aboutsummaryrefslogtreecommitdiffstats
path: root/net
diff options
context:
space:
mode:
authorDavid Woodhouse <David.Woodhouse@intel.com>2008-07-11 09:36:25 -0400
committerDavid Woodhouse <David.Woodhouse@intel.com>2008-07-11 09:36:25 -0400
commita8931ef380c92d121ae74ecfb03b2d63f72eea6f (patch)
tree980fb6b019e11e6cb1ece55b7faff184721a8053 /net
parent90574d0a4d4b73308ae54a2a57a4f3f1fa98e984 (diff)
parente5a5816f7875207cb0a0a7032e39a4686c5e10a4 (diff)
Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6
Diffstat (limited to 'net')
-rw-r--r--net/8021q/vlan.c28
-rw-r--r--net/8021q/vlan_dev.c2
-rw-r--r--net/8021q/vlanproc.c7
-rw-r--r--net/9p/Kconfig10
-rw-r--r--net/9p/Makefile3
-rw-r--r--net/9p/client.c32
-rw-r--r--net/9p/conv.c128
-rw-r--r--net/9p/error.c15
-rw-r--r--net/9p/fcprint.c8
-rw-r--r--net/9p/mod.c8
-rw-r--r--net/9p/trans_fd.c204
-rw-r--r--net/9p/trans_virtio.c175
-rw-r--r--net/9p/util.c36
-rw-r--r--net/atm/br2684.c82
-rw-r--r--net/atm/proc.c6
-rw-r--r--net/ax25/ax25_subr.c11
-rw-r--r--net/bluetooth/bnep/core.c8
-rw-r--r--net/bluetooth/hci_event.c9
-rw-r--r--net/bluetooth/hci_sock.c2
-rw-r--r--net/bluetooth/l2cap.c2
-rw-r--r--net/bluetooth/rfcomm/core.c2
-rw-r--r--net/bluetooth/rfcomm/tty.c13
-rw-r--r--net/bridge/br_fdb.c4
-rw-r--r--net/bridge/br_if.c22
-rw-r--r--net/bridge/br_stp_bpdu.c4
-rw-r--r--net/can/af_can.c26
-rw-r--r--net/can/bcm.c29
-rw-r--r--net/can/raw.c3
-rw-r--r--net/core/dev.c83
-rw-r--r--net/core/fib_rules.c4
-rw-r--r--net/core/filter.c5
-rw-r--r--net/core/neighbour.c14
-rw-r--r--net/core/net-sysfs.c7
-rw-r--r--net/core/net-sysfs.h2
-rw-r--r--net/core/net_namespace.c3
-rw-r--r--net/core/netpoll.c2
-rw-r--r--net/core/pktgen.c16
-rw-r--r--net/core/rtnetlink.c3
-rw-r--r--net/core/skbuff.c26
-rw-r--r--net/core/sock.c5
-rw-r--r--net/core/user_dma.c2
-rw-r--r--net/dccp/ackvec.c29
-rw-r--r--net/dccp/ccids/ccid2.c2
-rw-r--r--net/dccp/ccids/ccid3.c31
-rw-r--r--net/dccp/ccids/lib/tfrc.c8
-rw-r--r--net/dccp/ccids/lib/tfrc.h25
-rw-r--r--net/dccp/ccids/lib/tfrc_equation.c8
-rw-r--r--net/dccp/feat.c2
-rw-r--r--net/dccp/ipv4.c7
-rw-r--r--net/dccp/ipv6.c1
-rw-r--r--net/dccp/minisocks.c8
-rw-r--r--net/dccp/options.c4
-rw-r--r--net/dccp/output.c2
-rw-r--r--net/dccp/probe.c2
-rw-r--r--net/decnet/dn_route.c12
-rw-r--r--net/econet/af_econet.c2
-rw-r--r--net/ipv4/arp.c7
-rw-r--r--net/ipv4/cipso_ipv4.c22
-rw-r--r--net/ipv4/devinet.c9
-rw-r--r--net/ipv4/fib_frontend.c1
-rw-r--r--net/ipv4/fib_semantics.c5
-rw-r--r--net/ipv4/fib_trie.c17
-rw-r--r--net/ipv4/igmp.c4
-rw-r--r--net/ipv4/inet_connection_sock.c11
-rw-r--r--net/ipv4/inet_fragment.c16
-rw-r--r--net/ipv4/inet_lro.c3
-rw-r--r--net/ipv4/ip_fragment.c2
-rw-r--r--net/ipv4/ip_gre.c146
-rw-r--r--net/ipv4/ip_input.c2
-rw-r--r--net/ipv4/ipconfig.c6
-rw-r--r--net/ipv4/ipip.c130
-rw-r--r--net/ipv4/netfilter/ipt_CLUSTERIP.c6
-rw-r--r--net/ipv4/netfilter/nf_nat_core.c3
-rw-r--r--net/ipv4/netfilter/nf_nat_snmp_basic.c16
-rw-r--r--net/ipv4/raw.c21
-rw-r--r--net/ipv4/route.c22
-rw-r--r--net/ipv4/syncookies.c3
-rw-r--r--net/ipv4/tcp.c40
-rw-r--r--net/ipv4/tcp_hybla.c6
-rw-r--r--net/ipv4/tcp_input.c143
-rw-r--r--net/ipv4/tcp_ipv4.c30
-rw-r--r--net/ipv4/tcp_minisocks.c32
-rw-r--r--net/ipv4/tcp_output.c12
-rw-r--r--net/ipv4/tcp_probe.c2
-rw-r--r--net/ipv4/tcp_timer.c5
-rw-r--r--net/ipv4/tcp_vegas.c6
-rw-r--r--net/ipv4/tcp_veno.c6
-rw-r--r--net/ipv4/tunnel4.c2
-rw-r--r--net/ipv4/udp.c10
-rw-r--r--net/ipv4/xfrm4_mode_tunnel.c2
-rw-r--r--net/ipv6/addrconf.c146
-rw-r--r--net/ipv6/af_inet6.c2
-rw-r--r--net/ipv6/datagram.c50
-rw-r--r--net/ipv6/exthdrs.c2
-rw-r--r--net/ipv6/ip6_flowlabel.c2
-rw-r--r--net/ipv6/ip6_input.c9
-rw-r--r--net/ipv6/ip6_output.c2
-rw-r--r--net/ipv6/ip6mr.c2
-rw-r--r--net/ipv6/ipv6_sockglue.c44
-rw-r--r--net/ipv6/mcast.c4
-rw-r--r--net/ipv6/ndisc.c12
-rw-r--r--net/ipv6/netfilter/ip6table_mangle.c2
-rw-r--r--net/ipv6/netfilter/nf_conntrack_reasm.c9
-rw-r--r--net/ipv6/proc.c6
-rw-r--r--net/ipv6/raw.c23
-rw-r--r--net/ipv6/reassembly.c6
-rw-r--r--net/ipv6/route.c38
-rw-r--r--net/ipv6/sit.c135
-rw-r--r--net/ipv6/syncookies.c1
-rw-r--r--net/ipv6/tcp_ipv6.c7
-rw-r--r--net/ipv6/tunnel6.c2
-rw-r--r--net/ipv6/udp.c8
-rw-r--r--net/irda/af_irda.c12
-rw-r--r--net/irda/discovery.c8
-rw-r--r--net/irda/iriap.c8
-rw-r--r--net/irda/irlmp.c5
-rw-r--r--net/irda/irnet/irnet_irda.c5
-rw-r--r--net/irda/irnetlink.c4
-rw-r--r--net/key/af_key.c5
-rw-r--r--net/llc/llc_sap.c10
-rw-r--r--net/mac80211/cfg.c4
-rw-r--r--net/mac80211/debugfs_key.c15
-rw-r--r--net/mac80211/ieee80211_i.h4
-rw-r--r--net/mac80211/iface.c9
-rw-r--r--net/mac80211/key.c9
-rw-r--r--net/mac80211/main.c61
-rw-r--r--net/mac80211/mesh.c19
-rw-r--r--net/mac80211/mesh.h2
-rw-r--r--net/mac80211/mesh_hwmp.c2
-rw-r--r--net/mac80211/mesh_pathtbl.c17
-rw-r--r--net/mac80211/mlme.c143
-rw-r--r--net/mac80211/rc80211_pid.h5
-rw-r--r--net/mac80211/rc80211_pid_algo.c31
-rw-r--r--net/mac80211/rc80211_pid_debugfs.c4
-rw-r--r--net/mac80211/rx.c16
-rw-r--r--net/mac80211/tkip.c6
-rw-r--r--net/mac80211/tx.c10
-rw-r--r--net/mac80211/util.c51
-rw-r--r--net/mac80211/wext.c35
-rw-r--r--net/mac80211/wme.c8
-rw-r--r--net/netfilter/Kconfig4
-rw-r--r--net/netfilter/nf_conntrack_core.c3
-rw-r--r--net/netfilter/nf_conntrack_expect.c4
-rw-r--r--net/netfilter/nf_conntrack_extend.c9
-rw-r--r--net/netfilter/nf_conntrack_h323_main.c22
-rw-r--r--net/netfilter/nf_conntrack_netlink.c3
-rw-r--r--net/netfilter/nf_conntrack_proto_tcp.c23
-rw-r--r--net/netfilter/nf_conntrack_sip.c22
-rw-r--r--net/netfilter/nf_conntrack_standalone.c6
-rw-r--r--net/netfilter/nf_log.c4
-rw-r--r--net/netfilter/x_tables.c13
-rw-r--r--net/netfilter/xt_connlimit.c3
-rw-r--r--net/netfilter/xt_hashlimit.c12
-rw-r--r--net/netfilter/xt_iprange.c2
-rw-r--r--net/netlabel/netlabel_cipso_v4.c7
-rw-r--r--net/netlabel/netlabel_mgmt.c12
-rw-r--r--net/netlabel/netlabel_unlabeled.c8
-rw-r--r--net/netlink/af_netlink.c2
-rw-r--r--net/netlink/attr.c19
-rw-r--r--net/netlink/genetlink.c21
-rw-r--r--net/packet/af_packet.c2
-rw-r--r--net/rose/rose_route.c8
-rw-r--r--net/rxrpc/ar-transport.c3
-rw-r--r--net/sched/Kconfig11
-rw-r--r--net/sched/act_simple.c45
-rw-r--r--net/sched/cls_api.c2
-rw-r--r--net/sched/sch_api.c6
-rw-r--r--net/sched/sch_atm.c7
-rw-r--r--net/sched/sch_cbq.c8
-rw-r--r--net/sched/sch_dsmark.c8
-rw-r--r--net/sched/sch_generic.c3
-rw-r--r--net/sched/sch_gred.c3
-rw-r--r--net/sched/sch_hfsc.c8
-rw-r--r--net/sched/sch_htb.c35
-rw-r--r--net/sched/sch_ingress.c2
-rw-r--r--net/sched/sch_prio.c2
-rw-r--r--net/sched/sch_red.c3
-rw-r--r--net/sched/sch_sfq.c2
-rw-r--r--net/sctp/associola.c34
-rw-r--r--net/sctp/ipv6.c11
-rw-r--r--net/sctp/output.c2
-rw-r--r--net/sctp/outqueue.c120
-rw-r--r--net/sctp/protocol.c26
-rw-r--r--net/sctp/sm_make_chunk.c16
-rw-r--r--net/sctp/sm_statefuns.c9
-rw-r--r--net/sctp/socket.c4
-rw-r--r--net/sctp/transport.c50
-rw-r--r--net/sctp/ulpevent.c5
-rw-r--r--net/sunrpc/auth_generic.c8
-rw-r--r--net/sunrpc/auth_gss/svcauth_gss.c12
-rw-r--r--net/sunrpc/cache.c15
-rw-r--r--net/sunrpc/rpcb_clnt.c23
-rw-r--r--net/sunrpc/stats.c8
-rw-r--r--net/sunrpc/svc.c1
-rw-r--r--net/sunrpc/svc_xprt.c23
-rw-r--r--net/sunrpc/svcauth_unix.c4
-rw-r--r--net/sunrpc/xprtrdma/svc_rdma_recvfrom.c102
-rw-r--r--net/sunrpc/xprtrdma/svc_rdma_sendto.c11
-rw-r--r--net/sunrpc/xprtrdma/svc_rdma_transport.c290
-rw-r--r--net/sysctl_net.c2
-rw-r--r--net/tipc/core.h11
-rw-r--r--net/unix/af_unix.c75
-rw-r--r--net/wireless/nl80211.c12
-rw-r--r--net/wireless/reg.c18
-rw-r--r--net/xfrm/xfrm_algo.c4
-rw-r--r--net/xfrm/xfrm_output.c6
-rw-r--r--net/xfrm/xfrm_user.c14
207 files changed, 2417 insertions, 1818 deletions
diff --git a/net/8021q/vlan.c b/net/8021q/vlan.c
index 2a739adaa92b..ab2225da0ee2 100644
--- a/net/8021q/vlan.c
+++ b/net/8021q/vlan.c
@@ -382,6 +382,18 @@ static void vlan_sync_address(struct net_device *dev,
382 memcpy(vlan->real_dev_addr, dev->dev_addr, ETH_ALEN); 382 memcpy(vlan->real_dev_addr, dev->dev_addr, ETH_ALEN);
383} 383}
384 384
385static void vlan_transfer_features(struct net_device *dev,
386 struct net_device *vlandev)
387{
388 unsigned long old_features = vlandev->features;
389
390 vlandev->features &= ~dev->vlan_features;
391 vlandev->features |= dev->features & dev->vlan_features;
392
393 if (old_features != vlandev->features)
394 netdev_features_change(vlandev);
395}
396
385static void __vlan_device_event(struct net_device *dev, unsigned long event) 397static void __vlan_device_event(struct net_device *dev, unsigned long event)
386{ 398{
387 switch (event) { 399 switch (event) {
@@ -410,10 +422,8 @@ static int vlan_device_event(struct notifier_block *unused, unsigned long event,
410 int i, flgs; 422 int i, flgs;
411 struct net_device *vlandev; 423 struct net_device *vlandev;
412 424
413 if (is_vlan_dev(dev)) { 425 if (is_vlan_dev(dev))
414 __vlan_device_event(dev, event); 426 __vlan_device_event(dev, event);
415 goto out;
416 }
417 427
418 grp = __vlan_find_group(dev); 428 grp = __vlan_find_group(dev);
419 if (!grp) 429 if (!grp)
@@ -450,6 +460,18 @@ static int vlan_device_event(struct notifier_block *unused, unsigned long event,
450 } 460 }
451 break; 461 break;
452 462
463 case NETDEV_FEAT_CHANGE:
464 /* Propagate device features to underlying device */
465 for (i = 0; i < VLAN_GROUP_ARRAY_LEN; i++) {
466 vlandev = vlan_group_get_device(grp, i);
467 if (!vlandev)
468 continue;
469
470 vlan_transfer_features(dev, vlandev);
471 }
472
473 break;
474
453 case NETDEV_DOWN: 475 case NETDEV_DOWN:
454 /* Put all VLANs for this dev in the down state too. */ 476 /* Put all VLANs for this dev in the down state too. */
455 for (i = 0; i < VLAN_GROUP_ARRAY_LEN; i++) { 477 for (i = 0; i < VLAN_GROUP_ARRAY_LEN; i++) {
diff --git a/net/8021q/vlan_dev.c b/net/8021q/vlan_dev.c
index c961f0826005..5d055c242ed8 100644
--- a/net/8021q/vlan_dev.c
+++ b/net/8021q/vlan_dev.c
@@ -663,6 +663,8 @@ static int vlan_dev_init(struct net_device *dev)
663 (1<<__LINK_STATE_DORMANT))) | 663 (1<<__LINK_STATE_DORMANT))) |
664 (1<<__LINK_STATE_PRESENT); 664 (1<<__LINK_STATE_PRESENT);
665 665
666 dev->features |= real_dev->features & real_dev->vlan_features;
667
666 /* ipv6 shared card related stuff */ 668 /* ipv6 shared card related stuff */
667 dev->dev_id = real_dev->dev_id; 669 dev->dev_id = real_dev->dev_id;
668 670
diff --git a/net/8021q/vlanproc.c b/net/8021q/vlanproc.c
index daad0064e2c2..08b54b593d56 100644
--- a/net/8021q/vlanproc.c
+++ b/net/8021q/vlanproc.c
@@ -176,12 +176,11 @@ int vlan_proc_add_dev(struct net_device *vlandev)
176 struct vlan_dev_info *dev_info = vlan_dev_info(vlandev); 176 struct vlan_dev_info *dev_info = vlan_dev_info(vlandev);
177 struct vlan_net *vn = net_generic(dev_net(vlandev), vlan_net_id); 177 struct vlan_net *vn = net_generic(dev_net(vlandev), vlan_net_id);
178 178
179 dev_info->dent = proc_create(vlandev->name, S_IFREG|S_IRUSR|S_IWUSR, 179 dev_info->dent =
180 vn->proc_vlan_dir, &vlandev_fops); 180 proc_create_data(vlandev->name, S_IFREG|S_IRUSR|S_IWUSR,
181 vn->proc_vlan_dir, &vlandev_fops, vlandev);
181 if (!dev_info->dent) 182 if (!dev_info->dent)
182 return -ENOBUFS; 183 return -ENOBUFS;
183
184 dev_info->dent->data = vlandev;
185 return 0; 184 return 0;
186} 185}
187 186
diff --git a/net/9p/Kconfig b/net/9p/Kconfig
index bafc50c9e6ff..ff34c5acc130 100644
--- a/net/9p/Kconfig
+++ b/net/9p/Kconfig
@@ -13,16 +13,6 @@ menuconfig NET_9P
13 13
14 If unsure, say N. 14 If unsure, say N.
15 15
16config NET_9P_FD
17 depends on NET_9P
18 default y if NET_9P
19 tristate "9P File Descriptor Transports (Experimental)"
20 help
21 This builds support for file descriptor transports for 9p
22 which includes support for TCP/IP, named pipes, or passed
23 file descriptors. TCP/IP is the default transport for 9p,
24 so if you are going to use 9p, you'll likely want this.
25
26config NET_9P_VIRTIO 16config NET_9P_VIRTIO
27 depends on NET_9P && EXPERIMENTAL && VIRTIO 17 depends on NET_9P && EXPERIMENTAL && VIRTIO
28 tristate "9P Virtio Transport (Experimental)" 18 tristate "9P Virtio Transport (Experimental)"
diff --git a/net/9p/Makefile b/net/9p/Makefile
index 8a1051101898..519219480db1 100644
--- a/net/9p/Makefile
+++ b/net/9p/Makefile
@@ -1,5 +1,4 @@
1obj-$(CONFIG_NET_9P) := 9pnet.o 1obj-$(CONFIG_NET_9P) := 9pnet.o
2obj-$(CONFIG_NET_9P_FD) += 9pnet_fd.o
3obj-$(CONFIG_NET_9P_VIRTIO) += 9pnet_virtio.o 2obj-$(CONFIG_NET_9P_VIRTIO) += 9pnet_virtio.o
4 3
59pnet-objs := \ 49pnet-objs := \
@@ -9,8 +8,6 @@ obj-$(CONFIG_NET_9P_VIRTIO) += 9pnet_virtio.o
9 error.o \ 8 error.o \
10 fcprint.o \ 9 fcprint.o \
11 util.o \ 10 util.o \
12
139pnet_fd-objs := \
14 trans_fd.o \ 11 trans_fd.o \
15 12
169pnet_virtio-objs := \ 139pnet_virtio-objs := \
diff --git a/net/9p/client.c b/net/9p/client.c
index 84e087e24146..2ffe40cf2f01 100644
--- a/net/9p/client.c
+++ b/net/9p/client.c
@@ -64,21 +64,30 @@ static match_table_t tokens = {
64 * @options: options string passed from mount 64 * @options: options string passed from mount
65 * @v9ses: existing v9fs session information 65 * @v9ses: existing v9fs session information
66 * 66 *
67 * Return 0 upon success, -ERRNO upon failure
67 */ 68 */
68 69
69static void parse_opts(char *options, struct p9_client *clnt) 70static int parse_opts(char *opts, struct p9_client *clnt)
70{ 71{
72 char *options;
71 char *p; 73 char *p;
72 substring_t args[MAX_OPT_ARGS]; 74 substring_t args[MAX_OPT_ARGS];
73 int option; 75 int option;
74 int ret; 76 int ret = 0;
75 77
76 clnt->trans_mod = v9fs_default_trans(); 78 clnt->trans_mod = v9fs_default_trans();
77 clnt->dotu = 1; 79 clnt->dotu = 1;
78 clnt->msize = 8192; 80 clnt->msize = 8192;
79 81
80 if (!options) 82 if (!opts)
81 return; 83 return 0;
84
85 options = kstrdup(opts, GFP_KERNEL);
86 if (!options) {
87 P9_DPRINTK(P9_DEBUG_ERROR,
88 "failed to allocate copy of option string\n");
89 return -ENOMEM;
90 }
82 91
83 while ((p = strsep(&options, ",")) != NULL) { 92 while ((p = strsep(&options, ",")) != NULL) {
84 int token; 93 int token;
@@ -86,10 +95,11 @@ static void parse_opts(char *options, struct p9_client *clnt)
86 continue; 95 continue;
87 token = match_token(p, tokens, args); 96 token = match_token(p, tokens, args);
88 if (token < Opt_trans) { 97 if (token < Opt_trans) {
89 ret = match_int(&args[0], &option); 98 int r = match_int(&args[0], &option);
90 if (ret < 0) { 99 if (r < 0) {
91 P9_DPRINTK(P9_DEBUG_ERROR, 100 P9_DPRINTK(P9_DEBUG_ERROR,
92 "integer field, but no integer?\n"); 101 "integer field, but no integer?\n");
102 ret = r;
93 continue; 103 continue;
94 } 104 }
95 } 105 }
@@ -107,6 +117,8 @@ static void parse_opts(char *options, struct p9_client *clnt)
107 continue; 117 continue;
108 } 118 }
109 } 119 }
120 kfree(options);
121 return ret;
110} 122}
111 123
112 124
@@ -138,16 +150,20 @@ struct p9_client *p9_client_create(const char *dev_name, char *options)
138 if (!clnt) 150 if (!clnt)
139 return ERR_PTR(-ENOMEM); 151 return ERR_PTR(-ENOMEM);
140 152
153 clnt->trans = NULL;
141 spin_lock_init(&clnt->lock); 154 spin_lock_init(&clnt->lock);
142 INIT_LIST_HEAD(&clnt->fidlist); 155 INIT_LIST_HEAD(&clnt->fidlist);
143 clnt->fidpool = p9_idpool_create(); 156 clnt->fidpool = p9_idpool_create();
144 if (!clnt->fidpool) { 157 if (IS_ERR(clnt->fidpool)) {
145 err = PTR_ERR(clnt->fidpool); 158 err = PTR_ERR(clnt->fidpool);
146 clnt->fidpool = NULL; 159 clnt->fidpool = NULL;
147 goto error; 160 goto error;
148 } 161 }
149 162
150 parse_opts(options, clnt); 163 err = parse_opts(options, clnt);
164 if (err < 0)
165 goto error;
166
151 if (clnt->trans_mod == NULL) { 167 if (clnt->trans_mod == NULL) {
152 err = -EPROTONOSUPPORT; 168 err = -EPROTONOSUPPORT;
153 P9_DPRINTK(P9_DEBUG_ERROR, 169 P9_DPRINTK(P9_DEBUG_ERROR,
diff --git a/net/9p/conv.c b/net/9p/conv.c
index 3fe35d532c87..44547201f5bc 100644
--- a/net/9p/conv.c
+++ b/net/9p/conv.c
@@ -197,7 +197,7 @@ static void buf_get_qid(struct cbuf *bufp, struct p9_qid *qid)
197 197
198/** 198/**
199 * p9_size_wstat - calculate the size of a variable length stat struct 199 * p9_size_wstat - calculate the size of a variable length stat struct
200 * @stat: metadata (stat) structure 200 * @wstat: metadata (stat) structure
201 * @dotu: non-zero if 9P2000.u 201 * @dotu: non-zero if 9P2000.u
202 * 202 *
203 */ 203 */
@@ -511,6 +511,12 @@ p9_create_common(struct cbuf *bufp, u32 size, u8 id)
511 return fc; 511 return fc;
512} 512}
513 513
514/**
515 * p9_set_tag - set the tag field of an &p9_fcall structure
516 * @fc: fcall structure to set tag within
517 * @tag: tag id to set
518 */
519
514void p9_set_tag(struct p9_fcall *fc, u16 tag) 520void p9_set_tag(struct p9_fcall *fc, u16 tag)
515{ 521{
516 fc->tag = tag; 522 fc->tag = tag;
@@ -518,6 +524,12 @@ void p9_set_tag(struct p9_fcall *fc, u16 tag)
518} 524}
519EXPORT_SYMBOL(p9_set_tag); 525EXPORT_SYMBOL(p9_set_tag);
520 526
527/**
528 * p9_create_tversion - allocates and creates a T_VERSION request
529 * @msize: requested maximum data size
530 * @version: version string to negotiate
531 *
532 */
521struct p9_fcall *p9_create_tversion(u32 msize, char *version) 533struct p9_fcall *p9_create_tversion(u32 msize, char *version)
522{ 534{
523 int size; 535 int size;
@@ -542,6 +554,16 @@ error:
542} 554}
543EXPORT_SYMBOL(p9_create_tversion); 555EXPORT_SYMBOL(p9_create_tversion);
544 556
557/**
558 * p9_create_tauth - allocates and creates a T_AUTH request
559 * @afid: handle to use for authentication protocol
560 * @uname: user name attempting to authenticate
561 * @aname: mount specifier for remote server
562 * @n_uname: numeric id for user attempting to authneticate
563 * @dotu: 9P2000.u extension flag
564 *
565 */
566
545struct p9_fcall *p9_create_tauth(u32 afid, char *uname, char *aname, 567struct p9_fcall *p9_create_tauth(u32 afid, char *uname, char *aname,
546 u32 n_uname, int dotu) 568 u32 n_uname, int dotu)
547{ 569{
@@ -580,6 +602,18 @@ error:
580} 602}
581EXPORT_SYMBOL(p9_create_tauth); 603EXPORT_SYMBOL(p9_create_tauth);
582 604
605/**
606 * p9_create_tattach - allocates and creates a T_ATTACH request
607 * @fid: handle to use for the new mount point
608 * @afid: handle to use for authentication protocol
609 * @uname: user name attempting to attach
610 * @aname: mount specifier for remote server
611 * @n_uname: numeric id for user attempting to attach
612 * @n_uname: numeric id for user attempting to attach
613 * @dotu: 9P2000.u extension flag
614 *
615 */
616
583struct p9_fcall * 617struct p9_fcall *
584p9_create_tattach(u32 fid, u32 afid, char *uname, char *aname, 618p9_create_tattach(u32 fid, u32 afid, char *uname, char *aname,
585 u32 n_uname, int dotu) 619 u32 n_uname, int dotu)
@@ -616,6 +650,12 @@ error:
616} 650}
617EXPORT_SYMBOL(p9_create_tattach); 651EXPORT_SYMBOL(p9_create_tattach);
618 652
653/**
654 * p9_create_tflush - allocates and creates a T_FLUSH request
655 * @oldtag: tag id for the transaction we are attempting to cancel
656 *
657 */
658
619struct p9_fcall *p9_create_tflush(u16 oldtag) 659struct p9_fcall *p9_create_tflush(u16 oldtag)
620{ 660{
621 int size; 661 int size;
@@ -639,6 +679,15 @@ error:
639} 679}
640EXPORT_SYMBOL(p9_create_tflush); 680EXPORT_SYMBOL(p9_create_tflush);
641 681
682/**
683 * p9_create_twalk - allocates and creates a T_FLUSH request
684 * @fid: handle we are traversing from
685 * @newfid: a new handle for this transaction
686 * @nwname: number of path elements to traverse
687 * @wnames: array of path elements
688 *
689 */
690
642struct p9_fcall *p9_create_twalk(u32 fid, u32 newfid, u16 nwname, 691struct p9_fcall *p9_create_twalk(u32 fid, u32 newfid, u16 nwname,
643 char **wnames) 692 char **wnames)
644{ 693{
@@ -677,6 +726,13 @@ error:
677} 726}
678EXPORT_SYMBOL(p9_create_twalk); 727EXPORT_SYMBOL(p9_create_twalk);
679 728
729/**
730 * p9_create_topen - allocates and creates a T_OPEN request
731 * @fid: handle we are trying to open
732 * @mode: what mode we are trying to open the file in
733 *
734 */
735
680struct p9_fcall *p9_create_topen(u32 fid, u8 mode) 736struct p9_fcall *p9_create_topen(u32 fid, u8 mode)
681{ 737{
682 int size; 738 int size;
@@ -701,6 +757,19 @@ error:
701} 757}
702EXPORT_SYMBOL(p9_create_topen); 758EXPORT_SYMBOL(p9_create_topen);
703 759
760/**
761 * p9_create_tcreate - allocates and creates a T_CREATE request
762 * @fid: handle of directory we are trying to create in
763 * @name: name of the file we are trying to create
764 * @perm: permissions for the file we are trying to create
765 * @mode: what mode we are trying to open the file in
766 * @extension: 9p2000.u extension string (for special files)
767 * @dotu: 9p2000.u enabled flag
768 *
769 * Note: Plan 9 create semantics include opening the resulting file
770 * which is why mode is included.
771 */
772
704struct p9_fcall *p9_create_tcreate(u32 fid, char *name, u32 perm, u8 mode, 773struct p9_fcall *p9_create_tcreate(u32 fid, char *name, u32 perm, u8 mode,
705 char *extension, int dotu) 774 char *extension, int dotu)
706{ 775{
@@ -736,6 +805,13 @@ error:
736} 805}
737EXPORT_SYMBOL(p9_create_tcreate); 806EXPORT_SYMBOL(p9_create_tcreate);
738 807
808/**
809 * p9_create_tread - allocates and creates a T_READ request
810 * @fid: handle of the file we are trying to read
811 * @offset: offset to start reading from
812 * @count: how many bytes to read
813 */
814
739struct p9_fcall *p9_create_tread(u32 fid, u64 offset, u32 count) 815struct p9_fcall *p9_create_tread(u32 fid, u64 offset, u32 count)
740{ 816{
741 int size; 817 int size;
@@ -761,6 +837,17 @@ error:
761} 837}
762EXPORT_SYMBOL(p9_create_tread); 838EXPORT_SYMBOL(p9_create_tread);
763 839
840/**
841 * p9_create_twrite - allocates and creates a T_WRITE request from the kernel
842 * @fid: handle of the file we are trying to write
843 * @offset: offset to start writing at
844 * @count: how many bytes to write
845 * @data: data to write
846 *
847 * This function will create a requst with data buffers from the kernel
848 * such as the page cache.
849 */
850
764struct p9_fcall *p9_create_twrite(u32 fid, u64 offset, u32 count, 851struct p9_fcall *p9_create_twrite(u32 fid, u64 offset, u32 count,
765 const char *data) 852 const char *data)
766{ 853{
@@ -794,6 +881,16 @@ error:
794} 881}
795EXPORT_SYMBOL(p9_create_twrite); 882EXPORT_SYMBOL(p9_create_twrite);
796 883
884/**
885 * p9_create_twrite_u - allocates and creates a T_WRITE request from userspace
886 * @fid: handle of the file we are trying to write
887 * @offset: offset to start writing at
888 * @count: how many bytes to write
889 * @data: data to write
890 *
891 * This function will create a request with data buffers from userspace
892 */
893
797struct p9_fcall *p9_create_twrite_u(u32 fid, u64 offset, u32 count, 894struct p9_fcall *p9_create_twrite_u(u32 fid, u64 offset, u32 count,
798 const char __user *data) 895 const char __user *data)
799{ 896{
@@ -827,6 +924,14 @@ error:
827} 924}
828EXPORT_SYMBOL(p9_create_twrite_u); 925EXPORT_SYMBOL(p9_create_twrite_u);
829 926
927/**
928 * p9_create_tclunk - allocate a request to forget about a file handle
929 * @fid: handle of the file we closing or forgetting about
930 *
931 * clunk is used both to close open files and to discard transient handles
932 * which may be created during meta-data operations and hierarchy traversal.
933 */
934
830struct p9_fcall *p9_create_tclunk(u32 fid) 935struct p9_fcall *p9_create_tclunk(u32 fid)
831{ 936{
832 int size; 937 int size;
@@ -850,6 +955,12 @@ error:
850} 955}
851EXPORT_SYMBOL(p9_create_tclunk); 956EXPORT_SYMBOL(p9_create_tclunk);
852 957
958/**
959 * p9_create_tremove - allocate and create a request to remove a file
960 * @fid: handle of the file or directory we are removing
961 *
962 */
963
853struct p9_fcall *p9_create_tremove(u32 fid) 964struct p9_fcall *p9_create_tremove(u32 fid)
854{ 965{
855 int size; 966 int size;
@@ -873,6 +984,12 @@ error:
873} 984}
874EXPORT_SYMBOL(p9_create_tremove); 985EXPORT_SYMBOL(p9_create_tremove);
875 986
987/**
988 * p9_create_tstat - allocate and populate a request for attributes
989 * @fid: handle of the file or directory we are trying to get the attributes of
990 *
991 */
992
876struct p9_fcall *p9_create_tstat(u32 fid) 993struct p9_fcall *p9_create_tstat(u32 fid)
877{ 994{
878 int size; 995 int size;
@@ -896,6 +1013,14 @@ error:
896} 1013}
897EXPORT_SYMBOL(p9_create_tstat); 1014EXPORT_SYMBOL(p9_create_tstat);
898 1015
1016/**
1017 * p9_create_tstat - allocate and populate a request to change attributes
1018 * @fid: handle of the file or directory we are trying to change
1019 * @wstat: &p9_stat structure with attributes we wish to set
1020 * @dotu: 9p2000.u enabled flag
1021 *
1022 */
1023
899struct p9_fcall *p9_create_twstat(u32 fid, struct p9_wstat *wstat, 1024struct p9_fcall *p9_create_twstat(u32 fid, struct p9_wstat *wstat,
900 int dotu) 1025 int dotu)
901{ 1026{
@@ -922,3 +1047,4 @@ error:
922 return fc; 1047 return fc;
923} 1048}
924EXPORT_SYMBOL(p9_create_twstat); 1049EXPORT_SYMBOL(p9_create_twstat);
1050
diff --git a/net/9p/error.c b/net/9p/error.c
index 64104b9cb422..fdebe4314062 100644
--- a/net/9p/error.c
+++ b/net/9p/error.c
@@ -33,6 +33,13 @@
33#include <linux/errno.h> 33#include <linux/errno.h>
34#include <net/9p/9p.h> 34#include <net/9p/9p.h>
35 35
36/**
37 * struct errormap - map string errors from Plan 9 to Linux numeric ids
38 * @name: string sent over 9P
39 * @val: numeric id most closely representing @name
40 * @namelen: length of string
41 * @list: hash-table list for string lookup
42 */
36struct errormap { 43struct errormap {
37 char *name; 44 char *name;
38 int val; 45 int val;
@@ -177,8 +184,7 @@ static struct errormap errmap[] = {
177}; 184};
178 185
179/** 186/**
180 * p9_error_init - preload 187 * p9_error_init - preload mappings into hash list
181 * @errstr: error string
182 * 188 *
183 */ 189 */
184 190
@@ -206,6 +212,7 @@ EXPORT_SYMBOL(p9_error_init);
206/** 212/**
207 * errstr2errno - convert error string to error number 213 * errstr2errno - convert error string to error number
208 * @errstr: error string 214 * @errstr: error string
215 * @len: length of error string
209 * 216 *
210 */ 217 */
211 218
@@ -230,8 +237,8 @@ int p9_errstr2errno(char *errstr, int len)
230 if (errno == 0) { 237 if (errno == 0) {
231 /* TODO: if error isn't found, add it dynamically */ 238 /* TODO: if error isn't found, add it dynamically */
232 errstr[len] = 0; 239 errstr[len] = 0;
233 printk(KERN_ERR "%s: errstr :%s: not found\n", __func__, 240 printk(KERN_ERR "%s: server reported unknown error %s\n",
234 errstr); 241 __func__, errstr);
235 errno = 1; 242 errno = 1;
236 } 243 }
237 244
diff --git a/net/9p/fcprint.c b/net/9p/fcprint.c
index 40244fbd9b0d..53dd8e28dd8a 100644
--- a/net/9p/fcprint.c
+++ b/net/9p/fcprint.c
@@ -142,6 +142,14 @@ p9_printdata(char *buf, int buflen, u8 *data, int datalen)
142 return p9_dumpdata(buf, buflen, data, datalen < 16?datalen:16); 142 return p9_dumpdata(buf, buflen, data, datalen < 16?datalen:16);
143} 143}
144 144
145/**
146 * p9_printfcall - decode and print a protocol structure into a buffer
147 * @buf: buffer to deposit decoded structure into
148 * @buflen: available space in buffer
149 * @fc: protocol rpc structure of type &p9_fcall
150 * @extended: whether or not session is operating with extended protocol
151 */
152
145int 153int
146p9_printfcall(char *buf, int buflen, struct p9_fcall *fc, int extended) 154p9_printfcall(char *buf, int buflen, struct p9_fcall *fc, int extended)
147{ 155{
diff --git a/net/9p/mod.c b/net/9p/mod.c
index c285aab2af04..bdee1fb7cc62 100644
--- a/net/9p/mod.c
+++ b/net/9p/mod.c
@@ -39,9 +39,6 @@ module_param_named(debug, p9_debug_level, uint, 0);
39MODULE_PARM_DESC(debug, "9P debugging level"); 39MODULE_PARM_DESC(debug, "9P debugging level");
40#endif 40#endif
41 41
42extern int p9_mux_global_init(void);
43extern void p9_mux_global_exit(void);
44
45/* 42/*
46 * Dynamic Transport Registration Routines 43 * Dynamic Transport Registration Routines
47 * 44 *
@@ -52,7 +49,7 @@ static struct p9_trans_module *v9fs_default_transport;
52 49
53/** 50/**
54 * v9fs_register_trans - register a new transport with 9p 51 * v9fs_register_trans - register a new transport with 9p
55 * @m - structure describing the transport module and entry points 52 * @m: structure describing the transport module and entry points
56 * 53 *
57 */ 54 */
58void v9fs_register_trans(struct p9_trans_module *m) 55void v9fs_register_trans(struct p9_trans_module *m)
@@ -65,7 +62,7 @@ EXPORT_SYMBOL(v9fs_register_trans);
65 62
66/** 63/**
67 * v9fs_match_trans - match transport versus registered transports 64 * v9fs_match_trans - match transport versus registered transports
68 * @arg: string identifying transport 65 * @name: string identifying transport
69 * 66 *
70 */ 67 */
71struct p9_trans_module *v9fs_match_trans(const substring_t *name) 68struct p9_trans_module *v9fs_match_trans(const substring_t *name)
@@ -110,6 +107,7 @@ static int __init init_p9(void)
110 107
111 p9_error_init(); 108 p9_error_init();
112 printk(KERN_INFO "Installing 9P2000 support\n"); 109 printk(KERN_INFO "Installing 9P2000 support\n");
110 p9_trans_fd_init();
113 111
114 return ret; 112 return ret;
115} 113}
diff --git a/net/9p/trans_fd.c b/net/9p/trans_fd.c
index f624dff76852..4507f744f44e 100644
--- a/net/9p/trans_fd.c
+++ b/net/9p/trans_fd.c
@@ -47,12 +47,29 @@
47#define SCHED_TIMEOUT 10 47#define SCHED_TIMEOUT 10
48#define MAXPOLLWADDR 2 48#define MAXPOLLWADDR 2
49 49
50/**
51 * struct p9_fd_opts - per-transport options
52 * @rfd: file descriptor for reading (trans=fd)
53 * @wfd: file descriptor for writing (trans=fd)
54 * @port: port to connect to (trans=tcp)
55 *
56 */
57
50struct p9_fd_opts { 58struct p9_fd_opts {
51 int rfd; 59 int rfd;
52 int wfd; 60 int wfd;
53 u16 port; 61 u16 port;
54}; 62};
55 63
64
65/**
66 * struct p9_trans_fd - transport state
67 * @rd: reference to file to read from
68 * @wr: reference of file to write to
69 * @conn: connection state reference
70 *
71 */
72
56struct p9_trans_fd { 73struct p9_trans_fd {
57 struct file *rd; 74 struct file *rd;
58 struct file *wr; 75 struct file *wr;
@@ -90,10 +107,24 @@ enum {
90}; 107};
91 108
92struct p9_req; 109struct p9_req;
93
94typedef void (*p9_conn_req_callback)(struct p9_req *req, void *a); 110typedef void (*p9_conn_req_callback)(struct p9_req *req, void *a);
111
112/**
113 * struct p9_req - fd mux encoding of an rpc transaction
114 * @lock: protects req_list
115 * @tag: numeric tag for rpc transaction
116 * @tcall: request &p9_fcall structure
117 * @rcall: response &p9_fcall structure
118 * @err: error state
119 * @cb: callback for when response is received
120 * @cba: argument to pass to callback
121 * @flush: flag to indicate RPC has been flushed
122 * @req_list: list link for higher level objects to chain requests
123 *
124 */
125
95struct p9_req { 126struct p9_req {
96 spinlock_t lock; /* protect request structure */ 127 spinlock_t lock;
97 int tag; 128 int tag;
98 struct p9_fcall *tcall; 129 struct p9_fcall *tcall;
99 struct p9_fcall *rcall; 130 struct p9_fcall *rcall;
@@ -104,7 +135,39 @@ struct p9_req {
104 struct list_head req_list; 135 struct list_head req_list;
105}; 136};
106 137
107struct p9_mux_poll_task; 138struct p9_mux_poll_task {
139 struct task_struct *task;
140 struct list_head mux_list;
141 int muxnum;
142};
143
144/**
145 * struct p9_conn - fd mux connection state information
146 * @lock: protects mux_list (?)
147 * @mux_list: list link for mux to manage multiple connections (?)
148 * @poll_task: task polling on this connection
149 * @msize: maximum size for connection (dup)
150 * @extended: 9p2000.u flag (dup)
151 * @trans: reference to transport instance for this connection
152 * @tagpool: id accounting for transactions
153 * @err: error state
154 * @equeue: event wait_q (?)
155 * @req_list: accounting for requests which have been sent
156 * @unsent_req_list: accounting for requests that haven't been sent
157 * @rcall: current response &p9_fcall structure
158 * @rpos: read position in current frame
159 * @rbuf: current read buffer
160 * @wpos: write position for current frame
161 * @wsize: amount of data to write for current frame
162 * @wbuf: current write buffer
163 * @poll_wait: array of wait_q's for various worker threads
164 * @poll_waddr: ????
165 * @pt: poll state
166 * @rq: current read work
167 * @wq: current write work
168 * @wsched: ????
169 *
170 */
108 171
109struct p9_conn { 172struct p9_conn {
110 spinlock_t lock; /* protect lock structure */ 173 spinlock_t lock; /* protect lock structure */
@@ -132,11 +195,16 @@ struct p9_conn {
132 unsigned long wsched; 195 unsigned long wsched;
133}; 196};
134 197
135struct p9_mux_poll_task { 198/**
136 struct task_struct *task; 199 * struct p9_mux_rpc - fd mux rpc accounting structure
137 struct list_head mux_list; 200 * @m: connection this request was issued on
138 int muxnum; 201 * @err: error state
139}; 202 * @tcall: request &p9_fcall
203 * @rcall: response &p9_fcall
204 * @wqueue: wait queue that client is blocked on for this rpc
205 *
206 * Bug: isn't this information duplicated elsewhere like &p9_req
207 */
140 208
141struct p9_mux_rpc { 209struct p9_mux_rpc {
142 struct p9_conn *m; 210 struct p9_conn *m;
@@ -207,10 +275,12 @@ static void p9_mux_put_tag(struct p9_conn *m, u16 tag)
207 275
208/** 276/**
209 * p9_mux_calc_poll_procs - calculates the number of polling procs 277 * p9_mux_calc_poll_procs - calculates the number of polling procs
210 * based on the number of mounted v9fs filesystems. 278 * @muxnum: number of mounts
211 * 279 *
280 * Calculation is based on the number of mounted v9fs filesystems.
