx->gc_win = 0: */
208 if (ctx->gc_win == 0) { 208 if (ctx->gc_win == 0) {
209 /* in which case, p points to an error code which we ignore */ 209 /*
210 p = ERR_PTR(-EACCES); 210 * in which case, p points to an error code. Anything other
211 * than -EKEYEXPIRED gets converted to -EACCES.
212 */
213 p = simple_get_bytes(p, end, &ret, sizeof(ret));
214 if (!IS_ERR(p))
215 p = (ret == -EKEYEXPIRED) ? ERR_PTR(-EKEYEXPIRED) :
216 ERR_PTR(-EACCES);
211 goto err; 217 goto err;
212 } 218 }
213 /* copy the opaque wire context */ 219 /* copy the opaque wire context */
@@ -646,6 +652,7 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
646 err = PTR_ERR(p); 652 err = PTR_ERR(p);
647 switch (err) { 653 switch (err) {
648 case -EACCES: 654 case -EACCES:
655 case -EKEYEXPIRED:
649 gss_msg->msg.errno = err; 656 gss_msg->msg.errno = err;
650 err = mlen; 657 err = mlen;
651 break; 658 break;
diff --git a/net/sunrpc/auth_gss/gss_spkm3_token.c b/net/sunrpc/auth_gss/gss_spkm3_token.c
index 3308157436d2..a99825d7caa0 100644
--- a/net/sunrpc/auth_gss/gss_spkm3_token.c
+++ b/net/sunrpc/auth_gss/gss_spkm3_token.c
@@ -223,7 +223,7 @@ spkm3_verify_mic_token(unsigned char **tokp, int *mic_hdrlen, unsigned char **ck
223 223
224 /* only support SPKM_MIC_TOK */ 224 /* only support SPKM_MIC_TOK */
225 if((ptr[6] != 0x01) || (ptr[7] != 0x01)) { 225 if((ptr[6] != 0x01) || (ptr[7] != 0x01)) {
226 dprintk("RPC: ERROR unsupported SPKM3 token \n"); 226 dprintk("RPC: ERROR unsupported SPKM3 token\n");
227 goto out; 227 goto out;
228 } 228 }
229 229
diff --git a/net/sunrpc/bc_svc.c b/net/sunrpc/bc_svc.c
index 13f214f53120..8a610fb0cfec 100644
--- a/net/sunrpc/bc_svc.c
+++ b/net/sunrpc/bc_svc.c
@@ -75,7 +75,7 @@ int bc_send(struct rpc_rqst *req)
75 rpc_put_task(task); 75 rpc_put_task(task);
76 } 76 }
77 return ret; 77 return ret;
78 dprintk("RPC: bc_send ret= %d \n", ret); 78 dprintk("RPC: bc_send ret= %d\n", ret);
79} 79}
80 80
81#endif /* CONFIG_NFS_V4_1 */ 81#endif /* CONFIG_NFS_V4_1 */
diff --git a/net/sunrpc/rpc_pipe.c b/net/sunrpc/rpc_pipe.c
index 9ea45383480e..8d63f8fd29b7 100644
--- a/net/sunrpc/rpc_pipe.c
+++ b/net/sunrpc/rpc_pipe.c
@@ -999,19 +999,14 @@ rpc_fill_super(struct super_block *sb, void *data, int silent)
999 inode = rpc_get_inode(sb, S_IFDIR | 0755); 999 inode = rpc_get_inode(sb, S_IFDIR | 0755);
1000 if (!inode) 1000 if (!inode)
1001 return -ENOMEM; 1001 return -ENOMEM;
1002 root = d_alloc_root(inode); 1002 sb->s_root = root = d_alloc_root(inode);
1003 if (!root) { 1003 if (!root) {
1004 iput(inode); 1004 iput(inode);
1005 return -ENOMEM; 1005 return -ENOMEM;
1006 } 1006 }
1007 if (rpc_populate(root, files, RPCAUTH_lockd, RPCAUTH_RootEOF, NULL)) 1007 if (rpc_populate(root, files, RPCAUTH_lockd, RPCAUTH_RootEOF, NULL))
1008 goto out; 1008 return -ENOMEM;
1009 sb->s_root = root;
1010 return 0; 1009 return 0;
1011out:
1012 d_genocide(root);
1013 dput(root);
1014 return -ENOMEM;
1015} 1010}
1016 1011
1017static int 1012static int
diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
index 538ca433a56c..8420a4205b76 100644
--- a/net/sunrpc/svc.c
+++ b/net/sunrpc/svc.c
@@ -133,7 +133,7 @@ svc_pool_map_choose_mode(void)
133 return SVC_POOL_PERNODE; 133 return SVC_POOL_PERNODE;
134 } 134 }
135 135
136 node = any_online_node(node_online_map); 136 node = first_online_node;
137 if (nr_cpus_node(node) > 2) { 137 if (nr_cpus_node(node) > 2) {
138 /* 138 /*
139 * Non-trivial SMP, or CONFIG_NUMA on 139 * Non-trivial SMP, or CONFIG_NUMA on
@@ -506,6 +506,10 @@ svc_init_buffer(struct svc_rqst *rqstp, unsigned int size)
506{ 506{
507 unsigned int pages, arghi; 507 unsigned int pages, arghi;
508 508
509 /* bc_xprt uses fore channel allocated buffers */
510 if (svc_is_backchannel(rqstp))
511 return 1;
512
509 pages = size / PAGE_SIZE + 1; /* extra page as we hold both request and reply. 513 pages = size / PAGE_SIZE + 1; /* extra page as we hold both request and reply.
510 * We assume one is at most one page 514 * We assume one is at most one page
511 */ 515 */
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 7d1f9e928f69..8f0f1fb3dc52 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -173,11 +173,13 @@ static struct svc_xprt *__svc_xpo_create(struct svc_xprt_class *xcl,
173 .sin_addr.s_addr = htonl(INADDR_ANY), 173 .sin_addr.s_addr = htonl(INADDR_ANY),
174 .sin_port = htons(port), 174 .sin_port = htons(port),
175 }; 175 };
176#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
176 struct sockaddr_in6 sin6 = { 177 struct sockaddr_in6 sin6 = {
177 .sin6_family = AF_INET6, 178 .sin6_family = AF_INET6,
178 .sin6_addr = IN6ADDR_ANY_INIT, 179 .sin6_addr = IN6ADDR_ANY_INIT,
179 .sin6_port = htons(port), 180 .sin6_port = htons(port),
180 }; 181 };
182#endif /* defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) */
181 struct sockaddr *sap; 183 struct sockaddr *sap;
182 size_t len; 184 size_t len;
183 185
@@ -186,10 +188,12 @@ static struct svc_xprt *__svc_xpo_create(struct svc_xprt_class *xcl,
186 sap = (struct sockaddr *)&sin; 188 sap = (struct sockaddr *)&sin;
187 len = sizeof(sin); 189 len = sizeof(sin);
188 break; 190 break;
191#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
189 case PF_INET6: 192 case PF_INET6:
190 sap = (struct sockaddr *)&sin6; 193 sap = (struct sockaddr *)&sin6;
191 len = sizeof(sin6); 194 len = sizeof(sin6);
192 break; 195 break;
196#endif /* defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) */
193 default: 197 default:
194 return ERR_PTR(-EAFNOSUPPORT); 198 return ERR_PTR(-EAFNOSUPPORT);
195 } 199 }
@@ -231,7 +235,10 @@ int svc_create_xprt(struct svc_serv *serv, const char *xprt_name,
231 err: 235 err:
232 spin_unlock(&svc_xprt_class_lock); 236 spin_unlock(&svc_xprt_class_lock);
233 dprintk("svc: transport %s not found\n", xprt_name); 237 dprintk("svc: transport %s not found\n", xprt_name);
234 return -ENOENT; 238
239 /* This errno is exposed to user space. Provide a reasonable
240 * perror msg for a bad transport. */
241 return -EPROTONOSUPPORT;
235} 242}
236EXPORT_SYMBOL_GPL(svc_create_xprt); 243EXPORT_SYMBOL_GPL(svc_create_xprt);
237 244
@@ -699,8 +706,10 @@ int svc_recv(struct svc_rqst *rqstp, long timeout)
699 spin_unlock_bh(&pool->sp_lock); 706 spin_unlock_bh(&pool->sp_lock);
700 707
701 len = 0; 708 len = 0;
702 if (test_bit(XPT_LISTENER, &xprt->xpt_flags) && 709 if (test_bit(XPT_CLOSE, &xprt->xpt_flags)) {
703 !test_bit(XPT_CLOSE, &xprt->xpt_flags)) { 710 dprintk("svc_recv: found XPT_CLOSE\n");
711 svc_delete_xprt(xprt);
712 } else if (test_bit(XPT_LISTENER, &xprt->xpt_flags)) {
704 struct svc_xprt *newxpt; 713 struct svc_xprt *newxpt;
705 newxpt = xprt->xpt_ops->xpo_accept(xprt); 714 newxpt = xprt->xpt_ops->xpo_accept(xprt);
706 if (newxpt) { 715 if (newxpt) {
@@ -726,7 +735,7 @@ int svc_recv(struct svc_rqst *rqstp, long timeout)
726 svc_xprt_received(newxpt); 735 svc_xprt_received(newxpt);
727 } 736 }
728 svc_xprt_received(xprt); 737 svc_xprt_received(xprt);
729 } else if (!test_bit(XPT_CLOSE, &xprt->xpt_flags)) { 738 } else {
730 dprintk("svc: server %p, pool %u, transport %p, inuse=%d\n", 739 dprintk("svc: server %p, pool %u, transport %p, inuse=%d\n",
731 rqstp, pool->sp_id, xprt, 740 rqstp, pool->sp_id, xprt,
732 atomic_read(&xprt->xpt_ref.refcount)); 741 atomic_read(&xprt->xpt_ref.refcount));
@@ -739,11 +748,6 @@ int svc_recv(struct svc_rqst *rqstp, long timeout)
739 dprintk("svc: got len=%d\n", len); 748 dprintk("svc: got len=%d\n", len);
740 } 749 }
741 750
742 if (test_bit(XPT_CLOSE, &xprt->xpt_flags)) {
743 dprintk("svc_recv: found XPT_CLOSE\n");
744 svc_delete_xprt(xprt);
745 }
746
747 /* No data, incomplete (TCP) read, or accept() */ 751 /* No data, incomplete (TCP) read, or accept() */
748 if (len == 0 || len == -EAGAIN) { 752 if (len == 0 || len == -EAGAIN) {
749 rqstp->rq_res.len = 0; 753 rqstp->rq_res.len = 0;
@@ -889,11 +893,8 @@ void svc_delete_xprt(struct svc_xprt *xprt)
889 if (test_bit(XPT_TEMP, &xprt->xpt_flags)) 893 if (test_bit(XPT_TEMP, &xprt->xpt_flags))
890 serv->sv_tmpcnt--; 894 serv->sv_tmpcnt--;
891 895
892 for (dr = svc_deferred_dequeue(xprt); dr; 896 while ((dr = svc_deferred_dequeue(xprt)) != NULL)
893 dr = svc_deferred_dequeue(xprt)) {
894 svc_xprt_put(xprt);
895 kfree(dr); 897 kfree(dr);
896 }
897 898
898 svc_xprt_put(xprt); 899 svc_xprt_put(xprt);
899 spin_unlock_bh(&serv->sv_lock); 900 spin_unlock_bh(&serv->sv_lock);
diff --git a/net/sunrpc/svcauth_unix.c b/net/sunrpc/svcauth_unix.c
index d8c041114497..afdcb0459a83 100644
--- a/net/sunrpc/svcauth_unix.c
+++ b/net/sunrpc/svcauth_unix.c
@@ -15,6 +15,7 @@
15#include <linux/kernel.h> 15#include <linux/kernel.h>
16#define RPCDBG_FACILITY RPCDBG_AUTH 16#define RPCDBG_FACILITY RPCDBG_AUTH
17 17
18#include <linux/sunrpc/clnt.h>
18 19
19/* 20/*
20 * AUTHUNIX and AUTHNULL credentials are both handled here. 21 * AUTHUNIX and AUTHNULL credentials are both handled here.
@@ -187,10 +188,13 @@ static int ip_map_parse(struct cache_detail *cd,
187 * for scratch: */ 188 * for scratch: */
188 char *buf = mesg; 189 char *buf = mesg;
189 int len; 190 int len;
190 int b1, b2, b3, b4, b5, b6, b7, b8;
191 char c;
192 char class[8]; 191 char class[8];
193 struct in6_addr addr; 192 union {
193 struct sockaddr sa;
194 struct sockaddr_in s4;
195 struct sockaddr_in6 s6;
196 } address;
197 struct sockaddr_in6 sin6;
194 int err; 198 int err;
195 199
196 struct ip_map *ipmp; 200 struct ip_map *ipmp;
@@ -209,24 +213,24 @@ static int ip_map_parse(struct cache_detail *cd,
209 len = qword_get(&mesg, buf, mlen); 213 len = qword_get(&mesg, buf, mlen);
210 if (len <= 0) return -EINVAL; 214 if (len <= 0) return -EINVAL;
211 215
212 if (sscanf(buf, "%u.%u.%u.%u%c", &b1, &b2, &b3, &b4, &c) == 4) { 216 if (rpc_pton(buf, len, &address.sa, sizeof(address)) == 0)
213 addr.s6_addr32[0] = 0;
214 addr.s6_addr32[1] = 0;
215 addr.s6_addr32[2] = htonl(0xffff);
216 addr.s6_addr32[3] =
217 htonl((((((b1<<8)|b2)<<8)|b3)<<8)|b4);
218 } else if (sscanf(buf, "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x%c",
219 &b1, &b2, &b3, &b4, &b5, &b6, &b7, &b8, &c) == 8) {
220 addr.s6_addr16[0] = htons(b1);
221 addr.s6_addr16[1] = htons(b2);
222 addr.s6_addr16[2] = htons(b3);