212 * The current implementation returns sqrt of the number of mounts. 281 * The current implementation returns sqrt of the number of mounts.
213 */ 282 */
283
214static int p9_mux_calc_poll_procs(int muxnum) 284static int p9_mux_calc_poll_procs(int muxnum)
215{ 285{
216 int n; 286 int n;
@@ -331,12 +401,11 @@ static void p9_mux_poll_stop(struct p9_conn *m)
331 401
332/** 402/**
333 * p9_conn_create - allocate and initialize the per-session mux data 403 * p9_conn_create - allocate and initialize the per-session mux data
334 * Creates the polling task if this is the first session. 404 * @trans: transport structure
335 * 405 *
336 * @trans - transport structure 406 * Note: Creates the polling task if this is the first session.
337 * @msize - maximum message size
338 * @extended - extended flag
339 */ 407 */
408
340static struct p9_conn *p9_conn_create(struct p9_trans *trans) 409static struct p9_conn *p9_conn_create(struct p9_trans *trans)
341{ 410{
342 int i, n; 411 int i, n;
@@ -406,7 +475,10 @@ static struct p9_conn *p9_conn_create(struct p9_trans *trans)
406 475
407/** 476/**
408 * p9_mux_destroy - cancels all pending requests and frees mux resources 477 * p9_mux_destroy - cancels all pending requests and frees mux resources
478 * @m: mux to destroy
479 *
409 */ 480 */
481
410static void p9_conn_destroy(struct p9_conn *m) 482static void p9_conn_destroy(struct p9_conn *m)
411{ 483{
412 P9_DPRINTK(P9_DEBUG_MUX, "mux %p prev %p next %p\n", m, 484 P9_DPRINTK(P9_DEBUG_MUX, "mux %p prev %p next %p\n", m,
@@ -429,9 +501,14 @@ static void p9_conn_destroy(struct p9_conn *m)
429} 501}
430 502
431/** 503/**
432 * p9_pollwait - called by files poll operation to add v9fs-poll task 504 * p9_pollwait - add poll task to the wait queue
433 * to files wait queue 505 * @filp: file pointer being polled
506 * @wait_address: wait_q to block on
507 * @p: poll state
508 *
509 * called by files poll operation to add v9fs-poll task to files wait queue
434 */ 510 */
511
435static void 512static void
436p9_pollwait(struct file *filp, wait_queue_head_t *wait_address, poll_table *p) 513p9_pollwait(struct file *filp, wait_queue_head_t *wait_address, poll_table *p)
437{ 514{
@@ -462,7 +539,10 @@ p9_pollwait(struct file *filp, wait_queue_head_t *wait_address, poll_table *p)
462 539
463/** 540/**
464 * p9_poll_mux - polls a mux and schedules read or write works if necessary 541 * p9_poll_mux - polls a mux and schedules read or write works if necessary
542 * @m: connection to poll
543 *
465 */ 544 */
545
466static void p9_poll_mux(struct p9_conn *m) 546static void p9_poll_mux(struct p9_conn *m)
467{ 547{
468 int n; 548 int n;
@@ -499,9 +579,14 @@ static void p9_poll_mux(struct p9_conn *m)
499} 579}
500 580
501/** 581/**
502 * p9_poll_proc - polls all v9fs transports for new events and queues 582 * p9_poll_proc - poll worker thread
503 * the appropriate work to the work queue 583 * @a: thread state and arguments
584 *
585 * polls all v9fs transports for new events and queues the appropriate
586 * work to the work queue
587 *
504 */ 588 */
589
505static int p9_poll_proc(void *a) 590static int p9_poll_proc(void *a)
506{ 591{
507 struct p9_conn *m, *mtmp; 592 struct p9_conn *m, *mtmp;
@@ -527,7 +612,10 @@ static int p9_poll_proc(void *a)
527 612
528/** 613/**
529 * p9_write_work - called when a transport can send some data 614 * p9_write_work - called when a transport can send some data
615 * @work: container for work to be done
616 *
530 */ 617 */
618
531static void p9_write_work(struct work_struct *work) 619static void p9_write_work(struct work_struct *work)
532{ 620{
533 int n, err; 621 int n, err;
@@ -638,7 +726,10 @@ static void process_request(struct p9_conn *m, struct p9_req *req)
638 726
639/** 727/**
640 * p9_read_work - called when there is some data to be read from a transport 728 * p9_read_work - called when there is some data to be read from a transport
729 * @work: container of work to be done
730 *
641 */ 731 */
732
642static void p9_read_work(struct work_struct *work) 733static void p9_read_work(struct work_struct *work)
643{ 734{
644 int n, err; 735 int n, err;
@@ -793,7 +884,9 @@ error:
793 * @tc: request to be sent 884 * @tc: request to be sent
794 * @cb: callback function to call when response is received 885 * @cb: callback function to call when response is received
795 * @cba: parameter to pass to the callback function 886 * @cba: parameter to pass to the callback function
887 *
796 */ 888 */
889
797static struct p9_req *p9_send_request(struct p9_conn *m, 890static struct p9_req *p9_send_request(struct p9_conn *m,
798 struct p9_fcall *tc, 891 struct p9_fcall *tc,
799 p9_conn_req_callback cb, void *cba) 892 p9_conn_req_callback cb, void *cba)
@@ -961,10 +1054,12 @@ p9_conn_rpc_cb(struct p9_req *req, void *a)
961/** 1054/**
962 * p9_fd_rpc- sends 9P request and waits until a response is available. 1055 * p9_fd_rpc- sends 9P request and waits until a response is available.
963 * The function can be interrupted. 1056 * The function can be interrupted.
964 * @m: mux data 1057 * @t: transport data
965 * @tc: request to be sent 1058 * @tc: request to be sent
966 * @rc: pointer where a pointer to the response is stored 1059 * @rc: pointer where a pointer to the response is stored
1060 *
967 */ 1061 */
1062
968int 1063int
969p9_fd_rpc(struct p9_trans *t, struct p9_fcall *tc, struct p9_fcall **rc) 1064p9_fd_rpc(struct p9_trans *t, struct p9_fcall *tc, struct p9_fcall **rc)
970{ 1065{
@@ -1041,8 +1136,10 @@ p9_fd_rpc(struct p9_trans *t, struct p9_fcall *tc, struct p9_fcall **rc)
1041 * @m: mux data 1136 * @m: mux data
1042 * @tc: request to be sent 1137 * @tc: request to be sent
1043 * @cb: callback function to be called when response arrives 1138 * @cb: callback function to be called when response arrives
1044 * @cba: value to pass to the callback function 1139 * @a: value to pass to the callback function
1140 *
1045 */ 1141 */
1142
1046int p9_conn_rpcnb(struct p9_conn *m, struct p9_fcall *tc, 1143int p9_conn_rpcnb(struct p9_conn *m, struct p9_fcall *tc,
1047 p9_conn_req_callback cb, void *a) 1144 p9_conn_req_callback cb, void *a)
1048{ 1145{
@@ -1065,7 +1162,9 @@ int p9_conn_rpcnb(struct p9_conn *m, struct p9_fcall *tc,
1065 * p9_conn_cancel - cancel all pending requests with error 1162 * p9_conn_cancel - cancel all pending requests with error
1066 * @m: mux data 1163 * @m: mux data
1067 * @err: error code 1164 * @err: error code
1165 *
1068 */ 1166 */
1167
1069void p9_conn_cancel(struct p9_conn *m, int err) 1168void p9_conn_cancel(struct p9_conn *m, int err)
1070{ 1169{
1071 struct p9_req *req, *rtmp; 1170 struct p9_req *req, *rtmp;
@@ -1097,35 +1196,46 @@ void p9_conn_cancel(struct p9_conn *m, int err)
1097} 1196}
1098 1197
1099/** 1198/**
1100 * v9fs_parse_options - parse mount options into session structure 1199 * parse_options - parse mount options into session structure
1101 * @options: options string passed from mount 1200 * @options: options string passed from mount
1102 * @v9ses: existing v9fs session information 1201 * @opts: transport-specific structure to parse options into
1103 * 1202 *
1203 * Returns 0 upon success, -ERRNO upon failure
1104 */ 1204 */
1105 1205
1106static void parse_opts(char *options, struct p9_fd_opts *opts) 1206static int parse_opts(char *params, struct p9_fd_opts *opts)
1107{ 1207{
1108 char *p; 1208 char *p;
1109 substring_t args[MAX_OPT_ARGS]; 1209 substring_t args[MAX_OPT_ARGS];
1110 int option; 1210 int option;
1211 char *options;
1111 int ret; 1212 int ret;
1112 1213
1113 opts->port = P9_PORT; 1214 opts->port = P9_PORT;
1114 opts->rfd = ~0; 1215 opts->rfd = ~0;
1115 opts->wfd = ~0; 1216 opts->wfd = ~0;
1116 1217
1117 if (!options) 1218 if (!params)
1118 return; 1219 return 0;
1220
1221 options = kstrdup(params, GFP_KERNEL);
1222 if (!options) {
1223 P9_DPRINTK(P9_DEBUG_ERROR,
1224 "failed to allocate copy of option string\n");
1225 return -ENOMEM;
1226 }
1119 1227
1120 while ((p = strsep(&options, ",")) != NULL) { 1228 while ((p = strsep(&options, ",")) != NULL) {
1121 int token; 1229 int token;
1230 int r;
1122 if (!*p) 1231 if (!*p)
1123 continue; 1232 continue;
1124 token = match_token(p, tokens, args); 1233 token = match_token(p, tokens, args);
1125 ret = match_int(&args[0], &option); 1234 r = match_int(&args[0], &option);
1126 if (ret < 0) { 1235 if (r < 0) {
1127 P9_DPRINTK(P9_DEBUG_ERROR, 1236 P9_DPRINTK(P9_DEBUG_ERROR,
1128 "integer field, but no integer?\n"); 1237 "integer field, but no integer?\n");
1238 ret = r;
1129 continue; 1239 continue;
1130 } 1240 }
1131 switch (token) { 1241 switch (token) {
@@ -1142,6 +1252,8 @@ static void parse_opts(char *options, struct p9_fd_opts *opts)
1142 continue; 1252 continue;
1143 } 1253 }
1144 } 1254 }
1255 kfree(options);
1256 return 0;
1145} 1257}
1146 1258
1147static int p9_fd_open(struct p9_trans *trans, int rfd, int wfd) 1259static int p9_fd_open(struct p9_trans *trans, int rfd, int wfd)
@@ -1193,11 +1305,12 @@ static int p9_socket_open(struct p9_trans *trans, struct socket *csocket)
1193 1305
1194/** 1306/**
1195 * p9_fd_read- read from a fd 1307 * p9_fd_read- read from a fd
1196 * @v9ses: session information 1308 * @trans: transport instance state
1197 * @v: buffer to receive data into 1309 * @v: buffer to receive data into
1198 * @len: size of receive buffer 1310 * @len: size of receive buffer
1199 * 1311 *
1200 */ 1312 */
1313
1201static int p9_fd_read(struct p9_trans *trans, void *v, int len) 1314static int p9_fd_read(struct p9_trans *trans, void *v, int len)
1202{ 1315{
1203 int ret; 1316 int ret;
@@ -1220,11 +1333,12 @@ static int p9_fd_read(struct p9_trans *trans, void *v, int len)
1220 1333
1221/** 1334/**
1222 * p9_fd_write - write to a socket 1335 * p9_fd_write - write to a socket
1223 * @v9ses: session information 1336 * @trans: transport instance state
1224 * @v: buffer to send data from 1337 * @v: buffer to send data from
1225 * @len: size of send buffer 1338 * @len: size of send buffer
1226 * 1339 *
1227 */ 1340 */
1341
1228static int p9_fd_write(struct p9_trans *trans, void *v, int len) 1342static int p9_fd_write(struct p9_trans *trans, void *v, int len)
1229{ 1343{
1230 int ret; 1344 int ret;
@@ -1296,6 +1410,7 @@ end:
1296 * @trans: private socket structure 1410 * @trans: private socket structure
1297 * 1411 *
1298 */ 1412 */
1413
1299static void p9_fd_close(struct p9_trans *trans) 1414static void p9_fd_close(struct p9_trans *trans)
1300{ 1415{
1301 struct p9_trans_fd *ts; 1416 struct p9_trans_fd *ts;
@@ -1318,6 +1433,23 @@ static void p9_fd_close(struct p9_trans *trans)
1318 kfree(ts); 1433 kfree(ts);
1319} 1434}
1320 1435
1436/*
1437 * stolen from NFS - maybe should be made a generic function?
1438 */
1439static inline int valid_ipaddr4(const char *buf)
1440{
1441 int rc, count, in[4];
1442
1443 rc = sscanf(buf, "%d.%d.%d.%d", &in[0], &in[1], &in[2], &in[3]);
1444 if (rc != 4)
1445 return -EINVAL;
1446 for (count = 0; count < 4; count++) {
1447 if (in[count] > 255)
1448 return -EINVAL;
1449 }
1450 return 0;
1451}
1452
1321static struct p9_trans * 1453static struct p9_trans *
1322p9_trans_create_tcp(const char *addr, char *args, int msize, unsigned char dotu) 1454p9_trans_create_tcp(const char *addr, char *args, int msize, unsigned char dotu)
1323{ 1455{
@@ -1328,7 +1460,12 @@ p9_trans_create_tcp(const char *addr, char *args, int msize, unsigned char dotu)
1328 struct p9_fd_opts opts; 1460 struct p9_fd_opts opts;
1329 struct p9_trans_fd *p; 1461 struct p9_trans_fd *p;
1330 1462
1331 parse_opts(args, &opts); 1463 err = parse_opts(args, &opts);
1464 if (err < 0)
1465 return ERR_PTR(err);
1466
1467 if (valid_ipaddr4(addr) < 0)
1468 return ERR_PTR(-EINVAL);
1332 1469
1333 csocket = NULL; 1470 csocket = NULL;
1334 trans = kmalloc(sizeof(struct p9_trans), GFP_KERNEL); 1471 trans = kmalloc(sizeof(struct p9_trans), GFP_KERNEL);
@@ -1508,7 +1645,7 @@ static struct p9_trans_module p9_fd_trans = {
1508 .create = p9_trans_create_fd, 1645 .create = p9_trans_create_fd,
1509}; 1646};
1510 1647
1511static int __init p9_trans_fd_init(void) 1648int p9_trans_fd_init(void)
1512{ 1649{
1513 int ret = p9_mux_global_init(); 1650 int ret = p9_mux_global_init();
1514 if (ret) { 1651 if (ret) {
@@ -1522,9 +1659,4 @@ static int __init p9_trans_fd_init(void)
1522 1659
1523 return 0; 1660 return 0;
1524} 1661}
1525 1662EXPORT_SYMBOL(p9_trans_fd_init);
1526module_init(p9_trans_fd_init);
1527
1528MODULE_AUTHOR("Latchesar Ionkov <lucho@ionkov.net>");
1529MODULE_AUTHOR("Eric Van Hensbergen <ericvh@gmail.com>");
1530MODULE_LICENSE("GPL");
diff --git a/net/9p/trans_virtio.c b/net/9p/trans_virtio.c
index de7a9f532edc..42adc052b149 100644
--- a/net/9p/trans_virtio.c
+++ b/net/9p/trans_virtio.c
@@ -49,29 +49,75 @@
49#define VIRTQUEUE_NUM 128 49#define VIRTQUEUE_NUM 128
50 50
51/* a single mutex to manage channel initialization and attachment */ 51/* a single mutex to manage channel initialization and attachment */
52static DECLARE_MUTEX(virtio_9p_lock); 52static DEFINE_MUTEX(virtio_9p_lock);
53/* global which tracks highest initialized channel */ 53/* global which tracks highest initialized channel */
54static int chan_index; 54static int chan_index;
55 55
56#define P9_INIT_MAXTAG 16 56#define P9_INIT_MAXTAG 16
57 57
58#define REQ_STATUS_IDLE 0 58
59#define REQ_STATUS_SENT 1 59/**
60#define REQ_STATUS_RCVD 2 60 * enum p9_req_status_t - virtio request status
61#define REQ_STATUS_FLSH 3 61 * @REQ_STATUS_IDLE: request slot unused
62 * @REQ_STATUS_SENT: request sent to server
63 * @REQ_STATUS_RCVD: response received from server
64 * @REQ_STATUS_FLSH: request has been flushed
65 *
66 * The @REQ_STATUS_IDLE state is used to mark a request slot as unused
67 * but use is actually tracked by the idpool structure which handles tag
68 * id allocation.
69 *
70 */
71
72enum p9_req_status_t {
73 REQ_STATUS_IDLE,
74 REQ_STATUS_SENT,
75 REQ_STATUS_RCVD,
76 REQ_STATUS_FLSH,
77};
78
79/**
80 * struct p9_req_t - virtio request slots
81 * @status: status of this request slot
82 * @wq: wait_queue for the client to block on for this request
83 *
84 * The virtio transport uses an array to track outstanding requests
85 * instead of a list. While this may incurr overhead during initial
86 * allocation or expansion, it makes request lookup much easier as the
87 * tag id is a index into an array. (We use tag+1 so that we can accomodate
88 * the -1 tag for the T_VERSION request).
89 * This also has the nice effect of only having to allocate wait_queues
90 * once, instead of constantly allocating and freeing them. Its possible
91 * other resources could benefit from this scheme as well.
92 *
93 */
62 94
63struct p9_req_t { 95struct p9_req_t {
64 int status; 96 int status;
65 wait_queue_head_t *wq; 97 wait_queue_head_t *wq;
66}; 98};
67 99
68/* We keep all per-channel information in a structure. 100/**
101 * struct virtio_chan - per-instance transport information
102 * @initialized: whether the channel is initialized
103 * @inuse: whether the channel is in use
104 * @lock: protects multiple elements within this structure
105 * @vdev: virtio dev associated with this channel
106 * @vq: virtio queue associated with this channel
107 * @tagpool: accounting for tag ids (and request slots)
108 * @reqs: array of request slots
109 * @max_tag: current number of request_slots allocated
110 * @sg: scatter gather list which is used to pack a request (protected?)
111 *
112 * We keep all per-channel information in a structure.
69 * This structure is allocated within the devices dev->mem space. 113 * This structure is allocated within the devices dev->mem space.
70 * A pointer to the structure will get put in the transport private. 114 * A pointer to the structure will get put in the transport private.
115 *
71 */ 116 */
117
72static struct virtio_chan { 118static struct virtio_chan {
73 bool initialized; /* channel is initialized */ 119 bool initialized;
74 bool inuse; /* channel is in use */ 120 bool inuse;
75 121
76 spinlock_t lock; 122 spinlock_t lock;
77 123
@@ -86,7 +132,19 @@ static struct virtio_chan {
86 struct scatterlist sg[VIRTQUEUE_NUM]; 132 struct scatterlist sg[VIRTQUEUE_NUM];
87} channels[MAX_9P_CHAN]; 133} channels[MAX_9P_CHAN];
88 134
89/* Lookup requests by tag */ 135/**
136 * p9_lookup_tag - Lookup requests by tag
137 * @c: virtio channel to lookup tag within
138 * @tag: numeric id for transaction
139 *
140 * this is a simple array lookup, but will grow the
141 * request_slots as necessary to accomodate transaction
142 * ids which did not previously have a slot.
143 *
144 * Bugs: there is currently no upper limit on request slots set
145 * here, but that should be constrained by the id accounting.
146 */
147
90static struct p9_req_t *p9_lookup_tag(struct virtio_chan *c, u16 tag) 148static struct p9_req_t *p9_lookup_tag(struct virtio_chan *c, u16 tag)
91{ 149{
92 /* This looks up the original request by tag so we know which 150 /* This looks up the original request by tag so we know which
@@ -130,11 +188,20 @@ static unsigned int rest_of_page(void *data)
130 return PAGE_SIZE - ((unsigned long)data % PAGE_SIZE); 188 return PAGE_SIZE - ((unsigned long)data % PAGE_SIZE);
131} 189}
132 190
191/**
192 * p9_virtio_close - reclaim resources of a channel
193 * @trans: transport state
194 *
195 * This reclaims a channel by freeing its resources and
196 * reseting its inuse flag.
197 *
198 */
199
133static void p9_virtio_close(struct p9_trans *trans) 200static void p9_virtio_close(struct p9_trans *trans)
134{ 201{
135 struct virtio_chan *chan = trans->priv; 202 struct virtio_chan *chan = trans->priv;
136 int count; 203 int count;
137 unsigned int flags; 204 unsigned long flags;
138 205
139 spin_lock_irqsave(&chan->lock, flags); 206 spin_lock_irqsave(&chan->lock, flags);
140 p9_idpool_destroy(chan->tagpool); 207 p9_idpool_destroy(chan->tagpool);
@@ -144,13 +211,26 @@ static void p9_virtio_close(struct p9_trans *trans)
144 chan->max_tag = 0; 211 chan->max_tag = 0;
145 spin_unlock_irqrestore(&chan->lock, flags); 212 spin_unlock_irqrestore(&chan->lock, flags);
146 213
147 down(&virtio_9p_lock); 214 mutex_lock(&virtio_9p_lock);
148 chan->inuse = false; 215 chan->inuse = false;
149 up(&virtio_9p_lock); 216 mutex_unlock(&virtio_9p_lock);
150 217
151 kfree(trans); 218 kfree(trans);
152} 219}
153 220
221/**
222 * req_done - callback which signals activity from the server
223 * @vq: virtio queue activity was received on
224 *
225 * This notifies us that the server has triggered some activity
226 * on the virtio channel - most likely a response to request we
227 * sent. Figure out which requests now have responses and wake up
228 * those threads.
229 *
230 * Bugs: could do with some additional sanity checking, but appears to work.
231 *
232 */
233
154static void req_done(struct virtqueue *vq) 234static void req_done(struct virtqueue *vq)
155{ 235{
156 struct virtio_chan *chan = vq->vdev->priv; 236 struct virtio_chan *chan = vq->vdev->priv;
@@ -169,6 +249,20 @@ static void req_done(struct virtqueue *vq)
169 spin_unlock_irqrestore(&chan->lock, flags); 249 spin_unlock_irqrestore(&chan->lock, flags);
170} 250}
171 251
252/**
253 * pack_sg_list - pack a scatter gather list from a linear buffer
254 * @sg: scatter/gather list to pack into
255 * @start: which segment of the sg_list to start at
256 * @limit: maximum segment to pack data to
257 * @data: data to pack into scatter/gather list
258 * @count: amount of data to pack into the scatter/gather list
259 *
260 * sg_lists have multiple segments of various sizes. This will pack
261 * arbitrary data into an existing scatter gather list, segmenting the
262 * data as necessary within constraints.
263 *
264 */
265
172static int 266static int
173pack_sg_list(struct scatterlist *sg, int start, int limit, char *data, 267pack_sg_list(struct scatterlist *sg, int start, int limit, char *data,
174 int count) 268 int count)
@@ -189,6 +283,14 @@ pack_sg_list(struct scatterlist *sg, int start, int limit, char *data,
189 return index-start; 283 return index-start;
190} 284}
191 285
286/**
287 * p9_virtio_rpc - issue a request and wait for a response
288 * @t: transport state
289 * @tc: &p9_fcall request to transmit
290 * @rc: &p9_fcall to put reponse into
291 *
292 */
293
192static int 294static int
193p9_virtio_rpc(struct p9_trans *t, struct p9_fcall *tc, struct p9_fcall **rc) 295p9_virtio_rpc(struct p9_trans *t, struct p9_fcall *tc, struct p9_fcall **rc)
194{ 296{
@@ -263,16 +365,26 @@ p9_virtio_rpc(struct p9_trans *t, struct p9_fcall *tc, struct p9_fcall **rc)
263 return 0; 365 return 0;
264} 366}
265 367
368/**
369 * p9_virtio_probe - probe for existence of 9P virtio channels
370 * @vdev: virtio device to probe
371 *
372 * This probes for existing virtio channels. At present only
373 * a single channel is in use, so in the future more work may need
374 * to be done here.
375 *
376 */
377
266static int p9_virtio_probe(struct virtio_device *vdev) 378static int p9_virtio_probe(struct virtio_device *vdev)
267{ 379{
268 int err; 380 int err;
269 struct virtio_chan *chan; 381 struct virtio_chan *chan;
270 int index; 382 int index;
271 383
272 down(&virtio_9p_lock); 384 mutex_lock(&virtio_9p_lock);
273 index = chan_index++; 385 index = chan_index++;
274 chan = &channels[index]; 386 chan = &channels[index];
275 up(&virtio_9p_lock); 387 mutex_unlock(&virtio_9p_lock);
276 388
277 if (chan_index > MAX_9P_CHAN) { 389 if (chan_index > MAX_9P_CHAN) {
278 printk(KERN_ERR "9p: virtio: Maximum channels exceeded\n"); 390 printk(KERN_ERR "9p: virtio: Maximum channels exceeded\n");
@@ -301,17 +413,34 @@ static int p9_virtio_probe(struct virtio_device *vdev)
301out_free_vq: 413out_free_vq:
302 vdev->config->del_vq(chan->vq); 414 vdev->config->del_vq(chan->vq);
303fail: 415fail:
304 down(&virtio_9p_lock); 416 mutex_lock(&virtio_9p_lock);
305 chan_index--; 417 chan_index--;
306 up(&virtio_9p_lock); 418 mutex_unlock(&virtio_9p_lock);
307 return err; 419 return err;
308} 420}
309 421
310/* This sets up a transport channel for 9p communication. Right now 422
423/**
424 * p9_virtio_create - allocate a new virtio channel
425 * @devname: string identifying the channel to connect to (unused)
426 * @args: args passed from sys_mount() for per-transport options (unused)
427 * @msize: requested maximum packet size
428 * @extended: 9p2000.u enabled flag
429 *
430 * This sets up a transport channel for 9p communication. Right now
311 * we only match the first available channel, but eventually we couldlook up 431 * we only match the first available channel, but eventually we couldlook up
312 * alternate channels by matching devname versus a virtio_config entry. 432 * alternate channels by matching devname versus a virtio_config entry.
313 * We use a simple reference count mechanism to ensure that only a single 433 * We use a simple reference count mechanism to ensure that only a single
314 * mount has a channel open at a time. */ 434 * mount has a channel open at a time.
435 *
436 * Bugs: doesn't allow identification of a specific channel
437 * to allocate, channels are allocated sequentially. This was
438 * a pragmatic decision to get things rolling, but ideally some
439 * way of identifying the channel to attach to would be nice
440 * if we are going to support multiple channels.
441 *
442 */
443
315static struct p9_trans * 444static struct p9_trans *
316p9_virtio_create(const char *devname, char *args, int msize, 445p9_virtio_create(const char *devname, char *args, int msize,
317 unsigned char extended) 446 unsigned char extended)
@@ -320,7 +449,7 @@ p9_virtio_create(const char *devname, char *args, int msize,
320 struct virtio_chan *chan = channels; 449 struct virtio_chan *chan = channels;
321 int index = 0; 450 int index = 0;
322 451
323 down(&virtio_9p_lock); 452 mutex_lock(&virtio_9p_lock);
324 while (index < MAX_9P_CHAN) { 453 while (index < MAX_9P_CHAN) {
325 if (chan->initialized && !chan->inuse) { 454 if (chan->initialized && !chan->inuse) {
326 chan->inuse = true; 455 chan->inuse = true;
@@ -330,7 +459,7 @@ p9_virtio_create(const char *devname, char *args, int msize,
330 chan = &channels[index]; 459 chan = &channels[index];
331 } 460 }
332 } 461 }
333 up(&virtio_9p_lock); 462 mutex_unlock(&virtio_9p_lock);
334 463
335 if (index >= MAX_9P_CHAN) { 464 if (index >= MAX_9P_CHAN) {
336 printk(KERN_ERR "9p: no channels available\n"); 465 printk(KERN_ERR "9p: no channels available\n");
@@ -360,6 +489,12 @@ p9_virtio_create(const char *devname, char *args, int msize,
360 return trans; 489 return trans;
361} 490}
362 491
492/**
493 * p9_virtio_remove - clean up resources associated with a virtio device
494 * @vdev: virtio device to remove
495 *
496 */
497
363static void p9_virtio_remove(struct virtio_device *vdev) 498static void p9_virtio_remove(struct virtio_device *vdev)
364{ 499{
365 struct virtio_chan *chan = vdev->priv; 500 struct virtio_chan *chan = vdev->priv;
diff --git a/net/9p/util.c b/net/9p/util.c
index ef7215565d88..958fc58cd1ff 100644
--- a/net/9p/util.c
+++ b/net/9p/util.c
@@ -32,11 +32,23 @@
32#include <linux/idr.h> 32#include <linux/idr.h>
33#include <net/9p/9p.h> 33#include <net/9p/9p.h>
34 34
35/**
36 * struct p9_idpool - per-connection accounting for tag idpool
37 * @lock: protects the pool
38 * @pool: idr to allocate tag id from
39 *
40 */
41
35struct p9_idpool { 42struct p9_idpool {
36 spinlock_t lock; 43 spinlock_t lock;
37 struct idr pool; 44 struct idr pool;
38}; 45};
39 46
47/**
48 * p9_idpool_create - create a new per-connection id pool
49 *
50 */
51
40struct p9_idpool *p9_idpool_create(void) 52struct p9_idpool *p9_idpool_create(void)
41{ 53{
42 struct p9_idpool *p; 54 struct p9_idpool *p;
@@ -52,6 +64,11 @@ struct p9_idpool *p9_idpool_create(void)
52} 64}
53EXPORT_SYMBOL(p9_idpool_create); 65EXPORT_SYMBOL(p9_idpool_create);
54 66
67/**
68 * p9_idpool_destroy - create a new per-connection id pool
69 * @p: idpool to destory
70 */
71
55void p9_idpool_destroy(struct p9_idpool *p) 72void p9_idpool_destroy(struct p9_idpool *p)
56{ 73{
57 idr_destroy(&p->pool); 74 idr_destroy(&p->pool);
@@ -61,9 +78,9 @@ EXPORT_SYMBOL(p9_idpool_destroy);
61 78
62/** 79/**
63 * p9_idpool_get - allocate numeric id from pool 80 * p9_idpool_get - allocate numeric id from pool
64 * @p - pool to allocate from 81 * @p: pool to allocate from
65 * 82 *
66 * XXX - This seems to be an awful generic function, should it be in idr.c with 83 * Bugs: This seems to be an awful generic function, should it be in idr.c with
67 * the lock included in struct idr? 84 * the lock included in struct idr?
68 */ 85 */
69 86
@@ -71,7 +88,7 @@ int p9_idpool_get(struct p9_idpool *p)
71{ 88{
72 int i = 0; 89 int i = 0;
73 int error; 90 int error;
74 unsigned int flags; 91 unsigned long flags;
75 92
76retry: 93retry:
77 if (idr_pre_get(&p->pool, GFP_KERNEL) == 0) 94 if (idr_pre_get(&p->pool, GFP_KERNEL) == 0)
@@ -94,15 +111,16 @@ EXPORT_SYMBOL(p9_idpool_get);
94 111
95/** 112/**
96 * p9_idpool_put - release numeric id from pool 113 * p9_idpool_put - release numeric id from pool
97 * @p - pool to allocate from 114 * @id: numeric id which is being released
115 * @p: pool to release id into
98 * 116 *
99 * XXX - This seems to be an awful generic function, should it be in idr.c with 117 * Bugs: This seems to be an awful generic function, should it be in idr.c with
100 * the lock included in struct idr? 118 * the lock included in struct idr?
101 */ 119 */
102 120
103void p9_idpool_put(int id, struct p9_idpool *p) 121void p9_idpool_put(int id, struct p9_idpool *p)
104{ 122{
105 unsigned int flags; 123 unsigned long flags;
106 spin_lock_irqsave(&p->lock, flags); 124 spin_lock_irqsave(&p->lock, flags);
107 idr_remove(&p->pool, id); 125 idr_remove(&p->pool, id);
108 spin_unlock_irqrestore(&p->lock, flags); 126 spin_unlock_irqrestore(&p->lock, flags);
@@ -111,11 +129,13 @@ EXPORT_SYMBOL(p9_idpool_put);
111 129
112/** 130/**
113 * p9_idpool_check - check if the specified id is available 131 * p9_idpool_check - check if the specified id is available
114 * @id - id to check 132 * @id: id to check
115 * @p - pool 133 * @p: pool to check
116 */ 134 */
135
117int p9_idpool_check(int id, struct p9_idpool *p) 136int p9_idpool_check(int id, struct p9_idpool *p)
118{ 137{
119 return idr_find(&p->pool, id) != NULL; 138 return idr_find(&p->pool, id) != NULL;
120} 139}
121EXPORT_SYMBOL(p9_idpool_check); 140EXPORT_SYMBOL(p9_idpool_check);
141
diff --git a/net/atm/br2684.c b/net/atm/br2684.c
index 1b228065e745..05fafdc2eea3 100644
--- a/net/atm/br2684.c
+++ b/net/atm/br2684.c
@@ -188,10 +188,13 @@ static int br2684_xmit_vcc(struct sk_buff *skb, struct br2684_dev *brdev,
188 return 0; 188 return 0;
189 } 189 }
190 } 190 }
191 } else { 191 } else { /* e_vc */
192 skb_push(skb, 2); 192 if (brdev->payload == p_bridged) {
193 if (brdev->payload == p_bridged) 193 skb_push(skb, 2);
194 memset(skb->data, 0, 2); 194 memset(skb->data, 0, 2);
195 } else { /* p_routed */
196 skb_pull(skb, ETH_HLEN);
197 }
195 } 198 }
196 skb_debug(skb); 199 skb_debug(skb);
197 200
@@ -346,9 +349,9 @@ static void br2684_push(struct atm_vcc *atmvcc, struct sk_buff *skb)
346 /* skb==NULL means VCC is being destroyed */ 349 /* skb==NULL means VCC is being destroyed */
347 br2684_close_vcc(brvcc); 350 br2684_close_vcc(brvcc);
348 if (list_empty(&brdev->brvccs)) { 351 if (list_empty(&brdev->brvccs)) {
349 read_lock(&devs_lock); 352 write_lock_irq(&devs_lock);
350 list_del(&brdev->br2684_devs); 353 list_del(&brdev->br2684_devs);
351 read_unlock(&devs_lock); 354 write_unlock_irq(&devs_lock);
352 unregister_netdev(net_dev); 355 unregister_netdev(net_dev);
353 free_netdev(net_dev); 356 free_netdev(net_dev);
354 } 357 }
@@ -377,11 +380,8 @@ static void br2684_push(struct atm_vcc *atmvcc, struct sk_buff *skb)
377 (skb->data + 6, ethertype_ipv4, 380 (skb->data + 6, ethertype_ipv4,
378 sizeof(ethertype_ipv4)) == 0) 381 sizeof(ethertype_ipv4)) == 0)
379 skb->protocol = __constant_htons(ETH_P_IP); 382 skb->protocol = __constant_htons(ETH_P_IP);
380 else { 383 else
381 brdev->stats.rx_errors++; 384 goto error;
382 dev_kfree_skb(skb);
383 return;
384 }
385 skb_pull(skb, sizeof(llc_oui_ipv4)); 385 skb_pull(skb, sizeof(llc_oui_ipv4));
386 skb_reset_network_header(skb); 386 skb_reset_network_header(skb);
387 skb->pkt_type = PACKET_HOST; 387 skb->pkt_type = PACKET_HOST;
@@ -394,44 +394,56 @@ static void br2684_push(struct atm_vcc *atmvcc, struct sk_buff *skb)
394 (memcmp(skb->data, llc_oui_pid_pad, 7) == 0)) { 394 (memcmp(skb->data, llc_oui_pid_pad, 7) == 0)) {
395 skb_pull(skb, sizeof(llc_oui_pid_pad)); 395 skb_pull(skb, sizeof(llc_oui_pid_pad));
396 skb->protocol = eth_type_trans(skb, net_dev); 396 skb->protocol = eth_type_trans(skb, net_dev);
397 } else { 397 } else
398 brdev->stats.rx_errors++; 398 goto error;
399 dev_kfree_skb(skb);
400 return;
401 }
402 399
403 } else { 400 } else { /* e_vc */
404 /* first 2 chars should be 0 */ 401 if (brdev->payload == p_routed) {
405 if (*((u16 *) (skb->data)) != 0) { 402 struct iphdr *iph;
406 brdev->stats.rx_errors++; 403
407 dev_kfree_skb(skb); 404 skb_reset_network_header(skb);
408 return; 405 iph = ip_hdr(skb);
406 if (iph->version == 4)
407 skb->protocol = __constant_htons(ETH_P_IP);
408 else if (iph->version == 6)
409 skb->protocol = __constant_htons(ETH_P_IPV6);
410 else
411 goto error;
412 skb->pkt_type = PACKET_HOST;
413 } else { /* p_bridged */
414 /* first 2 chars should be 0 */
415 if (*((u16 *) (skb->data)) != 0)
416 goto error;
417 skb_pull(skb, BR2684_PAD_LEN);
418 skb->protocol = eth_type_trans(skb, net_dev);
409 } 419 }
410 skb_pull(skb, BR2684_PAD_LEN + ETH_HLEN); /* pad, dstmac, srcmac, ethtype */
411 skb->protocol = eth_type_trans(skb, net_dev);
412 } 420 }
413 421
414#ifdef CONFIG_ATM_BR2684_IPFILTER 422#ifdef CONFIG_ATM_BR2684_IPFILTER
415 if (unlikely(packet_fails_filter(skb->protocol, brvcc, skb))) { 423 if (unlikely(packet_fails_filter(skb->protocol, brvcc, skb)))
416 brdev->stats.rx_dropped++; 424 goto dropped;
417 dev_kfree_skb(skb);
418 return;
419 }
420#endif /* CONFIG_ATM_BR2684_IPFILTER */ 425#endif /* CONFIG_ATM_BR2684_IPFILTER */
421 skb->dev = net_dev; 426 skb->dev = net_dev;
422 ATM_SKB(skb)->vcc = atmvcc; /* needed ? */ 427 ATM_SKB(skb)->vcc = atmvcc; /* needed ? */
423 pr_debug("received packet's protocol: %x\n", ntohs(skb->protocol)); 428 pr_debug("received packet's protocol: %x\n", ntohs(skb->protocol));
424 skb_debug(skb); 429 skb_debug(skb);
425 if (unlikely(!(net_dev->flags & IFF_UP))) { 430 /* sigh, interface is down? */
426 /* sigh, interface is down */ 431 if (unlikely(!(net_dev->flags & IFF_UP)))
427 brdev->stats.rx_dropped++; 432 goto dropped;
428 dev_kfree_skb(skb);
429 return;
430 }
431 brdev->stats.rx_packets++; 433 brdev->stats.rx_packets++;
432 brdev->stats.rx_bytes += skb->len; 434 brdev->stats.rx_bytes += skb->len;
433 memset(ATM_SKB(skb), 0, sizeof(struct atm_skb_data)); 435 memset(ATM_SKB(skb), 0, sizeof(struct atm_skb_data));
434 netif_rx(skb); 436 netif_rx(skb);
437 return;
438
439dropped:
440 brdev->stats.rx_dropped++;
441 goto free_skb;
442error:
443 brdev->stats.rx_errors++;
444free_skb:
445 dev_kfree_skb(skb);
446 return;
435} 447}
436 448
437/* 449/*
@@ -518,9 +530,9 @@ static int br2684_regvcc(struct atm_vcc *atmvcc, void __user * arg)
518 struct sk_buff *next = skb->next; 530 struct sk_buff *next = skb->next;
519 531
520 skb->next = skb->prev = NULL; 532 skb->next = skb->prev = NULL;
533 br2684_push(atmvcc, skb);
521 BRPRIV(skb->dev)->stats.rx_bytes -= skb->len; 534 BRPRIV(skb->dev)->stats.rx_bytes -= skb->len;
522 BRPRIV(skb->dev)->stats.rx_packets--; 535 BRPRIV(skb->dev)->stats.rx_packets--;
523 br2684_push(atmvcc, skb);
524 536
525 skb = next; 537 skb = next;
526 } 538 }
diff --git a/net/atm/proc.c b/net/atm/proc.c
index 5c9f3d148135..49487b313f22 100644
--- a/net/atm/proc.c
+++ b/net/atm/proc.c
@@ -417,12 +417,10 @@ int atm_proc_dev_register(struct atm_dev *dev)
417 goto err_out; 417 goto err_out;
418 sprintf(dev->proc_name,"%s:%d",dev->type, dev->number); 418 sprintf(dev->proc_name,"%s:%d",dev->type, dev->number);
419 419
420 dev->proc_entry = proc_create(dev->proc_name, 0, atm_proc_root, 420 dev->proc_entry = proc_create_data(dev->proc_name, 0, atm_proc_root,
421 &proc_atm_dev_ops); 421 &proc_atm_dev_ops, dev);
422 if (!dev->proc_entry) 422 if (!dev->proc_entry)
423 goto err_free_name; 423 goto err_free_name;
424 dev->proc_entry->data = dev;
425 dev->proc_entry->owner = THIS_MODULE;
426 return 0; 424 return 0;
427err_free_name: 425err_free_name:
428 kfree(dev->proc_name); 426 kfree(dev->proc_name);
diff --git a/net/ax25/ax25_subr.c b/net/ax25/ax25_subr.c
index d8f215733175..034aa10a5198 100644
--- a/net/ax25/ax25_subr.c
+++ b/net/ax25/ax25_subr.c
@@ -64,20 +64,15 @@ void ax25_frames_acked(ax25_cb *ax25, unsigned short nr)
64 64
65void ax25_requeue_frames(ax25_cb *ax25) 65void ax25_requeue_frames(ax25_cb *ax25)
66{ 66{
67 struct sk_buff *skb, *skb_prev = NULL; 67 struct sk_buff *skb;
68 68
69 /* 69 /*
70 * Requeue all the un-ack-ed frames on the output queue to be picked 70 * Requeue all the un-ack-ed frames on the output queue to be picked
71 * up by ax25_kick called from the timer. This arrangement handles the 71 * up by ax25_kick called from the timer. This arrangement handles the
72 * possibility of an empty output queue. 72 * possibility of an empty output queue.