223 addr.s6_addr16[3] = htons(b4);
224 addr.s6_addr16[4] = htons(b5);
225 addr.s6_addr16[5] = htons(b6);
226 addr.s6_addr16[6] = htons(b7);
227 addr.s6_addr16[7] = htons(b8);
228 } else
229 return -EINVAL; 217 return -EINVAL;
218 switch (address.sa.sa_family) {
219 case AF_INET:
220 /* Form a mapped IPv4 address in sin6 */
221 memset(&sin6, 0, sizeof(sin6));
222 sin6.sin6_family = AF_INET6;
223 sin6.sin6_addr.s6_addr32[2] = htonl(0xffff);
224 sin6.sin6_addr.s6_addr32[3] = address.s4.sin_addr.s_addr;
225 break;
226#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
227 case AF_INET6:
228 memcpy(&sin6, &address.s6, sizeof(sin6));
229 break;
230#endif
231 default:
232 return -EINVAL;
233 }
230 234
231 expiry = get_expiry(&mesg); 235 expiry = get_expiry(&mesg);
232 if (expiry ==0) 236 if (expiry ==0)
@@ -243,7 +247,8 @@ static int ip_map_parse(struct cache_detail *cd,
243 } else 247 } else
244 dom = NULL; 248 dom = NULL;
245 249
246 ipmp = ip_map_lookup(class, &addr); 250 /* IPv6 scope IDs are ignored for now */
251 ipmp = ip_map_lookup(class, &sin6.sin6_addr);
247 if (ipmp) { 252 if (ipmp) {
248 err = ip_map_update(ipmp, 253 err = ip_map_update(ipmp,
249 container_of(dom, struct unix_domain, h), 254 container_of(dom, struct unix_domain, h),
@@ -619,7 +624,7 @@ static int unix_gid_show(struct seq_file *m,
619 else 624 else
620 glen = 0; 625 glen = 0;
621 626
622 seq_printf(m, "%d %d:", ug->uid, glen); 627 seq_printf(m, "%u %d:", ug->uid, glen);
623 for (i = 0; i < glen; i++) 628 for (i = 0; i < glen; i++)
624 seq_printf(m, " %d", GROUP_AT(ug->gi, i)); 629 seq_printf(m, " %d", GROUP_AT(ug->gi, i));
625 seq_printf(m, "\n"); 630 seq_printf(m, "\n");
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 870929e08e5d..a29f259204e6 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -968,6 +968,7 @@ static int svc_tcp_recv_record(struct svc_sock *svsk, struct svc_rqst *rqstp)
968 return len; 968 return len;
969 err_delete: 969 err_delete:
970 set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags); 970 set_bit(XPT_CLOSE, &svsk->sk_xprt.xpt_flags);
971 svc_xprt_received(&svsk->sk_xprt);
971 err_again: 972 err_again:
972 return -EAGAIN; 973 return -EAGAIN;
973} 974}
@@ -1357,7 +1358,7 @@ int svc_addsock(struct svc_serv *serv, const int fd, char *name_return,
1357 1358
1358 if (!so) 1359 if (!so)
1359 return err; 1360 return err;
1360 if (so->sk->sk_family != AF_INET) 1361 if ((so->sk->sk_family != PF_INET) && (so->sk->sk_family != PF_INET6))
1361 err = -EAFNOSUPPORT; 1362 err = -EAFNOSUPPORT;
1362 else if (so->sk->sk_protocol != IPPROTO_TCP && 1363 else if (so->sk->sk_protocol != IPPROTO_TCP &&
1363 so->sk->sk_protocol != IPPROTO_UDP) 1364 so->sk->sk_protocol != IPPROTO_UDP)
diff --git a/net/sunrpc/xprtrdma/transport.c b/net/sunrpc/xprtrdma/transport.c
index 7018eef1dcdd..f96c2fe6137b 100644
--- a/net/sunrpc/xprtrdma/transport.c
+++ b/net/sunrpc/xprtrdma/transport.c
@@ -160,16 +160,15 @@ xprt_rdma_format_addresses(struct rpc_xprt *xprt)
160 (void)rpc_ntop(sap, buf, sizeof(buf)); 160 (void)rpc_ntop(sap, buf, sizeof(buf));
161 xprt->address_strings[RPC_DISPLAY_ADDR] = kstrdup(buf, GFP_KERNEL); 161 xprt->address_strings[RPC_DISPLAY_ADDR] = kstrdup(buf, GFP_KERNEL);
162 162
163 (void)snprintf(buf, sizeof(buf), "%u", rpc_get_port(sap)); 163 snprintf(buf, sizeof(buf), "%u", rpc_get_port(sap));
164 xprt->address_strings[RPC_DISPLAY_PORT] = kstrdup(buf, GFP_KERNEL); 164 xprt->address_strings[RPC_DISPLAY_PORT] = kstrdup(buf, GFP_KERNEL);
165 165
166 xprt->address_strings[RPC_DISPLAY_PROTO] = "rdma"; 166 xprt->address_strings[RPC_DISPLAY_PROTO] = "rdma";
167 167
168 (void)snprintf(buf, sizeof(buf), "%02x%02x%02x%02x", 168 snprintf(buf, sizeof(buf), "%08x", ntohl(sin->sin_addr.s_addr));
169 NIPQUAD(sin->sin_addr.s_addr));
170 xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = kstrdup(buf, GFP_KERNEL); 169 xprt->address_strings[RPC_DISPLAY_HEX_ADDR] = kstrdup(buf, GFP_KERNEL);
171 170
172 (void)snprintf(buf, sizeof(buf), "%4hx", rpc_get_port(sap)); 171 snprintf(buf, sizeof(buf), "%4hx", rpc_get_port(sap));
173 xprt->address_strings[RPC_DISPLAY_HEX_PORT] = kstrdup(buf, GFP_KERNEL); 172 xprt->address_strings[RPC_DISPLAY_HEX_PORT] = kstrdup(buf, GFP_KERNEL);
174 173
175 /* netid */ 174 /* netid */
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 3d739e5d15d8..75ab08eac66b 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -297,12 +297,11 @@ static void xs_format_common_peer_addresses(struct rpc_xprt *xprt)
297 switch (sap->sa_family) { 297 switch (sap->sa_family) {
298 case AF_INET: 298 case AF_INET:
299 sin = xs_addr_in(xprt); 299 sin = xs_addr_in(xprt);
300 (void)snprintf(buf, sizeof(buf), "%02x%02x%02x%02x", 300 snprintf(buf, sizeof(buf), "%08x", ntohl(sin->sin_addr.s_addr));
301 NIPQUAD(sin->sin_addr.s_addr));
302 break; 301 break;
303 case AF_INET6: 302 case AF_INET6:
304 sin6 = xs_addr_in6(xprt); 303 sin6 = xs_addr_in6(xprt);
305 (void)snprintf(buf, sizeof(buf), "%pi6", &sin6->sin6_addr); 304 snprintf(buf, sizeof(buf), "%pi6", &sin6->sin6_addr);
306 break; 305 break;
307 default: 306 default:
308 BUG(); 307 BUG();
@@ -315,10 +314,10 @@ static void xs_format_common_peer_ports(struct rpc_xprt *xprt)
315 struct sockaddr *sap = xs_addr(xprt); 314 struct sockaddr *sap = xs_addr(xprt);
316 char buf[128]; 315 char buf[128];
317 316
318 (void)snprintf(buf, sizeof(buf), "%u", rpc_get_port(sap)); 317 snprintf(buf, sizeof(buf), "%u", rpc_get_port(sap));
319 xprt->address_strings[RPC_DISPLAY_PORT] = kstrdup(buf, GFP_KERNEL); 318 xprt->address_strings[RPC_DISPLAY_PORT] = kstrdup(buf, GFP_KERNEL);
320 319
321 (void)snprintf(buf, sizeof(buf), "%4hx", rpc_get_port(sap)); 320 snprintf(buf, sizeof(buf), "%4hx", rpc_get_port(sap));
322 xprt->address_strings[RPC_DISPLAY_HEX_PORT] = kstrdup(buf, GFP_KERNEL); 321 xprt->address_strings[RPC_DISPLAY_HEX_PORT] = kstrdup(buf, GFP_KERNEL);
323} 322}
324 323
@@ -1912,6 +1911,11 @@ static void xs_tcp_setup_socket(struct rpc_xprt *xprt,
1912 case -EALREADY: 1911 case -EALREADY:
1913 xprt_clear_connecting(xprt); 1912 xprt_clear_connecting(xprt);
1914 return; 1913 return;
1914 case -EINVAL:
1915 /* Happens, for instance, if the user specified a link
1916 * local IPv6 address without a scope-id.
1917 */
1918 goto out;
1915 } 1919 }
1916out_eagain: 1920out_eagain:
1917 status = -EAGAIN; 1921 status = -EAGAIN;
@@ -2100,7 +2104,7 @@ static void xs_tcp_print_stats(struct rpc_xprt *xprt, struct seq_file *seq)
2100 * we allocate pages instead doing a kmalloc like rpc_malloc is because we want 2104 * we allocate pages instead doing a kmalloc like rpc_malloc is because we want
2101 * to use the server side send routines. 2105 * to use the server side send routines.
2102 */ 2106 */
2103void *bc_malloc(struct rpc_task *task, size_t size) 2107static void *bc_malloc(struct rpc_task *task, size_t size)
2104{ 2108{
2105 struct page *page; 2109 struct page *page;
2106 struct rpc_buffer *buf; 2110 struct rpc_buffer *buf;
@@ -2120,7 +2124,7 @@ void *bc_malloc(struct rpc_task *task, size_t size)
2120/* 2124/*
2121 * Free the space allocated in the bc_alloc routine 2125 * Free the space allocated in the bc_alloc routine
2122 */ 2126 */
2123void bc_free(void *buffer) 2127static void bc_free(void *buffer)
2124{ 2128{
2125 struct rpc_buffer *buf; 2129 struct rpc_buffer *buf;
2126 2130
diff --git a/net/tipc/bcast.c b/net/tipc/bcast.c
index a3bfd4064912..90a051912c03 100644
--- a/net/tipc/bcast.c
+++ b/net/tipc/bcast.c
@@ -558,10 +558,7 @@ static int tipc_bcbearer_send(struct sk_buff *buf,
558 struct tipc_bearer *unused1, 558 struct tipc_bearer *unused1,
559 struct tipc_media_addr *unused2) 559 struct tipc_media_addr *unused2)
560{ 560{
561 static int send_count = 0;
562
563 int bp_index; 561 int bp_index;
564 int swap_time;
565 562
566 /* Prepare buffer for broadcasting (if first time trying to send it) */ 563 /* Prepare buffer for broadcasting (if first time trying to send it) */
567 564
@@ -575,11 +572,6 @@ static int tipc_bcbearer_send(struct sk_buff *buf,
575 msg_set_mc_netid(msg, tipc_net_id); 572 msg_set_mc_netid(msg, tipc_net_id);
576 } 573 }
577 574
578 /* Determine if bearer pairs should be swapped following this attempt */
579
580 if ((swap_time = (++send_count >= 10)))
581 send_count = 0;
582
583 /* Send buffer over bearers until all targets reached */ 575 /* Send buffer over bearers until all targets reached */
584 576
585 bcbearer->remains = tipc_cltr_bcast_nodes; 577 bcbearer->remains = tipc_cltr_bcast_nodes;
@@ -595,21 +587,22 @@ static int tipc_bcbearer_send(struct sk_buff *buf,
595 if (bcbearer->remains_new.count == bcbearer->remains.count) 587 if (bcbearer->remains_new.count == bcbearer->remains.count)
596 continue; /* bearer pair doesn't add anything */ 588 continue; /* bearer pair doesn't add anything */
597 589
598 if (!p->publ.blocked && 590 if (p->publ.blocked ||
599 !p->media->send_msg(buf, &p->publ, &p->media->bcast_addr)) { 591 p->media->send_msg(buf, &p->publ, &p->media->bcast_addr)) {
600 if (swap_time && s && !s->publ.blocked) 592 /* unable to send on primary bearer */
601 goto swap; 593 if (!s || s->publ.blocked ||
602 else 594 s->media->send_msg(buf, &s->publ,
603 goto update; 595 &s->media->bcast_addr)) {
596 /* unable to send on either bearer */
597 continue;
598 }
599 }
600
601 if (s) {
602 bcbearer->bpairs[bp_index].primary = s;
603 bcbearer->bpairs[bp_index].secondary = p;
604 } 604 }
605 605
606 if (!s || s->publ.blocked ||
607 s->media->send_msg(buf, &s->publ, &s->media->bcast_addr))
608 continue; /* unable to send using bearer pair */
609swap:
610 bcbearer->bpairs[bp_index].primary = s;
611 bcbearer->bpairs[bp_index].secondary = p;
612update:
613 if (bcbearer->remains_new.count == 0) 606 if (bcbearer->remains_new.count == 0)
614 return 0; 607 return 0;
615 608
diff --git a/net/tipc/bearer.c b/net/tipc/bearer.c
index 327011fcc407..78091375ca12 100644
--- a/net/tipc/bearer.c
+++ b/net/tipc/bearer.c
@@ -45,10 +45,10 @@
45 45
46#define MAX_ADDR_STR 32 46#define MAX_ADDR_STR 32
47 47
48static struct media *media_list = NULL; 48static struct media media_list[MAX_MEDIA];
49static u32 media_count = 0; 49static u32 media_count = 0;
50 50
51struct bearer *tipc_bearers = NULL; 51struct bearer tipc_bearers[MAX_BEARERS];
52 52
53/** 53/**
54 * media_name_valid - validate media name 54 * media_name_valid - validate media name
@@ -108,9 +108,11 @@ int tipc_register_media(u32 media_type,
108 int res = -EINVAL; 108 int res = -EINVAL;
109 109
110 write_lock_bh(&tipc_net_lock); 110 write_lock_bh(&tipc_net_lock);
111 if (!media_list)
112 goto exit;
113 111
112 if (tipc_mode != TIPC_NET_MODE) {
113 warn("Media <%s> rejected, not in networked mode yet\n", name);
114 goto exit;
115 }
114 if (!media_name_valid(name)) { 116 if (!media_name_valid(name)) {
115 warn("Media <%s> rejected, illegal name\n", name); 117 warn("Media <%s> rejected, illegal name\n", name);
116 goto exit; 118 goto exit;
@@ -660,33 +662,10 @@ int tipc_disable_bearer(const char *name)
660 662
661 663
662 664
663int tipc_bearer_init(void)
664{
665 int res;
666
667 write_lock_bh(&tipc_net_lock);
668 tipc_bearers = kcalloc(MAX_BEARERS, sizeof(struct bearer), GFP_ATOMIC);
669 media_list = kcalloc(MAX_MEDIA, sizeof(struct media), GFP_ATOMIC);
670 if (tipc_bearers && media_list) {
671 res = 0;
672 } else {
673 kfree(tipc_bearers);
674 kfree(media_list);
675 tipc_bearers = NULL;
676 media_list = NULL;
677 res = -ENOMEM;
678 }
679 write_unlock_bh(&tipc_net_lock);
680 return res;
681}
682
683void tipc_bearer_stop(void) 665void tipc_bearer_stop(void)
684{ 666{
685 u32 i; 667 u32 i;
686 668
687 if (!tipc_bearers)
688 return;
689
690 for (i = 0; i < MAX_BEARERS; i++) { 669 for (i = 0; i < MAX_BEARERS; i++) {
691 if (tipc_bearers[i].active) 670 if (tipc_bearers[i].active)
692 tipc_bearers[i].publ.blocked = 1; 671 tipc_bearers[i].publ.blocked = 1;
@@ -695,10 +674,6 @@ void tipc_bearer_stop(void)
695 if (tipc_bearers[i].active) 674 if (tipc_bearers[i].active)