73 */ 73 */
74 while ((skb = skb_dequeue(&ax25->ack_queue)) != NULL) { 74 while ((skb = skb_dequeue_tail(&ax25->ack_queue)) != NULL)
75 if (skb_prev == NULL) 75 skb_queue_head(&ax25->write_queue, skb);
76 skb_queue_head(&ax25->write_queue, skb);
77 else
78 skb_append(skb_prev, skb, &ax25->write_queue);
79 skb_prev = skb;
80 }
81} 76}
82 77
83/* 78/*
diff --git a/net/bluetooth/bnep/core.c b/net/bluetooth/bnep/core.c
index 347e935faaf0..f85d94643aaf 100644
--- a/net/bluetooth/bnep/core.c
+++ b/net/bluetooth/bnep/core.c
@@ -135,7 +135,7 @@ static int bnep_ctrl_set_netfilter(struct bnep_session *s, __be16 *data, int len
135 if (len < 2) 135 if (len < 2)
136 return -EILSEQ; 136 return -EILSEQ;
137 137
138 n = ntohs(get_unaligned(data)); 138 n = get_unaligned_be16(data);
139 data++; len -= 2; 139 data++; len -= 2;
140 140
141 if (len < n) 141 if (len < n)
@@ -150,8 +150,8 @@ static int bnep_ctrl_set_netfilter(struct bnep_session *s, __be16 *data, int len
150 int i; 150 int i;
151 151
152 for (i = 0; i < n; i++) { 152 for (i = 0; i < n; i++) {
153 f[i].start = ntohs(get_unaligned(data++)); 153 f[i].start = get_unaligned_be16(data++);
154 f[i].end = ntohs(get_unaligned(data++)); 154 f[i].end = get_unaligned_be16(data++);
155 155
156 BT_DBG("proto filter start %d end %d", 156 BT_DBG("proto filter start %d end %d",
157 f[i].start, f[i].end); 157 f[i].start, f[i].end);
@@ -180,7 +180,7 @@ static int bnep_ctrl_set_mcfilter(struct bnep_session *s, u8 *data, int len)
180 if (len < 2) 180 if (len < 2)
181 return -EILSEQ; 181 return -EILSEQ;
182 182
183 n = ntohs(get_unaligned((__be16 *) data)); 183 n = get_unaligned_be16(data);
184 data += 2; len -= 2; 184 data += 2; len -= 2;
185 185
186 if (len < n) 186 if (len < n)
diff --git a/net/bluetooth/hci_event.c b/net/bluetooth/hci_event.c
index 46df2e403df8..6aef8f24e581 100644
--- a/net/bluetooth/hci_event.c
+++ b/net/bluetooth/hci_event.c
@@ -129,8 +129,7 @@ static void hci_cc_write_link_policy(struct hci_dev *hdev, struct sk_buff *skb)
129 129
130 conn = hci_conn_hash_lookup_handle(hdev, __le16_to_cpu(rp->handle)); 130 conn = hci_conn_hash_lookup_handle(hdev, __le16_to_cpu(rp->handle));
131 if (conn) { 131 if (conn) {
132 __le16 policy = get_unaligned((__le16 *) (sent + 2)); 132 conn->link_policy = get_unaligned_le16(sent + 2);
133 conn->link_policy = __le16_to_cpu(policy);
134 } 133 }
135 134
136 hci_dev_unlock(hdev); 135 hci_dev_unlock(hdev);
@@ -313,7 +312,7 @@ static void hci_cc_write_voice_setting(struct hci_dev *hdev, struct sk_buff *skb
313 return; 312 return;
314 313
315 if (!status) { 314 if (!status) {
316 __u16 setting = __le16_to_cpu(get_unaligned((__le16 *) sent)); 315 __u16 setting = get_unaligned_le16(sent);
317 316
318 if (hdev->voice_setting != setting) { 317 if (hdev->voice_setting != setting) {
319 hdev->voice_setting = setting; 318 hdev->voice_setting = setting;
@@ -1152,8 +1151,8 @@ static inline void hci_num_comp_pkts_evt(struct hci_dev *hdev, struct sk_buff *s
1152 struct hci_conn *conn; 1151 struct hci_conn *conn;
1153 __u16 handle, count; 1152 __u16 handle, count;
1154 1153
1155 handle = __le16_to_cpu(get_unaligned(ptr++)); 1154 handle = get_unaligned_le16(ptr++);
1156 count = __le16_to_cpu(get_unaligned(ptr++)); 1155 count = get_unaligned_le16(ptr++);
1157 1156
1158 conn = hci_conn_hash_lookup_handle(hdev, handle); 1157 conn = hci_conn_hash_lookup_handle(hdev, handle);
1159 if (conn) { 1158 if (conn) {
diff --git a/net/bluetooth/hci_sock.c b/net/bluetooth/hci_sock.c
index 1d36c093523b..747fabd735d2 100644
--- a/net/bluetooth/hci_sock.c
+++ b/net/bluetooth/hci_sock.c
@@ -440,7 +440,7 @@ static int hci_sock_sendmsg(struct kiocb *iocb, struct socket *sock,
440 skb->dev = (void *) hdev; 440 skb->dev = (void *) hdev;
441 441
442 if (bt_cb(skb)->pkt_type == HCI_COMMAND_PKT) { 442 if (bt_cb(skb)->pkt_type == HCI_COMMAND_PKT) {
443 u16 opcode = __le16_to_cpu(get_unaligned((__le16 *) skb->data)); 443 u16 opcode = get_unaligned_le16(skb->data);
444 u16 ogf = hci_opcode_ogf(opcode); 444 u16 ogf = hci_opcode_ogf(opcode);
445 u16 ocf = hci_opcode_ocf(opcode); 445 u16 ocf = hci_opcode_ocf(opcode);
446 446
diff --git a/net/bluetooth/l2cap.c b/net/bluetooth/l2cap.c
index a4849f2c1d81..6e180d255505 100644
--- a/net/bluetooth/l2cap.c
+++ b/net/bluetooth/l2cap.c
@@ -1827,7 +1827,7 @@ static inline int l2cap_information_rsp(struct l2cap_conn *conn, struct l2cap_cm
1827 del_timer(&conn->info_timer); 1827 del_timer(&conn->info_timer);
1828 1828
1829 if (type == L2CAP_IT_FEAT_MASK) 1829 if (type == L2CAP_IT_FEAT_MASK)
1830 conn->feat_mask = __le32_to_cpu(get_unaligned((__le32 *) rsp->data)); 1830 conn->feat_mask = get_unaligned_le32(rsp->data);
1831 1831
1832 l2cap_conn_start(conn); 1832 l2cap_conn_start(conn);
1833 1833
diff --git a/net/bluetooth/rfcomm/core.c b/net/bluetooth/rfcomm/core.c
index eb62558e9b09..0c2c93735e93 100644
--- a/net/bluetooth/rfcomm/core.c
+++ b/net/bluetooth/rfcomm/core.c
@@ -423,8 +423,8 @@ static int __rfcomm_dlc_close(struct rfcomm_dlc *d, int err)
423 423
424 rfcomm_dlc_lock(d); 424 rfcomm_dlc_lock(d);
425 d->state = BT_CLOSED; 425 d->state = BT_CLOSED;
426 rfcomm_dlc_unlock(d);
427 d->state_change(d, err); 426 d->state_change(d, err);
427 rfcomm_dlc_unlock(d);
428 428
429 skb_queue_purge(&d->tx_queue); 429 skb_queue_purge(&d->tx_queue);
430 rfcomm_dlc_unlink(d); 430 rfcomm_dlc_unlink(d);
diff --git a/net/bluetooth/rfcomm/tty.c b/net/bluetooth/rfcomm/tty.c
index c3f749abb2d0..c9191871c1e0 100644
--- a/net/bluetooth/rfcomm/tty.c
+++ b/net/bluetooth/rfcomm/tty.c
@@ -566,11 +566,22 @@ static void rfcomm_dev_state_change(struct rfcomm_dlc *dlc, int err)
566 if (dlc->state == BT_CLOSED) { 566 if (dlc->state == BT_CLOSED) {
567 if (!dev->tty) { 567 if (!dev->tty) {
568 if (test_bit(RFCOMM_RELEASE_ONHUP, &dev->flags)) { 568 if (test_bit(RFCOMM_RELEASE_ONHUP, &dev->flags)) {
569 if (rfcomm_dev_get(dev->id) == NULL) 569 /* Drop DLC lock here to avoid deadlock
570 * 1. rfcomm_dev_get will take rfcomm_dev_lock
571 * but in rfcomm_dev_add there's lock order:
572 * rfcomm_dev_lock -> dlc lock
573 * 2. rfcomm_dev_put will deadlock if it's
574 * the last reference
575 */
576 rfcomm_dlc_unlock(dlc);
577 if (rfcomm_dev_get(dev->id) == NULL) {
578 rfcomm_dlc_lock(dlc);
570 return; 579 return;
580 }
571 581
572 rfcomm_dev_del(dev); 582 rfcomm_dev_del(dev);
573 rfcomm_dev_put(dev); 583 rfcomm_dev_put(dev);
584 rfcomm_dlc_lock(dlc);
574 } 585 }
575 } else 586 } else
576 tty_hangup(dev->tty); 587 tty_hangup(dev->tty);
diff --git a/net/bridge/br_fdb.c b/net/bridge/br_fdb.c
index 9326c377822e..72c5976a5ce3 100644
--- a/net/bridge/br_fdb.c
+++ b/net/bridge/br_fdb.c
@@ -285,7 +285,11 @@ int br_fdb_fillbuf(struct net_bridge *br, void *buf,
285 285
286 /* convert from internal format to API */ 286 /* convert from internal format to API */
287 memcpy(fe->mac_addr, f->addr.addr, ETH_ALEN); 287 memcpy(fe->mac_addr, f->addr.addr, ETH_ALEN);
288
289 /* due to ABI compat need to split into hi/lo */
288 fe->port_no = f->dst->port_no; 290 fe->port_no = f->dst->port_no;
291 fe->port_hi = f->dst->port_no >> 8;
292
289 fe->is_local = f->is_local; 293 fe->is_local = f->is_local;
290 if (!f->is_static) 294 if (!f->is_static)
291 fe->ageing_timer_value = jiffies_to_clock_t(jiffies - f->ageing_timer); 295 fe->ageing_timer_value = jiffies_to_clock_t(jiffies - f->ageing_timer);
diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c
index 77a981a1ee52..f38cc5317b88 100644
--- a/net/bridge/br_if.c
+++ b/net/bridge/br_if.c
@@ -273,15 +273,13 @@ int br_add_bridge(const char *name)
273 rtnl_lock(); 273 rtnl_lock();
274 if (strchr(dev->name, '%')) { 274 if (strchr(dev->name, '%')) {
275 ret = dev_alloc_name(dev, dev->name); 275 ret = dev_alloc_name(dev, dev->name);
276 if (ret < 0) { 276 if (ret < 0)
277 free_netdev(dev); 277 goto out_free;
278 goto out;
279 }
280 } 278 }
281 279
282 ret = register_netdevice(dev); 280 ret = register_netdevice(dev);
283 if (ret) 281 if (ret)
284 goto out; 282 goto out_free;
285 283
286 ret = br_sysfs_addbr(dev); 284 ret = br_sysfs_addbr(dev);
287 if (ret) 285 if (ret)
@@ -289,6 +287,10 @@ int br_add_bridge(const char *name)
289 out: 287 out:
290 rtnl_unlock(); 288 rtnl_unlock();
291 return ret; 289 return ret;
290
291out_free:
292 free_netdev(dev);
293 goto out;
292} 294}
293 295
294int br_del_bridge(const char *name) 296int br_del_bridge(const char *name)
@@ -440,12 +442,16 @@ int br_del_if(struct net_bridge *br, struct net_device *dev)
440 442
441void __exit br_cleanup_bridges(void) 443void __exit br_cleanup_bridges(void)
442{ 444{
443 struct net_device *dev, *nxt; 445 struct net_device *dev;
444 446
445 rtnl_lock(); 447 rtnl_lock();
446 for_each_netdev_safe(&init_net, dev, nxt) 448restart:
447 if (dev->priv_flags & IFF_EBRIDGE) 449 for_each_netdev(&init_net, dev) {
450 if (dev->priv_flags & IFF_EBRIDGE) {
448 del_br(dev->priv); 451 del_br(dev->priv);
452 goto restart;
453 }
454 }
449 rtnl_unlock(); 455 rtnl_unlock();
450 456
451} 457}
diff --git a/net/bridge/br_stp_bpdu.c b/net/bridge/br_stp_bpdu.c
index 8deab645ef75..ddeb6e5d45d6 100644
--- a/net/bridge/br_stp_bpdu.c
+++ b/net/bridge/br_stp_bpdu.c
@@ -58,12 +58,12 @@ static inline void br_set_ticks(unsigned char *dest, int j)
58{ 58{
59 unsigned long ticks = (STP_HZ * j)/ HZ; 59 unsigned long ticks = (STP_HZ * j)/ HZ;
60 60
61 put_unaligned(htons(ticks), (__be16 *)dest); 61 put_unaligned_be16(ticks, dest);
62} 62}
63 63
64static inline int br_get_ticks(const unsigned char *src) 64static inline int br_get_ticks(const unsigned char *src)
65{ 65{
66 unsigned long ticks = ntohs(get_unaligned((__be16 *)src)); 66 unsigned long ticks = get_unaligned_be16(src);
67 67
68 return DIV_ROUND_UP(ticks * HZ, STP_HZ); 68 return DIV_ROUND_UP(ticks * HZ, STP_HZ);
69} 69}
diff --git a/net/can/af_can.c b/net/can/af_can.c
index 2759b76f731c..484bbf6dd032 100644
--- a/net/can/af_can.c
+++ b/net/can/af_can.c
@@ -205,11 +205,19 @@ static int can_create(struct net *net, struct socket *sock, int protocol)
205 * -ENOBUFS on full driver queue (see net_xmit_errno()) 205 * -ENOBUFS on full driver queue (see net_xmit_errno())
206 * -ENOMEM when local loopback failed at calling skb_clone() 206 * -ENOMEM when local loopback failed at calling skb_clone()
207 * -EPERM when trying to send on a non-CAN interface 207 * -EPERM when trying to send on a non-CAN interface
208 * -EINVAL when the skb->data does not contain a valid CAN frame
208 */ 209 */
209int can_send(struct sk_buff *skb, int loop) 210int can_send(struct sk_buff *skb, int loop)
210{ 211{
212 struct sk_buff *newskb = NULL;
213 struct can_frame *cf = (struct can_frame *)skb->data;
211 int err; 214 int err;
212 215
216 if (skb->len != sizeof(struct can_frame) || cf->can_dlc > 8) {
217 kfree_skb(skb);
218 return -EINVAL;
219 }
220
213 if (skb->dev->type != ARPHRD_CAN) { 221 if (skb->dev->type != ARPHRD_CAN) {
214 kfree_skb(skb); 222 kfree_skb(skb);
215 return -EPERM; 223 return -EPERM;
@@ -244,8 +252,7 @@ int can_send(struct sk_buff *skb, int loop)
244 * If the interface is not capable to do loopback 252 * If the interface is not capable to do loopback
245 * itself, we do it here. 253 * itself, we do it here.
246 */ 254 */
247 struct sk_buff *newskb = skb_clone(skb, GFP_ATOMIC); 255 newskb = skb_clone(skb, GFP_ATOMIC);
248
249 if (!newskb) { 256 if (!newskb) {
250 kfree_skb(skb); 257 kfree_skb(skb);
251 return -ENOMEM; 258 return -ENOMEM;
@@ -254,7 +261,6 @@ int can_send(struct sk_buff *skb, int loop)
254 newskb->sk = skb->sk; 261 newskb->sk = skb->sk;
255 newskb->ip_summed = CHECKSUM_UNNECESSARY; 262 newskb->ip_summed = CHECKSUM_UNNECESSARY;
256 newskb->pkt_type = PACKET_BROADCAST; 263 newskb->pkt_type = PACKET_BROADCAST;
257 netif_rx(newskb);
258 } 264 }
259 } else { 265 } else {
260 /* indication for the CAN driver: no loopback required */ 266 /* indication for the CAN driver: no loopback required */
@@ -266,11 +272,20 @@ int can_send(struct sk_buff *skb, int loop)
266 if (err > 0) 272 if (err > 0)
267 err = net_xmit_errno(err); 273 err = net_xmit_errno(err);
268 274
275 if (err) {
276 if (newskb)
277 kfree_skb(newskb);
278 return err;
279 }
280
281 if (newskb)
282 netif_rx(newskb);
283
269 /* update statistics */ 284 /* update statistics */
270 can_stats.tx_frames++; 285 can_stats.tx_frames++;
271 can_stats.tx_frames_delta++; 286 can_stats.tx_frames_delta++;
272 287
273 return err; 288 return 0;
274} 289}
275EXPORT_SYMBOL(can_send); 290EXPORT_SYMBOL(can_send);
276 291
@@ -597,6 +612,7 @@ static int can_rcv(struct sk_buff *skb, struct net_device *dev,
597 struct packet_type *pt, struct net_device *orig_dev) 612 struct packet_type *pt, struct net_device *orig_dev)
598{ 613{
599 struct dev_rcv_lists *d; 614 struct dev_rcv_lists *d;
615 struct can_frame *cf = (struct can_frame *)skb->data;
600 int matches; 616 int matches;
601 617
602 if (dev->type != ARPHRD_CAN || dev_net(dev) != &init_net) { 618 if (dev->type != ARPHRD_CAN || dev_net(dev) != &init_net) {
@@ -604,6 +620,8 @@ static int can_rcv(struct sk_buff *skb, struct net_device *dev,
604 return 0; 620 return 0;
605 } 621 }
606 622
623 BUG_ON(skb->len != sizeof(struct can_frame) || cf->can_dlc > 8);
624
607 /* update statistics */ 625 /* update statistics */
608 can_stats.rx_frames++; 626 can_stats.rx_frames++;
609 can_stats.rx_frames_delta++; 627 can_stats.rx_frames_delta++;
diff --git a/net/can/bcm.c b/net/can/bcm.c
index 74fd2d33aff4..72c2ce904f83 100644
--- a/net/can/bcm.c
+++ b/net/can/bcm.c
@@ -298,7 +298,7 @@ static void bcm_send_to_user(struct bcm_op *op, struct bcm_msg_head *head,
298 298
299 if (head->nframes) { 299 if (head->nframes) {
300 /* can_frames starting here */ 300 /* can_frames starting here */
301 firstframe = (struct can_frame *) skb_tail_pointer(skb); 301 firstframe = (struct can_frame *)skb_tail_pointer(skb);
302 302
303 memcpy(skb_put(skb, datalen), frames, datalen); 303 memcpy(skb_put(skb, datalen), frames, datalen);
304 304
@@ -412,12 +412,6 @@ static void bcm_rx_changed(struct bcm_op *op, struct can_frame *data)
412 bcm_send_to_user(op, &head, data, 1); 412 bcm_send_to_user(op, &head, data, 1);
413} 413}
414 414
415/* TODO: move to linux/hrtimer.h */
416static inline int hrtimer_callback_running(struct hrtimer *timer)
417{
418 return timer->state & HRTIMER_STATE_CALLBACK;
419}
420
421/* 415/*
422 * bcm_rx_update_and_send - process a detected relevant receive content change 416 * bcm_rx_update_and_send - process a detected relevant receive content change
423 * 1. update the last received data 417 * 1. update the last received data
@@ -832,6 +826,10 @@ static int bcm_tx_setup(struct bcm_msg_head *msg_head, struct msghdr *msg,
832 for (i = 0; i < msg_head->nframes; i++) { 826 for (i = 0; i < msg_head->nframes; i++) {
833 err = memcpy_fromiovec((u8 *)&op->frames[i], 827 err = memcpy_fromiovec((u8 *)&op->frames[i],
834 msg->msg_iov, CFSIZ); 828 msg->msg_iov, CFSIZ);
829
830 if (op->frames[i].can_dlc > 8)
831 err = -EINVAL;
832
835 if (err < 0) 833 if (err < 0)
836 return err; 834 return err;
837 835
@@ -864,6 +862,10 @@ static int bcm_tx_setup(struct bcm_msg_head *msg_head, struct msghdr *msg,
864 for (i = 0; i < msg_head->nframes; i++) { 862 for (i = 0; i < msg_head->nframes; i++) {
865 err = memcpy_fromiovec((u8 *)&op->frames[i], 863 err = memcpy_fromiovec((u8 *)&op->frames[i],
866 msg->msg_iov, CFSIZ); 864 msg->msg_iov, CFSIZ);
865
866 if (op->frames[i].can_dlc > 8)
867 err = -EINVAL;
868
867 if (err < 0) { 869 if (err < 0) {
868 if (op->frames != &op->sframe) 870 if (op->frames != &op->sframe)
869 kfree(op->frames); 871 kfree(op->frames);
@@ -1170,9 +1172,12 @@ static int bcm_tx_send(struct msghdr *msg, int ifindex, struct sock *sk)
1170 1172
1171 skb->dev = dev; 1173 skb->dev = dev;
1172 skb->sk = sk; 1174 skb->sk = sk;
1173 can_send(skb, 1); /* send with loopback */ 1175 err = can_send(skb, 1); /* send with loopback */
1174 dev_put(dev); 1176 dev_put(dev);
1175 1177
1178 if (err)
1179 return err;
1180
1176 return CFSIZ + MHSIZ; 1181 return CFSIZ + MHSIZ;
1177} 1182}
1178 1183
@@ -1191,6 +1196,10 @@ static int bcm_sendmsg(struct kiocb *iocb, struct socket *sock,
1191 if (!bo->bound) 1196 if (!bo->bound)
1192 return -ENOTCONN; 1197 return -ENOTCONN;
1193 1198
1199 /* check for valid message length from userspace */
1200 if (size < MHSIZ || (size - MHSIZ) % CFSIZ)
1201 return -EINVAL;
1202
1194 /* check for alternative ifindex for this bcm_op */ 1203 /* check for alternative ifindex for this bcm_op */
1195 1204
1196 if (!ifindex && msg->msg_name) { 1205 if (!ifindex && msg->msg_name) {
@@ -1265,8 +1274,8 @@ static int bcm_sendmsg(struct kiocb *iocb, struct socket *sock,
1265 break; 1274 break;
1266 1275
1267 case TX_SEND: 1276 case TX_SEND:
1268 /* we need at least one can_frame */ 1277 /* we need exactly one can_frame behind the msg head */
1269 if (msg_head.nframes < 1) 1278 if ((msg_head.nframes != 1) || (size != CFSIZ + MHSIZ))
1270 ret = -EINVAL; 1279 ret = -EINVAL;
1271 else 1280 else
1272 ret = bcm_tx_send(msg, ifindex, sk); 1281 ret = bcm_tx_send(msg, ifindex, sk);
diff --git a/net/can/raw.c b/net/can/raw.c
index 69877b8e7e9c..3e46ee36a1aa 100644
--- a/net/can/raw.c
+++ b/net/can/raw.c
@@ -632,6 +632,9 @@ static int raw_sendmsg(struct kiocb *iocb, struct socket *sock,
632 } else 632 } else
633 ifindex = ro->ifindex; 633 ifindex = ro->ifindex;
634 634
635 if (size != sizeof(struct can_frame))
636 return -EINVAL;
637
635 dev = dev_get_by_index(&init_net, ifindex); 638 dev = dev_get_by_index(&init_net, ifindex);
636 if (!dev) 639 if (!dev)
637 return -ENXIO; 640 return -ENXIO;
diff --git a/net/core/dev.c b/net/core/dev.c
index ed49da592051..fca23a3bf12c 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -119,6 +119,7 @@
119#include <linux/err.h> 119#include <linux/err.h>
120#include <linux/ctype.h> 120#include <linux/ctype.h>
121#include <linux/if_arp.h> 121#include <linux/if_arp.h>
122#include <linux/if_vlan.h>
122 123
123#include "net-sysfs.h" 124#include "net-sysfs.h"
124 125
@@ -162,7 +163,7 @@ struct net_dma {
162 struct dma_client client; 163 struct dma_client client;
163 spinlock_t lock; 164 spinlock_t lock;
164 cpumask_t channel_mask; 165 cpumask_t channel_mask;
165 struct dma_chan *channels[NR_CPUS]; 166 struct dma_chan **channels;
166}; 167};
167 168
168static enum dma_state_client 169static enum dma_state_client
@@ -453,7 +454,7 @@ static int netdev_boot_setup_add(char *name, struct ifmap *map)
453 for (i = 0; i < NETDEV_BOOT_SETUP_MAX; i++) { 454 for (i = 0; i < NETDEV_BOOT_SETUP_MAX; i++) {
454 if (s[i].name[0] == '\0' || s[i].name[0] == ' ') { 455 if (s[i].name[0] == '\0' || s[i].name[0] == ' ') {
455 memset(s[i].name, 0, sizeof(s[i].name)); 456 memset(s[i].name, 0, sizeof(s[i].name));
456 strcpy(s[i].name, name); 457 strlcpy(s[i].name, name, IFNAMSIZ);
457 memcpy(&s[i].map, map, sizeof(s[i].map)); 458 memcpy(&s[i].map, map, sizeof(s[i].map));
458 break; 459 break;
459 } 460 }
@@ -478,7 +479,7 @@ int netdev_boot_setup_check(struct net_device *dev)
478 479
479 for (i = 0; i < NETDEV_BOOT_SETUP_MAX; i++) { 480 for (i = 0; i < NETDEV_BOOT_SETUP_MAX; i++) {
480 if (s[i].name[0] != '\0' && s[i].name[0] != ' ' && 481 if (s[i].name[0] != '\0' && s[i].name[0] != ' ' &&
481 !strncmp(dev->name, s[i].name, strlen(s[i].name))) { 482 !strcmp(dev->name, s[i].name)) {
482 dev->irq = s[i].map.irq; 483 dev->irq = s[i].map.irq;
483 dev->base_addr = s[i].map.base_addr; 484 dev->base_addr = s[i].map.base_addr;
484 dev->mem_start = s[i].map.mem_start; 485 dev->mem_start = s[i].map.mem_start;
@@ -903,7 +904,11 @@ int dev_change_name(struct net_device *dev, char *newname)
903 strlcpy(dev->name, newname, IFNAMSIZ); 904 strlcpy(dev->name, newname, IFNAMSIZ);
904 905
905rollback: 906rollback:
906 device_rename(&dev->dev, dev->name); 907 err = device_rename(&dev->dev, dev->name);
908 if (err) {
909 memcpy(dev->name, oldname, IFNAMSIZ);
910 return err;
911 }
907 912
908 write_lock_bh(&dev_base_lock); 913 write_lock_bh(&dev_base_lock);
909 hlist_del(&dev->name_hlist); 914 hlist_del(&dev->name_hlist);
@@ -994,6 +999,8 @@ int dev_open(struct net_device *dev)
994{ 999{
995 int ret = 0; 1000 int ret = 0;
996 1001
1002 ASSERT_RTNL();
1003
997 /* 1004 /*
998 * Is it already up? 1005 * Is it already up?
999 */ 1006 */
@@ -1060,6 +1067,8 @@ int dev_open(struct net_device *dev)
1060 */ 1067 */
1061int dev_close(struct net_device *dev) 1068int dev_close(struct net_device *dev)
1062{ 1069{
1070 ASSERT_RTNL();
1071
1063 might_sleep(); 1072 might_sleep();
1064 1073
1065 if (!(dev->flags & IFF_UP)) 1074 if (!(dev->flags & IFF_UP))
@@ -1354,6 +1363,29 @@ void netif_device_attach(struct net_device *dev)
1354} 1363}
1355EXPORT_SYMBOL(netif_device_attach); 1364EXPORT_SYMBOL(netif_device_attach);
1356 1365
1366static bool can_checksum_protocol(unsigned long features, __be16 protocol)
1367{
1368 return ((features & NETIF_F_GEN_CSUM) ||
1369 ((features & NETIF_F_IP_CSUM) &&
1370 protocol == htons(ETH_P_IP)) ||
1371 ((features & NETIF_F_IPV6_CSUM) &&
1372 protocol == htons(ETH_P_IPV6)));
1373}
1374
1375static bool dev_can_checksum(struct net_device *dev, struct sk_buff *skb)
1376{
1377 if (can_checksum_protocol(dev->features, skb->protocol))
1378 return true;
1379
1380 if (skb->protocol == htons(ETH_P_8021Q)) {
1381 struct vlan_ethhdr *veh = (struct vlan_ethhdr *)skb->data;
1382 if (can_checksum_protocol(dev->features & dev->vlan_features,
1383 veh->h_vlan_encapsulated_proto))
1384 return true;
1385 }
1386
1387 return false;
1388}
1357 1389
1358/* 1390/*
1359 * Invalidate hardware checksum when packet is to be mangled, and 1391 * Invalidate hardware checksum when packet is to be mangled, and
@@ -1632,14 +1664,8 @@ int dev_queue_xmit(struct sk_buff *skb)
1632 if (skb->ip_summed == CHECKSUM_PARTIAL) { 1664 if (skb->ip_summed == CHECKSUM_PARTIAL) {
1633 skb_set_transport_header(skb, skb->csum_start - 1665 skb_set_transport_header(skb, skb->csum_start -
1634 skb_headroom(skb)); 1666 skb_headroom(skb));
1635 1667 if (!dev_can_checksum(dev, skb) && skb_checksum_help(skb))
1636 if (!(dev->features & NETIF_F_GEN_CSUM) && 1668 goto out_kfree_skb;
1637 !((dev->features & NETIF_F_IP_CSUM) &&
1638 skb->protocol == htons(ETH_P_IP)) &&
1639 !((dev->features & NETIF_F_IPV6_CSUM) &&
1640 skb->protocol == htons(ETH_P_IPV6)))
1641 if (skb_checksum_help(skb))
1642 goto out_kfree_skb;
1643 } 1669 }
1644 1670
1645gso: 1671gso:
@@ -2051,6 +2077,10 @@ int netif_receive_skb(struct sk_buff *skb)
2051 2077
2052 rcu_read_lock(); 2078 rcu_read_lock();
2053 2079
2080 /* Don't receive packets in an exiting network namespace */
2081 if (!net_alive(dev_net(skb->dev)))
2082 goto out;
2083
2054#ifdef CONFIG_NET_CLS_ACT 2084#ifdef CONFIG_NET_CLS_ACT
2055 if (skb->tc_verd & TC_NCLS) { 2085 if (skb->tc_verd & TC_NCLS) {
2056 skb->tc_verd = CLR_TC_NCLS(skb->tc_verd); 2086 skb->tc_verd = CLR_TC_NCLS(skb->tc_verd);
@@ -2444,7 +2474,7 @@ static struct netif_rx_stats *softnet_get_online(loff_t *pos)
2444{ 2474{
2445 struct netif_rx_stats *rc = NULL; 2475 struct netif_rx_stats *rc = NULL;
2446 2476
2447 while (*pos < NR_CPUS) 2477 while (*pos < nr_cpu_ids)
2448 if (cpu_online(*pos)) { 2478 if (cpu_online(*pos)) {
2449 rc = &per_cpu(netdev_rx_stat, *pos); 2479 rc = &per_cpu(netdev_rx_stat, *pos);
2450 break; 2480 break;
@@ -2943,7 +2973,7 @@ EXPORT_SYMBOL(dev_unicast_delete);
2943/** 2973/**
2944 * dev_unicast_add - add a secondary unicast address 2974 * dev_unicast_add - add a secondary unicast address
2945 * @dev: device 2975 * @dev: device
2946 * @addr: address to delete 2976 * @addr: address to add
2947 * @alen: length of @addr 2977 * @alen: length of @addr
2948 * 2978 *
2949 * Add a secondary unicast address to the device or increase 2979 * Add a secondary unicast address to the device or increase
@@ -3133,7 +3163,7 @@ int dev_change_flags(struct net_device *dev, unsigned flags)
3133 * Load in the correct multicast list now the flags have changed. 3163 * Load in the correct multicast list now the flags have changed.
3134 */ 3164 */
3135 3165
3136 if (dev->change_rx_flags && (dev->flags ^ flags) & IFF_MULTICAST) 3166 if (dev->change_rx_flags && (old_flags ^ flags) & IFF_MULTICAST)
3137 dev->change_rx_flags(dev, IFF_MULTICAST); 3167 dev->change_rx_flags(dev, IFF_MULTICAST);
3138 3168
3139 dev_set_rx_mode(dev); 3169 dev_set_rx_mode(dev);
@@ -3776,6 +3806,7 @@ int register_netdevice(struct net_device *dev)
3776 } 3806 }
3777 } 3807 }
3778 3808
3809 netdev_initialize_kobject(dev);
3779 ret = netdev_register_kobject(dev); 3810 ret = netdev_register_kobject(dev);
3780 if (ret) 3811 if (ret)
3781 goto err_uninit; 3812 goto err_uninit;
@@ -4208,7 +4239,8 @@ int dev_change_net_namespace(struct net_device *dev, struct net *net, const char
4208 } 4239 }
4209 4240
4210 /* Fixup kobjects */ 4241 /* Fixup kobjects */
4211 err = device_rename(&dev->dev, dev->name); 4242 netdev_unregister_kobject(dev);
4243 err = netdev_register_kobject(dev);
4212 WARN_ON(err); 4244 WARN_ON(err);
4213 4245
4214 /* Add the device back in the hashes */ 4246 /* Add the device back in the hashes */
@@ -4324,7 +4356,7 @@ netdev_dma_event(struct dma_client *client, struct dma_chan *chan,
4324 spin_lock(&net_dma->lock); 4356 spin_lock(&net_dma->lock);
4325 switch (state) { 4357 switch (state) {
4326 case DMA_RESOURCE_AVAILABLE: 4358 case DMA_RESOURCE_AVAILABLE:
4327 for (i = 0; i < NR_CPUS; i++) 4359 for (i = 0; i < nr_cpu_ids; i++)
4328 if (net_dma->channels[i] == chan) { 4360 if (net_dma->channels[i] == chan) {
4329 found = 1; 4361 found = 1;
4330 break; 4362 break;
@@ -4339,7 +4371,7 @@ netdev_dma_event(struct dma_client *client, struct dma_chan *chan,
4339 } 4371 }
4340 break; 4372 break;
4341 case DMA_RESOURCE_REMOVED: 4373 case DMA_RESOURCE_REMOVED:
4342 for (i = 0; i < NR_CPUS; i++) 4374 for (i = 0; i < nr_cpu_ids; i++)
4343 if (net_dma->channels[i] == chan) { 4375 if (net_dma->channels[i] == chan) {
4344 found = 1; 4376 found = 1;
4345 pos = i; 4377 pos = i;
@@ -4366,6 +4398,13 @@ netdev_dma_event(struct dma_client *client, struct dma_chan *chan,
4366 */ 4398 */
4367static int __init netdev_dma_register(void) 4399static int __init netdev_dma_register(void)
4368{ 4400{
4401 net_dma.channels = kzalloc(nr_cpu_ids * sizeof(struct net_dma),
4402 GFP_KERNEL);
4403 if (unlikely(!net_dma.channels)) {
4404 printk(KERN_NOTICE
4405 "netdev_dma: no memory for net_dma.channels\n");
4406 return -ENOMEM;
4407 }
4369 spin_lock_init(&net_dma.lock); 4408 spin_lock_init(&net_dma.lock);
4370 dma_cap_set(DMA_MEMCPY, net_dma.client.cap_mask); 4409 dma_cap_set(DMA_MEMCPY, net_dma.client.cap_mask);
4371 dma_async_client_register(&net_dma.client); 4410 dma_async_client_register(&net_dma.client);
@@ -4471,17 +4510,19 @@ static void __net_exit default_device_exit(struct net *net)
4471 rtnl_lock(); 4510 rtnl_lock();
4472 for_each_netdev_safe(net, dev, next) { 4511 for_each_netdev_safe(net, dev, next) {
4473 int err; 4512 int err;
4513 char fb_name[IFNAMSIZ];
4474 4514
4475 /* Ignore unmoveable devices (i.e. loopback) */ 4515 /* Ignore unmoveable devices (i.e. loopback) */
4476 if (dev->features & NETIF_F_NETNS_LOCAL) 4516 if (dev->features & NETIF_F_NETNS_LOCAL)
4477 continue; 4517 continue;
4478 4518
4479 /* Push remaing network devices to init_net */ 4519 /* Push remaing network devices to init_net */
4480 err = dev_change_net_namespace(dev, &init_net, "dev%d"); 4520 snprintf(fb_name, IFNAMSIZ, "dev%d", dev->ifindex);
4521 err = dev_change_net_namespace(dev, &init_net, fb_name);
4481 if (err) { 4522 if (err) {
4482 printk(KERN_WARNING "%s: failed to move %s to init_net: %d\n", 4523 printk(KERN_EMERG "%s: failed to move %s to init_net: %d\n",
4483 __func__, dev->name, err); 4524 __func__, dev->name, err);
4484 unregister_netdevice(dev); 4525 BUG();
4485 } 4526 }
4486 } 4527 }
4487 rtnl_unlock(); 4528 rtnl_unlock();
diff --git a/net/core/fib_rules.c b/net/core/fib_rules.c
index e3e9ab0f74e3..277a2302eb3a 100644
--- a/net/core/fib_rules.c
+++ b/net/core/fib_rules.c
@@ -226,7 +226,7 @@ static int fib_nl_newrule(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
226 226
227 ops = lookup_rules_ops(net, frh->family); 227 ops = lookup_rules_ops(net, frh->family);
228 if (ops == NULL) { 228 if (ops == NULL) {
229 err = EAFNOSUPPORT; 229 err = -EAFNOSUPPORT;
230 goto errout; 230 goto errout;
231 } 231 }
232 232
@@ -365,7 +365,7 @@ static int fib_nl_delrule(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
365 365
366 ops = lookup_rules_ops(net, frh->family); 366 ops = lookup_rules_ops(net, frh->family);
367 if (ops == NULL) { 367 if (ops == NULL) {
368 err = EAFNOSUPPORT; 368 err = -EAFNOSUPPORT;
369 goto errout; 369 goto errout;
370 } 370 }
371 371
diff --git a/net/core/filter.c b/net/core/filter.c
index f5f3cf603064..df3744355839 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -68,7 +68,6 @@ static inline void *load_pointer(struct sk_buff *skb, int k,
68 * sk_filter - run a packet through a socket filter 68 * sk_filter - run a packet through a socket filter
69 * @sk: sock associated with &sk_buff 69 * @sk: sock associated with &sk_buff
70 * @skb: buffer to filter 70 * @skb: buffer to filter
71 * @needlock: set to 1 if the sock is not locked by caller.
72 * 71 *
73 * Run the filter code and then cut skb->data to correct size returned by 72 * Run the filter code and then cut skb->data to correct size returned by
74 * sk_run_filter. If pkt_len is 0 we toss packet. If skb->len is smaller 73 * sk_run_filter. If pkt_len is 0 we toss packet. If skb->len is smaller
@@ -213,7 +212,7 @@ unsigned int sk_run_filter(struct sk_buff *skb, struct sock_filter *filter, int
213load_w: 212load_w:
214 ptr = load_pointer(skb, k, 4, &tmp); 213 ptr = load_pointer(skb, k, 4, &tmp);
215 if (ptr != NULL) { 214 if (ptr != NULL) {
216 A = ntohl(get_unaligned((__be32 *)ptr)); 215 A = get_unaligned_be32(ptr);
217 continue; 216 continue;
218 } 217 }
219 break; 218 break;
@@ -222,7 +221,7 @@ load_w:
222load_h: 221load_h:
223 ptr = load_pointer(skb, k, 2, &tmp); 222 ptr = load_pointer(skb, k, 2, &tmp);
224 if (ptr != NULL) { 223 if (ptr != NULL) {
225 A = ntohs(get_unaligned((__be16 *)ptr)); 224 A = get_unaligned_be16(ptr);
226 continue; 225 continue;
227 } 226 }
228 break; 227 break;
diff --git a/net/core/neighbour.c b/net/core/neighbour.c
index 75075c303c44..65f01f71b3f3 100644
--- a/net/core/neighbour.c
+++ b/net/core/neighbour.c
@@ -1430,11 +1430,10 @@ void neigh_table_init_no_netlink(struct neigh_table *tbl)
1430 panic("cannot create neighbour cache statistics"); 1430 panic("cannot create neighbour cache statistics");
1431 1431
1432#ifdef CONFIG_PROC_FS 1432#ifdef CONFIG_PROC_FS
1433 tbl->pde = proc_create(tbl->id, 0, init_net.proc_net_stat, 1433 tbl->pde = proc_create_data(tbl->id, 0, init_net.proc_net_stat,
1434 &neigh_stat_seq_fops); 1434 &neigh_stat_seq_fops, tbl);
1435 if (!tbl->pde) 1435 if (!tbl->pde)
1436 panic("cannot create neighbour proc dir entry"); 1436 panic("cannot create neighbour proc dir entry");
1437 tbl->pde->data = tbl;
1438#endif 1437#endif
1439 1438
1440 tbl->hash_mask = 1; 1439 tbl->hash_mask = 1;
@@ -1715,7 +1714,8 @@ static int neightbl_fill_parms(struct sk_buff *skb, struct neigh_parms *parms)
1715 return nla_nest_end(skb, nest); 1714 return nla_nest_end(skb, nest);
1716 1715
1717nla_put_failure: 1716nla_put_failure:
1718 return nla_nest_cancel(skb, nest); 1717 nla_nest_cancel(skb, nest);
1718 return -EMSGSIZE;
1719} 1719}
1720 1720
1721static int neightbl_fill_info(struct sk_buff *skb, struct neigh_table *tbl, 1721static int neightbl_fill_info(struct sk_buff *skb, struct neigh_table *tbl,
@@ -2058,9 +2058,9 @@ static int neigh_fill_info(struct sk_buff *skb, struct neighbour *neigh,
2058 goto nla_put_failure; 2058 goto nla_put_failure;
2059 } 2059 }
2060 2060
2061 ci.ndm_used = now - neigh->used; 2061 ci.ndm_used = jiffies_to_clock_t(now - neigh->used);
2062 ci.ndm_confirmed = now - neigh->confirmed; 2062 ci.ndm_confirmed = jiffies_to_clock_t(now - neigh->confirmed);
2063 ci.ndm_updated = now - neigh->updated; 2063 ci.ndm_updated = jiffies_to_clock_t(now - neigh->updated);
2064 ci.ndm_refcnt = atomic_read(&neigh->refcnt) - 1; 2064 ci.ndm_refcnt = atomic_read(&neigh->refcnt) - 1;
2065 read_unlock_bh(&neigh->lock); 2065 read_unlock_bh(&neigh->lock);
2066 2066
diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c
index 4e7b847347f7..90e2177af081 100644
--- a/net/core/net-sysfs.c
+++ b/net/core/net-sysfs.c
@@ -449,7 +449,6 @@ int netdev_register_kobject(struct net_device *net)
449 struct device *dev = &(net->dev); 449 struct device *dev = &(net->dev);
450 struct attribute_group **groups = net->sysfs_groups; 450 struct attribute_group **groups = net->sysfs_groups;
451 451
452 device_initialize(dev);
453 dev->class = &net_class; 452 dev->class = &net_class;
454 dev->platform_data = net; 453 dev->platform_data = net;
455 dev->groups = groups; 454 dev->groups = groups;
@@ -470,6 +469,12 @@ int netdev_register_kobject(struct net_device *net)
470 return device_add(dev); 469 return device_add(dev);
471} 470}
472 471
472void netdev_initialize_kobject(struct net_device *net)
473{
474 struct device *device = &(net->dev);
475 device_initialize(device);
476}
477
473int netdev_kobject_init(void) 478int netdev_kobject_init(void)
474{ 479{
475 return class_register(&net_class); 480 return class_register(&net_class);
diff --git a/net/core/net-sysfs.h b/net/core/net-sysfs.h
index f5f108db3924..14e7524260b3 100644
--- a/net/core/net-sysfs.h
+++ b/net/core/net-sysfs.h
@@ -4,5 +4,5 @@
4int netdev_kobject_init(void); 4int netdev_kobject_init(void);
5int netdev_register_kobject(struct net_device *); 5int netdev_register_kobject(struct net_device *);
6void netdev_unregister_kobject(struct net_device *); 6void netdev_unregister_kobject(struct net_device *);
7 7void netdev_initialize_kobject(struct net_device *);
8#endif 8#endif
diff --git a/net/core/net_namespace.c b/net/core/net_namespace.c
index 72b4c184dd84..7c52fe277b62 100644
--- a/net/core/net_namespace.c
+++ b/net/core/net_namespace.c
@@ -140,6 +140,9 @@ static void cleanup_net(struct work_struct *work)
140 struct pernet_operations *ops; 140 struct pernet_operations *ops;
141 struct net *net; 141 struct net *net;
142 142
143 /* Be very certain incoming network packets will not find us */
144 rcu_barrier();
145
143 net = container_of(work, struct net, work); 146 net = container_of(work, struct net, work);
144 147
145 mutex_lock(&net_mutex); 148 mutex_lock(&net_mutex);
diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index b04d643fc3c7..8fb134da0346 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -419,7 +419,7 @@ static void arp_reply(struct sk_buff *skb)
419 return; 419 return;
420 420
421 size = arp_hdr_len(skb->dev); 421 size = arp_hdr_len(skb->dev);
422 send_skb = find_skb(np, size + LL_RESERVED_SPACE(np->dev), 422 send_skb = find_skb(np, size + LL_ALLOCATED_SPACE(np->dev),
423 LL_RESERVED_SPACE(np->dev)); 423 LL_RESERVED_SPACE(np->dev));
424 424
425 if (!send_skb) 425 if (!send_skb)
diff --git a/net/core/pktgen.c b/net/core/pktgen.c
index a803b442234c..fdf537707e51 100644
--- a/net/core/pktgen.c
+++ b/net/core/pktgen.c
@@ -390,6 +390,7 @@ struct pktgen_thread {
390 int cpu; 390 int cpu;
391 391
392 wait_queue_head_t queue; 392 wait_queue_head_t queue;
393 struct completion start_done;
393}; 394};
394 395
395#define REMOVE 1 396#define REMOVE 1
@@ -3414,6 +3415,7 @@ static int pktgen_thread_worker(void *arg)
3414 BUG_ON(smp_processor_id() != cpu); 3415 BUG_ON(smp_processor_id() != cpu);
3415 3416
3416 init_waitqueue_head(&t->queue); 3417 init_waitqueue_head(&t->queue);
3418 complete(&t->start_done);
3417 3419
3418 pr_debug("pktgen: starting pktgen/%d: pid=%d\n", cpu, task_pid_nr(current)); 3420 pr_debug("pktgen: starting pktgen/%d: pid=%d\n", cpu, task_pid_nr(current));
3419 3421
@@ -3570,15 +3572,14 @@ static int pktgen_add_device(struct pktgen_thread *t, const char *ifname)
3570 if (err) 3572 if (err)
3571 goto out1; 3573 goto out1;
3572 3574
3573 pkt_dev->entry = proc_create(ifname, 0600, 3575 pkt_dev->entry = proc_create_data(ifname, 0600, pg_proc_dir,
3574 pg_proc_dir, &pktgen_if_fops); 3576 &pktgen_if_fops, pkt_dev);
3575 if (!pkt_dev->entry) { 3577 if (!pkt_dev->entry) {
3576 printk(KERN_ERR "pktgen: cannot create %s/%s procfs entry.\n", 3578 printk(KERN_ERR "pktgen: cannot create %s/%s procfs entry.\n",
3577 PG_PROC_DIR, ifname); 3579 PG_PROC_DIR, ifname);
3578 err = -EINVAL; 3580 err = -EINVAL;
3579 goto out2; 3581 goto out2;
3580 } 3582 }
3581 pkt_dev->entry->data = pkt_dev;
3582#ifdef CONFIG_XFRM 3583#ifdef CONFIG_XFRM
3583 pkt_dev->ipsmode = XFRM_MODE_TRANSPORT; 3584 pkt_dev->ipsmode = XFRM_MODE_TRANSPORT;
3584 pkt_dev->ipsproto = IPPROTO_ESP; 3585 pkt_dev->ipsproto = IPPROTO_ESP;
@@ -3616,6 +3617,7 @@ static int __init pktgen_create_thread(int cpu)
3616 INIT_LIST_HEAD(&t->if_list); 3617 INIT_LIST_HEAD(&t->if_list);
3617 3618
3618 list_add_tail(&t->th_list, &pktgen_threads); 3619 list_add_tail(&t->th_list, &pktgen_threads);
3620 init_completion(&t->start_done);
3619 3621
3620 p = kthread_create(pktgen_thread_worker, t, "kpktgend_%d", cpu); 3622 p = kthread_create(pktgen_thread_worker, t, "kpktgend_%d", cpu);
3621 if (IS_ERR(p)) { 3623 if (IS_ERR(p)) {
@@ -3628,7 +3630,8 @@ static int __init pktgen_create_thread(int cpu)
3628 kthread_bind(p, cpu); 3630 kthread_bind(p, cpu);
3629 t->tsk = p; 3631 t->tsk = p;
3630 3632
3631 pe = proc_create(t->tsk->comm, 0600, pg_proc_dir, &pktgen_thread_fops); 3633 pe = proc_create_data(t->tsk->comm, 0600, pg_proc_dir,
3634 &pktgen_thread_fops, t);
3632 if (!pe) { 3635 if (!pe) {
3633 printk(KERN_ERR "pktgen: cannot create %s/%s procfs entry.\n", 3636 printk(KERN_ERR "pktgen: cannot create %s/%s procfs entry.\n",
3634 PG_PROC_DIR, t->tsk->comm); 3637 PG_PROC_DIR, t->tsk->comm);
@@ -3638,9 +3641,8 @@ static int __init pktgen_create_thread(int cpu)
3638 return -EINVAL; 3641 return -EINVAL;
3639 } 3642 }
3640 3643
3641 pe->data = t;
3642
3643 wake_up_process(p); 3644 wake_up_process(p);
3645 wait_for_completion(&t->start_done);
3644 3646
3645 return 0; 3647 return 0;
3646} 3648}
@@ -3716,8 +3718,6 @@ static int __init pg_init(void)
3716 return -EINVAL; 3718 return -EINVAL;
3717 } 3719 }
3718 3720
3719 pe->data = NULL;
3720
3721 /* Register us to receive netdevice events */ 3721 /* Register us to receive netdevice events */
3722 register_netdevice_notifier(&pktgen_notifier_block); 3722 register_netdevice_notifier(&pktgen_notifier_block);
3723 3723
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index cf857c4dc7b1..a9a77216310e 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -498,7 +498,8 @@ int rtnetlink_put_metrics(struct sk_buff *skb, u32 *metrics)
498 return nla_nest_end(skb, mx); 498 return nla_nest_end(skb, mx);
499 499
500nla_put_failure: 500nla_put_failure:
501 return nla_nest_cancel(skb, mx); 501 nla_nest_cancel(skb, mx);
502 return -EMSGSIZE;
502} 503}
503 504
504int rtnl_put_cacheinfo(struct sk_buff *skb, struct dst_entry *dst, u32 id, 505int rtnl_put_cacheinfo(struct sk_buff *skb, struct dst_entry *dst, u32 id,
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 4fe605fa6f8a..366621610e76 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -200,7 +200,9 @@ struct sk_buff *__alloc_skb(unsigned int size, gfp_t gfp_mask,
200 goto nodata; 200 goto nodata;
201 201
202 /* 202 /*
203 * See comment in sk_buff definition, just before the 'tail' member 203 * Only clear those fields we need to clear, not those that we will
204 * actually initialise below. Hence, don't put any more fields after
205 * the tail pointer in struct sk_buff!
204 */ 206 */
205 memset(skb, 0, offsetof(struct sk_buff, tail)); 207 memset(skb, 0, offsetof(struct sk_buff, tail));
206 skb->truesize = size + sizeof(struct sk_buff); 208 skb->truesize = size + sizeof(struct sk_buff);
@@ -1290,12 +1292,14 @@ static int __skb_splice_bits(struct sk_buff *skb, unsigned int *offset,
1290{ 1292{
1291 unsigned int nr_pages = spd->nr_pages; 1293 unsigned int nr_pages = spd->nr_pages;
1292 unsigned int poff, plen, len, toff, tlen; 1294 unsigned int poff, plen, len, toff, tlen;
1293 int headlen, seg; 1295 int headlen, seg, error = 0;
1294 1296
1295 toff = *offset; 1297 toff = *offset;
1296 tlen = *total_len; 1298 tlen = *total_len;
1297 if (!tlen) 1299 if (!tlen) {
1300 error = 1;
1298 goto err; 1301 goto err;
1302 }
1299 1303
1300 /* 1304 /*
1301 * if the offset is greater than the linear part, go directly to 1305 * if the offset is greater than the linear part, go directly to
@@ -1337,7 +1341,8 @@ static int __skb_splice_bits(struct sk_buff *skb, unsigned int *offset,
1337 * just jump directly to update and return, no point 1341 * just jump directly to update and return, no point
1338 * in going over fragments when the output is full. 1342 * in going over fragments when the output is full.
1339 */ 1343 */
1340 if (spd_fill_page(spd, virt_to_page(p), plen, poff, skb)) 1344 error = spd_fill_page(spd, virt_to_page(p), plen, poff, skb);
1345 if (error)
1341 goto done; 1346 goto done;
1342 1347
1343 tlen -= plen; 1348 tlen -= plen;
@@ -1367,7 +1372,8 @@ map_frag:
1367 if (!plen) 1372 if (!plen)
1368 break; 1373 break;
1369 1374
1370 if (spd_fill_page(spd, f->page, plen, poff, skb)) 1375 error = spd_fill_page(spd, f->page, plen, poff, skb);
1376 if (error)
1371 break; 1377 break;
1372 1378
1373 tlen -= plen; 1379 tlen -= plen;
@@ -1380,7 +1386,10 @@ done:
1380 return 0; 1386 return 0;
1381 } 1387 }
1382err: 1388err:
1383 return 1; 1389 /* update the offset to reflect the linear part skip, if any */
1390 if (!error)
1391 *offset = toff;
1392 return error;
1384} 1393}
1385 1394
1386/* 1395/*
@@ -1443,6 +1452,7 @@ done:
1443 1452
1444 if (spd.nr_pages) { 1453 if (spd.nr_pages) {
1445 int ret; 1454 int ret;
1455 struct sock *sk = __skb->sk;
1446 1456
1447 /* 1457 /*
1448 * Drop the socket lock, otherwise we have reverse 1458 * Drop the socket lock, otherwise we have reverse
@@ -1453,9 +1463,9 @@ done:
1453 * we call into ->sendpage() with the i_mutex lock held 1463 * we call into ->sendpage() with the i_mutex lock held
1454 * and networking will grab the socket lock. 1464 * and networking will grab the socket lock.
1455 */ 1465 */
1456 release_sock(__skb->sk); 1466 release_sock(sk);
1457 ret = splice_to_pipe(pipe, &spd); 1467 ret = splice_to_pipe(pipe, &spd);
1458 lock_sock(__skb->sk); 1468 lock_sock(sk);
1459 return ret; 1469 return ret;
1460 } 1470 }
1461 1471
diff --git a/net/core/sock.c b/net/core/sock.c
index 5dbb81bc9673..88094cb09c06 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -228,11 +228,12 @@ static int sock_set_timeout(long *timeo_p, char __user *optval, int optlen)
228 static int warned __read_mostly; 228 static int warned __read_mostly;
229 229
230 *timeo_p = 0; 230 *timeo_p = 0;
231 if (warned < 10 && net_ratelimit()) 231 if (warned < 10 && net_ratelimit()) {
232 warned++; 232 warned++;
233 printk(KERN_INFO "sock_set_timeout: `%s' (pid %d) " 233 printk(KERN_INFO "sock_set_timeout: `%s' (pid %d) "
234 "tries to set negative timeout\n", 234 "tries to set negative timeout\n",
235 current->comm, task_pid_nr(current)); 235 current->comm, task_pid_nr(current));
236 }
236 return 0; 237 return 0;
237 } 238 }
238 *timeo_p = MAX_SCHEDULE_TIMEOUT; 239 *timeo_p = MAX_SCHEDULE_TIMEOUT;
@@ -269,7 +270,7 @@ int sock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
269 int err = 0; 270 int err = 0;
270 int skb_len; 271 int skb_len;
271 272
272 /* Cast skb->rcvbuf to unsigned... It's pointless, but reduces 273 /* Cast sk->rcvbuf to unsigned... It's pointless, but reduces
273 number of warnings when compiling with -W --ANK 274 number of warnings when compiling with -W --ANK
274 */ 275 */
275 if (atomic_read(&sk->sk_rmem_alloc) + skb->truesize >= 276 if (atomic_read(&sk->sk_rmem_alloc) + skb->truesize >=
diff --git a/net/core/user_dma.c b/net/core/user_dma.c
index 0ad1cd57bc39..c77aff9c6eb3 100644
--- a/net/core/user_dma.c
+++ b/net/core/user_dma.c
@@ -75,7 +75,7 @@ int dma_skb_copy_datagram_iovec(struct dma_chan *chan,
75 75
76 end = start + skb_shinfo(skb)->frags[i].size; 76 end = start + skb_shinfo(skb)->frags[i].size;
77 copy = end - offset; 77 copy = end - offset;
78 if ((copy = end - offset) > 0) { 78 if (copy > 0) {
79 skb_frag_t *frag = &skb_shinfo(skb)->frags[i]; 79 skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
80 struct page *page = frag->page; 80 struct page *page = frag->page;
81 81
diff --git a/net/dccp/ackvec.c b/net/dccp/ackvec.c
index 6de4bd195d28..1e8be246ad15 100644
--- a/net/dccp/ackvec.c
+++ b/net/dccp/ackvec.c
@@ -290,12 +290,12 @@ int dccp_ackvec_add(struct dccp_ackvec *av, const struct sock *sk,
290 290
291 while (1) { 291 while (1) {
292 const u8 len = dccp_ackvec_len(av, index); 292 const u8 len = dccp_ackvec_len(av, index);
293 const u8 state = dccp_ackvec_state(av, index); 293 const u8 av_state = dccp_ackvec_state(av, index);
294 /* 294 /*
295 * valid packets not yet in av_buf have a reserved 295 * valid packets not yet in av_buf have a reserved
296 * entry, with a len equal to 0. 296 * entry, with a len equal to 0.
297 */ 297 */
298 if (state == DCCP_ACKVEC_STATE_NOT_RECEIVED && 298 if (av_state == DCCP_ACKVEC_STATE_NOT_RECEIVED &&
299 len == 0 && delta == 0) { /* Found our 299 len == 0 && delta == 0) { /* Found our
300 reserved seat! */ 300 reserved seat! */
301 dccp_pr_debug("Found %llu reserved seat!\n", 301 dccp_pr_debug("Found %llu reserved seat!\n",
@@ -325,31 +325,6 @@ out_duplicate:
325 return -EILSEQ; 325 return -EILSEQ;
326} 326}
327 327
328#ifdef CONFIG_IP_DCCP_DEBUG
329void dccp_ackvector_print(const u64 ackno, const unsigned char *vector, int len)
330{
331 dccp_pr_debug_cat("ACK vector len=%d, ackno=%llu |", len,
332 (unsigned long long)ackno);
333
334 while (len--) {
335 const u8 state = (*vector & DCCP_ACKVEC_STATE_MASK) >> 6;
336 const u8 rl = *vector & DCCP_ACKVEC_LEN_MASK;
337
338 dccp_pr_debug_cat("%d,%d|", state, rl);
339 ++vector;
340 }
341
342 dccp_pr_debug_cat("\n");
343}
344
345void dccp_ackvec_print(const struct dccp_ackvec *av)
346{
347 dccp_ackvector_print(av->av_buf_ackno,
348 av->av_buf + av->av_buf_head,
349 av->av_vec_len);
350}
351#endif
352
353static void dccp_ackvec_throw_record(struct dccp_ackvec *av, 328static void dccp_ackvec_throw_record(struct dccp_ackvec *av,
354 struct dccp_ackvec_record *avr) 329 struct dccp_ackvec_record *avr)
355{ 330{
diff --git a/net/dccp/ccids/ccid2.c b/net/dccp/ccids/ccid2.c
index b5b52ebb2693..8e9580874216 100644
--- a/net/dccp/ccids/ccid2.c
+++ b/net/dccp/ccids/ccid2.c
@@ -716,7 +716,7 @@ static int ccid2_hc_tx_init(struct ccid *ccid, struct sock *sk)
716 * packets for new connections, following the rules from [RFC3390]". 716 * packets for new connections, following the rules from [RFC3390]".
717 * We need to convert the bytes of RFC3390 into the packets of RFC 4341. 717 * We need to convert the bytes of RFC3390 into the packets of RFC 4341.
718 */ 718 */
719 hctx->ccid2hctx_cwnd = min(4U, max(2U, 4380U / dp->dccps_mss_cache)); 719 hctx->ccid2hctx_cwnd = clamp(4380U / dp->dccps_mss_cache, 2U, 4U);
720 720
721 /* Make sure that Ack Ratio is enabled and within bounds. */ 721 /* Make sure that Ack Ratio is enabled and within bounds. */
722 max_ratio = DIV_ROUND_UP(hctx->ccid2hctx_cwnd, 2); 722 max_ratio = DIV_ROUND_UP(hctx->ccid2hctx_cwnd, 2);
diff --git a/net/dccp/ccids/ccid3.c b/net/dccp/ccids/ccid3.c
index e76f460af0ea..a1929f33d703 100644
--- a/net/dccp/ccids/ccid3.c
+++ b/net/dccp/ccids/ccid3.c
@@ -88,8 +88,8 @@ static void ccid3_hc_tx_set_state(struct sock *sk,
88static inline u64 rfc3390_initial_rate(struct sock *sk) 88static inline u64 rfc3390_initial_rate(struct sock *sk)
89{ 89{
90 const struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk); 90 const struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
91 const __u32 w_init = min_t(__u32, 4 * hctx->ccid3hctx_s, 91 const __u32 w_init = clamp_t(__u32, 4380U,
92 max_t(__u32, 2 * hctx->ccid3hctx_s, 4380)); 92 2 * hctx->ccid3hctx_s, 4 * hctx->ccid3hctx_s);
93 93
94 return scaled_div(w_init << 6, hctx->ccid3hctx_rtt); 94 return scaled_div(w_init << 6, hctx->ccid3hctx_rtt);
95} 95}
@@ -159,8 +159,8 @@ static void ccid3_hc_tx_update_x(struct sock *sk, ktime_t *stamp)
159 } else if (ktime_us_delta(now, hctx->ccid3hctx_t_ld) 159 } else if (ktime_us_delta(now, hctx->ccid3hctx_t_ld)
160 - (s64)hctx->ccid3hctx_rtt >= 0) { 160 - (s64)hctx->ccid3hctx_rtt >= 0) {
161 161
162 hctx->ccid3hctx_x = 162 hctx->ccid3hctx_x = min(2 * hctx->ccid3hctx_x, min_rate);
163 max(min(2 * hctx->ccid3hctx_x, min_rate), 163 hctx->ccid3hctx_x = max(hctx->ccid3hctx_x,
164 scaled_div(((__u64)hctx->ccid3hctx_s) << 6, 164 scaled_div(((__u64)hctx->ccid3hctx_s) << 6,
165 hctx->ccid3hctx_rtt)); 165 hctx->ccid3hctx_rtt));
166 hctx->ccid3hctx_t_ld = now; 166 hctx->ccid3hctx_t_ld = now;
@@ -193,22 +193,17 @@ static inline void ccid3_hc_tx_update_s(struct ccid3_hc_tx_sock *hctx, int len)
193 193
194/* 194/*
195 * Update Window Counter using the algorithm from [RFC 4342, 8.1]. 195 * Update Window Counter using the algorithm from [RFC 4342, 8.1].
196 * The algorithm is not applicable if RTT < 4 microseconds. 196 * As elsewhere, RTT > 0 is assumed by using dccp_sample_rtt().
197 */ 197 */
198static inline void ccid3_hc_tx_update_win_count(struct ccid3_hc_tx_sock *hctx, 198static inline void ccid3_hc_tx_update_win_count(struct ccid3_hc_tx_sock *hctx,
199 ktime_t now) 199 ktime_t now)
200{ 200{
201 u32 quarter_rtts; 201 u32 delta = ktime_us_delta(now, hctx->ccid3hctx_t_last_win_count),
202 202 quarter_rtts = (4 * delta) / hctx->ccid3hctx_rtt;
203 if (unlikely(hctx->ccid3hctx_rtt < 4)) /* avoid divide-by-zero */
204 return;
205
206 quarter_rtts = ktime_us_delta(now, hctx->ccid3hctx_t_last_win_count);
207 quarter_rtts /= hctx->ccid3hctx_rtt / 4;
208 203
209 if (quarter_rtts > 0) { 204 if (quarter_rtts > 0) {
210 hctx->ccid3hctx_t_last_win_count = now; 205 hctx->ccid3hctx_t_last_win_count = now;
211 hctx->ccid3hctx_last_win_count += min_t(u32, quarter_rtts, 5); 206 hctx->ccid3hctx_last_win_count += min(quarter_rtts, 5U);
212 hctx->ccid3hctx_last_win_count &= 0xF; /* mod 16 */ 207 hctx->ccid3hctx_last_win_count &= 0xF; /* mod 16 */
213 } 208 }
214} 209}
@@ -334,8 +329,14 @@ static int ccid3_hc_tx_send_packet(struct sock *sk, struct sk_buff *skb)
334 hctx->ccid3hctx_x = rfc3390_initial_rate(sk); 329 hctx->ccid3hctx_x = rfc3390_initial_rate(sk);
335 hctx->ccid3hctx_t_ld = now; 330 hctx->ccid3hctx_t_ld = now;
336 } else { 331 } else {
337 /* Sender does not have RTT sample: X_pps = 1 pkt/sec */ 332 /*
338 hctx->ccid3hctx_x = hctx->ccid3hctx_s; 333 * Sender does not have RTT sample:
334 * - set fallback RTT (RFC 4340, 3.4) since a RTT value
335 * is needed in several parts (e.g. window counter);
336 * - set sending rate X_pps = 1pps as per RFC 3448, 4.2.
337 */
338 hctx->ccid3hctx_rtt = DCCP_FALLBACK_RTT;
339 hctx->ccid3hctx_x = hctx->ccid3hctx_s;
339 hctx->ccid3hctx_x <<= 6; 340 hctx->ccid3hctx_x <<= 6;
340 } 341 }
341 ccid3_update_send_interval(hctx); 342 ccid3_update_send_interval(hctx);
diff --git a/net/dccp/ccids/lib/tfrc.c b/net/dccp/ccids/lib/tfrc.c
index d1dfbb8de64c..97ecec0a8e76 100644
--- a/net/dccp/ccids/lib/tfrc.c
+++ b/net/dccp/ccids/lib/tfrc.c
@@ -14,14 +14,6 @@ module_param(tfrc_debug, bool, 0444);
14MODULE_PARM_DESC(tfrc_debug, "Enable debug messages"); 14MODULE_PARM_DESC(tfrc_debug, "Enable debug messages");
15#endif 15#endif
16 16
17extern int tfrc_tx_packet_history_init(void);
18extern void tfrc_tx_packet_history_exit(void);
19extern int tfrc_rx_packet_history_init(void);
20extern void tfrc_rx_packet_history_exit(void);
21
22extern int tfrc_li_init(void);
23extern void tfrc_li_exit(void);
24
25static int __init tfrc_module_init(void) 17static int __init tfrc_module_init(void)
26{ 18{
27 int rc = tfrc_li_init(); 19 int rc = tfrc_li_init();
diff --git a/net/dccp/ccids/lib/tfrc.h b/net/dccp/ccids/lib/tfrc.h
index 1fb1187bbf1c..ed9857527acf 100644
--- a/net/dccp/ccids/lib/tfrc.h
+++ b/net/dccp/ccids/lib/tfrc.h
@@ -15,7 +15,7 @@
15 * (at your option) any later version. 15 * (at your option) any later version.
16 */ 16 */
17#include <linux/types.h> 17#include <linux/types.h>
18#include <asm/div64.h> 18#include <linux/math64.h>
19#include "../../dccp.h" 19#include "../../dccp.h"
20/* internal includes that this module exports: */ 20/* internal includes that this module exports: */
21#include "loss_interval.h" 21#include "loss_interval.h"
@@ -29,21 +29,19 @@ extern int tfrc_debug;
29#endif 29#endif
30 30
31/* integer-arithmetic divisions of type (a * 1000000)/b */ 31/* integer-arithmetic divisions of type (a * 1000000)/b */
32static inline u64 scaled_div(u64 a, u32 b) 32static inline u64 scaled_div(u64 a, u64 b)
33{ 33{
34 BUG_ON(b==0); 34 BUG_ON(b==0);
35 a *= 1000000; 35 return div64_u64(a * 1000000, b);
36 do_div(a, b);
37 return a;
38} 36}
39 37
40static inline u32 scaled_div32(u64 a, u32 b) 38static inline u32 scaled_div32(u64 a, u64 b)
41{ 39{
42 u64 result = scaled_div(a, b); 40 u64 result = scaled_div(a, b);
43 41
44 if (result > UINT_MAX) { 42 if (result > UINT_MAX) {
45 DCCP_CRIT("Overflow: a(%llu)/b(%u) > ~0U", 43 DCCP_CRIT("Overflow: %llu/%llu > UINT_MAX",
46 (unsigned long long)a, b); 44 (unsigned long long)a, (unsigned long long)b);
47 return UINT_MAX; 45 return UINT_MAX;
48 } 46 }
49 return result; 47 return result;
@@ -58,7 +56,14 @@ static inline u32 tfrc_ewma(const u32 avg, const u32 newval, const u8 weight)
58 return avg ? (weight * avg + (10 - weight) * newval) / 10 : newval; 56 return avg ? (weight * avg + (10 - weight) * newval) / 10 : newval;
59} 57}
60 58
61extern u32 tfrc_calc_x(u16 s, u32 R, u32 p); 59extern u32 tfrc_calc_x(u16 s, u32 R, u32 p);
62extern u32 tfrc_calc_x_reverse_lookup(u32 fvalue); 60extern u32 tfrc_calc_x_reverse_lookup(u32 fvalue);
63 61
62extern int tfrc_tx_packet_history_init(void);
63extern void tfrc_tx_packet_history_exit(void);
64extern int tfrc_rx_packet_history_init(void);
65extern void tfrc_rx_packet_history_exit(void);
66
67extern int tfrc_li_init(void);
68extern void tfrc_li_exit(void);
64#endif /* _TFRC_H_ */ 69#endif /* _TFRC_H_ */
diff --git a/net/dccp/ccids/lib/tfrc_equation.c b/net/dccp/ccids/lib/tfrc_equation.c
index e4e64b76c10c..2f20a29cffe4 100644
--- a/net/dccp/ccids/lib/tfrc_equation.c
+++ b/net/dccp/ccids/lib/tfrc_equation.c
@@ -661,7 +661,7 @@ u32 tfrc_calc_x(u16 s, u32 R, u32 p)
661 661
662EXPORT_SYMBOL_GPL(tfrc_calc_x); 662EXPORT_SYMBOL_GPL(tfrc_calc_x);
663 663
664/* 664/**
665 * tfrc_calc_x_reverse_lookup - try to find p given f(p) 665 * tfrc_calc_x_reverse_lookup - try to find p given f(p)
666 * 666 *
667 * @fvalue: function value to match, scaled by 1000000 667 * @fvalue: function value to match, scaled by 1000000
@@ -676,11 +676,11 @@ u32 tfrc_calc_x_reverse_lookup(u32 fvalue)
676 676
677 /* Error cases. */ 677 /* Error cases. */
678 if (fvalue < tfrc_calc_x_lookup[0][1]) { 678 if (fvalue < tfrc_calc_x_lookup[0][1]) {
679 DCCP_WARN("fvalue %d smaller than resolution\n", fvalue); 679 DCCP_WARN("fvalue %u smaller than resolution\n", fvalue);
680 return tfrc_calc_x_lookup[0][1]; 680 return TFRC_SMALLEST_P;
681 } 681 }
682 if (fvalue > tfrc_calc_x_lookup[TFRC_CALC_X_ARRSIZE - 1][0]) { 682 if (fvalue > tfrc_calc_x_lookup[TFRC_CALC_X_ARRSIZE - 1][0]) {
683 DCCP_WARN("fvalue %d exceeds bounds!\n", fvalue); 683 DCCP_WARN("fvalue %u exceeds bounds!\n", fvalue);
684 return 1000000; 684 return 1000000;
685 } 685 }
686 686
diff --git a/net/dccp/feat.c b/net/dccp/feat.c
index 4a4f6ce4498d..933a0ecf8d46 100644
--- a/net/dccp/feat.c
+++ b/net/dccp/feat.c
@@ -32,7 +32,7 @@ int dccp_feat_change(struct dccp_minisock *dmsk, u8 type, u8 feature,
32 32
33 if (len > 3) { 33 if (len > 3) {
34 DCCP_WARN("invalid length %d\n", len); 34 DCCP_WARN("invalid length %d\n", len);
35 return 1; 35 return -EINVAL;
36 } 36 }
37 /* XXX add further sanity checks */ 37 /* XXX add further sanity checks */
38 38
diff --git a/net/dccp/ipv4.c b/net/dccp/ipv4.c
index b348dd70c685..37d27bcb361f 100644
--- a/net/dccp/ipv4.c
+++ b/net/dccp/ipv4.c
@@ -589,7 +589,7 @@ int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
589 if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1) 589 if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
590 goto drop; 590 goto drop;
591 591
592 req = reqsk_alloc(&dccp_request_sock_ops); 592 req = inet_reqsk_alloc(&dccp_request_sock_ops);
593 if (req == NULL) 593 if (req == NULL)
594 goto drop; 594 goto drop;
595 595
@@ -605,7 +605,6 @@ int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
605 ireq = inet_rsk(req); 605 ireq = inet_rsk(req);
606 ireq->loc_addr = ip_hdr(skb)->daddr; 606 ireq->loc_addr = ip_hdr(skb)->daddr;
607 ireq->rmt_addr = ip_hdr(skb)->saddr; 607 ireq->rmt_addr = ip_hdr(skb)->saddr;
608 ireq->opt = NULL;
609 608
610 /* 609 /*
611 * Step 3: Process LISTEN state 610 * Step 3: Process LISTEN state
@@ -739,8 +738,8 @@ int dccp_invalid_packet(struct sk_buff *skb)
739 * If P.type is not Data, Ack, or DataAck and P.X == 0 (the packet 738 * If P.type is not Data, Ack, or DataAck and P.X == 0 (the packet
740 * has short sequence numbers), drop packet and return 739 * has short sequence numbers), drop packet and return
741 */ 740 */
742 if (dh->dccph_type >= DCCP_PKT_DATA && 741 if ((dh->dccph_type < DCCP_PKT_DATA ||
743 dh->dccph_type <= DCCP_PKT_DATAACK && dh->dccph_x == 0) { 742 dh->dccph_type > DCCP_PKT_DATAACK) && dh->dccph_x == 0) {
744 DCCP_WARN("P.type (%s) not Data || [Data]Ack, while P.X == 0\n", 743 DCCP_WARN("P.type (%s) not Data || [Data]Ack, while P.X == 0\n",
745 dccp_packet_name(dh->dccph_type)); 744 dccp_packet_name(dh->dccph_type));
746 return 1; 745 return 1;
diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c
index 9b1129bb7ece..f7fe2a572d7b 100644
--- a/net/dccp/ipv6.c
+++ b/net/dccp/ipv6.c
@@ -421,7 +421,6 @@ static int dccp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
421 ireq6 = inet6_rsk(req); 421 ireq6 = inet6_rsk(req);
422 ipv6_addr_copy(&ireq6->rmt_addr, &ipv6_hdr(skb)->saddr); 422 ipv6_addr_copy(&ireq6->rmt_addr, &ipv6_hdr(skb)->saddr);
423 ipv6_addr_copy(&ireq6->loc_addr, &ipv6_hdr(skb)->daddr); 423 ipv6_addr_copy(&ireq6->loc_addr, &ipv6_hdr(skb)->daddr);
424 ireq6->pktopts = NULL;
425 424
426 if (ipv6_opt_accepted(sk, skb) || 425 if (ipv6_opt_accepted(sk, skb) ||
427 np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo || 426 np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo ||
diff --git a/net/dccp/minisocks.c b/net/dccp/minisocks.c
index 33ad48321b08..66dca5bba858 100644
--- a/net/dccp/minisocks.c
+++ b/net/dccp/minisocks.c
@@ -165,12 +165,12 @@ out_free:
165 /* See dccp_v4_conn_request */ 165 /* See dccp_v4_conn_request */
166 newdmsk->dccpms_sequence_window = req->rcv_wnd; 166 newdmsk->dccpms_sequence_window = req->rcv_wnd;
167 167
168 newdp->dccps_gar = newdp->dccps_isr = dreq->dreq_isr; 168 newdp->dccps_gar = newdp->dccps_iss = dreq->dreq_iss;
169 dccp_update_gsr(newsk, dreq->dreq_isr);
170
171 newdp->dccps_iss = dreq->dreq_iss;
172 dccp_update_gss(newsk, dreq->dreq_iss); 169 dccp_update_gss(newsk, dreq->dreq_iss);
173 170
171 newdp->dccps_isr = dreq->dreq_isr;
172 dccp_update_gsr(newsk, dreq->dreq_isr);
173
174 /* 174 /*
175 * SWL and AWL are initially adjusted so that they are not less than 175 * SWL and AWL are initially adjusted so that they are not less than
176 * the initial Sequence Numbers received and sent, respectively: 176 * the initial Sequence Numbers received and sent, respectively:
diff --git a/net/dccp/options.c b/net/dccp/options.c
index d2a84a2fecee..43bc24e761d0 100644
--- a/net/dccp/options.c
+++ b/net/dccp/options.c
@@ -107,9 +107,11 @@ int dccp_parse_options(struct sock *sk, struct dccp_request_sock *dreq,
107 * 107 *
108 * CCID-specific options are ignored during connection setup, as 108 * CCID-specific options are ignored during connection setup, as
109 * negotiation may still be in progress (see RFC 4340, 10.3). 109 * negotiation may still be in progress (see RFC 4340, 10.3).
110 * The same applies to Ack Vectors, as these depend on the CCID.