696 bearer_disable(tipc_bearers[i].publ.name); 675 bearer_disable(tipc_bearers[i].publ.name);
697 } 676 }
698 kfree(tipc_bearers);
699 kfree(media_list);
700 tipc_bearers = NULL;
701 media_list = NULL;
702 media_count = 0; 677 media_count = 0;
703} 678}
704 679
diff --git a/net/tipc/bearer.h b/net/tipc/bearer.h
index ca5734892713..000228e93f9e 100644
--- a/net/tipc/bearer.h
+++ b/net/tipc/bearer.h
@@ -114,7 +114,7 @@ struct bearer_name {
114 114
115struct link; 115struct link;
116 116
117extern struct bearer *tipc_bearers; 117extern struct bearer tipc_bearers[];
118 118
119void tipc_media_addr_printf(struct print_buf *pb, struct tipc_media_addr *a); 119void tipc_media_addr_printf(struct print_buf *pb, struct tipc_media_addr *a);
120struct sk_buff *tipc_media_get_names(void); 120struct sk_buff *tipc_media_get_names(void);
diff --git a/net/tipc/link.c b/net/tipc/link.c
index 6f50f6423f63..c76e82e5f982 100644
--- a/net/tipc/link.c
+++ b/net/tipc/link.c
@@ -877,7 +877,7 @@ static void link_state_event(struct link *l_ptr, unsigned event)
877 case TIMEOUT_EVT: 877 case TIMEOUT_EVT:
878 dbg_link("TIM "); 878 dbg_link("TIM ");
879 if (l_ptr->next_in_no != l_ptr->checkpoint) { 879 if (l_ptr->next_in_no != l_ptr->checkpoint) {
880 dbg_link("-> WW \n"); 880 dbg_link("-> WW\n");
881 l_ptr->state = WORKING_WORKING; 881 l_ptr->state = WORKING_WORKING;
882 l_ptr->fsm_msg_cnt = 0; 882 l_ptr->fsm_msg_cnt = 0;
883 l_ptr->checkpoint = l_ptr->next_in_no; 883 l_ptr->checkpoint = l_ptr->next_in_no;
@@ -934,7 +934,7 @@ static void link_state_event(struct link *l_ptr, unsigned event)
934 link_set_timer(l_ptr, cont_intv); 934 link_set_timer(l_ptr, cont_intv);
935 break; 935 break;
936 case RESET_MSG: 936 case RESET_MSG:
937 dbg_link("RES \n"); 937 dbg_link("RES\n");
938 dbg_link(" -> RR\n"); 938 dbg_link(" -> RR\n");
939 l_ptr->state = RESET_RESET; 939 l_ptr->state = RESET_RESET;
940 l_ptr->fsm_msg_cnt = 0; 940 l_ptr->fsm_msg_cnt = 0;
@@ -947,7 +947,7 @@ static void link_state_event(struct link *l_ptr, unsigned event)
947 l_ptr->started = 1; 947 l_ptr->started = 1;
948 /* fall through */ 948 /* fall through */
949 case TIMEOUT_EVT: 949 case TIMEOUT_EVT:
950 dbg_link("TIM \n"); 950 dbg_link("TIM\n");
951 tipc_link_send_proto_msg(l_ptr, RESET_MSG, 0, 0, 0, 0, 0); 951 tipc_link_send_proto_msg(l_ptr, RESET_MSG, 0, 0, 0, 0, 0);
952 l_ptr->fsm_msg_cnt++; 952 l_ptr->fsm_msg_cnt++;
953 link_set_timer(l_ptr, cont_intv); 953 link_set_timer(l_ptr, cont_intv);
@@ -1553,7 +1553,7 @@ u32 tipc_link_push_packet(struct link *l_ptr)
1553 1553
1554 /* Continue retransmission now, if there is anything: */ 1554 /* Continue retransmission now, if there is anything: */
1555 1555
1556 if (r_q_size && buf && !skb_cloned(buf)) { 1556 if (r_q_size && buf) {
1557 msg_set_ack(buf_msg(buf), mod(l_ptr->next_in_no - 1)); 1557 msg_set_ack(buf_msg(buf), mod(l_ptr->next_in_no - 1));
1558 msg_set_bcast_ack(buf_msg(buf), l_ptr->owner->bclink.last_in); 1558 msg_set_bcast_ack(buf_msg(buf), l_ptr->owner->bclink.last_in);
1559 if (tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr)) { 1559 if (tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr)) {
@@ -1722,15 +1722,16 @@ void tipc_link_retransmit(struct link *l_ptr, struct sk_buff *buf,
1722 dbg("Retransmitting %u in link %x\n", retransmits, l_ptr); 1722 dbg("Retransmitting %u in link %x\n", retransmits, l_ptr);
1723 1723
1724 if (tipc_bearer_congested(l_ptr->b_ptr, l_ptr)) { 1724 if (tipc_bearer_congested(l_ptr->b_ptr, l_ptr)) {
1725 if (!skb_cloned(buf)) { 1725 if (l_ptr->retransm_queue_size == 0) {
1726 msg_dbg(msg, ">NO_RETR->BCONG>"); 1726 msg_dbg(msg, ">NO_RETR->BCONG>");
1727 dbg_print_link(l_ptr, " "); 1727 dbg_print_link(l_ptr, " ");
1728 l_ptr->retransm_queue_head = msg_seqno(msg); 1728 l_ptr->retransm_queue_head = msg_seqno(msg);
1729 l_ptr->retransm_queue_size = retransmits; 1729 l_ptr->retransm_queue_size = retransmits;
1730 return;
1731 } else { 1730 } else {
1732 /* Don't retransmit if driver already has the buffer */ 1731 err("Unexpected retransmit on link %s (qsize=%d)\n",
1732 l_ptr->name, l_ptr->retransm_queue_size);
1733 } 1733 }
1734 return;
1734 } else { 1735 } else {
1735 /* Detect repeated retransmit failures on uncongested bearer */ 1736 /* Detect repeated retransmit failures on uncongested bearer */
1736 1737
@@ -1745,7 +1746,7 @@ void tipc_link_retransmit(struct link *l_ptr, struct sk_buff *buf,
1745 } 1746 }
1746 } 1747 }
1747 1748
1748 while (retransmits && (buf != l_ptr->next_out) && buf && !skb_cloned(buf)) { 1749 while (retransmits && (buf != l_ptr->next_out) && buf) {
1749 msg = buf_msg(buf); 1750 msg = buf_msg(buf);
1750 msg_set_ack(msg, mod(l_ptr->next_in_no - 1)); 1751 msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1751 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in); 1752 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
@@ -1882,6 +1883,15 @@ void tipc_recv_msg(struct sk_buff *head, struct tipc_bearer *tb_ptr)
1882 (msg_destnode(msg) != tipc_own_addr))) 1883 (msg_destnode(msg) != tipc_own_addr)))
1883 goto cont; 1884 goto cont;
1884 1885
1886 /* Discard non-routeable messages destined for another node */
1887
1888 if (unlikely(!msg_isdata(msg) &&
1889 (msg_destnode(msg) != tipc_own_addr))) {
1890 if ((msg_user(msg) != CONN_MANAGER) &&
1891 (msg_user(msg) != MSG_FRAGMENTER))
1892 goto cont;
1893 }
1894
1885 /* Locate unicast link endpoint that should handle message */ 1895 /* Locate unicast link endpoint that should handle message */
1886 1896
1887 n_ptr = tipc_node_find(msg_prevnode(msg)); 1897 n_ptr = tipc_node_find(msg_prevnode(msg));
@@ -3285,7 +3295,7 @@ static void link_dump_rec_queue(struct link *l_ptr)
3285 info("buffer %x invalid\n", crs); 3295 info("buffer %x invalid\n", crs);
3286 return; 3296 return;
3287 } 3297 }
3288 msg_dbg(buf_msg(crs), "In rec queue: \n"); 3298 msg_dbg(buf_msg(crs), "In rec queue:\n");
3289 crs = crs->next; 3299 crs = crs->next;
3290 } 3300 }
3291} 3301}
diff --git a/net/tipc/net.c b/net/tipc/net.c
index 7906608bf510..79ce8fa2b77a 100644
--- a/net/tipc/net.c
+++ b/net/tipc/net.c
@@ -116,7 +116,8 @@
116*/ 116*/
117 117
118DEFINE_RWLOCK(tipc_net_lock); 118DEFINE_RWLOCK(tipc_net_lock);
119struct network tipc_net = { NULL }; 119struct _zone *tipc_zones[256] = { NULL, };
120struct network tipc_net = { tipc_zones };
120 121
121struct tipc_node *tipc_net_select_remote_node(u32 addr, u32 ref) 122struct tipc_node *tipc_net_select_remote_node(u32 addr, u32 ref)
122{ 123{
@@ -158,28 +159,12 @@ void tipc_net_send_external_routes(u32 dest)
158 } 159 }
159} 160}
160 161
161static int net_init(void)
162{
163 memset(&tipc_net, 0, sizeof(tipc_net));
164 tipc_net.zones = kcalloc(tipc_max_zones + 1, sizeof(struct _zone *), GFP_ATOMIC);
165 if (!tipc_net.zones) {
166 return -ENOMEM;
167 }
168 return 0;
169}
170
171static void net_stop(void) 162static void net_stop(void)
172{ 163{
173 u32 z_num; 164 u32 z_num;
174 165
175 if (!tipc_net.zones) 166 for (z_num = 1; z_num <= tipc_max_zones; z_num++)
176 return;
177
178 for (z_num = 1; z_num <= tipc_max_zones; z_num++) {
179 tipc_zone_delete(tipc_net.zones[z_num]); 167 tipc_zone_delete(tipc_net.zones[z_num]);
180 }
181 kfree(tipc_net.zones);
182 tipc_net.zones = NULL;
183} 168}
184 169
185static void net_route_named_msg(struct sk_buff *buf) 170static void net_route_named_msg(struct sk_buff *buf)
@@ -282,9 +267,7 @@ int tipc_net_start(u32 addr)
282 tipc_named_reinit(); 267 tipc_named_reinit();
283 tipc_port_reinit(); 268 tipc_port_reinit();
284 269
285 if ((res = tipc_bearer_init()) || 270 if ((res = tipc_cltr_init()) ||
286 (res = net_init()) ||
287 (res = tipc_cltr_init()) ||
288 (res = tipc_bclink_init())) { 271 (res = tipc_bclink_init())) {
289 return res; 272 return res;
290 } 273 }
@@ -308,6 +291,6 @@ void tipc_net_stop(void)
308 tipc_bclink_stop(); 291 tipc_bclink_stop();
309 net_stop(); 292 net_stop();
310 write_unlock_bh(&tipc_net_lock); 293 write_unlock_bh(&tipc_net_lock);
311 info("Left network mode \n"); 294 info("Left network mode\n");
312} 295}
313 296
diff --git a/net/tipc/node.c b/net/tipc/node.c
index 2c24e7d6d950..17cc394f424f 100644
--- a/net/tipc/node.c
+++ b/net/tipc/node.c
@@ -278,7 +278,7 @@ struct tipc_node *tipc_node_attach_link(struct link *l_ptr)
278 n_ptr->link_cnt++; 278 n_ptr->link_cnt++;
279 return n_ptr; 279 return n_ptr;
280 } 280 }
281 err("Attempt to establish second link on <%s> to %s \n", 281 err("Attempt to establish second link on <%s> to %s\n",
282 l_ptr->b_ptr->publ.name, 282 l_ptr->b_ptr->publ.name,
283 addr_string_fill(addr_string, l_ptr->addr)); 283 addr_string_fill(addr_string, l_ptr->addr));
284 } 284 }
diff --git a/net/tipc/ref.c b/net/tipc/ref.c
index 414fc34b8bea..8dea66500cf5 100644
--- a/net/tipc/ref.c
+++ b/net/tipc/ref.c
@@ -153,11 +153,11 @@ void tipc_ref_table_stop(void)
153 153
154u32 tipc_ref_acquire(void *object, spinlock_t **lock) 154u32 tipc_ref_acquire(void *object, spinlock_t **lock)
155{ 155{
156 struct reference *entry;
157 u32 index; 156 u32 index;
158 u32 index_mask; 157 u32 index_mask;
159 u32 next_plus_upper; 158 u32 next_plus_upper;
160 u32 ref; 159 u32 ref;
160 struct reference *entry = NULL;
161 161
162 if (!object) { 162 if (!object) {
163 err("Attempt to acquire reference to non-existent object\n"); 163 err("Attempt to acquire reference to non-existent object\n");
@@ -175,30 +175,36 @@ u32 tipc_ref_acquire(void *object, spinlock_t **lock)
175 index = tipc_ref_table.first_free; 175 index = tipc_ref_table.first_free;
176 entry = &(tipc_ref_table.entries[index]); 176 entry = &(tipc_ref_table.entries[index]);
177 index_mask = tipc_ref_table.index_mask; 177 index_mask = tipc_ref_table.index_mask;
178 /* take lock in case a previous user of entry still holds it */
179 spin_lock_bh(&entry->lock);
180 next_plus_upper = entry->ref; 178 next_plus_upper = entry->ref;
181 tipc_ref_table.first_free = next_plus_upper & index_mask; 179 tipc_ref_table.first_free = next_plus_upper & index_mask;
182 ref = (next_plus_upper & ~index_mask) + index; 180 ref = (next_plus_upper & ~index_mask) + index;
183 entry->ref = ref;
184 entry->object = object;
185 *lock = &entry->lock;
186 } 181 }
187 else if (tipc_ref_table.init_point < tipc_ref_table.capacity) { 182 else if (tipc_ref_table.init_point < tipc_ref_table.capacity) {
188 index = tipc_ref_table.init_point++; 183 index = tipc_ref_table.init_point++;
189 entry = &(tipc_ref_table.entries[index]); 184 entry = &(tipc_ref_table.entries[index]);
190 spin_lock_init(&entry->lock); 185 spin_lock_init(&entry->lock);
191 spin_lock_bh(&entry->lock);
192 ref = tipc_ref_table.start_mask + index; 186 ref = tipc_ref_table.start_mask + index;
193 entry->ref = ref;
194 entry->object = object;
195 *lock = &entry->lock;
196 } 187 }
197 else { 188 else {
198 ref = 0; 189 ref = 0;
199 } 190 }
200 write_unlock_bh(&ref_table_lock); 191 write_unlock_bh(&ref_table_lock);
201 192
193 /*
194 * Grab the lock so no one else can modify this entry
195 * While we assign its ref value & object pointer
196 */
197 if (entry) {
198 spin_lock_bh(&entry->lock);
199 entry->ref = ref;
200 entry->object = object;
201 *lock = &entry->lock;
202 /*
203 * keep it locked, the caller is responsible
204 * for unlocking this when they're done with it
205 */
206 }
207
202 return ref; 208 return ref;
203} 209}
204 210