110 * 111 *
111 */ 112 */
112 if (dreq != NULL && opt >= 128) 113 if (dreq != NULL && (opt >= 128 ||
114 opt == DCCPO_ACK_VECTOR_0 || opt == DCCPO_ACK_VECTOR_1))
113 goto ignore_option; 115 goto ignore_option;
114 116
115 switch (opt) { 117 switch (opt) {
diff --git a/net/dccp/output.c b/net/dccp/output.c
index 1f8a9b64c083..fe20068c5d8e 100644
--- a/net/dccp/output.c
+++ b/net/dccp/output.c
@@ -508,6 +508,7 @@ void dccp_send_ack(struct sock *sk)
508 508
509EXPORT_SYMBOL_GPL(dccp_send_ack); 509EXPORT_SYMBOL_GPL(dccp_send_ack);
510 510
511#if 0
511/* FIXME: Is this still necessary (11.3) - currently nowhere used by DCCP. */ 512/* FIXME: Is this still necessary (11.3) - currently nowhere used by DCCP. */
512void dccp_send_delayed_ack(struct sock *sk) 513void dccp_send_delayed_ack(struct sock *sk)
513{ 514{
@@ -538,6 +539,7 @@ void dccp_send_delayed_ack(struct sock *sk)
538 icsk->icsk_ack.timeout = timeout; 539 icsk->icsk_ack.timeout = timeout;
539 sk_reset_timer(sk, &icsk->icsk_delack_timer, timeout); 540 sk_reset_timer(sk, &icsk->icsk_delack_timer, timeout);
540} 541}
542#endif
541 543
542void dccp_send_sync(struct sock *sk, const u64 ackno, 544void dccp_send_sync(struct sock *sk, const u64 ackno,
543 const enum dccp_pkt_type pkt_type) 545 const enum dccp_pkt_type pkt_type)
diff --git a/net/dccp/probe.c b/net/dccp/probe.c
index 0bcdc9250279..81368a7f5379 100644
--- a/net/dccp/probe.c
+++ b/net/dccp/probe.c
@@ -42,7 +42,7 @@ static int bufsize = 64 * 1024;
42 42
43static const char procname[] = "dccpprobe"; 43static const char procname[] = "dccpprobe";
44 44
45struct { 45static struct {
46 struct kfifo *fifo; 46 struct kfifo *fifo;
47 spinlock_t lock; 47 spinlock_t lock;
48 wait_queue_head_t wait; 48 wait_queue_head_t wait;
diff --git a/net/decnet/dn_route.c b/net/decnet/dn_route.c
index 2f665a516476..f50e88bf2661 100644
--- a/net/decnet/dn_route.c
+++ b/net/decnet/dn_route.c
@@ -235,14 +235,14 @@ static void dn_dst_update_pmtu(struct dst_entry *dst, u32 mtu)
235 else 235 else
236 min_mtu -= 21; 236 min_mtu -= 21;
237 237
238 if (dst->metrics[RTAX_MTU-1] > mtu && mtu >= min_mtu) { 238 if (dst_metric(dst, RTAX_MTU) > mtu && mtu >= min_mtu) {
239 if (!(dst_metric_locked(dst, RTAX_MTU))) { 239 if (!(dst_metric_locked(dst, RTAX_MTU))) {
240 dst->metrics[RTAX_MTU-1] = mtu; 240 dst->metrics[RTAX_MTU-1] = mtu;
241 dst_set_expires(dst, dn_rt_mtu_expires); 241 dst_set_expires(dst, dn_rt_mtu_expires);
242 } 242 }
243 if (!(dst_metric_locked(dst, RTAX_ADVMSS))) { 243 if (!(dst_metric_locked(dst, RTAX_ADVMSS))) {
244 u32 mss = mtu - DN_MAX_NSP_DATA_HEADER; 244 u32 mss = mtu - DN_MAX_NSP_DATA_HEADER;
245 if (dst->metrics[RTAX_ADVMSS-1] > mss) 245 if (dst_metric(dst, RTAX_ADVMSS) > mss)
246 dst->metrics[RTAX_ADVMSS-1] = mss; 246 dst->metrics[RTAX_ADVMSS-1] = mss;
247 } 247 }
248 } 248 }
@@ -805,12 +805,12 @@ static int dn_rt_set_next_hop(struct dn_route *rt, struct dn_fib_res *res)
805 rt->u.dst.neighbour = n; 805 rt->u.dst.neighbour = n;
806 } 806 }
807 807
808 if (rt->u.dst.metrics[RTAX_MTU-1] == 0 || 808 if (dst_metric(&rt->u.dst, RTAX_MTU) == 0 ||
809 rt->u.dst.metrics[RTAX_MTU-1] > rt->u.dst.dev->mtu) 809 dst_metric(&rt->u.dst, RTAX_MTU) > rt->u.dst.dev->mtu)
810 rt->u.dst.metrics[RTAX_MTU-1] = rt->u.dst.dev->mtu; 810 rt->u.dst.metrics[RTAX_MTU-1] = rt->u.dst.dev->mtu;
811 mss = dn_mss_from_pmtu(dev, dst_mtu(&rt->u.dst)); 811 mss = dn_mss_from_pmtu(dev, dst_mtu(&rt->u.dst));
812 if (rt->u.dst.metrics[RTAX_ADVMSS-1] == 0 || 812 if (dst_metric(&rt->u.dst, RTAX_ADVMSS) == 0 ||
813 rt->u.dst.metrics[RTAX_ADVMSS-1] > mss) 813 dst_metric(&rt->u.dst, RTAX_ADVMSS) > mss)
814 rt->u.dst.metrics[RTAX_ADVMSS-1] = mss; 814 rt->u.dst.metrics[RTAX_ADVMSS-1] = mss;
815 return 0; 815 return 0;
816} 816}
diff --git a/net/econet/af_econet.c b/net/econet/af_econet.c
index 68d154480043..7c9bb13b1539 100644
--- a/net/econet/af_econet.c
+++ b/net/econet/af_econet.c
@@ -340,7 +340,7 @@ static int econet_sendmsg(struct kiocb *iocb, struct socket *sock,
340 340
341 dev_hold(dev); 341 dev_hold(dev);
342 342
343 skb = sock_alloc_send_skb(sk, len+LL_RESERVED_SPACE(dev), 343 skb = sock_alloc_send_skb(sk, len+LL_ALLOCATED_SPACE(dev),
344 msg->msg_flags & MSG_DONTWAIT, &err); 344 msg->msg_flags & MSG_DONTWAIT, &err);
345 if (skb==NULL) 345 if (skb==NULL)
346 goto out_unlock; 346 goto out_unlock;
diff --git a/net/ipv4/arp.c b/net/ipv4/arp.c
index 68b72a7a1806..9b539fa9fe18 100644
--- a/net/ipv4/arp.c
+++ b/net/ipv4/arp.c
@@ -570,7 +570,7 @@ struct sk_buff *arp_create(int type, int ptype, __be32 dest_ip,
570 * Allocate a buffer 570 * Allocate a buffer
571 */ 571 */
572 572
573 skb = alloc_skb(arp_hdr_len(dev) + LL_RESERVED_SPACE(dev), GFP_ATOMIC); 573 skb = alloc_skb(arp_hdr_len(dev) + LL_ALLOCATED_SPACE(dev), GFP_ATOMIC);
574 if (skb == NULL) 574 if (skb == NULL)
575 return NULL; 575 return NULL;
576 576
@@ -1288,7 +1288,6 @@ static void arp_format_neigh_entry(struct seq_file *seq,
1288 struct neighbour *n) 1288 struct neighbour *n)
1289{ 1289{
1290 char hbuffer[HBUFFERLEN]; 1290 char hbuffer[HBUFFERLEN];
1291 const char hexbuf[] = "0123456789ABCDEF";
1292 int k, j; 1291 int k, j;
1293 char tbuf[16]; 1292 char tbuf[16];
1294 struct net_device *dev = n->dev; 1293 struct net_device *dev = n->dev;
@@ -1302,8 +1301,8 @@ static void arp_format_neigh_entry(struct seq_file *seq,
1302 else { 1301 else {
1303#endif 1302#endif
1304 for (k = 0, j = 0; k < HBUFFERLEN - 3 && j < dev->addr_len; j++) { 1303 for (k = 0, j = 0; k < HBUFFERLEN - 3 && j < dev->addr_len; j++) {
1305 hbuffer[k++] = hexbuf[(n->ha[j] >> 4) & 15]; 1304 hbuffer[k++] = hex_asc_hi(n->ha[j]);
1306 hbuffer[k++] = hexbuf[n->ha[j] & 15]; 1305 hbuffer[k++] = hex_asc_lo(n->ha[j]);
1307 hbuffer[k++] = ':'; 1306 hbuffer[k++] = ':';
1308 } 1307 }
1309 hbuffer[--k] = 0; 1308 hbuffer[--k] = 0;
diff --git a/net/ipv4/cipso_ipv4.c b/net/ipv4/cipso_ipv4.c
index 4637ded3dba8..2c0e4572cc90 100644
--- a/net/ipv4/cipso_ipv4.c
+++ b/net/ipv4/cipso_ipv4.c
@@ -338,7 +338,7 @@ static int cipso_v4_cache_check(const unsigned char *key,
338 return -ENOENT; 338 return -ENOENT;
339 339
340 hash = cipso_v4_map_cache_hash(key, key_len); 340 hash = cipso_v4_map_cache_hash(key, key_len);
341 bkt = hash & (CIPSO_V4_CACHE_BUCKETBITS - 1); 341 bkt = hash & (CIPSO_V4_CACHE_BUCKETS - 1);
342 spin_lock_bh(&cipso_v4_cache[bkt].lock); 342 spin_lock_bh(&cipso_v4_cache[bkt].lock);
343 list_for_each_entry(entry, &cipso_v4_cache[bkt].list, list) { 343 list_for_each_entry(entry, &cipso_v4_cache[bkt].list, list) {
344 if (entry->hash == hash && 344 if (entry->hash == hash &&
@@ -417,7 +417,7 @@ int cipso_v4_cache_add(const struct sk_buff *skb,
417 atomic_inc(&secattr->cache->refcount); 417 atomic_inc(&secattr->cache->refcount);
418 entry->lsm_data = secattr->cache; 418 entry->lsm_data = secattr->cache;
419 419
420 bkt = entry->hash & (CIPSO_V4_CACHE_BUCKETBITS - 1); 420 bkt = entry->hash & (CIPSO_V4_CACHE_BUCKETS - 1);
421 spin_lock_bh(&cipso_v4_cache[bkt].lock); 421 spin_lock_bh(&cipso_v4_cache[bkt].lock);
422 if (cipso_v4_cache[bkt].size < cipso_v4_cache_bucketsize) { 422 if (cipso_v4_cache[bkt].size < cipso_v4_cache_bucketsize) {
423 list_add(&entry->list, &cipso_v4_cache[bkt].list); 423 list_add(&entry->list, &cipso_v4_cache[bkt].list);
@@ -983,7 +983,7 @@ static int cipso_v4_map_cat_enum_valid(const struct cipso_v4_doi *doi_def,
983 return -EFAULT; 983 return -EFAULT;
984 984
985 for (iter = 0; iter < enumcat_len; iter += 2) { 985 for (iter = 0; iter < enumcat_len; iter += 2) {
986 cat = ntohs(get_unaligned((__be16 *)&enumcat[iter])); 986 cat = get_unaligned_be16(&enumcat[iter]);
987 if (cat <= cat_prev) 987 if (cat <= cat_prev)
988 return -EFAULT; 988 return -EFAULT;
989 cat_prev = cat; 989 cat_prev = cat;
@@ -1052,7 +1052,7 @@ static int cipso_v4_map_cat_enum_ntoh(const struct cipso_v4_doi *doi_def,
1052 1052
1053 for (iter = 0; iter < net_cat_len; iter += 2) { 1053 for (iter = 0; iter < net_cat_len; iter += 2) {
1054 ret_val = netlbl_secattr_catmap_setbit(secattr->attr.mls.cat, 1054 ret_val = netlbl_secattr_catmap_setbit(secattr->attr.mls.cat,
1055 ntohs(get_unaligned((__be16 *)&net_cat[iter])), 1055 get_unaligned_be16(&net_cat[iter]),
1056 GFP_ATOMIC); 1056 GFP_ATOMIC);
1057 if (ret_val != 0) 1057 if (ret_val != 0)
1058 return ret_val; 1058 return ret_val;
@@ -1086,10 +1086,9 @@ static int cipso_v4_map_cat_rng_valid(const struct cipso_v4_doi *doi_def,
1086 return -EFAULT; 1086 return -EFAULT;
1087 1087
1088 for (iter = 0; iter < rngcat_len; iter += 4) { 1088 for (iter = 0; iter < rngcat_len; iter += 4) {
1089 cat_high = ntohs(get_unaligned((__be16 *)&rngcat[iter])); 1089 cat_high = get_unaligned_be16(&rngcat[iter]);
1090 if ((iter + 4) <= rngcat_len) 1090 if ((iter + 4) <= rngcat_len)
1091 cat_low = ntohs( 1091 cat_low = get_unaligned_be16(&rngcat[iter + 2]);
1092 get_unaligned((__be16 *)&rngcat[iter + 2]));
1093 else 1092 else
1094 cat_low = 0; 1093 cat_low = 0;
1095 1094
@@ -1188,10 +1187,9 @@ static int cipso_v4_map_cat_rng_ntoh(const struct cipso_v4_doi *doi_def,
1188 u16 cat_high; 1187 u16 cat_high;
1189 1188
1190 for (net_iter = 0; net_iter < net_cat_len; net_iter += 4) { 1189 for (net_iter = 0; net_iter < net_cat_len; net_iter += 4) {
1191 cat_high = ntohs(get_unaligned((__be16 *)&net_cat[net_iter])); 1190 cat_high = get_unaligned_be16(&net_cat[net_iter]);
1192 if ((net_iter + 4) <= net_cat_len) 1191 if ((net_iter + 4) <= net_cat_len)
1193 cat_low = ntohs( 1192 cat_low = get_unaligned_be16(&net_cat[net_iter + 2]);
1194 get_unaligned((__be16 *)&net_cat[net_iter + 2]));
1195 else 1193 else
1196 cat_low = 0; 1194 cat_low = 0;
1197 1195
@@ -1562,7 +1560,7 @@ int cipso_v4_validate(unsigned char **option)
1562 } 1560 }
1563 1561
1564 rcu_read_lock(); 1562 rcu_read_lock();
1565 doi_def = cipso_v4_doi_search(ntohl(get_unaligned((__be32 *)&opt[2]))); 1563 doi_def = cipso_v4_doi_search(get_unaligned_be32(&opt[2]));
1566 if (doi_def == NULL) { 1564 if (doi_def == NULL) {
1567 err_offset = 2; 1565 err_offset = 2;
1568 goto validate_return_locked; 1566 goto validate_return_locked;
@@ -1843,7 +1841,7 @@ static int cipso_v4_getattr(const unsigned char *cipso,
1843 if (cipso_v4_cache_check(cipso, cipso[1], secattr) == 0) 1841 if (cipso_v4_cache_check(cipso, cipso[1], secattr) == 0)
1844 return 0; 1842 return 0;
1845 1843
1846 doi = ntohl(get_unaligned((__be32 *)&cipso[2])); 1844 doi = get_unaligned_be32(&cipso[2]);
1847 rcu_read_lock(); 1845 rcu_read_lock();
1848 doi_def = cipso_v4_doi_search(doi); 1846 doi_def = cipso_v4_doi_search(doi);
1849 if (doi_def == NULL) 1847 if (doi_def == NULL)
diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index 6848e4760f34..79a7ef6209ff 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -90,7 +90,6 @@ static const struct nla_policy ifa_ipv4_policy[IFA_MAX+1] = {
90 [IFA_LOCAL] = { .type = NLA_U32 }, 90 [IFA_LOCAL] = { .type = NLA_U32 },
91 [IFA_ADDRESS] = { .type = NLA_U32 }, 91 [IFA_ADDRESS] = { .type = NLA_U32 },
92 [IFA_BROADCAST] = { .type = NLA_U32 }, 92 [IFA_BROADCAST] = { .type = NLA_U32 },
93 [IFA_ANYCAST] = { .type = NLA_U32 },
94 [IFA_LABEL] = { .type = NLA_STRING, .len = IFNAMSIZ - 1 }, 93 [IFA_LABEL] = { .type = NLA_STRING, .len = IFNAMSIZ - 1 },
95}; 94};
96 95
@@ -536,9 +535,6 @@ static struct in_ifaddr *rtm_to_ifaddr(struct net *net, struct nlmsghdr *nlh)
536 if (tb[IFA_BROADCAST]) 535 if (tb[IFA_BROADCAST])
537 ifa->ifa_broadcast = nla_get_be32(tb[IFA_BROADCAST]); 536 ifa->ifa_broadcast = nla_get_be32(tb[IFA_BROADCAST]);
538 537
539 if (tb[IFA_ANYCAST])
540 ifa->ifa_anycast = nla_get_be32(tb[IFA_ANYCAST]);
541
542 if (tb[IFA_LABEL]) 538 if (tb[IFA_LABEL])
543 nla_strlcpy(ifa->ifa_label, tb[IFA_LABEL], IFNAMSIZ); 539 nla_strlcpy(ifa->ifa_label, tb[IFA_LABEL], IFNAMSIZ);
544 else 540 else
@@ -745,7 +741,6 @@ int devinet_ioctl(struct net *net, unsigned int cmd, void __user *arg)
745 break; 741 break;
746 inet_del_ifa(in_dev, ifap, 0); 742 inet_del_ifa(in_dev, ifap, 0);
747 ifa->ifa_broadcast = 0; 743 ifa->ifa_broadcast = 0;
748 ifa->ifa_anycast = 0;
749 ifa->ifa_scope = 0; 744 ifa->ifa_scope = 0;
750 } 745 }
751 746
@@ -1113,7 +1108,6 @@ static inline size_t inet_nlmsg_size(void)
1113 + nla_total_size(4) /* IFA_ADDRESS */ 1108 + nla_total_size(4) /* IFA_ADDRESS */
1114 + nla_total_size(4) /* IFA_LOCAL */ 1109 + nla_total_size(4) /* IFA_LOCAL */
1115 + nla_total_size(4) /* IFA_BROADCAST */ 1110 + nla_total_size(4) /* IFA_BROADCAST */
1116 + nla_total_size(4) /* IFA_ANYCAST */
1117 + nla_total_size(IFNAMSIZ); /* IFA_LABEL */ 1111 + nla_total_size(IFNAMSIZ); /* IFA_LABEL */
1118} 1112}
1119 1113
@@ -1143,9 +1137,6 @@ static int inet_fill_ifaddr(struct sk_buff *skb, struct in_ifaddr *ifa,
1143 if (ifa->ifa_broadcast) 1137 if (ifa->ifa_broadcast)
1144 NLA_PUT_BE32(skb, IFA_BROADCAST, ifa->ifa_broadcast); 1138 NLA_PUT_BE32(skb, IFA_BROADCAST, ifa->ifa_broadcast);
1145 1139
1146 if (ifa->ifa_anycast)
1147 NLA_PUT_BE32(skb, IFA_ANYCAST, ifa->ifa_anycast);
1148
1149 if (ifa->ifa_label[0]) 1140 if (ifa->ifa_label[0])
1150 NLA_PUT_STRING(skb, IFA_LABEL, ifa->ifa_label); 1141 NLA_PUT_STRING(skb, IFA_LABEL, ifa->ifa_label);
1151 1142
diff --git a/net/ipv4/fib_frontend.c b/net/ipv4/fib_frontend.c
index 0f1557a4ac7a..0b2ac6a3d903 100644
--- a/net/ipv4/fib_frontend.c
+++ b/net/ipv4/fib_frontend.c
@@ -506,7 +506,6 @@ const struct nla_policy rtm_ipv4_policy[RTA_MAX+1] = {
506 [RTA_PREFSRC] = { .type = NLA_U32 }, 506 [RTA_PREFSRC] = { .type = NLA_U32 },
507 [RTA_METRICS] = { .type = NLA_NESTED }, 507 [RTA_METRICS] = { .type = NLA_NESTED },
508 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) }, 508 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
509 [RTA_PROTOINFO] = { .type = NLA_U32 },
510 [RTA_FLOW] = { .type = NLA_U32 }, 509 [RTA_FLOW] = { .type = NLA_U32 },
511}; 510};
512 511
diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c
index 3b83c34019fc..0d4d72827e4b 100644
--- a/net/ipv4/fib_semantics.c
+++ b/net/ipv4/fib_semantics.c
@@ -960,7 +960,10 @@ int fib_dump_info(struct sk_buff *skb, u32 pid, u32 seq, int event,
960 rtm->rtm_dst_len = dst_len; 960 rtm->rtm_dst_len = dst_len;
961 rtm->rtm_src_len = 0; 961 rtm->rtm_src_len = 0;
962 rtm->rtm_tos = tos; 962 rtm->rtm_tos = tos;
963 rtm->rtm_table = tb_id; 963 if (tb_id < 256)
964 rtm->rtm_table = tb_id;
965 else
966 rtm->rtm_table = RT_TABLE_COMPAT;
964 NLA_PUT_U32(skb, RTA_TABLE, tb_id); 967 NLA_PUT_U32(skb, RTA_TABLE, tb_id);
965 rtm->rtm_type = type; 968 rtm->rtm_type = type;
966 rtm->rtm_flags = fi->fib_flags; 969 rtm->rtm_flags = fi->fib_flags;
diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
index 4b02d14e7ab9..e1600ad8fb0e 100644
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -1359,17 +1359,17 @@ static int check_leaf(struct trie *t, struct leaf *l,
1359 t->stats.semantic_match_miss++; 1359 t->stats.semantic_match_miss++;
1360#endif 1360#endif
1361 if (err <= 0) 1361 if (err <= 0)
1362 return plen; 1362 return err;
1363 } 1363 }
1364 1364
1365 return -1; 1365 return 1;
1366} 1366}
1367 1367
1368static int fn_trie_lookup(struct fib_table *tb, const struct flowi *flp, 1368static int fn_trie_lookup(struct fib_table *tb, const struct flowi *flp,
1369 struct fib_result *res) 1369 struct fib_result *res)
1370{ 1370{
1371 struct trie *t = (struct trie *) tb->tb_data; 1371 struct trie *t = (struct trie *) tb->tb_data;
1372 int plen, ret = 0; 1372 int ret;
1373 struct node *n; 1373 struct node *n;
1374 struct tnode *pn; 1374 struct tnode *pn;
1375 int pos, bits; 1375 int pos, bits;
@@ -1393,10 +1393,7 @@ static int fn_trie_lookup(struct fib_table *tb, const struct flowi *flp,
1393 1393
1394 /* Just a leaf? */ 1394 /* Just a leaf? */
1395 if (IS_LEAF(n)) { 1395 if (IS_LEAF(n)) {
1396 plen = check_leaf(t, (struct leaf *)n, key, flp, res); 1396 ret = check_leaf(t, (struct leaf *)n, key, flp, res);
1397 if (plen < 0)
1398 goto failed;
1399 ret = 0;
1400 goto found; 1397 goto found;
1401 } 1398 }
1402 1399
@@ -1421,11 +1418,9 @@ static int fn_trie_lookup(struct fib_table *tb, const struct flowi *flp,
1421 } 1418 }
1422 1419
1423 if (IS_LEAF(n)) { 1420 if (IS_LEAF(n)) {
1424 plen = check_leaf(t, (struct leaf *)n, key, flp, res); 1421 ret = check_leaf(t, (struct leaf *)n, key, flp, res);
1425 if (plen < 0) 1422 if (ret > 0)
1426 goto backtrace; 1423 goto backtrace;
1427
1428 ret = 0;
1429 goto found; 1424 goto found;
1430 } 1425 }
1431 1426
diff --git a/net/ipv4/igmp.c b/net/ipv4/igmp.c
index 6250f4239b61..2769dc4a4c84 100644
--- a/net/ipv4/igmp.c
+++ b/net/ipv4/igmp.c
@@ -292,7 +292,7 @@ static struct sk_buff *igmpv3_newpack(struct net_device *dev, int size)
292 struct iphdr *pip; 292 struct iphdr *pip;
293 struct igmpv3_report *pig; 293 struct igmpv3_report *pig;
294 294
295 skb = alloc_skb(size + LL_RESERVED_SPACE(dev), GFP_ATOMIC); 295 skb = alloc_skb(size + LL_ALLOCATED_SPACE(dev), GFP_ATOMIC);
296 if (skb == NULL) 296 if (skb == NULL)
297 return NULL; 297 return NULL;
298 298
@@ -653,7 +653,7 @@ static int igmp_send_report(struct in_device *in_dev, struct ip_mc_list *pmc,
653 return -1; 653 return -1;
654 } 654 }
655 655
656 skb=alloc_skb(IGMP_SIZE+LL_RESERVED_SPACE(dev), GFP_ATOMIC); 656 skb=alloc_skb(IGMP_SIZE+LL_ALLOCATED_SPACE(dev), GFP_ATOMIC);
657 if (skb == NULL) { 657 if (skb == NULL) {
658 ip_rt_put(rt); 658 ip_rt_put(rt);
659 return -1; 659 return -1;
diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c
index 828ea211ff21..ec834480abe7 100644
--- a/net/ipv4/inet_connection_sock.c
+++ b/net/ipv4/inet_connection_sock.c
@@ -419,7 +419,8 @@ void inet_csk_reqsk_queue_prune(struct sock *parent,
419 struct inet_connection_sock *icsk = inet_csk(parent); 419 struct inet_connection_sock *icsk = inet_csk(parent);
420 struct request_sock_queue *queue = &icsk->icsk_accept_queue; 420 struct request_sock_queue *queue = &icsk->icsk_accept_queue;
421 struct listen_sock *lopt = queue->listen_opt; 421 struct listen_sock *lopt = queue->listen_opt;
422 int thresh = icsk->icsk_syn_retries ? : sysctl_tcp_synack_retries; 422 int max_retries = icsk->icsk_syn_retries ? : sysctl_tcp_synack_retries;
423 int thresh = max_retries;
423 unsigned long now = jiffies; 424 unsigned long now = jiffies;
424 struct request_sock **reqp, *req; 425 struct request_sock **reqp, *req;
425 int i, budget; 426 int i, budget;
@@ -455,6 +456,9 @@ void inet_csk_reqsk_queue_prune(struct sock *parent,
455 } 456 }
456 } 457 }
457 458
459 if (queue->rskq_defer_accept)
460 max_retries = queue->rskq_defer_accept;
461
458 budget = 2 * (lopt->nr_table_entries / (timeout / interval)); 462 budget = 2 * (lopt->nr_table_entries / (timeout / interval));
459 i = lopt->clock_hand; 463 i = lopt->clock_hand;
460 464
@@ -462,8 +466,9 @@ void inet_csk_reqsk_queue_prune(struct sock *parent,
462 reqp=&lopt->syn_table[i]; 466 reqp=&lopt->syn_table[i];
463 while ((req = *reqp) != NULL) { 467 while ((req = *reqp) != NULL) {
464 if (time_after_eq(now, req->expires)) { 468 if (time_after_eq(now, req->expires)) {
465 if (req->retrans < thresh && 469 if ((req->retrans < thresh ||
466 !req->rsk_ops->rtx_syn_ack(parent, req)) { 470 (inet_rsk(req)->acked && req->retrans < max_retries))
471 && !req->rsk_ops->rtx_syn_ack(parent, req)) {
467 unsigned long timeo; 472 unsigned long timeo;
468 473
469 if (req->retrans++ == 0) 474 if (req->retrans++ == 0)
diff --git a/net/ipv4/inet_fragment.c b/net/ipv4/inet_fragment.c
index 4ed429bd5951..0546a0bc97ea 100644
--- a/net/ipv4/inet_fragment.c
+++ b/net/ipv4/inet_fragment.c
@@ -192,14 +192,21 @@ EXPORT_SYMBOL(inet_frag_evictor);
192 192
193static struct inet_frag_queue *inet_frag_intern(struct netns_frags *nf, 193static struct inet_frag_queue *inet_frag_intern(struct netns_frags *nf,
194 struct inet_frag_queue *qp_in, struct inet_frags *f, 194 struct inet_frag_queue *qp_in, struct inet_frags *f,
195 unsigned int hash, void *arg) 195 void *arg)
196{ 196{
197 struct inet_frag_queue *qp; 197 struct inet_frag_queue *qp;
198#ifdef CONFIG_SMP 198#ifdef CONFIG_SMP
199 struct hlist_node *n; 199 struct hlist_node *n;
200#endif 200#endif
201 unsigned int hash;
201 202
202 write_lock(&f->lock); 203 write_lock(&f->lock);
204 /*
205 * While we stayed w/o the lock other CPU could update
206 * the rnd seed, so we need to re-calculate the hash
207 * chain. Fortunatelly the qp_in can be used to get one.
208 */
209 hash = f->hashfn(qp_in);
203#ifdef CONFIG_SMP 210#ifdef CONFIG_SMP
204 /* With SMP race we have to recheck hash table, because 211 /* With SMP race we have to recheck hash table, because
205 * such entry could be created on other cpu, while we 212 * such entry could be created on other cpu, while we
@@ -247,7 +254,7 @@ static struct inet_frag_queue *inet_frag_alloc(struct netns_frags *nf,
247} 254}
248 255
249static struct inet_frag_queue *inet_frag_create(struct netns_frags *nf, 256static struct inet_frag_queue *inet_frag_create(struct netns_frags *nf,
250 struct inet_frags *f, void *arg, unsigned int hash) 257 struct inet_frags *f, void *arg)
251{ 258{
252 struct inet_frag_queue *q; 259 struct inet_frag_queue *q;
253 260
@@ -255,7 +262,7 @@ static struct inet_frag_queue *inet_frag_create(struct netns_frags *nf,
255 if (q == NULL) 262 if (q == NULL)
256 return NULL; 263 return NULL;
257 264
258 return inet_frag_intern(nf, q, f, hash, arg); 265 return inet_frag_intern(nf, q, f, arg);
259} 266}
260 267
261struct inet_frag_queue *inet_frag_find(struct netns_frags *nf, 268struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
@@ -264,7 +271,6 @@ struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
264 struct inet_frag_queue *q; 271 struct inet_frag_queue *q;
265 struct hlist_node *n; 272 struct hlist_node *n;
266 273
267 read_lock(&f->lock);
268 hlist_for_each_entry(q, n, &f->hash[hash], list) { 274 hlist_for_each_entry(q, n, &f->hash[hash], list) {
269 if (q->net == nf && f->match(q, key)) { 275 if (q->net == nf && f->match(q, key)) {
270 atomic_inc(&q->refcnt); 276 atomic_inc(&q->refcnt);
@@ -274,6 +280,6 @@ struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
274 } 280 }
275 read_unlock(&f->lock); 281 read_unlock(&f->lock);
276 282
277 return inet_frag_create(nf, f, key, hash); 283 return inet_frag_create(nf, f, key);
278} 284}
279EXPORT_SYMBOL(inet_frag_find); 285EXPORT_SYMBOL(inet_frag_find);
diff --git a/net/ipv4/inet_lro.c b/net/ipv4/inet_lro.c
index 4a4d49fca1f2..cfd034a2b96e 100644
--- a/net/ipv4/inet_lro.c
+++ b/net/ipv4/inet_lro.c
@@ -383,8 +383,7 @@ static int __lro_proc_skb(struct net_lro_mgr *lro_mgr, struct sk_buff *skb,
383out2: /* send aggregated SKBs to stack */ 383out2: /* send aggregated SKBs to stack */
384 lro_flush(lro_mgr, lro_desc); 384 lro_flush(lro_mgr, lro_desc);
385 385
386out: /* Original SKB has to be posted to stack */ 386out:
387 skb->ip_summed = lro_mgr->ip_summed;
388 return 1; 387 return 1;
389} 388}
390 389
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index cd6ce6ac6358..37221f659159 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -229,6 +229,8 @@ static inline struct ipq *ip_find(struct net *net, struct iphdr *iph, u32 user)
229 229
230 arg.iph = iph; 230 arg.iph = iph;
231 arg.user = user; 231 arg.user = user;
232
233 read_lock(&ip4_frags.lock);
232 hash = ipqhashfn(iph->id, iph->saddr, iph->daddr, iph->protocol); 234 hash = ipqhashfn(iph->id, iph->saddr, iph->daddr, iph->protocol);
233 235
234 q = inet_frag_find(&net->ipv4.frags, &ip4_frags, &arg, hash); 236 q = inet_frag_find(&net->ipv4.frags, &ip4_frags, &arg, hash);
diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index 2ada033406de..4342cba4ff82 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -313,9 +313,8 @@ static void ipgre_tunnel_uninit(struct net_device *dev)
313 313
314static void ipgre_err(struct sk_buff *skb, u32 info) 314static void ipgre_err(struct sk_buff *skb, u32 info)
315{ 315{
316#ifndef I_WISH_WORLD_WERE_PERFECT
317 316
318/* It is not :-( All the routers (except for Linux) return only 317/* All the routers (except for Linux) return only
319 8 bytes of packet payload. It means, that precise relaying of 318 8 bytes of packet payload. It means, that precise relaying of
320 ICMP in the real Internet is absolutely infeasible. 319 ICMP in the real Internet is absolutely infeasible.
321 320
@@ -398,149 +397,6 @@ static void ipgre_err(struct sk_buff *skb, u32 info)
398out: 397out:
399 read_unlock(&ipgre_lock); 398 read_unlock(&ipgre_lock);
400 return; 399 return;
401#else
402 struct iphdr *iph = (struct iphdr*)dp;
403 struct iphdr *eiph;
404 __be16 *p = (__be16*)(dp+(iph->ihl<<2));
405 const int type = icmp_hdr(skb)->type;
406 const int code = icmp_hdr(skb)->code;
407 int rel_type = 0;
408 int rel_code = 0;
409 __be32 rel_info = 0;
410 __u32 n = 0;
411 __be16 flags;
412 int grehlen = (iph->ihl<<2) + 4;
413 struct sk_buff *skb2;
414 struct flowi fl;
415 struct rtable *rt;
416
417 if (p[1] != htons(ETH_P_IP))
418 return;
419
420 flags = p[0];
421 if (flags&(GRE_CSUM|GRE_KEY|GRE_SEQ|GRE_ROUTING|GRE_VERSION)) {
422 if (flags&(GRE_VERSION|GRE_ROUTING))
423 return;
424 if (flags&GRE_CSUM)
425 grehlen += 4;
426 if (flags&GRE_KEY)
427 grehlen += 4;
428 if (flags&GRE_SEQ)
429 grehlen += 4;
430 }
431 if (len < grehlen + sizeof(struct iphdr))
432 return;
433 eiph = (struct iphdr*)(dp + grehlen);
434
435 switch (type) {
436 default:
437 return;
438 case ICMP_PARAMETERPROB:
439 n = ntohl(icmp_hdr(skb)->un.gateway) >> 24;
440 if (n < (iph->ihl<<2))
441 return;
442
443 /* So... This guy found something strange INSIDE encapsulated
444 packet. Well, he is fool, but what can we do ?
445 */
446 rel_type = ICMP_PARAMETERPROB;
447 n -= grehlen;
448 rel_info = htonl(n << 24);
449 break;
450
451 case ICMP_DEST_UNREACH:
452 switch (code) {
453 case ICMP_SR_FAILED:
454 case ICMP_PORT_UNREACH:
455 /* Impossible event. */
456 return;
457 case ICMP_FRAG_NEEDED:
458 /* And it is the only really necessary thing :-) */
459 n = ntohs(icmp_hdr(skb)->un.frag.mtu);
460 if (n < grehlen+68)
461 return;
462 n -= grehlen;
463 /* BSD 4.2 MORE DOES NOT EXIST IN NATURE. */
464 if (n > ntohs(eiph->tot_len))
465 return;
466 rel_info = htonl(n);
467 break;
468 default:
469 /* All others are translated to HOST_UNREACH.
470 rfc2003 contains "deep thoughts" about NET_UNREACH,
471 I believe, it is just ether pollution. --ANK
472 */
473 rel_type = ICMP_DEST_UNREACH;
474 rel_code = ICMP_HOST_UNREACH;
475 break;
476 }
477 break;
478 case ICMP_TIME_EXCEEDED:
479 if (code != ICMP_EXC_TTL)
480 return;
481 break;
482 }
483
484 /* Prepare fake skb to feed it to icmp_send */
485 skb2 = skb_clone(skb, GFP_ATOMIC);
486 if (skb2 == NULL)
487 return;
488 dst_release(skb2->dst);
489 skb2->dst = NULL;
490 skb_pull(skb2, skb->data - (u8*)eiph);
491 skb_reset_network_header(skb2);
492
493 /* Try to guess incoming interface */
494 memset(&fl, 0, sizeof(fl));
495 fl.fl4_dst = eiph->saddr;
496 fl.fl4_tos = RT_TOS(eiph->tos);
497 fl.proto = IPPROTO_GRE;
498 if (ip_route_output_key(dev_net(skb->dev), &rt, &fl)) {
499 kfree_skb(skb2);
500 return;
501 }
502 skb2->dev = rt->u.dst.dev;
503
504 /* route "incoming" packet */
505 if (rt->rt_flags&RTCF_LOCAL) {
506 ip_rt_put(rt);
507 rt = NULL;
508 fl.fl4_dst = eiph->daddr;
509 fl.fl4_src = eiph->saddr;
510 fl.fl4_tos = eiph->tos;
511 if (ip_route_output_key(dev_net(skb->dev), &rt, &fl) ||
512 rt->u.dst.dev->type != ARPHRD_IPGRE) {
513 ip_rt_put(rt);
514 kfree_skb(skb2);
515 return;
516 }
517 } else {
518 ip_rt_put(rt);
519 if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos, skb2->dev) ||
520 skb2->dst->dev->type != ARPHRD_IPGRE) {
521 kfree_skb(skb2);
522 return;
523 }
524 }
525
526 /* change mtu on this route */
527 if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED) {
528 if (n > dst_mtu(skb2->dst)) {
529 kfree_skb(skb2);
530 return;
531 }
532 skb2->dst->ops->update_pmtu(skb2->dst, n);
533 } else if (type == ICMP_TIME_EXCEEDED) {
534 struct ip_tunnel *t = netdev_priv(skb2->dev);
535 if (t->parms.iph.ttl) {
536 rel_type = ICMP_DEST_UNREACH;
537 rel_code = ICMP_HOST_UNREACH;
538 }
539 }
540
541 icmp_send(skb2, rel_type, rel_code, rel_info);
542 kfree_skb(skb2);
543#endif
544} 400}
545 401
546static inline void ipgre_ecn_decapsulate(struct iphdr *iph, struct sk_buff *skb) 402static inline void ipgre_ecn_decapsulate(struct iphdr *iph, struct sk_buff *skb)
diff --git a/net/ipv4/ip_input.c b/net/ipv4/ip_input.c
index 7b4bad6d572f..ff77a4a7f9ec 100644
--- a/net/ipv4/ip_input.c
+++ b/net/ipv4/ip_input.c
@@ -397,7 +397,7 @@ int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
397 iph = ip_hdr(skb); 397 iph = ip_hdr(skb);
398 398
399 /* 399 /*
400 * RFC1122: 3.1.2.2 MUST silently discard any IP frame that fails the checksum. 400 * RFC1122: 3.2.1.2 MUST silently discard any IP frame that fails the checksum.
401 * 401 *
402 * Is the datagram acceptable? 402 * Is the datagram acceptable?
403 * 403 *
diff --git a/net/ipv4/ipconfig.c b/net/ipv4/ipconfig.c
index 89dee4346f60..ed45037ce9be 100644
--- a/net/ipv4/ipconfig.c
+++ b/net/ipv4/ipconfig.c
@@ -710,14 +710,14 @@ static void __init ic_bootp_send_if(struct ic_device *d, unsigned long jiffies_d
710 struct net_device *dev = d->dev; 710 struct net_device *dev = d->dev;
711 struct sk_buff *skb; 711 struct sk_buff *skb;
712 struct bootp_pkt *b; 712 struct bootp_pkt *b;
713 int hh_len = LL_RESERVED_SPACE(dev);
714 struct iphdr *h; 713 struct iphdr *h;
715 714
716 /* Allocate packet */ 715 /* Allocate packet */
717 skb = alloc_skb(sizeof(struct bootp_pkt) + hh_len + 15, GFP_KERNEL); 716 skb = alloc_skb(sizeof(struct bootp_pkt) + LL_ALLOCATED_SPACE(dev) + 15,
717 GFP_KERNEL);
718 if (!skb) 718 if (!skb)
719 return; 719 return;
720 skb_reserve(skb, hh_len); 720 skb_reserve(skb, LL_RESERVED_SPACE(dev));
721 b = (struct bootp_pkt *) skb_put(skb, sizeof(struct bootp_pkt)); 721 b = (struct bootp_pkt *) skb_put(skb, sizeof(struct bootp_pkt));
722 memset(b, 0, sizeof(struct bootp_pkt)); 722 memset(b, 0, sizeof(struct bootp_pkt));
723 723
diff --git a/net/ipv4/ipip.c b/net/ipv4/ipip.c
index 149111f08e8d..af5cb53da5cc 100644
--- a/net/ipv4/ipip.c
+++ b/net/ipv4/ipip.c
@@ -278,9 +278,8 @@ static void ipip_tunnel_uninit(struct net_device *dev)
278 278
279static int ipip_err(struct sk_buff *skb, u32 info) 279static int ipip_err(struct sk_buff *skb, u32 info)
280{ 280{
281#ifndef I_WISH_WORLD_WERE_PERFECT
282 281
283/* It is not :-( All the routers (except for Linux) return only 282/* All the routers (except for Linux) return only
284 8 bytes of packet payload. It means, that precise relaying of 283 8 bytes of packet payload. It means, that precise relaying of
285 ICMP in the real Internet is absolutely infeasible. 284 ICMP in the real Internet is absolutely infeasible.
286 */ 285 */
@@ -337,133 +336,6 @@ static int ipip_err(struct sk_buff *skb, u32 info)
337out: 336out:
338 read_unlock(&ipip_lock); 337 read_unlock(&ipip_lock);
339 return err; 338 return err;
340#else
341 struct iphdr *iph = (struct iphdr*)dp;
342 int hlen = iph->ihl<<2;
343 struct iphdr *eiph;
344 const int type = icmp_hdr(skb)->type;
345 const int code = icmp_hdr(skb)->code;
346 int rel_type = 0;
347 int rel_code = 0;
348 __be32 rel_info = 0;
349 __u32 n = 0;
350 struct sk_buff *skb2;
351 struct flowi fl;
352 struct rtable *rt;
353
354 if (len < hlen + sizeof(struct iphdr))
355 return 0;
356 eiph = (struct iphdr*)(dp + hlen);
357
358 switch (type) {
359 default:
360 return 0;
361 case ICMP_PARAMETERPROB:
362 n = ntohl(icmp_hdr(skb)->un.gateway) >> 24;
363 if (n < hlen)
364 return 0;
365
366 /* So... This guy found something strange INSIDE encapsulated
367 packet. Well, he is fool, but what can we do ?
368 */
369 rel_type = ICMP_PARAMETERPROB;
370 rel_info = htonl((n - hlen) << 24);
371 break;
372
373 case ICMP_DEST_UNREACH:
374 switch (code) {
375 case ICMP_SR_FAILED:
376 case ICMP_PORT_UNREACH:
377 /* Impossible event. */
378 return 0;
379 case ICMP_FRAG_NEEDED:
380 /* And it is the only really necessary thing :-) */
381 n = ntohs(icmp_hdr(skb)->un.frag.mtu);
382 if (n < hlen+68)
383 return 0;
384 n -= hlen;
385 /* BSD 4.2 MORE DOES NOT EXIST IN NATURE. */
386 if (n > ntohs(eiph->tot_len))
387 return 0;
388 rel_info = htonl(n);
389 break;
390 default:
391 /* All others are translated to HOST_UNREACH.