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index 1ea64f09cc45..4b235fc1c70f 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -1322,8 +1322,10 @@ static u32 dispatch(struct tipc_port *tport, struct sk_buff *buf)
1322 if (!sock_owned_by_user(sk)) { 1322 if (!sock_owned_by_user(sk)) {
1323 res = filter_rcv(sk, buf); 1323 res = filter_rcv(sk, buf);
1324 } else { 1324 } else {
1325 sk_add_backlog(sk, buf); 1325 if (sk_add_backlog(sk, buf))
1326 res = TIPC_OK; 1326 res = TIPC_ERR_OVERLOAD;
1327 else
1328 res = TIPC_OK;
1327 } 1329 }
1328 bh_unlock_sock(sk); 1330 bh_unlock_sock(sk);
1329 1331
diff --git a/net/tipc/subscr.c b/net/tipc/subscr.c
index ac91f0dfa144..ff123e56114a 100644
--- a/net/tipc/subscr.c
+++ b/net/tipc/subscr.c
@@ -76,19 +76,6 @@ struct top_srv {
76static struct top_srv topsrv = { 0 }; 76static struct top_srv topsrv = { 0 };
77 77
78/** 78/**
79 * htohl - convert value to endianness used by destination
80 * @in: value to convert
81 * @swap: non-zero if endianness must be reversed
82 *
83 * Returns converted value
84 */
85
86static u32 htohl(u32 in, int swap)
87{
88 return swap ? swab32(in) : in;
89}
90
91/**
92 * subscr_send_event - send a message containing a tipc_event to the subscriber 79 * subscr_send_event - send a message containing a tipc_event to the subscriber
93 * 80 *
94 * Note: Must not hold subscriber's server port lock, since tipc_send() will 81 * Note: Must not hold subscriber's server port lock, since tipc_send() will
@@ -107,11 +94,11 @@ static void subscr_send_event(struct subscription *sub,
107 msg_sect.iov_base = (void *)&sub->evt; 94 msg_sect.iov_base = (void *)&sub->evt;
108 msg_sect.iov_len = sizeof(struct tipc_event); 95 msg_sect.iov_len = sizeof(struct tipc_event);
109 96
110 sub->evt.event = htohl(event, sub->swap); 97 sub->evt.event = htonl(event);
111 sub->evt.found_lower = htohl(found_lower, sub->swap); 98 sub->evt.found_lower = htonl(found_lower);
112 sub->evt.found_upper = htohl(found_upper, sub->swap); 99 sub->evt.found_upper = htonl(found_upper);
113 sub->evt.port.ref = htohl(port_ref, sub->swap); 100 sub->evt.port.ref = htonl(port_ref);
114 sub->evt.port.node = htohl(node, sub->swap); 101 sub->evt.port.node = htonl(node);
115 tipc_send(sub->server_ref, 1, &msg_sect); 102 tipc_send(sub->server_ref, 1, &msg_sect);
116} 103}
117 104
@@ -287,16 +274,23 @@ static void subscr_cancel(struct tipc_subscr *s,
287{ 274{
288 struct subscription *sub; 275 struct subscription *sub;
289 struct subscription *sub_temp; 276 struct subscription *sub_temp;
277 __u32 type, lower, upper;
290 int found = 0; 278 int found = 0;
291 279
292 /* Find first matching subscription, exit if not found */ 280 /* Find first matching subscription, exit if not found */
293 281
282 type = ntohl(s->seq.type);
283 lower = ntohl(s->seq.lower);
284 upper = ntohl(s->seq.upper);
285
294 list_for_each_entry_safe(sub, sub_temp, &subscriber->subscription_list, 286 list_for_each_entry_safe(sub, sub_temp, &subscriber->subscription_list,
295 subscription_list) { 287 subscription_list) {
296 if (!memcmp(s, &sub->evt.s, sizeof(struct tipc_subscr))) { 288 if ((type == sub->seq.type) &&
297 found = 1; 289 (lower == sub->seq.lower) &&
298 break; 290 (upper == sub->seq.upper)) {
299 } 291 found = 1;
292 break;
293 }
300 } 294 }
301 if (!found) 295 if (!found)
302 return; 296 return;
@@ -325,16 +319,10 @@ static struct subscription *subscr_subscribe(struct tipc_subscr *s,
325 struct subscriber *subscriber) 319 struct subscriber *subscriber)
326{ 320{
327 struct subscription *sub; 321 struct subscription *sub;
328 int swap;
329
330 /* Determine subscriber's endianness */
331
332 swap = !(s->filter & (TIPC_SUB_PORTS | TIPC_SUB_SERVICE));
333 322
334 /* Detect & process a subscription cancellation request */ 323 /* Detect & process a subscription cancellation request */
335 324
336 if (s->filter & htohl(TIPC_SUB_CANCEL, swap)) { 325 if (ntohl(s->filter) & TIPC_SUB_CANCEL) {
337 s->filter &= ~htohl(TIPC_SUB_CANCEL, swap);
338 subscr_cancel(s, subscriber); 326 subscr_cancel(s, subscriber);
339 return NULL; 327 return NULL;
340 } 328 }
@@ -359,11 +347,11 @@ static struct subscription *subscr_subscribe(struct tipc_subscr *s,
359 347
360 /* Initialize subscription object */ 348 /* Initialize subscription object */
361 349
362 sub->seq.type = htohl(s->seq.type, swap); 350 sub->seq.type = ntohl(s->seq.type);
363 sub->seq.lower = htohl(s->seq.lower, swap); 351 sub->seq.lower = ntohl(s->seq.lower);
364 sub->seq.upper = htohl(s->seq.upper, swap); 352 sub->seq.upper = ntohl(s->seq.upper);
365 sub->timeout = htohl(s->timeout, swap); 353 sub->timeout = ntohl(s->timeout);
366 sub->filter = htohl(s->filter, swap); 354 sub->filter = ntohl(s->filter);
367 if ((!(sub->filter & TIPC_SUB_PORTS) == 355 if ((!(sub->filter & TIPC_SUB_PORTS) ==
368 !(sub->filter & TIPC_SUB_SERVICE)) || 356 !(sub->filter & TIPC_SUB_SERVICE)) ||
369 (sub->seq.lower > sub->seq.upper)) { 357 (sub->seq.lower > sub->seq.upper)) {
@@ -376,7 +364,6 @@ static struct subscription *subscr_subscribe(struct tipc_subscr *s,
376 INIT_LIST_HEAD(&sub->nameseq_list); 364 INIT_LIST_HEAD(&sub->nameseq_list);
377 list_add(&sub->subscription_list, &subscriber->subscription_list); 365 list_add(&sub->subscription_list, &subscriber->subscription_list);
378 sub->server_ref = subscriber->port_ref; 366 sub->server_ref = subscriber->port_ref;
379 sub->swap = swap;
380 memcpy(&sub->evt.s, s, sizeof(struct tipc_subscr)); 367 memcpy(&sub->evt.s, s, sizeof(struct tipc_subscr));
381 atomic_inc(&topsrv.subscription_count); 368 atomic_inc(&topsrv.subscription_count);
382 if (sub->timeout != TIPC_WAIT_FOREVER) { 369 if (sub->timeout != TIPC_WAIT_FOREVER) {
diff --git a/net/tipc/subscr.h b/net/tipc/subscr.h
index 45d89bf4d202..c20f496d95b2 100644
--- a/net/tipc/subscr.h