392 rfc2003 contains "deep thoughts" about NET_UNREACH,
393 I believe, it is just ether pollution. --ANK
394 */
395 rel_type = ICMP_DEST_UNREACH;
396 rel_code = ICMP_HOST_UNREACH;
397 break;
398 }
399 break;
400 case ICMP_TIME_EXCEEDED:
401 if (code != ICMP_EXC_TTL)
402 return 0;
403 break;
404 }
405
406 /* Prepare fake skb to feed it to icmp_send */
407 skb2 = skb_clone(skb, GFP_ATOMIC);
408 if (skb2 == NULL)
409 return 0;
410 dst_release(skb2->dst);
411 skb2->dst = NULL;
412 skb_pull(skb2, skb->data - (u8*)eiph);
413 skb_reset_network_header(skb2);
414
415 /* Try to guess incoming interface */
416 memset(&fl, 0, sizeof(fl));
417 fl.fl4_daddr = eiph->saddr;
418 fl.fl4_tos = RT_TOS(eiph->tos);
419 fl.proto = IPPROTO_IPIP;
420 if (ip_route_output_key(dev_net(skb->dev), &rt, &key)) {
421 kfree_skb(skb2);
422 return 0;
423 }
424 skb2->dev = rt->u.dst.dev;
425
426 /* route "incoming" packet */
427 if (rt->rt_flags&RTCF_LOCAL) {
428 ip_rt_put(rt);
429 rt = NULL;
430 fl.fl4_daddr = eiph->daddr;
431 fl.fl4_src = eiph->saddr;
432 fl.fl4_tos = eiph->tos;
433 if (ip_route_output_key(dev_net(skb->dev), &rt, &fl) ||
434 rt->u.dst.dev->type != ARPHRD_TUNNEL) {
435 ip_rt_put(rt);
436 kfree_skb(skb2);
437 return 0;
438 }
439 } else {
440 ip_rt_put(rt);
441 if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos, skb2->dev) ||
442 skb2->dst->dev->type != ARPHRD_TUNNEL) {
443 kfree_skb(skb2);
444 return 0;
445 }
446 }
447
448 /* change mtu on this route */
449 if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED) {
450 if (n > dst_mtu(skb2->dst)) {
451 kfree_skb(skb2);
452 return 0;
453 }
454 skb2->dst->ops->update_pmtu(skb2->dst, n);
455 } else if (type == ICMP_TIME_EXCEEDED) {
456 struct ip_tunnel *t = netdev_priv(skb2->dev);
457 if (t->parms.iph.ttl) {
458 rel_type = ICMP_DEST_UNREACH;
459 rel_code = ICMP_HOST_UNREACH;
460 }
461 }
462
463 icmp_send(skb2, rel_type, rel_code, rel_info);
464 kfree_skb(skb2);
465 return 0;
466#endif
467} 339}
468 340
469static inline void ipip_ecn_decapsulate(const struct iphdr *outer_iph, 341static inline void ipip_ecn_decapsulate(const struct iphdr *outer_iph,
diff --git a/net/ipv4/netfilter/ipt_CLUSTERIP.c b/net/ipv4/netfilter/ipt_CLUSTERIP.c
index 22d8e7cd9197..1819ad7ab910 100644
--- a/net/ipv4/netfilter/ipt_CLUSTERIP.c
+++ b/net/ipv4/netfilter/ipt_CLUSTERIP.c
@@ -169,14 +169,14 @@ clusterip_config_init(const struct ipt_clusterip_tgt_info *i, __be32 ip,
169 169
170 /* create proc dir entry */ 170 /* create proc dir entry */
171 sprintf(buffer, "%u.%u.%u.%u", NIPQUAD(ip)); 171 sprintf(buffer, "%u.%u.%u.%u", NIPQUAD(ip));
172 c->pde = proc_create(buffer, S_IWUSR|S_IRUSR, 172 c->pde = proc_create_data(buffer, S_IWUSR|S_IRUSR,
173 clusterip_procdir, &clusterip_proc_fops); 173 clusterip_procdir,
174 &clusterip_proc_fops, c);
174 if (!c->pde) { 175 if (!c->pde) {
175 kfree(c); 176 kfree(c);
176 return NULL; 177 return NULL;
177 } 178 }
178 } 179 }
179 c->pde->data = c;
180#endif 180#endif
181 181
182 write_lock_bh(&clusterip_lock); 182 write_lock_bh(&clusterip_lock);
diff --git a/net/ipv4/netfilter/nf_nat_core.c b/net/ipv4/netfilter/nf_nat_core.c
index 04578593e100..d2a887fc8d9b 100644
--- a/net/ipv4/netfilter/nf_nat_core.c
+++ b/net/ipv4/netfilter/nf_nat_core.c
@@ -556,7 +556,6 @@ static void nf_nat_cleanup_conntrack(struct nf_conn *ct)
556 556
557 spin_lock_bh(&nf_nat_lock); 557 spin_lock_bh(&nf_nat_lock);
558 hlist_del_rcu(&nat->bysource); 558 hlist_del_rcu(&nat->bysource);
559 nat->ct = NULL;
560 spin_unlock_bh(&nf_nat_lock); 559 spin_unlock_bh(&nf_nat_lock);
561} 560}
562 561
@@ -570,8 +569,8 @@ static void nf_nat_move_storage(void *new, void *old)
570 return; 569 return;
571 570
572 spin_lock_bh(&nf_nat_lock); 571 spin_lock_bh(&nf_nat_lock);
573 hlist_replace_rcu(&old_nat->bysource, &new_nat->bysource);
574 new_nat->ct = ct; 572 new_nat->ct = ct;
573 hlist_replace_rcu(&old_nat->bysource, &new_nat->bysource);
575 spin_unlock_bh(&nf_nat_lock); 574 spin_unlock_bh(&nf_nat_lock);
576} 575}
577 576
diff --git a/net/ipv4/netfilter/nf_nat_snmp_basic.c b/net/ipv4/netfilter/nf_nat_snmp_basic.c
index 5daefad3d193..ffeaffc3fffe 100644
--- a/net/ipv4/netfilter/nf_nat_snmp_basic.c
+++ b/net/ipv4/netfilter/nf_nat_snmp_basic.c
@@ -232,6 +232,11 @@ static unsigned char asn1_length_decode(struct asn1_ctx *ctx,
232 } 232 }
233 } 233 }
234 } 234 }
235
236 /* don't trust len bigger than ctx buffer */
237 if (*len > ctx->end - ctx->pointer)
238 return 0;
239
235 return 1; 240 return 1;
236} 241}
237 242
@@ -250,6 +255,10 @@ static unsigned char asn1_header_decode(struct asn1_ctx *ctx,
250 if (!asn1_length_decode(ctx, &def, &len)) 255 if (!asn1_length_decode(ctx, &def, &len))
251 return 0; 256 return 0;
252 257
258 /* primitive shall be definite, indefinite shall be constructed */
259 if (*con == ASN1_PRI && !def)
260 return 0;
261
253 if (def) 262 if (def)
254 *eoc = ctx->pointer + len; 263 *eoc = ctx->pointer + len;
255 else 264 else
@@ -430,10 +439,15 @@ static unsigned char asn1_oid_decode(struct asn1_ctx *ctx,
430 unsigned int *len) 439 unsigned int *len)
431{ 440{
432 unsigned long subid; 441 unsigned long subid;
433 unsigned int size;
434 unsigned long *optr; 442 unsigned long *optr;
443 size_t size;
435 444
436 size = eoc - ctx->pointer + 1; 445 size = eoc - ctx->pointer + 1;
446
447 /* first subid actually encodes first two subids */
448 if (size < 2 || size > ULONG_MAX/sizeof(unsigned long))
449 return 0;
450
437 *oid = kmalloc(size * sizeof(unsigned long), GFP_ATOMIC); 451 *oid = kmalloc(size * sizeof(unsigned long), GFP_ATOMIC);
438 if (*oid == NULL) { 452 if (*oid == NULL) {
439 if (net_ratelimit()) 453 if (net_ratelimit())
diff --git a/net/ipv4/raw.c b/net/ipv4/raw.c
index 11d7f753a820..37a1ecd9d600 100644
--- a/net/ipv4/raw.c
+++ b/net/ipv4/raw.c
@@ -322,7 +322,6 @@ static int raw_send_hdrinc(struct sock *sk, void *from, size_t length,
322 unsigned int flags) 322 unsigned int flags)
323{ 323{
324 struct inet_sock *inet = inet_sk(sk); 324 struct inet_sock *inet = inet_sk(sk);
325 int hh_len;
326 struct iphdr *iph; 325 struct iphdr *iph;
327 struct sk_buff *skb; 326 struct sk_buff *skb;
328 unsigned int iphlen; 327 unsigned int iphlen;
@@ -336,13 +335,12 @@ static int raw_send_hdrinc(struct sock *sk, void *from, size_t length,
336 if (flags&MSG_PROBE) 335 if (flags&MSG_PROBE)
337 goto out; 336 goto out;
338 337
339 hh_len = LL_RESERVED_SPACE(rt->u.dst.dev); 338 skb = sock_alloc_send_skb(sk,
340 339 length + LL_ALLOCATED_SPACE(rt->u.dst.dev) + 15,
341 skb = sock_alloc_send_skb(sk, length+hh_len+15, 340 flags & MSG_DONTWAIT, &err);
342 flags&MSG_DONTWAIT, &err);
343 if (skb == NULL) 341 if (skb == NULL)
344 goto error; 342 goto error;
345 skb_reserve(skb, hh_len); 343 skb_reserve(skb, LL_RESERVED_SPACE(rt->u.dst.dev));
346 344
347 skb->priority = sk->sk_priority; 345 skb->priority = sk->sk_priority;
348 skb->mark = sk->sk_mark; 346 skb->mark = sk->sk_mark;
@@ -610,6 +608,14 @@ static void raw_close(struct sock *sk, long timeout)
610 sk_common_release(sk); 608 sk_common_release(sk);
611} 609}
612 610
611static int raw_destroy(struct sock *sk)
612{
613 lock_sock(sk);
614 ip_flush_pending_frames(sk);
615 release_sock(sk);
616 return 0;
617}
618
613/* This gets rid of all the nasties in af_inet. -DaveM */ 619/* This gets rid of all the nasties in af_inet. -DaveM */
614static int raw_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len) 620static int raw_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
615{ 621{
@@ -822,6 +828,7 @@ struct proto raw_prot = {
822 .name = "RAW", 828 .name = "RAW",
823 .owner = THIS_MODULE, 829 .owner = THIS_MODULE,
824 .close = raw_close, 830 .close = raw_close,
831 .destroy = raw_destroy,
825 .connect = ip4_datagram_connect, 832 .connect = ip4_datagram_connect,
826 .disconnect = udp_disconnect, 833 .disconnect = udp_disconnect,
827 .ioctl = raw_ioctl, 834 .ioctl = raw_ioctl,
@@ -927,7 +934,7 @@ static void raw_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
927 srcp = inet->num; 934 srcp = inet->num;
928 935
929 seq_printf(seq, "%4d: %08X:%04X %08X:%04X" 936 seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
930 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d", 937 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d\n",
931 i, src, srcp, dest, destp, sp->sk_state, 938 i, src, srcp, dest, destp, sp->sk_state,
932 atomic_read(&sp->sk_wmem_alloc), 939 atomic_read(&sp->sk_wmem_alloc),
933 atomic_read(&sp->sk_rmem_alloc), 940 atomic_read(&sp->sk_rmem_alloc),
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 5e3685c5c407..96be336064fb 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -160,7 +160,7 @@ static struct dst_ops ipv4_dst_ops = {
160 .negative_advice = ipv4_negative_advice, 160 .negative_advice = ipv4_negative_advice,
161 .link_failure = ipv4_link_failure, 161 .link_failure = ipv4_link_failure,
162 .update_pmtu = ip_rt_update_pmtu, 162 .update_pmtu = ip_rt_update_pmtu,
163 .local_out = ip_local_out, 163 .local_out = __ip_local_out,
164 .entry_size = sizeof(struct rtable), 164 .entry_size = sizeof(struct rtable),
165 .entries = ATOMIC_INIT(0), 165 .entries = ATOMIC_INIT(0),
166}; 166};
@@ -1468,14 +1468,14 @@ unsigned short ip_rt_frag_needed(struct net *net, struct iphdr *iph,
1468 1468
1469 /* BSD 4.2 compatibility hack :-( */ 1469 /* BSD 4.2 compatibility hack :-( */
1470 if (mtu == 0 && 1470 if (mtu == 0 &&
1471 old_mtu >= rth->u.dst.metrics[RTAX_MTU-1] && 1471 old_mtu >= dst_metric(&rth->u.dst, RTAX_MTU) &&
1472 old_mtu >= 68 + (iph->ihl << 2)) 1472 old_mtu >= 68 + (iph->ihl << 2))
1473 old_mtu -= iph->ihl << 2; 1473 old_mtu -= iph->ihl << 2;
1474 1474
1475 mtu = guess_mtu(old_mtu); 1475 mtu = guess_mtu(old_mtu);
1476 } 1476 }
1477 if (mtu <= rth->u.dst.metrics[RTAX_MTU-1]) { 1477 if (mtu <= dst_metric(&rth->u.dst, RTAX_MTU)) {
1478 if (mtu < rth->u.dst.metrics[RTAX_MTU-1]) { 1478 if (mtu < dst_metric(&rth->u.dst, RTAX_MTU)) {
1479 dst_confirm(&rth->u.dst); 1479 dst_confirm(&rth->u.dst);
1480 if (mtu < ip_rt_min_pmtu) { 1480 if (mtu < ip_rt_min_pmtu) {
1481 mtu = ip_rt_min_pmtu; 1481 mtu = ip_rt_min_pmtu;
@@ -1497,7 +1497,7 @@ unsigned short ip_rt_frag_needed(struct net *net, struct iphdr *iph,
1497 1497
1498static void ip_rt_update_pmtu(struct dst_entry *dst, u32 mtu) 1498static void ip_rt_update_pmtu(struct dst_entry *dst, u32 mtu)
1499{ 1499{
1500 if (dst->metrics[RTAX_MTU-1] > mtu && mtu >= 68 && 1500 if (dst_metric(dst, RTAX_MTU) > mtu && mtu >= 68 &&
1501 !(dst_metric_locked(dst, RTAX_MTU))) { 1501 !(dst_metric_locked(dst, RTAX_MTU))) {
1502 if (mtu < ip_rt_min_pmtu) { 1502 if (mtu < ip_rt_min_pmtu) {
1503 mtu = ip_rt_min_pmtu; 1503 mtu = ip_rt_min_pmtu;
@@ -1613,7 +1613,7 @@ static void rt_set_nexthop(struct rtable *rt, struct fib_result *res, u32 itag)
1613 sizeof(rt->u.dst.metrics)); 1613 sizeof(rt->u.dst.metrics));
1614 if (fi->fib_mtu == 0) { 1614 if (fi->fib_mtu == 0) {
1615 rt->u.dst.metrics[RTAX_MTU-1] = rt->u.dst.dev->mtu; 1615 rt->u.dst.metrics[RTAX_MTU-1] = rt->u.dst.dev->mtu;
1616 if (rt->u.dst.metrics[RTAX_LOCK-1] & (1 << RTAX_MTU) && 1616 if (dst_metric_locked(&rt->u.dst, RTAX_MTU) &&
1617 rt->rt_gateway != rt->rt_dst && 1617 rt->rt_gateway != rt->rt_dst &&
1618 rt->u.dst.dev->mtu > 576) 1618 rt->u.dst.dev->mtu > 576)
1619 rt->u.dst.metrics[RTAX_MTU-1] = 576; 1619 rt->u.dst.metrics[RTAX_MTU-1] = 576;
@@ -1624,14 +1624,14 @@ static void rt_set_nexthop(struct rtable *rt, struct fib_result *res, u32 itag)
1624 } else 1624 } else
1625 rt->u.dst.metrics[RTAX_MTU-1]= rt->u.dst.dev->mtu; 1625 rt->u.dst.metrics[RTAX_MTU-1]= rt->u.dst.dev->mtu;
1626 1626
1627 if (rt->u.dst.metrics[RTAX_HOPLIMIT-1] == 0) 1627 if (dst_metric(&rt->u.dst, RTAX_HOPLIMIT) == 0)
1628 rt->u.dst.metrics[RTAX_HOPLIMIT-1] = sysctl_ip_default_ttl; 1628 rt->u.dst.metrics[RTAX_HOPLIMIT-1] = sysctl_ip_default_ttl;
1629 if (rt->u.dst.metrics[RTAX_MTU-1] > IP_MAX_MTU) 1629 if (dst_metric(&rt->u.dst, RTAX_MTU) > IP_MAX_MTU)
1630 rt->u.dst.metrics[RTAX_MTU-1] = IP_MAX_MTU; 1630 rt->u.dst.metrics[RTAX_MTU-1] = IP_MAX_MTU;
1631 if (rt->u.dst.metrics[RTAX_ADVMSS-1] == 0) 1631 if (dst_metric(&rt->u.dst, RTAX_ADVMSS) == 0)
1632 rt->u.dst.metrics[RTAX_ADVMSS-1] = max_t(unsigned int, rt->u.dst.dev->mtu - 40, 1632 rt->u.dst.metrics[RTAX_ADVMSS-1] = max_t(unsigned int, rt->u.dst.dev->mtu - 40,
1633 ip_rt_min_advmss); 1633 ip_rt_min_advmss);
1634 if (rt->u.dst.metrics[RTAX_ADVMSS-1] > 65535 - 40) 1634 if (dst_metric(&rt->u.dst, RTAX_ADVMSS) > 65535 - 40)
1635 rt->u.dst.metrics[RTAX_ADVMSS-1] = 65535 - 40; 1635 rt->u.dst.metrics[RTAX_ADVMSS-1] = 65535 - 40;
1636 1636
1637#ifdef CONFIG_NET_CLS_ROUTE 1637#ifdef CONFIG_NET_CLS_ROUTE
@@ -1792,7 +1792,7 @@ static int __mkroute_input(struct sk_buff *skb,
1792 if (err) 1792 if (err)
1793 flags |= RTCF_DIRECTSRC; 1793 flags |= RTCF_DIRECTSRC;
1794 1794
1795 if (out_dev == in_dev && err && !(flags & RTCF_MASQ) && 1795 if (out_dev == in_dev && err &&
1796 (IN_DEV_SHARED_MEDIA(out_dev) || 1796 (IN_DEV_SHARED_MEDIA(out_dev) ||
1797 inet_addr_onlink(out_dev, saddr, FIB_RES_GW(*res)))) 1797 inet_addr_onlink(out_dev, saddr, FIB_RES_GW(*res))))
1798 flags |= RTCF_DOREDIRECT; 1798 flags |= RTCF_DOREDIRECT;
diff --git a/net/ipv4/syncookies.c b/net/ipv4/syncookies.c
index 73ba98921d64..d182a2a26291 100644
--- a/net/ipv4/syncookies.c
+++ b/net/ipv4/syncookies.c
@@ -285,7 +285,7 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb,
285 cookie_check_timestamp(&tcp_opt); 285 cookie_check_timestamp(&tcp_opt);
286 286
287 ret = NULL; 287 ret = NULL;
288 req = reqsk_alloc(&tcp_request_sock_ops); /* for safety */ 288 req = inet_reqsk_alloc(&tcp_request_sock_ops); /* for safety */
289 if (!req) 289 if (!req)
290 goto out; 290 goto out;
291 291
@@ -301,7 +301,6 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb,
301 ireq->rmt_port = th->source; 301 ireq->rmt_port = th->source;
302 ireq->loc_addr = ip_hdr(skb)->daddr; 302 ireq->loc_addr = ip_hdr(skb)->daddr;
303 ireq->rmt_addr = ip_hdr(skb)->saddr; 303 ireq->rmt_addr = ip_hdr(skb)->saddr;
304 ireq->opt = NULL;
305 ireq->snd_wscale = tcp_opt.snd_wscale; 304 ireq->snd_wscale = tcp_opt.snd_wscale;
306 ireq->rcv_wscale = tcp_opt.rcv_wscale; 305 ireq->rcv_wscale = tcp_opt.rcv_wscale;
307 ireq->sack_ok = tcp_opt.sack_ok; 306 ireq->sack_ok = tcp_opt.sack_ok;
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index f88653138621..1d723de18686 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -255,11 +255,14 @@
255#include <linux/init.h> 255#include <linux/init.h>
256#include <linux/fs.h> 256#include <linux/fs.h>
257#include <linux/skbuff.h> 257#include <linux/skbuff.h>
258#include <linux/scatterlist.h>
258#include <linux/splice.h> 259#include <linux/splice.h>
259#include <linux/net.h> 260#include <linux/net.h>
260#include <linux/socket.h> 261#include <linux/socket.h>
261#include <linux/random.h> 262#include <linux/random.h>
262#include <linux/bootmem.h> 263#include <linux/bootmem.h>
264#include <linux/highmem.h>
265#include <linux/swap.h>
263#include <linux/cache.h> 266#include <linux/cache.h>
264#include <linux/err.h> 267#include <linux/err.h>
265#include <linux/crypto.h> 268#include <linux/crypto.h>
@@ -1206,7 +1209,8 @@ int tcp_read_sock(struct sock *sk, read_descriptor_t *desc,
1206 return -ENOTCONN; 1209 return -ENOTCONN;
1207 while ((skb = tcp_recv_skb(sk, seq, &offset)) != NULL) { 1210 while ((skb = tcp_recv_skb(sk, seq, &offset)) != NULL) {
1208 if (offset < skb->len) { 1211 if (offset < skb->len) {
1209 size_t used, len; 1212 int used;
1213 size_t len;
1210 1214
1211 len = skb->len - offset; 1215 len = skb->len - offset;
1212 /* Stop reading if we hit a patch of urgent data */ 1216 /* Stop reading if we hit a patch of urgent data */
@@ -1227,7 +1231,14 @@ int tcp_read_sock(struct sock *sk, read_descriptor_t *desc,
1227 copied += used; 1231 copied += used;
1228 offset += used; 1232 offset += used;
1229 } 1233 }
1230 if (offset != skb->len) 1234 /*
1235 * If recv_actor drops the lock (e.g. TCP splice
1236 * receive) the skb pointer might be invalid when
1237 * getting here: tcp_collapse might have deleted it
1238 * while aggregating skbs from the socket queue.
1239 */
1240 skb = tcp_recv_skb(sk, seq-1, &offset);
1241 if (!skb || (offset+1 != skb->len))
1231 break; 1242 break;
1232 } 1243 }
1233 if (tcp_hdr(skb)->fin) { 1244 if (tcp_hdr(skb)->fin) {
@@ -2105,12 +2116,15 @@ static int do_tcp_setsockopt(struct sock *sk, int level,
2105 break; 2116 break;
2106 2117
2107 case TCP_DEFER_ACCEPT: 2118 case TCP_DEFER_ACCEPT:
2108 if (val < 0) { 2119 icsk->icsk_accept_queue.rskq_defer_accept = 0;
2109 err = -EINVAL; 2120 if (val > 0) {
2110 } else { 2121 /* Translate value in seconds to number of
2111 if (val > MAX_TCP_ACCEPT_DEFERRED) 2122 * retransmits */
2112 val = MAX_TCP_ACCEPT_DEFERRED; 2123 while (icsk->icsk_accept_queue.rskq_defer_accept < 32 &&
2113 icsk->icsk_accept_queue.rskq_defer_accept = val; 2124 val > ((TCP_TIMEOUT_INIT / HZ) <<
2125 icsk->icsk_accept_queue.rskq_defer_accept))
2126 icsk->icsk_accept_queue.rskq_defer_accept++;
2127 icsk->icsk_accept_queue.rskq_defer_accept++;
2114 } 2128 }
2115 break; 2129 break;
2116 2130
@@ -2292,7 +2306,8 @@ static int do_tcp_getsockopt(struct sock *sk, int level,
2292 val = (val ? : sysctl_tcp_fin_timeout) / HZ; 2306 val = (val ? : sysctl_tcp_fin_timeout) / HZ;
2293 break; 2307 break;
2294 case TCP_DEFER_ACCEPT: 2308 case TCP_DEFER_ACCEPT:
2295 val = icsk->icsk_accept_queue.rskq_defer_accept; 2309 val = !icsk->icsk_accept_queue.rskq_defer_accept ? 0 :
2310 ((TCP_TIMEOUT_INIT / HZ) << (icsk->icsk_accept_queue.rskq_defer_accept - 1));
2296 break; 2311 break;
2297 case TCP_WINDOW_CLAMP: 2312 case TCP_WINDOW_CLAMP:
2298 val = tp->window_clamp; 2313 val = tp->window_clamp;
@@ -2609,7 +2624,7 @@ __setup("thash_entries=", set_thash_entries);
2609void __init tcp_init(void) 2624void __init tcp_init(void)
2610{ 2625{
2611 struct sk_buff *skb = NULL; 2626 struct sk_buff *skb = NULL;
2612 unsigned long limit; 2627 unsigned long nr_pages, limit;
2613 int order, i, max_share; 2628 int order, i, max_share;
2614 2629
2615 BUILD_BUG_ON(sizeof(struct tcp_skb_cb) > sizeof(skb->cb)); 2630 BUILD_BUG_ON(sizeof(struct tcp_skb_cb) > sizeof(skb->cb));
@@ -2678,8 +2693,9 @@ void __init tcp_init(void)
2678 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of 2693 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of
2679 * memory, with a floor of 128 pages. 2694 * memory, with a floor of 128 pages.
2680 */ 2695 */
2681 limit = min(nr_all_pages, 1UL<<(28-PAGE_SHIFT)) >> (20-PAGE_SHIFT); 2696 nr_pages = totalram_pages - totalhigh_pages;
2682 limit = (limit * (nr_all_pages >> (20-PAGE_SHIFT))) >> (PAGE_SHIFT-11); 2697 limit = min(nr_pages, 1UL<<(28-PAGE_SHIFT)) >> (20-PAGE_SHIFT);
2698 limit = (limit * (nr_pages >> (20-PAGE_SHIFT))) >> (PAGE_SHIFT-11);
2683 limit = max(limit, 128UL); 2699 limit = max(limit, 128UL);
2684 sysctl_tcp_mem[0] = limit / 4 * 3; 2700 sysctl_tcp_mem[0] = limit / 4 * 3;
2685 sysctl_tcp_mem[1] = limit; 2701 sysctl_tcp_mem[1] = limit;
diff --git a/net/ipv4/tcp_hybla.c b/net/ipv4/tcp_hybla.c
index 44618b675916..bfcbd148a89d 100644
--- a/net/ipv4/tcp_hybla.c
+++ b/net/ipv4/tcp_hybla.c
@@ -101,8 +101,10 @@ static void hybla_cong_avoid(struct sock *sk, u32 ack, u32 in_flight)
101 if (!tcp_is_cwnd_limited(sk, in_flight)) 101 if (!tcp_is_cwnd_limited(sk, in_flight))
102 return; 102 return;
103 103
104 if (!ca->hybla_en) 104 if (!ca->hybla_en) {
105 return tcp_reno_cong_avoid(sk, ack, in_flight); 105 tcp_reno_cong_avoid(sk, ack, in_flight);
106 return;
107 }
106 108
107 if (ca->rho == 0) 109 if (ca->rho == 0)
108 hybla_recalc_param(sk); 110 hybla_recalc_param(sk);
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 0298f80681f2..cad73b7dfef0 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -66,6 +66,7 @@
66#include <linux/mm.h> 66#include <linux/mm.h>
67#include <linux/module.h> 67#include <linux/module.h>
68#include <linux/sysctl.h> 68#include <linux/sysctl.h>
69#include <net/dst.h>
69#include <net/tcp.h> 70#include <net/tcp.h>
70#include <net/inet_common.h> 71#include <net/inet_common.h>
71#include <linux/ipsec.h> 72#include <linux/ipsec.h>
@@ -113,8 +114,6 @@ int sysctl_tcp_abc __read_mostly;
113#define FLAG_FORWARD_PROGRESS (FLAG_ACKED|FLAG_DATA_SACKED) 114#define FLAG_FORWARD_PROGRESS (FLAG_ACKED|FLAG_DATA_SACKED)
114#define FLAG_ANY_PROGRESS (FLAG_FORWARD_PROGRESS|FLAG_SND_UNA_ADVANCED) 115#define FLAG_ANY_PROGRESS (FLAG_FORWARD_PROGRESS|FLAG_SND_UNA_ADVANCED)
115 116
116#define IsSackFrto() (sysctl_tcp_frto == 0x2)
117
118#define TCP_REMNANT (TCP_FLAG_FIN|TCP_FLAG_URG|TCP_FLAG_SYN|TCP_FLAG_PSH) 117#define TCP_REMNANT (TCP_FLAG_FIN|TCP_FLAG_URG|TCP_FLAG_SYN|TCP_FLAG_PSH)
119#define TCP_HP_BITS (~(TCP_RESERVED_BITS|TCP_FLAG_PSH)) 118#define TCP_HP_BITS (~(TCP_RESERVED_BITS|TCP_FLAG_PSH))
120 119
@@ -605,7 +604,7 @@ static u32 tcp_rto_min(struct sock *sk)
605 u32 rto_min = TCP_RTO_MIN; 604 u32 rto_min = TCP_RTO_MIN;
606 605
607 if (dst && dst_metric_locked(dst, RTAX_RTO_MIN)) 606 if (dst && dst_metric_locked(dst, RTAX_RTO_MIN))
608 rto_min = dst->metrics[RTAX_RTO_MIN - 1]; 607 rto_min = dst_metric(dst, RTAX_RTO_MIN);
609 return rto_min; 608 return rto_min;
610} 609}
611 610
@@ -769,7 +768,7 @@ void tcp_update_metrics(struct sock *sk)
769 dst->metrics[RTAX_RTTVAR - 1] = m; 768 dst->metrics[RTAX_RTTVAR - 1] = m;
770 else 769 else
771 dst->metrics[RTAX_RTTVAR-1] -= 770 dst->metrics[RTAX_RTTVAR-1] -=
772 (dst->metrics[RTAX_RTTVAR-1] - m)>>2; 771 (dst_metric(dst, RTAX_RTTVAR) - m)>>2;
773 } 772 }
774 773
775 if (tp->snd_ssthresh >= 0xFFFF) { 774 if (tp->snd_ssthresh >= 0xFFFF) {
@@ -788,21 +787,21 @@ void tcp_update_metrics(struct sock *sk)
788 dst->metrics[RTAX_SSTHRESH-1] = 787 dst->metrics[RTAX_SSTHRESH-1] =
789 max(tp->snd_cwnd >> 1, tp->snd_ssthresh); 788 max(tp->snd_cwnd >> 1, tp->snd_ssthresh);
790 if (!dst_metric_locked(dst, RTAX_CWND)) 789 if (!dst_metric_locked(dst, RTAX_CWND))
791 dst->metrics[RTAX_CWND-1] = (dst->metrics[RTAX_CWND-1] + tp->snd_cwnd) >> 1; 790 dst->metrics[RTAX_CWND-1] = (dst_metric(dst, RTAX_CWND) + tp->snd_cwnd) >> 1;
792 } else { 791 } else {
793 /* Else slow start did not finish, cwnd is non-sense, 792 /* Else slow start did not finish, cwnd is non-sense,
794 ssthresh may be also invalid. 793 ssthresh may be also invalid.
795 */ 794 */
796 if (!dst_metric_locked(dst, RTAX_CWND)) 795 if (!dst_metric_locked(dst, RTAX_CWND))
797 dst->metrics[RTAX_CWND-1] = (dst->metrics[RTAX_CWND-1] + tp->snd_ssthresh) >> 1; 796 dst->metrics[RTAX_CWND-1] = (dst_metric(dst, RTAX_CWND) + tp->snd_ssthresh) >> 1;
798 if (dst->metrics[RTAX_SSTHRESH-1] && 797 if (dst_metric(dst, RTAX_SSTHRESH) &&
799 !dst_metric_locked(dst, RTAX_SSTHRESH) && 798 !dst_metric_locked(dst, RTAX_SSTHRESH) &&
800 tp->snd_ssthresh > dst->metrics[RTAX_SSTHRESH-1]) 799 tp->snd_ssthresh > dst_metric(dst, RTAX_SSTHRESH))
801 dst->metrics[RTAX_SSTHRESH-1] = tp->snd_ssthresh; 800 dst->metrics[RTAX_SSTHRESH-1] = tp->snd_ssthresh;
802 } 801 }
803 802
804 if (!dst_metric_locked(dst, RTAX_REORDERING)) { 803 if (!dst_metric_locked(dst, RTAX_REORDERING)) {
805 if (dst->metrics[RTAX_REORDERING-1] < tp->reordering && 804 if (dst_metric(dst, RTAX_REORDERING) < tp->reordering &&
806 tp->reordering != sysctl_tcp_reordering) 805 tp->reordering != sysctl_tcp_reordering)
807 dst->metrics[RTAX_REORDERING-1] = tp->reordering; 806 dst->metrics[RTAX_REORDERING-1] = tp->reordering;
808 } 807 }
@@ -1172,8 +1171,8 @@ static int tcp_check_dsack(struct tcp_sock *tp, struct sk_buff *ack_skb,
1172 struct tcp_sack_block_wire *sp, int num_sacks, 1171 struct tcp_sack_block_wire *sp, int num_sacks,
1173 u32 prior_snd_una) 1172 u32 prior_snd_una)
1174{ 1173{
1175 u32 start_seq_0 = ntohl(get_unaligned(&sp[0].start_seq)); 1174 u32 start_seq_0 = get_unaligned_be32(&sp[0].start_seq);
1176 u32 end_seq_0 = ntohl(get_unaligned(&sp[0].end_seq)); 1175 u32 end_seq_0 = get_unaligned_be32(&sp[0].end_seq);
1177 int dup_sack = 0; 1176 int dup_sack = 0;
1178 1177
1179 if (before(start_seq_0, TCP_SKB_CB(ack_skb)->ack_seq)) { 1178 if (before(start_seq_0, TCP_SKB_CB(ack_skb)->ack_seq)) {
@@ -1181,8 +1180,8 @@ static int tcp_check_dsack(struct tcp_sock *tp, struct sk_buff *ack_skb,
1181 tcp_dsack_seen(tp); 1180 tcp_dsack_seen(tp);
1182 NET_INC_STATS_BH(LINUX_MIB_TCPDSACKRECV); 1181 NET_INC_STATS_BH(LINUX_MIB_TCPDSACKRECV);
1183 } else if (num_sacks > 1) { 1182 } else if (num_sacks > 1) {
1184 u32 end_seq_1 = ntohl(get_unaligned(&sp[1].end_seq)); 1183 u32 end_seq_1 = get_unaligned_be32(&sp[1].end_seq);
1185 u32 start_seq_1 = ntohl(get_unaligned(&sp[1].start_seq)); 1184 u32 start_seq_1 = get_unaligned_be32(&sp[1].start_seq);
1186 1185
1187 if (!after(end_seq_0, end_seq_1) && 1186 if (!after(end_seq_0, end_seq_1) &&
1188 !before(start_seq_0, start_seq_1)) { 1187 !before(start_seq_0, start_seq_1)) {
@@ -1393,9 +1392,9 @@ static struct sk_buff *tcp_maybe_skipping_dsack(struct sk_buff *skb,
1393 1392
1394 if (before(next_dup->start_seq, skip_to_seq)) { 1393 if (before(next_dup->start_seq, skip_to_seq)) {
1395 skb = tcp_sacktag_skip(skb, sk, next_dup->start_seq, fack_count); 1394 skb = tcp_sacktag_skip(skb, sk, next_dup->start_seq, fack_count);
1396 tcp_sacktag_walk(skb, sk, NULL, 1395 skb = tcp_sacktag_walk(skb, sk, NULL,
1397 next_dup->start_seq, next_dup->end_seq, 1396 next_dup->start_seq, next_dup->end_seq,
1398 1, fack_count, reord, flag); 1397 1, fack_count, reord, flag);
1399 } 1398 }
1400 1399
1401 return skb; 1400 return skb;
@@ -1453,8 +1452,8 @@ tcp_sacktag_write_queue(struct sock *sk, struct sk_buff *ack_skb,
1453 for (i = 0; i < num_sacks; i++) { 1452 for (i = 0; i < num_sacks; i++) {
1454 int dup_sack = !i && found_dup_sack; 1453 int dup_sack = !i && found_dup_sack;
1455 1454
1456 sp[used_sacks].start_seq = ntohl(get_unaligned(&sp_wire[i].start_seq)); 1455 sp[used_sacks].start_seq = get_unaligned_be32(&sp_wire[i].start_seq);
1457 sp[used_sacks].end_seq = ntohl(get_unaligned(&sp_wire[i].end_seq)); 1456 sp[used_sacks].end_seq = get_unaligned_be32(&sp_wire[i].end_seq);
1458 1457
1459 if (!tcp_is_sackblock_valid(tp, dup_sack, 1458 if (!tcp_is_sackblock_valid(tp, dup_sack,
1460 sp[used_sacks].start_seq, 1459 sp[used_sacks].start_seq,
@@ -1685,6 +1684,11 @@ static inline void tcp_reset_reno_sack(struct tcp_sock *tp)
1685 tp->sacked_out = 0; 1684 tp->sacked_out = 0;
1686} 1685}
1687 1686
1687static int tcp_is_sackfrto(const struct tcp_sock *tp)
1688{
1689 return (sysctl_tcp_frto == 0x2) && !tcp_is_reno(tp);
1690}
1691
1688/* F-RTO can only be used if TCP has never retransmitted anything other than 1692/* F-RTO can only be used if TCP has never retransmitted anything other than
1689 * head (SACK enhanced variant from Appendix B of RFC4138 is more robust here) 1693 * head (SACK enhanced variant from Appendix B of RFC4138 is more robust here)
1690 */ 1694 */
@@ -1701,7 +1705,7 @@ int tcp_use_frto(struct sock *sk)
1701 if (icsk->icsk_mtup.probe_size) 1705 if (icsk->icsk_mtup.probe_size)
1702 return 0; 1706 return 0;
1703 1707
1704 if (IsSackFrto()) 1708 if (tcp_is_sackfrto(tp))
1705 return 1; 1709 return 1;
1706 1710
1707 /* Avoid expensive walking of rexmit queue if possible */ 1711 /* Avoid expensive walking of rexmit queue if possible */
@@ -1791,7 +1795,7 @@ void tcp_enter_frto(struct sock *sk)
1791 /* Earlier loss recovery underway (see RFC4138; Appendix B). 1795 /* Earlier loss recovery underway (see RFC4138; Appendix B).
1792 * The last condition is necessary at least in tp->frto_counter case. 1796 * The last condition is necessary at least in tp->frto_counter case.
1793 */ 1797 */
1794 if (IsSackFrto() && (tp->frto_counter || 1798 if (tcp_is_sackfrto(tp) && (tp->frto_counter ||
1795 ((1 << icsk->icsk_ca_state) & (TCPF_CA_Recovery|TCPF_CA_Loss))) && 1799 ((1 << icsk->icsk_ca_state) & (TCPF_CA_Recovery|TCPF_CA_Loss))) &&
1796 after(tp->high_seq, tp->snd_una)) { 1800 after(tp->high_seq, tp->snd_una)) {
1797 tp->frto_highmark = tp->high_seq; 1801 tp->frto_highmark = tp->high_seq;
@@ -1838,9 +1842,16 @@ static void tcp_enter_frto_loss(struct sock *sk, int allowed_segments, int flag)
1838 TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_RETRANS; 1842 TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_RETRANS;
1839 } 1843 }
1840 1844
1841 /* Don't lost mark skbs that were fwd transmitted after RTO */ 1845 /* Marking forward transmissions that were made after RTO lost
1842 if (!(TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED) && 1846 * can cause unnecessary retransmissions in some scenarios,
1843 !after(TCP_SKB_CB(skb)->end_seq, tp->frto_highmark)) { 1847 * SACK blocks will mitigate that in some but not in all cases.
1848 * We used to not mark them but it was causing break-ups with
1849 * receivers that do only in-order receival.
1850 *
1851 * TODO: we could detect presence of such receiver and select
1852 * different behavior per flow.
1853 */
1854 if (!(TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED)) {
1844 TCP_SKB_CB(skb)->sacked |= TCPCB_LOST; 1855 TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
1845 tp->lost_out += tcp_skb_pcount(skb); 1856 tp->lost_out += tcp_skb_pcount(skb);
1846 } 1857 }
@@ -1856,7 +1867,7 @@ static void tcp_enter_frto_loss(struct sock *sk, int allowed_segments, int flag)
1856 tp->reordering = min_t(unsigned int, tp->reordering, 1867 tp->reordering = min_t(unsigned int, tp->reordering,
1857 sysctl_tcp_reordering); 1868 sysctl_tcp_reordering);
1858 tcp_set_ca_state(sk, TCP_CA_Loss); 1869 tcp_set_ca_state(sk, TCP_CA_Loss);
1859 tp->high_seq = tp->frto_highmark; 1870 tp->high_seq = tp->snd_nxt;
1860 TCP_ECN_queue_cwr(tp); 1871 TCP_ECN_queue_cwr(tp);
1861 1872
1862 tcp_clear_retrans_hints_partial(tp); 1873 tcp_clear_retrans_hints_partial(tp);
@@ -2472,28 +2483,34 @@ static inline void tcp_complete_cwr(struct sock *sk)
2472 tcp_ca_event(sk, CA_EVENT_COMPLETE_CWR); 2483 tcp_ca_event(sk, CA_EVENT_COMPLETE_CWR);
2473} 2484}
2474 2485
2486static void tcp_try_keep_open(struct sock *sk)
2487{
2488 struct tcp_sock *tp = tcp_sk(sk);
2489 int state = TCP_CA_Open;
2490
2491 if (tcp_left_out(tp) || tp->retrans_out || tp->undo_marker)
2492 state = TCP_CA_Disorder;
2493
2494 if (inet_csk(sk)->icsk_ca_state != state) {
2495 tcp_set_ca_state(sk, state);
2496 tp->high_seq = tp->snd_nxt;
2497 }
2498}
2499
2475static void tcp_try_to_open(struct sock *sk, int flag) 2500static void tcp_try_to_open(struct sock *sk, int flag)
2476{ 2501{
2477 struct tcp_sock *tp = tcp_sk(sk); 2502 struct tcp_sock *tp = tcp_sk(sk);
2478 2503
2479 tcp_verify_left_out(tp); 2504 tcp_verify_left_out(tp);
2480 2505
2481 if (tp->retrans_out == 0) 2506 if (!tp->frto_counter && tp->retrans_out == 0)
2482 tp->retrans_stamp = 0; 2507 tp->retrans_stamp = 0;
2483 2508
2484 if (flag & FLAG_ECE) 2509 if (flag & FLAG_ECE)
2485 tcp_enter_cwr(sk, 1); 2510 tcp_enter_cwr(sk, 1);
2486 2511
2487 if (inet_csk(sk)->icsk_ca_state != TCP_CA_CWR) { 2512 if (inet_csk(sk)->icsk_ca_state != TCP_CA_CWR) {
2488 int state = TCP_CA_Open; 2513 tcp_try_keep_open(sk);
2489
2490 if (tcp_left_out(tp) || tp->retrans_out || tp->undo_marker)
2491 state = TCP_CA_Disorder;
2492
2493 if (inet_csk(sk)->icsk_ca_state != state) {
2494 tcp_set_ca_state(sk, state);
2495 tp->high_seq = tp->snd_nxt;
2496 }
2497 tcp_moderate_cwnd(tp); 2514 tcp_moderate_cwnd(tp);
2498 } else { 2515 } else {
2499 tcp_cwnd_down(sk, flag); 2516 tcp_cwnd_down(sk, flag);
@@ -3123,7 +3140,7 @@ static int tcp_process_frto(struct sock *sk, int flag)
3123 return 1; 3140 return 1;
3124 } 3141 }
3125 3142
3126 if (!IsSackFrto() || tcp_is_reno(tp)) { 3143 if (!tcp_is_sackfrto(tp)) {
3127 /* RFC4138 shortcoming in step 2; should also have case c): 3144 /* RFC4138 shortcoming in step 2; should also have case c):
3128 * ACK isn't duplicate nor advances window, e.g., opposite dir 3145 * ACK isn't duplicate nor advances window, e.g., opposite dir
3129 * data, winupdate 3146 * data, winupdate
@@ -3299,8 +3316,11 @@ no_queue:
3299 return 1; 3316 return 1;
3300 3317
3301old_ack: 3318old_ack:
3302 if (TCP_SKB_CB(skb)->sacked) 3319 if (TCP_SKB_CB(skb)->sacked) {
3303 tcp_sacktag_write_queue(sk, skb, prior_snd_una); 3320 tcp_sacktag_write_queue(sk, skb, prior_snd_una);
3321 if (icsk->icsk_ca_state == TCP_CA_Open)
3322 tcp_try_keep_open(sk);
3323 }
3304 3324
3305uninteresting_ack: 3325uninteresting_ack:
3306 SOCK_DEBUG(sk, "Ack %u out of %u:%u\n", ack, tp->snd_una, tp->snd_nxt); 3326 SOCK_DEBUG(sk, "Ack %u out of %u:%u\n", ack, tp->snd_una, tp->snd_nxt);
@@ -3340,7 +3360,7 @@ void tcp_parse_options(struct sk_buff *skb, struct tcp_options_received *opt_rx,
3340 switch (opcode) { 3360 switch (opcode) {
3341 case TCPOPT_MSS: 3361 case TCPOPT_MSS:
3342 if (opsize == TCPOLEN_MSS && th->syn && !estab) { 3362 if (opsize == TCPOLEN_MSS && th->syn && !estab) {
3343 u16 in_mss = ntohs(get_unaligned((__be16 *)ptr)); 3363 u16 in_mss = get_unaligned_be16(ptr);
3344 if (in_mss) { 3364 if (in_mss) {
3345 if (opt_rx->user_mss && 3365 if (opt_rx->user_mss &&
3346 opt_rx->user_mss < in_mss) 3366 opt_rx->user_mss < in_mss)
@@ -3369,8 +3389,8 @@ void tcp_parse_options(struct sk_buff *skb, struct tcp_options_received *opt_rx,
3369 ((estab && opt_rx->tstamp_ok) || 3389 ((estab && opt_rx->tstamp_ok) ||
3370 (!estab && sysctl_tcp_timestamps))) { 3390 (!estab && sysctl_tcp_timestamps))) {
3371 opt_rx->saw_tstamp = 1; 3391 opt_rx->saw_tstamp = 1;
3372 opt_rx->rcv_tsval = ntohl(get_unaligned((__be32 *)ptr)); 3392 opt_rx->rcv_tsval = get_unaligned_be32(ptr);
3373 opt_rx->rcv_tsecr = ntohl(get_unaligned((__be32 *)(ptr+4))); 3393 opt_rx->rcv_tsecr = get_unaligned_be32(ptr + 4);
3374 } 3394 }
3375 break; 3395 break;
3376 case TCPOPT_SACK_PERM: 3396 case TCPOPT_SACK_PERM:
@@ -4521,49 +4541,6 @@ static void tcp_urg(struct sock *sk, struct sk_buff *skb, struct tcphdr *th)
4521 } 4541 }
4522} 4542}
4523 4543
4524static int tcp_defer_accept_check(struct sock *sk)
4525{
4526 struct tcp_sock *tp = tcp_sk(sk);
4527
4528 if (tp->defer_tcp_accept.request) {
4529 int queued_data = tp->rcv_nxt - tp->copied_seq;
4530 int hasfin = !skb_queue_empty(&sk->sk_receive_queue) ?
4531 tcp_hdr((struct sk_buff *)
4532 sk->sk_receive_queue.prev)->fin : 0;
4533
4534 if (queued_data && hasfin)
4535 queued_data--;
4536
4537 if (queued_data &&
4538 tp->defer_tcp_accept.listen_sk->sk_state == TCP_LISTEN) {
4539 if (sock_flag(sk, SOCK_KEEPOPEN)) {
4540 inet_csk_reset_keepalive_timer(sk,
4541 keepalive_time_when(tp));
4542 } else {
4543 inet_csk_delete_keepalive_timer(sk);
4544 }
4545
4546 inet_csk_reqsk_queue_add(
4547 tp->defer_tcp_accept.listen_sk,
4548 tp->defer_tcp_accept.request,
4549 sk);
4550
4551 tp->defer_tcp_accept.listen_sk->sk_data_ready(
4552 tp->defer_tcp_accept.listen_sk, 0);
4553
4554 sock_put(tp->defer_tcp_accept.listen_sk);
4555 sock_put(sk);
4556 tp->defer_tcp_accept.listen_sk = NULL;
4557 tp->defer_tcp_accept.request = NULL;
4558 } else if (hasfin ||
4559 tp->defer_tcp_accept.listen_sk->sk_state != TCP_LISTEN) {
4560 tcp_reset(sk);
4561 return -1;
4562 }
4563 }
4564 return 0;
4565}
4566
4567static int tcp_copy_to_iovec(struct sock *sk, struct sk_buff *skb, int hlen) 4544static int tcp_copy_to_iovec(struct sock *sk, struct sk_buff *skb, int hlen)
4568{ 4545{
4569 struct tcp_sock *tp = tcp_sk(sk); 4546 struct tcp_sock *tp = tcp_sk(sk);
@@ -4924,8 +4901,6 @@ step5:
4924 4901
4925 tcp_data_snd_check(sk); 4902 tcp_data_snd_check(sk);
4926 tcp_ack_snd_check(sk); 4903 tcp_ack_snd_check(sk);
4927
4928 tcp_defer_accept_check(sk);
4929 return 0; 4904 return 0;
4930 4905
4931csum_error: 4906csum_error:
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 0e9bc120707d..ffe869ac1bcf 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -85,10 +85,6 @@
85int sysctl_tcp_tw_reuse __read_mostly; 85int sysctl_tcp_tw_reuse __read_mostly;
86int sysctl_tcp_low_latency __read_mostly; 86int sysctl_tcp_low_latency __read_mostly;
87 87
88/* Check TCP sequence numbers in ICMP packets. */
89#define ICMP_MIN_LENGTH 8
90
91void tcp_v4_send_check(struct sock *sk, int len, struct sk_buff *skb);
92 88
93#ifdef CONFIG_TCP_MD5SIG 89#ifdef CONFIG_TCP_MD5SIG
94static struct tcp_md5sig_key *tcp_v4_md5_do_lookup(struct sock *sk, 90static struct tcp_md5sig_key *tcp_v4_md5_do_lookup(struct sock *sk,
@@ -1285,7 +1281,7 @@ int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
1285 if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1) 1281 if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
1286 goto drop; 1282 goto drop;
1287 1283
1288 req = reqsk_alloc(&tcp_request_sock_ops); 1284 req = inet_reqsk_alloc(&tcp_request_sock_ops);
1289 if (!req) 1285 if (!req)
1290 goto drop; 1286 goto drop;
1291 1287
@@ -1918,14 +1914,6 @@ int tcp_v4_destroy_sock(struct sock *sk)
1918 sk->sk_sndmsg_page = NULL; 1914 sk->sk_sndmsg_page = NULL;
1919 } 1915 }
1920 1916
1921 if (tp->defer_tcp_accept.request) {
1922 reqsk_free(tp->defer_tcp_accept.request);
1923 sock_put(tp->defer_tcp_accept.listen_sk);
1924 sock_put(sk);
1925 tp->defer_tcp_accept.listen_sk = NULL;
1926 tp->defer_tcp_accept.request = NULL;
1927 }
1928
1929 atomic_dec(&tcp_sockets_allocated); 1917 atomic_dec(&tcp_sockets_allocated);
1930 1918
1931 return 0; 1919 return 0;
@@ -2214,9 +2202,6 @@ static int tcp_seq_open(struct inode *inode, struct file *file)
2214 struct tcp_iter_state *s; 2202 struct tcp_iter_state *s;
2215 int err; 2203 int err;
2216 2204
2217 if (unlikely(afinfo == NULL))
2218 return -EINVAL;
2219
2220 err = seq_open_net(inode, file, &afinfo->seq_ops, 2205 err = seq_open_net(inode, file, &afinfo->seq_ops,
2221 sizeof(struct tcp_iter_state)); 2206 sizeof(struct tcp_iter_state));
2222 if (err < 0) 2207 if (err < 0)
@@ -2241,10 +2226,9 @@ int tcp_proc_register(struct net *net, struct tcp_seq_afinfo *afinfo)
2241 afinfo->seq_ops.next = tcp_seq_next; 2226 afinfo->seq_ops.next = tcp_seq_next;
2242 afinfo->seq_ops.stop = tcp_seq_stop; 2227 afinfo->seq_ops.stop = tcp_seq_stop;
2243 2228
2244 p = proc_net_fops_create(net, afinfo->name, S_IRUGO, &afinfo->seq_fops); 2229 p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
2245 if (p) 2230 &afinfo->seq_fops, afinfo);
2246 p->data = afinfo; 2231 if (!p)
2247 else
2248 rc = -ENOMEM; 2232 rc = -ENOMEM;
2249 return rc; 2233 return rc;
2250} 2234}
@@ -2307,7 +2291,7 @@ static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i, int *len)
2307 } 2291 }
2308 2292
2309 seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX " 2293 seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX "
2310 "%08X %5d %8d %lu %d %p %u %u %u %u %d%n", 2294 "%08X %5d %8d %lu %d %p %lu %lu %u %u %d%n",
2311 i, src, srcp, dest, destp, sk->sk_state, 2295 i, src, srcp, dest, destp, sk->sk_state,
2312 tp->write_seq - tp->snd_una, 2296 tp->write_seq - tp->snd_una,
2313 sk->sk_state == TCP_LISTEN ? sk->sk_ack_backlog : 2297 sk->sk_state == TCP_LISTEN ? sk->sk_ack_backlog :
@@ -2319,8 +2303,8 @@ static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i, int *len)
2319 icsk->icsk_probes_out, 2303 icsk->icsk_probes_out,
2320 sock_i_ino(sk), 2304 sock_i_ino(sk),
2321 atomic_read(&sk->sk_refcnt), sk, 2305 atomic_read(&sk->sk_refcnt), sk,
2322 icsk->icsk_rto, 2306 jiffies_to_clock_t(icsk->icsk_rto),
2323 icsk->icsk_ack.ato, 2307 jiffies_to_clock_t(icsk->icsk_ack.ato),
2324 (icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong, 2308 (icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong,
2325 tp->snd_cwnd, 2309 tp->snd_cwnd,
2326 tp->snd_ssthresh >= 0xFFFF ? -1 : tp->snd_ssthresh, 2310 tp->snd_ssthresh >= 0xFFFF ? -1 : tp->snd_ssthresh,
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index 019c8c16e5cc..8245247a6ceb 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -571,8 +571,10 @@ struct sock *tcp_check_req(struct sock *sk,struct sk_buff *skb,
571 does sequence test, SYN is truncated, and thus we consider 571 does sequence test, SYN is truncated, and thus we consider
572 it a bare ACK. 572 it a bare ACK.
573 573
574 Both ends (listening sockets) accept the new incoming 574 If icsk->icsk_accept_queue.rskq_defer_accept, we silently drop this
575 connection and try to talk to each other. 8-) 575 bare ACK. Otherwise, we create an established connection. Both
576 ends (listening sockets) accept the new incoming connection and try
577 to talk to each other. 8-)
576 578
577 Note: This case is both harmless, and rare. Possibility is about the 579 Note: This case is both harmless, and rare. Possibility is about the
578 same as us discovering intelligent life on another plant tomorrow. 580 same as us discovering intelligent life on another plant tomorrow.
@@ -640,6 +642,13 @@ struct sock *tcp_check_req(struct sock *sk,struct sk_buff *skb,
640 if (!(flg & TCP_FLAG_ACK)) 642 if (!(flg & TCP_FLAG_ACK))
641 return NULL; 643 return NULL;
642 644
645 /* If TCP_DEFER_ACCEPT is set, drop bare ACK. */
646 if (inet_csk(sk)->icsk_accept_queue.rskq_defer_accept &&
647 TCP_SKB_CB(skb)->end_seq == tcp_rsk(req)->rcv_isn + 1) {
648 inet_rsk(req)->acked = 1;
649 return NULL;
650 }
651
643 /* OK, ACK is valid, create big socket and 652 /* OK, ACK is valid, create big socket and
644 * feed this segment to it. It will repeat all 653 * feed this segment to it. It will repeat all
645 * the tests. THIS SEGMENT MUST MOVE SOCKET TO 654 * the tests. THIS SEGMENT MUST MOVE SOCKET TO
@@ -678,24 +687,7 @@ struct sock *tcp_check_req(struct sock *sk,struct sk_buff *skb,
678 inet_csk_reqsk_queue_unlink(sk, req, prev); 687 inet_csk_reqsk_queue_unlink(sk, req, prev);
679 inet_csk_reqsk_queue_removed(sk, req); 688 inet_csk_reqsk_queue_removed(sk, req);
680 689
681 if (inet_csk(sk)->icsk_accept_queue.rskq_defer_accept && 690 inet_csk_reqsk_queue_add(sk, req, child);
682 TCP_SKB_CB(skb)->end_seq == tcp_rsk(req)->rcv_isn + 1) {
683
684 /* the accept queue handling is done is est recv slow
685 * path so lets make sure to start there
686 */
687 tcp_sk(child)->pred_flags = 0;
688 sock_hold(sk);
689 sock_hold(child);
690 tcp_sk(child)->defer_tcp_accept.listen_sk = sk;
691 tcp_sk(child)->defer_tcp_accept.request = req;
692
693 inet_csk_reset_keepalive_timer(child,
694 inet_csk(sk)->icsk_accept_queue.rskq_defer_accept * HZ);
695 } else {
696 inet_csk_reqsk_queue_add(sk, req, child);
697 }
698
699 return child; 691 return child;
700 692
701 listen_overflow: 693 listen_overflow:
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index debf23581606..ad993ecb4810 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -1836,7 +1836,7 @@ int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
1836{ 1836{
1837 struct tcp_sock *tp = tcp_sk(sk); 1837 struct tcp_sock *tp = tcp_sk(sk);
1838 struct inet_connection_sock *icsk = inet_csk(sk); 1838 struct inet_connection_sock *icsk = inet_csk(sk);
1839 unsigned int cur_mss = tcp_current_mss(sk, 0); 1839 unsigned int cur_mss;
1840 int err; 1840 int err;
1841 1841
1842 /* Inconslusive MTU probe */ 1842 /* Inconslusive MTU probe */
@@ -1858,6 +1858,11 @@ int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
1858 return -ENOMEM; 1858 return -ENOMEM;
1859 } 1859 }
1860 1860
1861 if (inet_csk(sk)->icsk_af_ops->rebuild_header(sk))
1862 return -EHOSTUNREACH; /* Routing failure or similar. */
1863
1864 cur_mss = tcp_current_mss(sk, 0);
1865
1861 /* If receiver has shrunk his window, and skb is out of 1866 /* If receiver has shrunk his window, and skb is out of
1862 * new window, do not retransmit it. The exception is the 1867 * new window, do not retransmit it. The exception is the
1863 * case, when window is shrunk to zero. In this case 1868 * case, when window is shrunk to zero. In this case
@@ -1884,9 +1889,6 @@ int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
1884 (sysctl_tcp_retrans_collapse != 0)) 1889 (sysctl_tcp_retrans_collapse != 0))
1885 tcp_retrans_try_collapse(sk, skb, cur_mss); 1890 tcp_retrans_try_collapse(sk, skb, cur_mss);
1886 1891
1887 if (inet_csk(sk)->icsk_af_ops->rebuild_header(sk))
1888 return -EHOSTUNREACH; /* Routing failure or similar. */
1889
1890 /* Some Solaris stacks overoptimize and ignore the FIN on a 1892 /* Some Solaris stacks overoptimize and ignore the FIN on a
1891 * retransmit when old data is attached. So strip it off 1893 * retransmit when old data is attached. So strip it off
1892 * since it is cheap to do so and saves bytes on the network. 1894 * since it is cheap to do so and saves bytes on the network.
@@ -2129,6 +2131,8 @@ void tcp_send_active_reset(struct sock *sk, gfp_t priority)
2129 TCP_SKB_CB(skb)->when = tcp_time_stamp; 2131 TCP_SKB_CB(skb)->when = tcp_time_stamp;
2130 if (tcp_transmit_skb(sk, skb, 0, priority)) 2132 if (tcp_transmit_skb(sk, skb, 0, priority))
2131 NET_INC_STATS(LINUX_MIB_TCPABORTFAILED); 2133 NET_INC_STATS(LINUX_MIB_TCPABORTFAILED);
2134
2135 TCP_INC_STATS(TCP_MIB_OUTRSTS);
2132} 2136}
2133 2137
2134/* WARNING: This routine must only be called when we have already sent 2138/* WARNING: This routine must only be called when we have already sent
diff --git a/net/ipv4/tcp_probe.c b/net/ipv4/tcp_probe.c
index 5ff0ce6e9d39..7ddc30f0744f 100644
--- a/net/ipv4/tcp_probe.c
+++ b/net/ipv4/tcp_probe.c
@@ -224,7 +224,7 @@ static __init int tcpprobe_init(void)
224 if (bufsize < 0) 224 if (bufsize < 0)
225 return -EINVAL; 225 return -EINVAL;
226 226
227 tcp_probe.log = kcalloc(sizeof(struct tcp_log), bufsize, GFP_KERNEL); 227 tcp_probe.log = kcalloc(bufsize, sizeof(struct tcp_log), GFP_KERNEL);
228 if (!tcp_probe.log) 228 if (!tcp_probe.log)
229 goto err0; 229 goto err0;
230 230
diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
index 4de68cf5f2aa..63ed9d6830e7 100644
--- a/net/ipv4/tcp_timer.c
+++ b/net/ipv4/tcp_timer.c
@@ -489,11 +489,6 @@ static void tcp_keepalive_timer (unsigned long data)
489 goto death; 489 goto death;
490 } 490 }
491 491
492 if (tp->defer_tcp_accept.request && sk->sk_state == TCP_ESTABLISHED) {
493 tcp_send_active_reset(sk, GFP_ATOMIC);
494 goto death;
495 }
496
497 if (!sock_flag(sk, SOCK_KEEPOPEN) || sk->sk_state == TCP_CLOSE) 492 if (!sock_flag(sk, SOCK_KEEPOPEN) || sk->sk_state == TCP_CLOSE)
498 goto out; 493 goto out;
499 494
diff --git a/net/ipv4/tcp_vegas.c b/net/ipv4/tcp_vegas.c
index 0e1a8c91f78e..14504dada116 100644
--- a/net/ipv4/tcp_vegas.c
+++ b/net/ipv4/tcp_vegas.c
@@ -167,8 +167,10 @@ static void tcp_vegas_cong_avoid(struct sock *sk, u32 ack, u32 in_flight)
167 struct tcp_sock *tp = tcp_sk(sk); 167 struct tcp_sock *tp = tcp_sk(sk);
168 struct vegas *vegas = inet_csk_ca(sk); 168 struct vegas *vegas = inet_csk_ca(sk);
169 169
170 if (!vegas->doing_vegas_now) 170 if (!vegas->doing_vegas_now) {
171 return tcp_reno_cong_avoid(sk, ack, in_flight); 171 tcp_reno_cong_avoid(sk, ack, in_flight);
172 return;
173 }
172 174
173 /* The key players are v_beg_snd_una and v_beg_snd_nxt. 175 /* The key players are v_beg_snd_una and v_beg_snd_nxt.
174 * 176 *
diff --git a/net/ipv4/tcp_veno.c b/net/ipv4/tcp_veno.c
index 2bf618a3b00b..d08b2e855c22 100644
--- a/net/ipv4/tcp_veno.c
+++ b/net/ipv4/tcp_veno.c
@@ -119,8 +119,10 @@ static void tcp_veno_cong_avoid(struct sock *sk, u32 ack, u32 in_flight)
119 struct tcp_sock *tp = tcp_sk(sk); 119 struct tcp_sock *tp = tcp_sk(sk);
120 struct veno *veno = inet_csk_ca(sk); 120 struct veno *veno = inet_csk_ca(sk);
121 121
122 if (!veno->doing_veno_now) 122 if (!veno->doing_veno_now) {
123 return tcp_reno_cong_avoid(sk, ack, in_flight); 123 tcp_reno_cong_avoid(sk, ack, in_flight);
124 return;
125 }
124 126
125 /* limited by applications */ 127 /* limited by applications */
126 if (!tcp_is_cwnd_limited(sk, in_flight)) 128 if (!tcp_is_cwnd_limited(sk, in_flight))
diff --git a/net/ipv4/tunnel4.c b/net/ipv4/tunnel4.c
index d3b709a6f264..cb1f0e83830b 100644
--- a/net/ipv4/tunnel4.c
+++ b/net/ipv4/tunnel4.c
@@ -97,7 +97,7 @@ static int tunnel64_rcv(struct sk_buff *skb)
97{ 97{
98 struct xfrm_tunnel *handler; 98 struct xfrm_tunnel *handler;
99 99
100 if (!pskb_may_pull(skb, sizeof(struct iphdr))) 100 if (!pskb_may_pull(skb, sizeof(struct ipv6hdr)))
101 goto drop; 101 goto drop;
102 102
103 for (handler = tunnel64_handlers; handler; handler = handler->next) 103 for (handler = tunnel64_handlers; handler; handler = handler->next)
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index 1f535e315188..56fcda3694ba 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -420,7 +420,7 @@ void udp_err(struct sk_buff *skb, u32 info)
420/* 420/*
421 * Throw away all pending data and cancel the corking. Socket is locked. 421 * Throw away all pending data and cancel the corking. Socket is locked.
422 */ 422 */
423static void udp_flush_pending_frames(struct sock *sk) 423void udp_flush_pending_frames(struct sock *sk)
424{ 424{
425 struct udp_sock *up = udp_sk(sk); 425 struct udp_sock *up = udp_sk(sk);
426 426
@@ -430,6 +430,7 @@ static void udp_flush_pending_frames(struct sock *sk)
430 ip_flush_pending_frames(sk); 430 ip_flush_pending_frames(sk);
431 } 431 }
432} 432}
433EXPORT_SYMBOL(udp_flush_pending_frames);
433 434
434/** 435/**
435 * udp4_hwcsum_outgoing - handle outgoing HW checksumming 436 * udp4_hwcsum_outgoing - handle outgoing HW checksumming
@@ -1605,10 +1606,9 @@ int udp_proc_register(struct net *net, struct udp_seq_afinfo *afinfo)
1605 afinfo->seq_ops.next = udp_seq_next; 1606 afinfo->seq_ops.next = udp_seq_next;
1606 afinfo->seq_ops.stop = udp_seq_stop; 1607 afinfo->seq_ops.stop = udp_seq_stop;
1607 1608
1608 p = proc_net_fops_create(net, afinfo->name, S_IRUGO, &afinfo->seq_fops); 1609 p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
1609 if (p) 1610 &afinfo->seq_fops, afinfo);
1610 p->data = afinfo; 1611 if (!p)
1611 else
1612 rc = -ENOMEM; 1612 rc = -ENOMEM;
1613 return rc; 1613 return rc;
1614} 1614}
diff --git a/net/ipv4/xfrm4_mode_tunnel.c b/net/ipv4/xfrm4_mode_tunnel.c
index 584e6d74e3a9..7135279f3f84 100644
--- a/net/ipv4/xfrm4_mode_tunnel.c
+++ b/net/ipv4/xfrm4_mode_tunnel.c
@@ -52,7 +52,7 @@ static int xfrm4_mode_tunnel_output(struct xfrm_state *x, struct sk_buff *skb)
52 IP_ECN_clear(top_iph); 52 IP_ECN_clear(top_iph);
53 53
54 top_iph->frag_off = (flags & XFRM_STATE_NOPMTUDISC) ? 54 top_iph->frag_off = (flags & XFRM_STATE_NOPMTUDISC) ?
55 0 : XFRM_MODE_SKB_CB(skb)->frag_off; 55 0 : (XFRM_MODE_SKB_CB(skb)->frag_off & htons(IP_DF));
56 ip_select_ident(top_iph, dst->child, NULL); 56 ip_select_ident(top_iph, dst->child, NULL);
57 57
58 top_iph->ttl = dst_metric(dst->child, RTAX_HOPLIMIT); 58 top_iph->ttl = dst_metric(dst->child, RTAX_HOPLIMIT);
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index e591e09e5e4e..ff61a5cdb0b3 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -731,8 +731,13 @@ static void ipv6_del_addr(struct inet6_ifaddr *ifp)
731 onlink = -1; 731 onlink = -1;
732 732
733 spin_lock(&ifa->lock); 733 spin_lock(&ifa->lock);
734 lifetime = min_t(unsigned long, 734
735 ifa->valid_lft, 0x7fffffffUL/HZ); 735 lifetime = addrconf_timeout_fixup(ifa->valid_lft, HZ);
736 /*
737 * Note: Because this address is
738 * not permanent, lifetime <
739 * LONG_MAX / HZ here.
740 */
736 if (time_before(expires, 741 if (time_before(expires,
737 ifa->tstamp + lifetime * HZ)) 742 ifa->tstamp + lifetime * HZ))
738 expires = ifa->tstamp + lifetime * HZ; 743 expires = ifa->tstamp + lifetime * HZ;
@@ -744,12 +749,12 @@ static void ipv6_del_addr(struct inet6_ifaddr *ifp)
744 } 749 }
745 write_unlock_bh(&idev->lock); 750 write_unlock_bh(&idev->lock);
746 751
752 addrconf_del_timer(ifp);
753
747 ipv6_ifa_notify(RTM_DELADDR, ifp); 754 ipv6_ifa_notify(RTM_DELADDR, ifp);
748 755
749 atomic_notifier_call_chain(&inet6addr_chain, NETDEV_DOWN, ifp); 756 atomic_notifier_call_chain(&inet6addr_chain, NETDEV_DOWN, ifp);
750 757
751 addrconf_del_timer(ifp);
752
753 /* 758 /*
754 * Purge or update corresponding prefix 759 * Purge or update corresponding prefix
755 * 760 *
@@ -1722,7 +1727,6 @@ void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len)
1722 __u32 valid_lft; 1727 __u32 valid_lft;
1723 __u32 prefered_lft; 1728 __u32 prefered_lft;
1724 int addr_type; 1729 int addr_type;
1725 unsigned long rt_expires;
1726 struct inet6_dev *in6_dev; 1730 struct inet6_dev *in6_dev;
1727 1731
1728 pinfo = (struct prefix_info *) opt; 1732 pinfo = (struct prefix_info *) opt;
@@ -1764,41 +1768,49 @@ void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len)
1764 * 2) Configure prefixes with the auto flag set 1768 * 2) Configure prefixes with the auto flag set
1765 */ 1769 */
1766 1770
1767 /* Avoid arithmetic overflow. Really, we could
1768 save rt_expires in seconds, likely valid_lft,
1769 but it would require division in fib gc, that it
1770 not good.
1771 */
1772 if (valid_lft >= 0x7FFFFFFF/HZ)
1773 rt_expires = 0x7FFFFFFF - (0x7FFFFFFF % HZ);
1774 else
1775 rt_expires = valid_lft * HZ;
1776
1777 /*
1778 * We convert this (in jiffies) to clock_t later.
1779 * Avoid arithmetic overflow there as well.
1780 * Overflow can happen only if HZ < USER_HZ.
1781 */
1782 if (HZ < USER_HZ && rt_expires > 0x7FFFFFFF / USER_HZ)
1783 rt_expires = 0x7FFFFFFF / USER_HZ;
1784
1785 if (pinfo->onlink) { 1771 if (pinfo->onlink) {
1786 struct rt6_info *rt; 1772 struct rt6_info *rt;
1773 unsigned long rt_expires;
1774
1775 /* Avoid arithmetic overflow. Really, we could
1776 * save rt_expires in seconds, likely valid_lft,
1777 * but it would require division in fib gc, that it
1778 * not good.
1779 */
1780 if (HZ > USER_HZ)
1781 rt_expires = addrconf_timeout_fixup(valid_lft, HZ);
1782 else
1783 rt_expires = addrconf_timeout_fixup(valid_lft, USER_HZ);
1784
1785 if (addrconf_finite_timeout(rt_expires))
1786 rt_expires *= HZ;
1787
1787 rt = rt6_lookup(dev_net(dev), &pinfo->prefix, NULL, 1788 rt = rt6_lookup(dev_net(dev), &pinfo->prefix, NULL,
1788 dev->ifindex, 1); 1789 dev->ifindex, 1);
1789 1790
1790 if (rt && ((rt->rt6i_flags & (RTF_GATEWAY | RTF_DEFAULT)) == 0)) { 1791 if (rt && ((rt->rt6i_flags & (RTF_GATEWAY | RTF_DEFAULT)) == 0)) {
1791 if (rt->rt6i_flags&RTF_EXPIRES) { 1792 /* Autoconf prefix route */
1792 if (valid_lft == 0) { 1793 if (valid_lft == 0) {
1793 ip6_del_rt(rt); 1794 ip6_del_rt(rt);
1794 rt = NULL; 1795 rt = NULL;
1795 } else { 1796 } else if (addrconf_finite_timeout(rt_expires)) {
1796 rt->rt6i_expires = jiffies + rt_expires; 1797 /* not infinity */
1797 } 1798 rt->rt6i_expires = jiffies + rt_expires;
1799 rt->rt6i_flags |= RTF_EXPIRES;
1800 } else {
1801 rt->rt6i_flags &= ~RTF_EXPIRES;
1802 rt->rt6i_expires = 0;
1798 } 1803 }
1799 } else if (valid_lft) { 1804 } else if (valid_lft) {
1805 clock_t expires = 0;
1806 int flags = RTF_ADDRCONF | RTF_PREFIX_RT;
1807 if (addrconf_finite_timeout(rt_expires)) {
1808 /* not infinity */
1809 flags |= RTF_EXPIRES;
1810 expires = jiffies_to_clock_t(rt_expires);
1811 }
1800 addrconf_prefix_route(&pinfo->prefix, pinfo->prefix_len, 1812 addrconf_prefix_route(&pinfo->prefix, pinfo->prefix_len,
1801 dev, jiffies_to_clock_t(rt_expires), RTF_ADDRCONF|RTF_EXPIRES|RTF_PREFIX_RT); 1813 dev, expires, flags);
1802 } 1814 }
1803 if (rt) 1815 if (rt)
1804 dst_release(&rt->u.dst); 1816 dst_release(&rt->u.dst);
@@ -2014,17 +2026,22 @@ err_exit:
2014 * Manual configuration of address on an interface 2026 * Manual configuration of address on an interface
2015 */ 2027 */
2016static int inet6_addr_add(struct net *net, int ifindex, struct in6_addr *pfx, 2028static int inet6_addr_add(struct net *net, int ifindex, struct in6_addr *pfx,
2017 int plen, __u8 ifa_flags, __u32 prefered_lft, 2029 unsigned int plen, __u8 ifa_flags, __u32 prefered_lft,
2018 __u32 valid_lft) 2030 __u32 valid_lft)
2019{ 2031{
2020 struct inet6_ifaddr *ifp; 2032 struct inet6_ifaddr *ifp;
2021 struct inet6_dev *idev; 2033 struct inet6_dev *idev;
2022 struct net_device *dev; 2034 struct net_device *dev;
2023 int scope; 2035 int scope;
2024 u32 flags = RTF_EXPIRES; 2036 u32 flags;
2037 clock_t expires;
2038 unsigned long timeout;
2025 2039
2026 ASSERT_RTNL(); 2040 ASSERT_RTNL();
2027 2041
2042 if (plen > 128)
2043 return -EINVAL;
2044
2028 /* check the lifetime */ 2045 /* check the lifetime */
2029 if (!valid_lft || prefered_lft > valid_lft) 2046 if (!valid_lft || prefered_lft > valid_lft)
2030 return -EINVAL; 2047 return -EINVAL;
@@ -2038,17 +2055,23 @@ static int inet6_addr_add(struct net *net, int ifindex, struct in6_addr *pfx,
2038 2055
2039 scope = ipv6_addr_scope(pfx); 2056 scope = ipv6_addr_scope(pfx);
2040 2057
2041 if (valid_lft == INFINITY_LIFE_TIME) { 2058 timeout = addrconf_timeout_fixup(valid_lft, HZ);
2042 ifa_flags |= IFA_F_PERMANENT; 2059 if (addrconf_finite_timeout(timeout)) {
2060 expires = jiffies_to_clock_t(timeout * HZ);
2061 valid_lft = timeout;
2062 flags = RTF_EXPIRES;
2063 } else {
2064 expires = 0;
2043 flags = 0; 2065 flags = 0;
2044 } else if (valid_lft >= 0x7FFFFFFF/HZ) 2066 ifa_flags |= IFA_F_PERMANENT;
2045 valid_lft = 0x7FFFFFFF/HZ; 2067 }
2046 2068
2047 if (prefered_lft == 0) 2069 timeout = addrconf_timeout_fixup(prefered_lft, HZ);
2048 ifa_flags |= IFA_F_DEPRECATED; 2070 if (addrconf_finite_timeout(timeout)) {
2049 else if ((prefered_lft >= 0x7FFFFFFF/HZ) && 2071 if (timeout == 0)
2050 (prefered_lft != INFINITY_LIFE_TIME)) 2072 ifa_flags |= IFA_F_DEPRECATED;
2051 prefered_lft = 0x7FFFFFFF/HZ; 2073 prefered_lft = timeout;
2074 }
2052 2075
2053 ifp = ipv6_add_addr(idev, pfx, plen, scope, ifa_flags); 2076 ifp = ipv6_add_addr(idev, pfx, plen, scope, ifa_flags);
2054 2077
@@ -2060,7 +2083,7 @@ static int inet6_addr_add(struct net *net, int ifindex, struct in6_addr *pfx,
2060 spin_unlock_bh(&ifp->lock); 2083 spin_unlock_bh(&ifp->lock);
2061 2084
2062 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, dev, 2085 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, dev,
2063 jiffies_to_clock_t(valid_lft * HZ), flags); 2086 expires, flags);
2064 /* 2087 /*
2065 * Note that section 3.1 of RFC 4429 indicates 2088 * Note that section 3.1 of RFC 4429 indicates
2066 * that the Optimistic flag should not be set for 2089 * that the Optimistic flag should not be set for
@@ -2076,12 +2099,15 @@ static int inet6_addr_add(struct net *net, int ifindex, struct in6_addr *pfx,
2076} 2099}
2077 2100
2078static int inet6_addr_del(struct net *net, int ifindex, struct in6_addr *pfx, 2101static int inet6_addr_del(struct net *net, int ifindex, struct in6_addr *pfx,
2079 int plen) 2102 unsigned int plen)
2080{ 2103{
2081 struct inet6_ifaddr *ifp; 2104 struct inet6_ifaddr *ifp;
2082 struct inet6_dev *idev; 2105 struct inet6_dev *idev;
2083 struct net_device *dev; 2106 struct net_device *dev;
2084 2107
2108 if (plen > 128)
2109 return -EINVAL;
2110
2085 dev = __dev_get_by_index(net, ifindex); 2111 dev = __dev_get_by_index(net, ifindex);
2086 if (!dev) 2112 if (!dev)
2087 return -ENODEV; 2113 return -ENODEV;
@@ -3148,22 +3174,30 @@ inet6_rtm_deladdr(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
3148static int inet6_addr_modify(struct inet6_ifaddr *ifp, u8 ifa_flags, 3174static int inet6_addr_modify(struct inet6_ifaddr *ifp, u8 ifa_flags,
3149 u32 prefered_lft, u32 valid_lft) 3175 u32 prefered_lft, u32 valid_lft)
3150{ 3176{
3151 u32 flags = RTF_EXPIRES; 3177 u32 flags;
3178 clock_t expires;
3179 unsigned long timeout;
3152 3180
3153 if (!valid_lft || (prefered_lft > valid_lft)) 3181 if (!valid_lft || (prefered_lft > valid_lft))
3154 return -EINVAL; 3182 return -EINVAL;
3155 3183
3156 if (valid_lft == INFINITY_LIFE_TIME) { 3184 timeout = addrconf_timeout_fixup(valid_lft, HZ);
3157 ifa_flags |= IFA_F_PERMANENT; 3185 if (addrconf_finite_timeout(timeout)) {
3186 expires = jiffies_to_clock_t(timeout * HZ);
3187 valid_lft = timeout;
3188 flags = RTF_EXPIRES;
3189 } else {
3190 expires = 0;
3158 flags = 0; 3191 flags = 0;
3159 } else if (valid_lft >= 0x7FFFFFFF/HZ) 3192 ifa_flags |= IFA_F_PERMANENT;
3160 valid_lft = 0x7FFFFFFF/HZ; 3193 }
3161 3194
3162 if (prefered_lft == 0) 3195 timeout = addrconf_timeout_fixup(prefered_lft, HZ);
3163 ifa_flags |= IFA_F_DEPRECATED; 3196 if (addrconf_finite_timeout(timeout)) {
3164 else if ((prefered_lft >= 0x7FFFFFFF/HZ) && 3197 if (timeout == 0)
3165 (prefered_lft != INFINITY_LIFE_TIME)) 3198 ifa_flags |= IFA_F_DEPRECATED;
3166 prefered_lft = 0x7FFFFFFF/HZ; 3199 prefered_lft = timeout;
3200 }
3167 3201
3168 spin_lock_bh(&ifp->lock); 3202 spin_lock_bh(&ifp->lock);
3169 ifp->flags = (ifp->flags & ~(IFA_F_DEPRECATED | IFA_F_PERMANENT | IFA_F_NODAD | IFA_F_HOMEADDRESS)) | ifa_flags; 3203 ifp->flags = (ifp->flags & ~(IFA_F_DEPRECATED | IFA_F_PERMANENT | IFA_F_NODAD | IFA_F_HOMEADDRESS)) | ifa_flags;
@@ -3176,7 +3210,7 @@ static int inet6_addr_modify(struct inet6_ifaddr *ifp, u8 ifa_flags,
3176 ipv6_ifa_notify(0, ifp); 3210 ipv6_ifa_notify(0, ifp);
3177 3211
3178 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, ifp->idev->dev, 3212 addrconf_prefix_route(&ifp->addr, ifp->prefix_len, ifp->idev->dev,
3179 jiffies_to_clock_t(valid_lft * HZ), flags); 3213 expires, flags);
3180 addrconf_verify(0); 3214 addrconf_verify(0);
3181 3215
3182 return 0; 3216 return 0;
@@ -4242,7 +4276,7 @@ static void addrconf_sysctl_register(struct inet6_dev *idev)
4242 neigh_sysctl_register(idev->dev, idev->nd_parms, NET_IPV6, 4276 neigh_sysctl_register(idev->dev, idev->nd_parms, NET_IPV6,
4243 NET_IPV6_NEIGH, "ipv6", 4277 NET_IPV6_NEIGH, "ipv6",
4244 &ndisc_ifinfo_sysctl_change, 4278 &ndisc_ifinfo_sysctl_change,
4245 NULL); 4279 ndisc_ifinfo_sysctl_strategy);
4246 __addrconf_sysctl_register(dev_net(idev->dev), idev->dev->name, 4280 __addrconf_sysctl_register(dev_net(idev->dev), idev->dev->name,
4247 idev->dev->ifindex, idev, &idev->cnf); 4281 idev->dev->ifindex, idev, &idev->cnf);
4248} 4282}
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index 3c6aafb02183..e84b3fd17fb4 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -191,7 +191,7 @@ lookup_protocol:
191 np->mcast_hops = -1; 191 np->mcast_hops = -1;
192 np->mc_loop = 1; 192 np->mc_loop = 1;
193 np->pmtudisc = IPV6_PMTUDISC_WANT; 193 np->pmtudisc = IPV6_PMTUDISC_WANT;
194 np->ipv6only = init_net.ipv6.sysctl.bindv6only; 194 np->ipv6only = net->ipv6.sysctl.bindv6only;
195 195
196 /* Init the ipv4 part of the socket since we can have sockets 196 /* Init the ipv4 part of the socket since we can have sockets
197 * using v6 API for ipv4. 197 * using v6 API for ipv4.
diff --git a/net/ipv6/datagram.c b/net/ipv6/datagram.c
index 94fa6ae77cfe..0f0f94a40335 100644
--- a/net/ipv6/datagram.c
+++ b/net/ipv6/datagram.c
@@ -496,7 +496,8 @@ int datagram_recv_ctl(struct sock *sk, struct msghdr *msg, struct sk_buff *skb)
496 return 0; 496 return 0;
497} 497}
498 498
499int datagram_send_ctl(struct msghdr *msg, struct flowi *fl, 499int datagram_send_ctl(struct net *net,
500 struct msghdr *msg, struct flowi *fl,
500 struct ipv6_txoptions *opt, 501 struct ipv6_txoptions *opt,
501 int *hlimit, int *tclass) 502 int *hlimit, int *tclass)
502{ 503{
@@ -509,7 +510,6 @@ int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
509 510
510 for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) { 511 for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) {
511 int addr_type; 512 int addr_type;
512 struct net_device *dev = NULL;
513 513
514 if (!CMSG_OK(msg, cmsg)) { 514 if (!CMSG_OK(msg, cmsg)) {
515 err = -EINVAL; 515 err = -EINVAL;
@@ -522,6 +522,9 @@ int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
522 switch (cmsg->cmsg_type) { 522 switch (cmsg->cmsg_type) {
523 case IPV6_PKTINFO: 523 case IPV6_PKTINFO:
524 case IPV6_2292PKTINFO: 524 case IPV6_2292PKTINFO:
525 {
526 struct net_device *dev = NULL;
527
525 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct in6_pktinfo))) { 528 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct in6_pktinfo))) {
526 err = -EINVAL; 529 err = -EINVAL;
527 goto exit_f; 530 goto exit_f;
@@ -535,32 +538,32 @@ int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
535 fl->oif = src_info->ipi6_ifindex; 538 fl->oif = src_info->ipi6_ifindex;
536 } 539 }
537 540
538 addr_type = ipv6_addr_type(&src_info->ipi6_addr); 541 addr_type = __ipv6_addr_type(&src_info->ipi6_addr);
539 542
540 if (addr_type == IPV6_ADDR_ANY) 543 if (fl->oif) {
541 break; 544 dev = dev_get_by_index(net, fl->oif);
545 if (!dev)
546 return -ENODEV;
547 } else if (addr_type & IPV6_ADDR_LINKLOCAL)
548 return -EINVAL;
542 549
543 if (addr_type & IPV6_ADDR_LINKLOCAL) { 550 if (addr_type != IPV6_ADDR_ANY) {
544 if (!src_info->ipi6_ifindex) 551 int strict = __ipv6_addr_src_scope(addr_type) <= IPV6_ADDR_SCOPE_LINKLOCAL;
545 return -EINVAL; 552 if (!ipv6_chk_addr(net, &src_info->ipi6_addr,
546 else { 553 strict ? dev : NULL, 0))
547 dev = dev_get_by_index(&init_net, src_info->ipi6_ifindex); 554 err = -EINVAL;
548 if (!dev) 555 else
549 return -ENODEV; 556 ipv6_addr_copy(&fl->fl6_src, &src_info->ipi6_addr);
550 }
551 }
552 if (!ipv6_chk_addr(&init_net, &src_info->ipi6_addr,
553 dev, 0)) {
554 if (dev)
555 dev_put(dev);
556 err = -EINVAL;
557 goto exit_f;
558 } 557 }
558
559 if (dev) 559 if (dev)
560 dev_put(dev); 560 dev_put(dev);
561 561
562 ipv6_addr_copy(&fl->fl6_src, &src_info->ipi6_addr); 562 if (err)
563 goto exit_f;
564
563 break; 565 break;
566 }
564 567
565 case IPV6_FLOWINFO: 568 case IPV6_FLOWINFO:
566 if (cmsg->cmsg_len < CMSG_LEN(4)) { 569 if (cmsg->cmsg_len < CMSG_LEN(4)) {
@@ -702,6 +705,11 @@ int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
702 } 705 }
703 706
704 *hlimit = *(int *)CMSG_DATA(cmsg); 707 *hlimit = *(int *)CMSG_DATA(cmsg);
708 if (*hlimit < -1 || *hlimit > 0xff) {
709 err = -EINVAL;
710 goto exit_f;
711 }
712
705 break; 713 break;
706 714
707 case IPV6_TCLASS: 715 case IPV6_TCLASS:
diff --git a/net/ipv6/exthdrs.c b/net/ipv6/exthdrs.c
index 3cd1c993d52b..dcf94fdfb863 100644
--- a/net/ipv6/exthdrs.c
+++ b/net/ipv6/exthdrs.c
@@ -445,7 +445,7 @@ looped_back:
445 kfree_skb(skb); 445 kfree_skb(skb);
446 return -1; 446 return -1;
447 } 447 }
448 if (!ipv6_chk_home_addr(&init_net, addr)) { 448 if (!ipv6_chk_home_addr(dev_net(skb->dst->dev), addr)) {
449 IP6_INC_STATS_BH(ip6_dst_idev(skb->dst), 449 IP6_INC_STATS_BH(ip6_dst_idev(skb->dst),
450 IPSTATS_MIB_INADDRERRORS); 450 IPSTATS_MIB_INADDRERRORS);
451 kfree_skb(skb); 451 kfree_skb(skb);
diff --git a/net/ipv6/ip6_flowlabel.c b/net/ipv6/ip6_flowlabel.c
index eb7a940310f4..37a4e777e347 100644
--- a/net/ipv6/ip6_flowlabel.c
+++ b/net/ipv6/ip6_flowlabel.c
@@ -354,7 +354,7 @@ fl_create(struct net *net, struct in6_flowlabel_req *freq, char __user *optval,
354 msg.msg_control = (void*)(fl->opt+1); 354 msg.msg_control = (void*)(fl->opt+1);
355 flowi.oif = 0; 355 flowi.oif = 0;
356 356
357 err = datagram_send_ctl(&msg, &flowi, fl->opt, &junk, &junk); 357 err = datagram_send_ctl(net, &msg, &flowi, fl->opt, &junk, &junk);
358 if (err) 358 if (err)
359 goto done; 359 goto done;
360 err = -EINVAL; 360 err = -EINVAL;
diff --git a/net/ipv6/ip6_input.c b/net/ipv6/ip6_input.c
index 4e5c8615832c..17eb48b8e329 100644
--- a/net/ipv6/ip6_input.c
+++ b/net/ipv6/ip6_input.c
@@ -102,6 +102,15 @@ int ipv6_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt
102 if (hdr->version != 6) 102 if (hdr->version != 6)
103 goto err; 103 goto err;
104 104
105 /*
106 * RFC4291 2.5.3
107 * A packet received on an interface with a destination address
108 * of loopback must be dropped.
109 */
110 if (!(dev->flags & IFF_LOOPBACK) &&
111 ipv6_addr_loopback(&hdr->daddr))
112 goto err;
113
105 skb->transport_header = skb->network_header + sizeof(*hdr); 114 skb->transport_header = skb->network_header + sizeof(*hdr);
106 IP6CB(skb)->nhoff = offsetof(struct ipv6hdr, nexthdr); 115 IP6CB(skb)->nhoff = offsetof(struct ipv6hdr, nexthdr);
107 116
diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index 0af2e055f883..48cdce9c696c 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -780,7 +780,7 @@ slow_path:
780 * Allocate buffer. 780 * Allocate buffer.
781 */ 781 */
782 782
783 if ((frag = alloc_skb(len+hlen+sizeof(struct frag_hdr)+LL_RESERVED_SPACE(rt->u.dst.dev), GFP_ATOMIC)) == NULL) { 783 if ((frag = alloc_skb(len+hlen+sizeof(struct frag_hdr)+LL_ALLOCATED_SPACE(rt->u.dst.dev), GFP_ATOMIC)) == NULL) {
784 NETDEBUG(KERN_INFO "IPv6: frag: no memory for new fragment!\n"); 784 NETDEBUG(KERN_INFO "IPv6: frag: no memory for new fragment!\n");
785 IP6_INC_STATS(ip6_dst_idev(skb->dst), 785 IP6_INC_STATS(ip6_dst_idev(skb->dst),
786 IPSTATS_MIB_FRAGFAILS); 786 IPSTATS_MIB_FRAGFAILS);
diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c
index 2de3c464fe75..14796181e8b5 100644
--- a/net/ipv6/ip6mr.c
+++ b/net/ipv6/ip6mr.c
@@ -197,7 +197,7 @@ static int ip6mr_vif_seq_show(struct seq_file *seq, void *v)
197 const char *name = vif->dev ? vif->dev->name : "none"; 197 const char *name = vif->dev ? vif->dev->name : "none";
198 198
199 seq_printf(seq, 199 seq_printf(seq,
200 "%2Zd %-10s %8ld %7ld %8ld %7ld %05X\n", 200 "%2td %-10s %8ld %7ld %8ld %7ld %05X\n",
201 vif - vif6_table, 201 vif - vif6_table,
202 name, vif->bytes_in, vif->pkt_in, 202 name, vif->bytes_in, vif->pkt_in,
203 vif->bytes_out, vif->pkt_out, 203 vif->bytes_out, vif->pkt_out,
diff --git a/net/ipv6/ipv6_sockglue.c b/net/ipv6/ipv6_sockglue.c
index 56d55fecf8ec..86e28a75267f 100644
--- a/net/ipv6/ipv6_sockglue.c
+++ b/net/ipv6/ipv6_sockglue.c
@@ -67,7 +67,7 @@ int ip6_ra_control(struct sock *sk, int sel, void (*destructor)(struct sock *))
67 67
68 /* RA packet may be delivered ONLY to IPPROTO_RAW socket */ 68 /* RA packet may be delivered ONLY to IPPROTO_RAW socket */
69 if (sk->sk_type != SOCK_RAW || inet_sk(sk)->num != IPPROTO_RAW) 69 if (sk->sk_type != SOCK_RAW || inet_sk(sk)->num != IPPROTO_RAW)
70 return -EINVAL; 70 return -ENOPROTOOPT;
71 71
72 new_ra = (sel>=0) ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL; 72 new_ra = (sel>=0) ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL;
73 73
@@ -161,9 +161,17 @@ static int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
161 struct ipv6_txoptions *opt; 161 struct ipv6_txoptions *opt;
162 struct sk_buff *pktopt; 162 struct sk_buff *pktopt;
163 163
164 if (sk->sk_protocol != IPPROTO_UDP && 164 if (sk->sk_type == SOCK_RAW)
165 sk->sk_protocol != IPPROTO_UDPLITE && 165 break;
166 sk->sk_protocol != IPPROTO_TCP) 166
167 if (sk->sk_protocol == IPPROTO_UDP ||
168 sk->sk_protocol == IPPROTO_UDPLITE) {
169 struct udp_sock *up = udp_sk(sk);
170 if (up->pending == AF_INET6) {
171 retv = -EBUSY;
172 break;
173 }
174 } else if (sk->sk_protocol != IPPROTO_TCP)
167 break; 175 break;
168 176
169 if (sk->sk_state != TCP_ESTABLISHED) { 177 if (sk->sk_state != TCP_ESTABLISHED) {
@@ -337,18 +345,21 @@ static int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
337 case IPV6_DSTOPTS: 345 case IPV6_DSTOPTS:
338 { 346 {
339 struct ipv6_txoptions *opt; 347 struct ipv6_txoptions *opt;
348
349 /* remove any sticky options header with a zero option
350 * length, per RFC3542.