+++ b/net/tipc/subscr.h
@@ -53,7 +53,6 @@ typedef void (*tipc_subscr_event) (struct subscription *sub,
53 * @nameseq_list: adjacent subscriptions in name sequence's subscription list 53 * @nameseq_list: adjacent subscriptions in name sequence's subscription list
54 * @subscription_list: adjacent subscriptions in subscriber's subscription list 54 * @subscription_list: adjacent subscriptions in subscriber's subscription list
55 * @server_ref: object reference of server port associated with subscription 55 * @server_ref: object reference of server port associated with subscription
56 * @swap: indicates if subscriber uses opposite endianness in its messages
57 * @evt: template for events generated by subscription 56 * @evt: template for events generated by subscription
58 */ 57 */
59 58
@@ -66,7 +65,6 @@ struct subscription {
66 struct list_head nameseq_list; 65 struct list_head nameseq_list;
67 struct list_head subscription_list; 66 struct list_head subscription_list;
68 u32 server_ref; 67 u32 server_ref;
69 int swap;
70 struct tipc_event evt; 68 struct tipc_event evt;
71}; 69};
72 70
diff --git a/net/wimax/op-reset.c b/net/wimax/op-reset.c
index 4dc82a54ba30..68bedf3e5443 100644
--- a/net/wimax/op-reset.c
+++ b/net/wimax/op-reset.c
@@ -110,7 +110,6 @@ int wimax_gnl_doit_reset(struct sk_buff *skb, struct genl_info *info)
110{ 110{
111 int result, ifindex; 111 int result, ifindex;
112 struct wimax_dev *wimax_dev; 112 struct wimax_dev *wimax_dev;
113 struct device *dev;
114 113
115 d_fnstart(3, NULL, "(skb %p info %p)\n", skb, info); 114 d_fnstart(3, NULL, "(skb %p info %p)\n", skb, info);
116 result = -ENODEV; 115 result = -ENODEV;
@@ -123,7 +122,6 @@ int wimax_gnl_doit_reset(struct sk_buff *skb, struct genl_info *info)
123 wimax_dev = wimax_dev_get_by_genl_info(info, ifindex); 122 wimax_dev = wimax_dev_get_by_genl_info(info, ifindex);
124 if (wimax_dev == NULL) 123 if (wimax_dev == NULL)
125 goto error_no_wimax_dev; 124 goto error_no_wimax_dev;
126 dev = wimax_dev_to_dev(wimax_dev);
127 /* Execute the operation and send the result back to user space */ 125 /* Execute the operation and send the result back to user space */
128 result = wimax_reset(wimax_dev); 126 result = wimax_reset(wimax_dev);
129 dev_put(wimax_dev->net_dev); 127 dev_put(wimax_dev->net_dev);
diff --git a/net/wimax/op-state-get.c b/net/wimax/op-state-get.c
index 11ad3356eb56..aff8776e2d41 100644
--- a/net/wimax/op-state-get.c
+++ b/net/wimax/op-state-get.c
@@ -53,7 +53,6 @@ int wimax_gnl_doit_state_get(struct sk_buff *skb, struct genl_info *info)
53{ 53{
54 int result, ifindex; 54 int result, ifindex;
55 struct wimax_dev *wimax_dev; 55 struct wimax_dev *wimax_dev;
56 struct device *dev;
57 56
58 d_fnstart(3, NULL, "(skb %p info %p)\n", skb, info); 57 d_fnstart(3, NULL, "(skb %p info %p)\n", skb, info);
59 result = -ENODEV; 58 result = -ENODEV;
@@ -66,7 +65,6 @@ int wimax_gnl_doit_state_get(struct sk_buff *skb, struct genl_info *info)
66 wimax_dev = wimax_dev_get_by_genl_info(info, ifindex); 65 wimax_dev = wimax_dev_get_by_genl_info(info, ifindex);
67 if (wimax_dev == NULL) 66 if (wimax_dev == NULL)
68 goto error_no_wimax_dev; 67 goto error_no_wimax_dev;
69 dev = wimax_dev_to_dev(wimax_dev);
70 /* Execute the operation and send the result back to user space */ 68 /* Execute the operation and send the result back to user space */
71 result = wimax_state_get(wimax_dev); 69 result = wimax_state_get(wimax_dev);
72 dev_put(wimax_dev->net_dev); 70 dev_put(wimax_dev->net_dev);
diff --git a/net/x25/x25_dev.c b/net/x25/x25_dev.c
index 3e1efe534645..52e304212241 100644
--- a/net/x25/x25_dev.c
+++ b/net/x25/x25_dev.c
@@ -53,7 +53,7 @@ static int x25_receive_data(struct sk_buff *skb, struct x25_neigh *nb)
53 if (!sock_owned_by_user(sk)) { 53 if (!sock_owned_by_user(sk)) {
54 queued = x25_process_rx_frame(sk, skb); 54 queued = x25_process_rx_frame(sk, skb);
55 } else { 55 } else {
56 sk_add_backlog(sk, skb); 56 queued = !sk_add_backlog(sk, skb);
57 } 57 }
58 bh_unlock_sock(sk); 58 bh_unlock_sock(sk);
59 sock_put(sk); 59 sock_put(sk);
diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c
index 34a5ef8316e7..843e066649cb 100644
--- a/net/xfrm/xfrm_policy.c
+++ b/net/xfrm/xfrm_policy.c
@@ -1372,7 +1372,8 @@ static inline int xfrm_init_path(struct xfrm_dst *path, struct dst_entry *dst,
1372 return err; 1372 return err;
1373} 1373}
1374 1374
1375static inline int xfrm_fill_dst(struct xfrm_dst *xdst, struct net_device *dev) 1375static inline int xfrm_fill_dst(struct xfrm_dst *xdst, struct net_device *dev,
1376 struct flowi *fl)
1376{ 1377{
1377 struct xfrm_policy_afinfo *afinfo = 1378 struct xfrm_policy_afinfo *afinfo =
1378 xfrm_policy_get_afinfo(xdst->u.dst.ops->family); 1379 xfrm_policy_get_afinfo(xdst->u.dst.ops->family);
@@ -1381,7 +1382,7 @@ static inline int xfrm_fill_dst(struct xfrm_dst *xdst, struct net_device *dev)
1381 if (!afinfo) 1382 if (!afinfo)
1382 return -EINVAL; 1383 return -EINVAL;
1383 1384
1384 err = afinfo->fill_dst(xdst, dev); 1385 err = afinfo->fill_dst(xdst, dev, fl);
1385 1386
1386 xfrm_policy_put_afinfo(afinfo); 1387 xfrm_policy_put_afinfo(afinfo);
1387 1388
@@ -1486,7 +1487,7 @@ static struct dst_entry *xfrm_bundle_create(struct xfrm_policy *policy,
1486 for (dst_prev = dst0; dst_prev != dst; dst_prev = dst_prev->child) { 1487 for (dst_prev = dst0; dst_prev != dst; dst_prev = dst_prev->child) {
1487 struct xfrm_dst *xdst = (struct xfrm_dst *)dst_prev; 1488 struct xfrm_dst *xdst = (struct xfrm_dst *)dst_prev;
1488 1489
1489 err = xfrm_fill_dst(xdst, dev); 1490 err = xfrm_fill_dst(xdst, dev, fl);
1490 if (err) 1491 if (err)
1491 goto free_dst; 1492 goto free_dst;
1492 1493
generated by cgit v1.2.2 (git 2.25.0) at 2025-09-21 06:19:43 -0400