351 */
340 if (optlen == 0) 352 if (optlen == 0)
341 optval = NULL; 353 optval = NULL;
354 else if (optlen < sizeof(struct ipv6_opt_hdr) ||
355 optlen & 0x7 || optlen > 8 * 255)
356 goto e_inval;
342 357
343 /* hop-by-hop / destination options are privileged option */ 358 /* hop-by-hop / destination options are privileged option */
344 retv = -EPERM; 359 retv = -EPERM;
345 if (optname != IPV6_RTHDR && !capable(CAP_NET_RAW)) 360 if (optname != IPV6_RTHDR && !capable(CAP_NET_RAW))
346 break; 361 break;
347 362
348 if (optlen < sizeof(struct ipv6_opt_hdr) ||
349 optlen & 0x7 || optlen > 8 * 255)
350 goto e_inval;
351
352 opt = ipv6_renew_options(sk, np->opt, optname, 363 opt = ipv6_renew_options(sk, np->opt, optname,
353 (struct ipv6_opt_hdr __user *)optval, 364 (struct ipv6_opt_hdr __user *)optval,
354 optlen); 365 optlen);
@@ -416,7 +427,7 @@ sticky_done:
416 msg.msg_controllen = optlen; 427 msg.msg_controllen = optlen;
417 msg.msg_control = (void*)(opt+1); 428 msg.msg_control = (void*)(opt+1);
418 429
419 retv = datagram_send_ctl(&msg, &fl, opt, &junk, &junk); 430 retv = datagram_send_ctl(net, &msg, &fl, opt, &junk, &junk);
420 if (retv) 431 if (retv)
421 goto done; 432 goto done;
422update: 433update:
@@ -438,7 +449,7 @@ done:
438 449
439 case IPV6_MULTICAST_HOPS: 450 case IPV6_MULTICAST_HOPS:
440 if (sk->sk_type == SOCK_STREAM) 451 if (sk->sk_type == SOCK_STREAM)
441 goto e_inval; 452 break;
442 if (optlen < sizeof(int)) 453 if (optlen < sizeof(int))
443 goto e_inval; 454 goto e_inval;
444 if (val > 255 || val < -1) 455 if (val > 255 || val < -1)
@@ -450,13 +461,15 @@ done:
450 case IPV6_MULTICAST_LOOP: 461 case IPV6_MULTICAST_LOOP:
451 if (optlen < sizeof(int)) 462 if (optlen < sizeof(int))
452 goto e_inval; 463 goto e_inval;
464 if (val != valbool)
465 goto e_inval;
453 np->mc_loop = valbool; 466 np->mc_loop = valbool;
454 retv = 0; 467 retv = 0;
455 break; 468 break;
456 469
457 case IPV6_MULTICAST_IF: 470 case IPV6_MULTICAST_IF:
458 if (sk->sk_type == SOCK_STREAM) 471 if (sk->sk_type == SOCK_STREAM)
459 goto e_inval; 472 break;
460 if (optlen < sizeof(int)) 473 if (optlen < sizeof(int))
461 goto e_inval; 474 goto e_inval;
462 475
@@ -832,7 +845,7 @@ static int ipv6_getsockopt_sticky(struct sock *sk, struct ipv6_txoptions *opt,
832 len = min_t(unsigned int, len, ipv6_optlen(hdr)); 845 len = min_t(unsigned int, len, ipv6_optlen(hdr));
833 if (copy_to_user(optval, hdr, len)) 846 if (copy_to_user(optval, hdr, len))
834 return -EFAULT; 847 return -EFAULT;
835 return ipv6_optlen(hdr); 848 return len;
836} 849}
837 850
838static int do_ipv6_getsockopt(struct sock *sk, int level, int optname, 851static int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
@@ -852,7 +865,7 @@ static int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
852 if (sk->sk_protocol != IPPROTO_UDP && 865 if (sk->sk_protocol != IPPROTO_UDP &&
853 sk->sk_protocol != IPPROTO_UDPLITE && 866 sk->sk_protocol != IPPROTO_UDPLITE &&
854 sk->sk_protocol != IPPROTO_TCP) 867 sk->sk_protocol != IPPROTO_TCP)
855 return -EINVAL; 868 return -ENOPROTOOPT;
856 if (sk->sk_state != TCP_ESTABLISHED) 869 if (sk->sk_state != TCP_ESTABLISHED)
857 return -ENOTCONN; 870 return -ENOTCONN;
858 val = sk->sk_family; 871 val = sk->sk_family;
@@ -866,6 +879,8 @@ static int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
866 return -EINVAL; 879 return -EINVAL;
867 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) 880 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0)))
868 return -EFAULT; 881 return -EFAULT;
882 if (gsf.gf_group.ss_family != AF_INET6)
883 return -EADDRNOTAVAIL;
869 lock_sock(sk); 884 lock_sock(sk);
870 err = ip6_mc_msfget(sk, &gsf, 885 err = ip6_mc_msfget(sk, &gsf,
871 (struct group_filter __user *)optval, optlen); 886 (struct group_filter __user *)optval, optlen);
@@ -975,6 +990,9 @@ static int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
975 len = ipv6_getsockopt_sticky(sk, np->opt, 990 len = ipv6_getsockopt_sticky(sk, np->opt,
976 optname, optval, len); 991 optname, optval, len);
977 release_sock(sk); 992 release_sock(sk);
993 /* check if ipv6_getsockopt_sticky() returns err code */
994 if (len < 0)
995 return len;
978 return put_user(len, optlen); 996 return put_user(len, optlen);
979 } 997 }
980 998
diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
index 54f91efdae58..fd632dd7f98d 100644
--- a/net/ipv6/mcast.c
+++ b/net/ipv6/mcast.c
@@ -1411,7 +1411,7 @@ static struct sk_buff *mld_newpack(struct net_device *dev, int size)
1411 IPV6_TLV_PADN, 0 }; 1411 IPV6_TLV_PADN, 0 };
1412 1412
1413 /* we assume size > sizeof(ra) here */ 1413 /* we assume size > sizeof(ra) here */
1414 skb = sock_alloc_send_skb(sk, size + LL_RESERVED_SPACE(dev), 1, &err); 1414 skb = sock_alloc_send_skb(sk, size + LL_ALLOCATED_SPACE(dev), 1, &err);
1415 1415
1416 if (!skb) 1416 if (!skb)
1417 return NULL; 1417 return NULL;
@@ -1790,7 +1790,7 @@ static void igmp6_send(struct in6_addr *addr, struct net_device *dev, int type)
1790 payload_len = len + sizeof(ra); 1790 payload_len = len + sizeof(ra);
1791 full_len = sizeof(struct ipv6hdr) + payload_len; 1791 full_len = sizeof(struct ipv6hdr) + payload_len;
1792 1792
1793 skb = sock_alloc_send_skb(sk, LL_RESERVED_SPACE(dev) + full_len, 1, &err); 1793 skb = sock_alloc_send_skb(sk, LL_ALLOCATED_SPACE(dev) + full_len, 1, &err);
1794 1794
1795 if (skb == NULL) { 1795 if (skb == NULL) {
1796 rcu_read_lock(); 1796 rcu_read_lock();
diff --git a/net/ipv6/ndisc.c b/net/ipv6/ndisc.c
index 2c74885f8355..282fdb31f8ed 100644
--- a/net/ipv6/ndisc.c
+++ b/net/ipv6/ndisc.c
@@ -479,7 +479,7 @@ static void __ndisc_send(struct net_device *dev,
479 479
480 skb = sock_alloc_send_skb(sk, 480 skb = sock_alloc_send_skb(sk,
481 (MAX_HEADER + sizeof(struct ipv6hdr) + 481 (MAX_HEADER + sizeof(struct ipv6hdr) +
482 len + LL_RESERVED_SPACE(dev)), 482 len + LL_ALLOCATED_SPACE(dev)),
483 1, &err); 483 1, &err);
484 if (!skb) { 484 if (!skb) {
485 ND_PRINTK0(KERN_ERR 485 ND_PRINTK0(KERN_ERR
@@ -1521,7 +1521,7 @@ void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh,
1521 1521
1522 buff = sock_alloc_send_skb(sk, 1522 buff = sock_alloc_send_skb(sk,
1523 (MAX_HEADER + sizeof(struct ipv6hdr) + 1523 (MAX_HEADER + sizeof(struct ipv6hdr) +
1524 len + LL_RESERVED_SPACE(dev)), 1524 len + LL_ALLOCATED_SPACE(dev)),
1525 1, &err); 1525 1, &err);
1526 if (buff == NULL) { 1526 if (buff == NULL) {
1527 ND_PRINTK0(KERN_ERR 1527 ND_PRINTK0(KERN_ERR
@@ -1727,10 +1727,10 @@ int ndisc_ifinfo_sysctl_change(struct ctl_table *ctl, int write, struct file * f
1727 return ret; 1727 return ret;
1728} 1728}
1729 1729
1730static int ndisc_ifinfo_sysctl_strategy(ctl_table *ctl, int __user *name, 1730int ndisc_ifinfo_sysctl_strategy(ctl_table *ctl, int __user *name,
1731 int nlen, void __user *oldval, 1731 int nlen, void __user *oldval,
1732 size_t __user *oldlenp, 1732 size_t __user *oldlenp,
1733 void __user *newval, size_t newlen) 1733 void __user *newval, size_t newlen)
1734{ 1734{
1735 struct net_device *dev = ctl->extra1; 1735 struct net_device *dev = ctl->extra1;
1736 struct inet6_dev *idev; 1736 struct inet6_dev *idev;
diff --git a/net/ipv6/netfilter/ip6table_mangle.c b/net/ipv6/netfilter/ip6table_mangle.c
index 27a5e8b48d93..f405cea21a8b 100644
--- a/net/ipv6/netfilter/ip6table_mangle.c
+++ b/net/ipv6/netfilter/ip6table_mangle.c
@@ -129,7 +129,7 @@ static struct nf_hook_ops ip6t_ops[] __read_mostly = {
129 .priority = NF_IP6_PRI_MANGLE, 129 .priority = NF_IP6_PRI_MANGLE,
130 }, 130 },
131 { 131 {
132 .hook = ip6t_local_hook, 132 .hook = ip6t_route_hook,
133 .owner = THIS_MODULE, 133 .owner = THIS_MODULE,
134 .pf = PF_INET6, 134 .pf = PF_INET6,
135 .hooknum = NF_INET_LOCAL_IN, 135 .hooknum = NF_INET_LOCAL_IN,
diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c b/net/ipv6/netfilter/nf_conntrack_reasm.c
index 2dccad48058c..cf20bc4fd60d 100644
--- a/net/ipv6/netfilter/nf_conntrack_reasm.c
+++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
@@ -207,9 +207,12 @@ fq_find(__be32 id, struct in6_addr *src, struct in6_addr *dst)
207 arg.id = id; 207 arg.id = id;
208 arg.src = src; 208 arg.src = src;
209 arg.dst = dst; 209 arg.dst = dst;
210
211 read_lock_bh(&nf_frags.lock);
210 hash = ip6qhashfn(id, src, dst); 212 hash = ip6qhashfn(id, src, dst);
211 213
212 q = inet_frag_find(&nf_init_frags, &nf_frags, &arg, hash); 214 q = inet_frag_find(&nf_init_frags, &nf_frags, &arg, hash);
215 local_bh_enable();
213 if (q == NULL) 216 if (q == NULL)
214 goto oom; 217 goto oom;
215 218
@@ -638,10 +641,10 @@ struct sk_buff *nf_ct_frag6_gather(struct sk_buff *skb)
638 goto ret_orig; 641 goto ret_orig;
639 } 642 }
640 643
641 spin_lock(&fq->q.lock); 644 spin_lock_bh(&fq->q.lock);
642 645
643 if (nf_ct_frag6_queue(fq, clone, fhdr, nhoff) < 0) { 646 if (nf_ct_frag6_queue(fq, clone, fhdr, nhoff) < 0) {
644 spin_unlock(&fq->q.lock); 647 spin_unlock_bh(&fq->q.lock);
645 pr_debug("Can't insert skb to queue\n"); 648 pr_debug("Can't insert skb to queue\n");
646 fq_put(fq); 649 fq_put(fq);
647 goto ret_orig; 650 goto ret_orig;
@@ -653,7 +656,7 @@ struct sk_buff *nf_ct_frag6_gather(struct sk_buff *skb)
653 if (ret_skb == NULL) 656 if (ret_skb == NULL)
654 pr_debug("Can't reassemble fragmented packets\n"); 657 pr_debug("Can't reassemble fragmented packets\n");
655 } 658 }
656 spin_unlock(&fq->q.lock); 659 spin_unlock_bh(&fq->q.lock);
657 660
658 fq_put(fq); 661 fq_put(fq);
659 return ret_skb; 662 return ret_skb;
diff --git a/net/ipv6/proc.c b/net/ipv6/proc.c
index ca8b82f96fe5..df0736a4cafa 100644
--- a/net/ipv6/proc.c
+++ b/net/ipv6/proc.c
@@ -247,13 +247,11 @@ int snmp6_register_dev(struct inet6_dev *idev)
247 if (!proc_net_devsnmp6) 247 if (!proc_net_devsnmp6)
248 return -ENOENT; 248 return -ENOENT;
249 249
250 p = proc_create(idev->dev->name, S_IRUGO, 250 p = proc_create_data(idev->dev->name, S_IRUGO,
251 proc_net_devsnmp6, &snmp6_seq_fops); 251 proc_net_devsnmp6, &snmp6_seq_fops, idev);
252 if (!p) 252 if (!p)
253 return -ENOMEM; 253 return -ENOMEM;
254 254
255 p->data = idev;
256
257 idev->stats.proc_dir_entry = p; 255 idev->stats.proc_dir_entry = p;
258 return 0; 256 return 0;
259} 257}
diff --git a/net/ipv6/raw.c b/net/ipv6/raw.c
index 396f0ea11090..3aee12310d94 100644
--- a/net/ipv6/raw.c
+++ b/net/ipv6/raw.c
@@ -609,7 +609,6 @@ static int rawv6_send_hdrinc(struct sock *sk, void *from, int length,
609 struct ipv6_pinfo *np = inet6_sk(sk); 609 struct ipv6_pinfo *np = inet6_sk(sk);
610 struct ipv6hdr *iph; 610 struct ipv6hdr *iph;
611 struct sk_buff *skb; 611 struct sk_buff *skb;
612 unsigned int hh_len;
613 int err; 612 int err;
614 613
615 if (length > rt->u.dst.dev->mtu) { 614 if (length > rt->u.dst.dev->mtu) {
@@ -619,13 +618,12 @@ static int rawv6_send_hdrinc(struct sock *sk, void *from, int length,
619 if (flags&MSG_PROBE) 618 if (flags&MSG_PROBE)
620 goto out; 619 goto out;
621 620
622 hh_len = LL_RESERVED_SPACE(rt->u.dst.dev); 621 skb = sock_alloc_send_skb(sk,
623 622 length + LL_ALLOCATED_SPACE(rt->u.dst.dev) + 15,
624 skb = sock_alloc_send_skb(sk, length+hh_len+15, 623 flags & MSG_DONTWAIT, &err);
625 flags&MSG_DONTWAIT, &err);
626 if (skb == NULL) 624 if (skb == NULL)
627 goto error; 625 goto error;
628 skb_reserve(skb, hh_len); 626 skb_reserve(skb, LL_RESERVED_SPACE(rt->u.dst.dev));
629 627
630 skb->priority = sk->sk_priority; 628 skb->priority = sk->sk_priority;
631 skb->mark = sk->sk_mark; 629 skb->mark = sk->sk_mark;
@@ -815,7 +813,7 @@ static int rawv6_sendmsg(struct kiocb *iocb, struct sock *sk,
815 memset(opt, 0, sizeof(struct ipv6_txoptions)); 813 memset(opt, 0, sizeof(struct ipv6_txoptions));
816 opt->tot_len = sizeof(struct ipv6_txoptions); 814 opt->tot_len = sizeof(struct ipv6_txoptions);
817 815
818 err = datagram_send_ctl(msg, &fl, opt, &hlimit, &tclass); 816 err = datagram_send_ctl(sock_net(sk), msg, &fl, opt, &hlimit, &tclass);
819 if (err < 0) { 817 if (err < 0) {
820 fl6_sock_release(flowlabel); 818 fl6_sock_release(flowlabel);
821 return err; 819 return err;
@@ -1166,6 +1164,15 @@ static void rawv6_close(struct sock *sk, long timeout)
1166 sk_common_release(sk); 1164 sk_common_release(sk);
1167} 1165}
1168 1166
1167static int raw6_destroy(struct sock *sk)
1168{
1169 lock_sock(sk);
1170 ip6_flush_pending_frames(sk);
1171 release_sock(sk);
1172
1173 return inet6_destroy_sock(sk);
1174}
1175
1169static int rawv6_init_sk(struct sock *sk) 1176static int rawv6_init_sk(struct sock *sk)
1170{ 1177{
1171 struct raw6_sock *rp = raw6_sk(sk); 1178 struct raw6_sock *rp = raw6_sk(sk);
@@ -1189,11 +1196,11 @@ struct proto rawv6_prot = {
1189 .name = "RAWv6", 1196 .name = "RAWv6",
1190 .owner = THIS_MODULE, 1197 .owner = THIS_MODULE,
1191 .close = rawv6_close, 1198 .close = rawv6_close,
1199 .destroy = raw6_destroy,
1192 .connect = ip6_datagram_connect, 1200 .connect = ip6_datagram_connect,
1193 .disconnect = udp_disconnect, 1201 .disconnect = udp_disconnect,
1194 .ioctl = rawv6_ioctl, 1202 .ioctl = rawv6_ioctl,
1195 .init = rawv6_init_sk, 1203 .init = rawv6_init_sk,
1196 .destroy = inet6_destroy_sock,
1197 .setsockopt = rawv6_setsockopt, 1204 .setsockopt = rawv6_setsockopt,
1198 .getsockopt = rawv6_getsockopt, 1205 .getsockopt = rawv6_getsockopt,
1199 .sendmsg = rawv6_sendmsg, 1206 .sendmsg = rawv6_sendmsg,
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 7b247e3a16fe..a60d7d129713 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -197,6 +197,7 @@ static void ip6_frag_expire(unsigned long data)
197{ 197{
198 struct frag_queue *fq; 198 struct frag_queue *fq;
199 struct net_device *dev = NULL; 199 struct net_device *dev = NULL;
200 struct net *net;
200 201
201 fq = container_of((struct inet_frag_queue *)data, struct frag_queue, q); 202 fq = container_of((struct inet_frag_queue *)data, struct frag_queue, q);
202 203
@@ -207,7 +208,8 @@ static void ip6_frag_expire(unsigned long data)
207 208
208 fq_kill(fq); 209 fq_kill(fq);
209 210
210 dev = dev_get_by_index(&init_net, fq->iif); 211 net = container_of(fq->q.net, struct net, ipv6.frags);
212 dev = dev_get_by_index(net, fq->iif);
211 if (!dev) 213 if (!dev)
212 goto out; 214 goto out;
213 215
@@ -245,6 +247,8 @@ fq_find(struct net *net, __be32 id, struct in6_addr *src, struct in6_addr *dst,
245 arg.id = id; 247 arg.id = id;
246 arg.src = src; 248 arg.src = src;
247 arg.dst = dst; 249 arg.dst = dst;
250
251 read_lock(&ip6_frags.lock);
248 hash = ip6qhashfn(id, src, dst); 252 hash = ip6qhashfn(id, src, dst);
249 253
250 q = inet_frag_find(&net->ipv6.frags, &ip6_frags, &arg, hash); 254 q = inet_frag_find(&net->ipv6.frags, &ip6_frags, &arg, hash);
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index a493ad9b8914..7ff687020fa9 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -109,7 +109,7 @@ static struct dst_ops ip6_dst_ops_template = {
109 .negative_advice = ip6_negative_advice, 109 .negative_advice = ip6_negative_advice,
110 .link_failure = ip6_link_failure, 110 .link_failure = ip6_link_failure,
111 .update_pmtu = ip6_rt_update_pmtu, 111 .update_pmtu = ip6_rt_update_pmtu,
112 .local_out = ip6_local_out, 112 .local_out = __ip6_local_out,
113 .entry_size = sizeof(struct rt6_info), 113 .entry_size = sizeof(struct rt6_info),
114 .entries = ATOMIC_INIT(0), 114 .entries = ATOMIC_INIT(0),
115}; 115};
@@ -240,7 +240,7 @@ static inline int rt6_need_strict(struct in6_addr *daddr)
240static inline struct rt6_info *rt6_device_match(struct net *net, 240static inline struct rt6_info *rt6_device_match(struct net *net,
241 struct rt6_info *rt, 241 struct rt6_info *rt,
242 int oif, 242 int oif,
243 int strict) 243 int flags)
244{ 244{
245 struct rt6_info *local = NULL; 245 struct rt6_info *local = NULL;
246 struct rt6_info *sprt; 246 struct rt6_info *sprt;
@@ -253,7 +253,7 @@ static inline struct rt6_info *rt6_device_match(struct net *net,
253 if (dev->flags & IFF_LOOPBACK) { 253 if (dev->flags & IFF_LOOPBACK) {
254 if (sprt->rt6i_idev == NULL || 254 if (sprt->rt6i_idev == NULL ||
255 sprt->rt6i_idev->dev->ifindex != oif) { 255 sprt->rt6i_idev->dev->ifindex != oif) {
256 if (strict && oif) 256 if (flags & RT6_LOOKUP_F_IFACE && oif)
257 continue; 257 continue;
258 if (local && (!oif || 258 if (local && (!oif ||
259 local->rt6i_idev->dev->ifindex == oif)) 259 local->rt6i_idev->dev->ifindex == oif))
@@ -266,7 +266,7 @@ static inline struct rt6_info *rt6_device_match(struct net *net,
266 if (local) 266 if (local)
267 return local; 267 return local;
268 268
269 if (strict) 269 if (flags & RT6_LOOKUP_F_IFACE)
270 return net->ipv6.ip6_null_entry; 270 return net->ipv6.ip6_null_entry;
271 } 271 }
272 return rt; 272 return rt;
@@ -446,7 +446,7 @@ int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
446 struct route_info *rinfo = (struct route_info *) opt; 446 struct route_info *rinfo = (struct route_info *) opt;
447 struct in6_addr prefix_buf, *prefix; 447 struct in6_addr prefix_buf, *prefix;
448 unsigned int pref; 448 unsigned int pref;
449 u32 lifetime; 449 unsigned long lifetime;
450 struct rt6_info *rt; 450 struct rt6_info *rt;
451 451
452 if (len < sizeof(struct route_info)) { 452 if (len < sizeof(struct route_info)) {
@@ -472,13 +472,7 @@ int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
472 if (pref == ICMPV6_ROUTER_PREF_INVALID) 472 if (pref == ICMPV6_ROUTER_PREF_INVALID)
473 pref = ICMPV6_ROUTER_PREF_MEDIUM; 473 pref = ICMPV6_ROUTER_PREF_MEDIUM;
474 474
475 lifetime = ntohl(rinfo->lifetime); 475 lifetime = addrconf_timeout_fixup(ntohl(rinfo->lifetime), HZ);
476 if (lifetime == 0xffffffff) {
477 /* infinity */
478 } else if (lifetime > 0x7fffffff/HZ) {
479 /* Avoid arithmetic overflow */
480 lifetime = 0x7fffffff/HZ - 1;
481 }
482 476
483 if (rinfo->length == 3) 477 if (rinfo->length == 3)
484 prefix = (struct in6_addr *)rinfo->prefix; 478 prefix = (struct in6_addr *)rinfo->prefix;
@@ -506,7 +500,7 @@ int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
506 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref); 500 (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
507 501
508 if (rt) { 502 if (rt) {
509 if (lifetime == 0xffffffff) { 503 if (!addrconf_finite_timeout(lifetime)) {
510 rt->rt6i_flags &= ~RTF_EXPIRES; 504 rt->rt6i_flags &= ~RTF_EXPIRES;
511 } else { 505 } else {
512 rt->rt6i_expires = jiffies + HZ * lifetime; 506 rt->rt6i_expires = jiffies + HZ * lifetime;
@@ -1106,7 +1100,9 @@ int ip6_route_add(struct fib6_config *cfg)
1106 } 1100 }
1107 1101
1108 rt->u.dst.obsolete = -1; 1102 rt->u.dst.obsolete = -1;
1109 rt->rt6i_expires = jiffies + clock_t_to_jiffies(cfg->fc_expires); 1103 rt->rt6i_expires = (cfg->fc_flags & RTF_EXPIRES) ?
1104 jiffies + clock_t_to_jiffies(cfg->fc_expires) :
1105 0;
1110 1106
1111 if (cfg->fc_protocol == RTPROT_UNSPEC) 1107 if (cfg->fc_protocol == RTPROT_UNSPEC)
1112 cfg->fc_protocol = RTPROT_BOOT; 1108 cfg->fc_protocol = RTPROT_BOOT;
@@ -1243,11 +1239,11 @@ install_route:
1243 } 1239 }
1244 } 1240 }
1245 1241
1246 if (rt->u.dst.metrics[RTAX_HOPLIMIT-1] == 0) 1242 if (dst_metric(&rt->u.dst, RTAX_HOPLIMIT) == 0)
1247 rt->u.dst.metrics[RTAX_HOPLIMIT-1] = -1; 1243 rt->u.dst.metrics[RTAX_HOPLIMIT-1] = -1;
1248 if (!rt->u.dst.metrics[RTAX_MTU-1]) 1244 if (!dst_metric(&rt->u.dst, RTAX_MTU))
1249 rt->u.dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(dev); 1245 rt->u.dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(dev);
1250 if (!rt->u.dst.metrics[RTAX_ADVMSS-1]) 1246 if (!dst_metric(&rt->u.dst, RTAX_ADVMSS))
1251 rt->u.dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, dst_mtu(&rt->u.dst)); 1247 rt->u.dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(net, dst_mtu(&rt->u.dst));
1252 rt->u.dst.dev = dev; 1248 rt->u.dst.dev = dev;
1253 rt->rt6i_idev = idev; 1249 rt->rt6i_idev = idev;
@@ -2200,7 +2196,13 @@ static int rt6_fill_node(struct sk_buff *skb, struct rt6_info *rt,
2200 2196
2201 NLA_PUT_U32(skb, RTA_PRIORITY, rt->rt6i_metric); 2197 NLA_PUT_U32(skb, RTA_PRIORITY, rt->rt6i_metric);
2202 2198
2203 expires = rt->rt6i_expires ? rt->rt6i_expires - jiffies : 0; 2199 if (!(rt->rt6i_flags & RTF_EXPIRES))
2200 expires = 0;
2201 else if (rt->rt6i_expires - jiffies < INT_MAX)
2202 expires = rt->rt6i_expires - jiffies;
2203 else
2204 expires = INT_MAX;
2205
2204 if (rtnl_put_cacheinfo(skb, &rt->u.dst, 0, 0, 0, 2206 if (rtnl_put_cacheinfo(skb, &rt->u.dst, 0, 0, 0,
2205 expires, rt->u.dst.error) < 0) 2207 expires, rt->u.dst.error) < 0)
2206 goto nla_put_failure; 2208 goto nla_put_failure;
diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index 4b2f1033994e..32e871a6c25a 100644
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -222,15 +222,18 @@ __ipip6_tunnel_locate_prl(struct ip_tunnel *t, __be32 addr)
222 222
223} 223}
224 224
225static int ipip6_tunnel_get_prl(struct ip_tunnel *t, struct ip_tunnel_prl *a) 225static int ipip6_tunnel_get_prl(struct ip_tunnel *t,
226 struct ip_tunnel_prl __user *a)
226{ 227{
227 struct ip_tunnel_prl *kp; 228 struct ip_tunnel_prl kprl, *kp;
228 struct ip_tunnel_prl_entry *prl; 229 struct ip_tunnel_prl_entry *prl;
229 unsigned int cmax, c = 0, ca, len; 230 unsigned int cmax, c = 0, ca, len;
230 int ret = 0; 231 int ret = 0;
231 232
232 cmax = a->datalen / sizeof(*a); 233 if (copy_from_user(&kprl, a, sizeof(kprl)))
233 if (cmax > 1 && a->addr != htonl(INADDR_ANY)) 234 return -EFAULT;
235 cmax = kprl.datalen / sizeof(kprl);
236 if (cmax > 1 && kprl.addr != htonl(INADDR_ANY))
234 cmax = 1; 237 cmax = 1;
235 238
236 /* For simple GET or for root users, 239 /* For simple GET or for root users,
@@ -261,26 +264,25 @@ static int ipip6_tunnel_get_prl(struct ip_tunnel *t, struct ip_tunnel_prl *a)
261 for (prl = t->prl; prl; prl = prl->next) { 264 for (prl = t->prl; prl; prl = prl->next) {
262 if (c > cmax) 265 if (c > cmax)
263 break; 266 break;
264 if (a->addr != htonl(INADDR_ANY) && prl->addr != a->addr) 267 if (kprl.addr != htonl(INADDR_ANY) && prl->addr != kprl.addr)
265 continue; 268 continue;
266 kp[c].addr = prl->addr; 269 kp[c].addr = prl->addr;
267 kp[c].flags = prl->flags; 270 kp[c].flags = prl->flags;
268 c++; 271 c++;
269 if (a->addr != htonl(INADDR_ANY)) 272 if (kprl.addr != htonl(INADDR_ANY))
270 break; 273 break;
271 } 274 }
272out: 275out:
273 read_unlock(&ipip6_lock); 276 read_unlock(&ipip6_lock);
274 277
275 len = sizeof(*kp) * c; 278 len = sizeof(*kp) * c;
276 ret = len ? copy_to_user(a->data, kp, len) : 0; 279 ret = 0;
280 if ((len && copy_to_user(a + 1, kp, len)) || put_user(len, &a->datalen))
281 ret = -EFAULT;
277 282
278 kfree(kp); 283 kfree(kp);
279 if (ret)
280 return -EFAULT;
281 284
282 a->datalen = len; 285 return ret;
283 return 0;
284} 286}
285 287
286static int 288static int
@@ -403,9 +405,8 @@ static void ipip6_tunnel_uninit(struct net_device *dev)
403 405
404static int ipip6_err(struct sk_buff *skb, u32 info) 406static int ipip6_err(struct sk_buff *skb, u32 info)
405{ 407{
406#ifndef I_WISH_WORLD_WERE_PERFECT
407 408
408/* It is not :-( All the routers (except for Linux) return only 409/* All the routers (except for Linux) return only
409 8 bytes of packet payload. It means, that precise relaying of 410 8 bytes of packet payload. It means, that precise relaying of
410 ICMP in the real Internet is absolutely infeasible. 411 ICMP in the real Internet is absolutely infeasible.
411 */ 412 */
@@ -462,92 +463,6 @@ static int ipip6_err(struct sk_buff *skb, u32 info)
462out: 463out:
463 read_unlock(&ipip6_lock); 464 read_unlock(&ipip6_lock);
464 return err; 465 return err;
465#else
466 struct iphdr *iph = (struct iphdr*)dp;
467 int hlen = iph->ihl<<2;
468 struct ipv6hdr *iph6;
469 const int type = icmp_hdr(skb)->type;
470 const int code = icmp_hdr(skb)->code;
471 int rel_type = 0;
472 int rel_code = 0;
473 int rel_info = 0;
474 struct sk_buff *skb2;
475 struct rt6_info *rt6i;
476
477 if (len < hlen + sizeof(struct ipv6hdr))
478 return;
479 iph6 = (struct ipv6hdr*)(dp + hlen);
480
481 switch (type) {
482 default:
483 return;
484 case ICMP_PARAMETERPROB:
485 if (icmp_hdr(skb)->un.gateway < hlen)
486 return;
487
488 /* So... This guy found something strange INSIDE encapsulated
489 packet. Well, he is fool, but what can we do ?
490 */
491 rel_type = ICMPV6_PARAMPROB;
492 rel_info = icmp_hdr(skb)->un.gateway - hlen;
493 break;
494
495 case ICMP_DEST_UNREACH:
496 switch (code) {
497 case ICMP_SR_FAILED:
498 case ICMP_PORT_UNREACH:
499 /* Impossible event. */
500 return;
501 case ICMP_FRAG_NEEDED:
502 /* Too complicated case ... */
503 return;
504 default:
505 /* All others are translated to HOST_UNREACH.
506 rfc2003 contains "deep thoughts" about NET_UNREACH,
507 I believe, it is just ether pollution. --ANK
508 */
509 rel_type = ICMPV6_DEST_UNREACH;
510 rel_code = ICMPV6_ADDR_UNREACH;
511 break;
512 }
513 break;
514 case ICMP_TIME_EXCEEDED:
515 if (code != ICMP_EXC_TTL)
516 return;
517 rel_type = ICMPV6_TIME_EXCEED;
518 rel_code = ICMPV6_EXC_HOPLIMIT;
519 break;
520 }
521
522 /* Prepare fake skb to feed it to icmpv6_send */
523 skb2 = skb_clone(skb, GFP_ATOMIC);
524 if (skb2 == NULL)
525 return 0;
526 dst_release(skb2->dst);
527 skb2->dst = NULL;
528 skb_pull(skb2, skb->data - (u8*)iph6);
529 skb_reset_network_header(skb2);
530
531 /* Try to guess incoming interface */
532 rt6i = rt6_lookup(dev_net(skb->dev), &iph6->saddr, NULL, NULL, 0);
533 if (rt6i && rt6i->rt6i_dev) {
534 skb2->dev = rt6i->rt6i_dev;
535
536 rt6i = rt6_lookup(dev_net(skb->dev),
537 &iph6->daddr, &iph6->saddr, NULL, 0);
538
539 if (rt6i && rt6i->rt6i_dev && rt6i->rt6i_dev->type == ARPHRD_SIT) {
540 struct ip_tunnel *t = netdev_priv(rt6i->rt6i_dev);
541 if (rel_type == ICMPV6_TIME_EXCEED && t->parms.iph.ttl) {
542 rel_type = ICMPV6_DEST_UNREACH;
543 rel_code = ICMPV6_ADDR_UNREACH;
544 }
545 icmpv6_send(skb2, rel_type, rel_code, rel_info, skb2->dev);
546 }
547 }
548 kfree_skb(skb2);
549 return 0;
550#endif
551} 466}
552 467
553static inline void ipip6_ecn_decapsulate(struct iphdr *iph, struct sk_buff *skb) 468static inline void ipip6_ecn_decapsulate(struct iphdr *iph, struct sk_buff *skb)
@@ -596,9 +511,9 @@ static int ipip6_rcv(struct sk_buff *skb)
596 } 511 }
597 512
598 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_PORT_UNREACH, 0); 513 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_PORT_UNREACH, 0);
599 kfree_skb(skb);
600 read_unlock(&ipip6_lock); 514 read_unlock(&ipip6_lock);
601out: 515out:
516 kfree_skb(skb);
602 return 0; 517 return 0;
603} 518}
604 519
@@ -960,11 +875,20 @@ ipip6_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
960 break; 875 break;
961 876
962 case SIOCGETPRL: 877 case SIOCGETPRL:
878 err = -EINVAL;
879 if (dev == sitn->fb_tunnel_dev)
880 goto done;
881 err = -ENOENT;
882 if (!(t = netdev_priv(dev)))
883 goto done;
884 err = ipip6_tunnel_get_prl(t, ifr->ifr_ifru.ifru_data);
885 break;
886
963 case SIOCADDPRL: 887 case SIOCADDPRL:
964 case SIOCDELPRL: 888 case SIOCDELPRL:
965 case SIOCCHGPRL: 889 case SIOCCHGPRL:
966 err = -EPERM; 890 err = -EPERM;
967 if (cmd != SIOCGETPRL && !capable(CAP_NET_ADMIN)) 891 if (!capable(CAP_NET_ADMIN))
968 goto done; 892 goto done;
969 err = -EINVAL; 893 err = -EINVAL;
970 if (dev == sitn->fb_tunnel_dev) 894 if (dev == sitn->fb_tunnel_dev)
@@ -977,12 +901,6 @@ ipip6_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
977 goto done; 901 goto done;
978 902
979 switch (cmd) { 903 switch (cmd) {
980 case SIOCGETPRL:
981 err = ipip6_tunnel_get_prl(t, &prl);
982 if (!err && copy_to_user(ifr->ifr_ifru.ifru_data,
983 &prl, sizeof(prl)))
984 err = -EFAULT;
985 break;
986 case SIOCDELPRL: 904 case SIOCDELPRL:
987 err = ipip6_tunnel_del_prl(t, &prl); 905 err = ipip6_tunnel_del_prl(t, &prl);
988 break; 906 break;
@@ -991,8 +909,7 @@ ipip6_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
991 err = ipip6_tunnel_add_prl(t, &prl, cmd == SIOCCHGPRL); 909 err = ipip6_tunnel_add_prl(t, &prl, cmd == SIOCCHGPRL);
992 break; 910 break;
993 } 911 }
994 if (cmd != SIOCGETPRL) 912 netdev_state_change(dev);
995 netdev_state_change(dev);
996 break; 913 break;
997 914
998 default: 915 default:
diff --git a/net/ipv6/syncookies.c b/net/ipv6/syncookies.c
index 938ce4ecde55..3ecc1157994e 100644
--- a/net/ipv6/syncookies.c
+++ b/net/ipv6/syncookies.c
@@ -198,7 +198,6 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb)
198 ireq = inet_rsk(req); 198 ireq = inet_rsk(req);
199 ireq6 = inet6_rsk(req); 199 ireq6 = inet6_rsk(req);
200 treq = tcp_rsk(req); 200 treq = tcp_rsk(req);
201 ireq6->pktopts = NULL;
202 201
203 if (security_inet_conn_request(sk, skb, req)) { 202 if (security_inet_conn_request(sk, skb, req)) {
204 reqsk_free(req); 203 reqsk_free(req);
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 715965f0fac0..40ea9c36d24b 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -1299,7 +1299,6 @@ static int tcp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
1299 treq = inet6_rsk(req); 1299 treq = inet6_rsk(req);
1300 ipv6_addr_copy(&treq->rmt_addr, &ipv6_hdr(skb)->saddr); 1300 ipv6_addr_copy(&treq->rmt_addr, &ipv6_hdr(skb)->saddr);
1301 ipv6_addr_copy(&treq->loc_addr, &ipv6_hdr(skb)->daddr); 1301 ipv6_addr_copy(&treq->loc_addr, &ipv6_hdr(skb)->daddr);
1302 treq->pktopts = NULL;
1303 if (!want_cookie) 1302 if (!want_cookie)
1304 TCP_ECN_create_request(req, tcp_hdr(skb)); 1303 TCP_ECN_create_request(req, tcp_hdr(skb));
1305 1304
@@ -2037,7 +2036,7 @@ static void get_tcp6_sock(struct seq_file *seq, struct sock *sp, int i)
2037 2036
2038 seq_printf(seq, 2037 seq_printf(seq,
2039 "%4d: %08X%08X%08X%08X:%04X %08X%08X%08X%08X:%04X " 2038 "%4d: %08X%08X%08X%08X:%04X %08X%08X%08X%08X:%04X "
2040 "%02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %u %u %u %u %d\n", 2039 "%02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %lu %lu %u %u %d\n",
2041 i, 2040 i,
2042 src->s6_addr32[0], src->s6_addr32[1], 2041 src->s6_addr32[0], src->s6_addr32[1],
2043 src->s6_addr32[2], src->s6_addr32[3], srcp, 2042 src->s6_addr32[2], src->s6_addr32[3], srcp,
@@ -2053,8 +2052,8 @@ static void get_tcp6_sock(struct seq_file *seq, struct sock *sp, int i)
2053 icsk->icsk_probes_out, 2052 icsk->icsk_probes_out,
2054 sock_i_ino(sp), 2053 sock_i_ino(sp),
2055 atomic_read(&sp->sk_refcnt), sp, 2054 atomic_read(&sp->sk_refcnt), sp,
2056 icsk->icsk_rto, 2055 jiffies_to_clock_t(icsk->icsk_rto),
2057 icsk->icsk_ack.ato, 2056 jiffies_to_clock_t(icsk->icsk_ack.ato),
2058 (icsk->icsk_ack.quick << 1 ) | icsk->icsk_ack.pingpong, 2057 (icsk->icsk_ack.quick << 1 ) | icsk->icsk_ack.pingpong,
2059 tp->snd_cwnd, tp->snd_ssthresh>=0xFFFF?-1:tp->snd_ssthresh 2058 tp->snd_cwnd, tp->snd_ssthresh>=0xFFFF?-1:tp->snd_ssthresh
2060 ); 2059 );
diff --git a/net/ipv6/tunnel6.c b/net/ipv6/tunnel6.c
index 6323921b40be..669f280989c3 100644
--- a/net/ipv6/tunnel6.c
+++ b/net/ipv6/tunnel6.c
@@ -109,7 +109,7 @@ static int tunnel46_rcv(struct sk_buff *skb)
109{ 109{
110 struct xfrm6_tunnel *handler; 110 struct xfrm6_tunnel *handler;
111 111
112 if (!pskb_may_pull(skb, sizeof(struct ipv6hdr))) 112 if (!pskb_may_pull(skb, sizeof(struct iphdr)))
113 goto drop; 113 goto drop;
114 114
115 for (handler = tunnel46_handlers; handler; handler = handler->next) 115 for (handler = tunnel46_handlers; handler; handler = handler->next)
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index 1fd784f3e2ec..dd309626ae9a 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -534,7 +534,9 @@ static void udp_v6_flush_pending_frames(struct sock *sk)
534{ 534{
535 struct udp_sock *up = udp_sk(sk); 535 struct udp_sock *up = udp_sk(sk);
536 536
537 if (up->pending) { 537 if (up->pending == AF_INET)
538 udp_flush_pending_frames(sk);
539 else if (up->pending) {
538 up->len = 0; 540 up->len = 0;
539 up->pending = 0; 541 up->pending = 0;
540 ip6_flush_pending_frames(sk); 542 ip6_flush_pending_frames(sk);
@@ -731,7 +733,7 @@ do_udp_sendmsg:
731 memset(opt, 0, sizeof(struct ipv6_txoptions)); 733 memset(opt, 0, sizeof(struct ipv6_txoptions));
732 opt->tot_len = sizeof(*opt); 734 opt->tot_len = sizeof(*opt);
733 735
734 err = datagram_send_ctl(msg, &fl, opt, &hlimit, &tclass); 736 err = datagram_send_ctl(sock_net(sk), msg, &fl, opt, &hlimit, &tclass);
735 if (err < 0) { 737 if (err < 0) {
736 fl6_sock_release(flowlabel); 738 fl6_sock_release(flowlabel);
737 return err; 739 return err;
@@ -848,12 +850,14 @@ do_append_data:
848 } else { 850 } else {
849 dst_release(dst); 851 dst_release(dst);
850 } 852 }
853 dst = NULL;
851 } 854 }
852 855
853 if (err > 0) 856 if (err > 0)
854 err = np->recverr ? net_xmit_errno(err) : 0; 857 err = np->recverr ? net_xmit_errno(err) : 0;
855 release_sock(sk); 858 release_sock(sk);
856out: 859out:
860 dst_release(dst);
857 fl6_sock_release(flowlabel); 861 fl6_sock_release(flowlabel);
858 if (!err) 862 if (!err)
859 return len; 863 return len;
diff --git a/net/irda/af_irda.c b/net/irda/af_irda.c
index ae54b20d0470..3eb5bcc75f99 100644
--- a/net/irda/af_irda.c
+++ b/net/irda/af_irda.c
@@ -1093,11 +1093,6 @@ static int irda_create(struct net *net, struct socket *sock, int protocol)
1093 1093
1094 init_waitqueue_head(&self->query_wait); 1094 init_waitqueue_head(&self->query_wait);
1095 1095
1096 /* Initialise networking socket struct */
1097 sock_init_data(sock, sk); /* Note : set sk->sk_refcnt to 1 */
1098 sk->sk_family = PF_IRDA;
1099 sk->sk_protocol = protocol;
1100
1101 switch (sock->type) { 1096 switch (sock->type) {
1102 case SOCK_STREAM: 1097 case SOCK_STREAM:
1103 sock->ops = &irda_stream_ops; 1098 sock->ops = &irda_stream_ops;
@@ -1124,13 +1119,20 @@ static int irda_create(struct net *net, struct socket *sock, int protocol)
1124 self->max_sdu_size_rx = TTP_SAR_UNBOUND; 1119 self->max_sdu_size_rx = TTP_SAR_UNBOUND;
1125 break; 1120 break;
1126 default: 1121 default:
1122 sk_free(sk);
1127 return -ESOCKTNOSUPPORT; 1123 return -ESOCKTNOSUPPORT;
1128 } 1124 }
1129 break; 1125 break;
1130 default: 1126 default:
1127 sk_free(sk);
1131 return -ESOCKTNOSUPPORT; 1128 return -ESOCKTNOSUPPORT;
1132 } 1129 }
1133 1130
1131 /* Initialise networking socket struct */
1132 sock_init_data(sock, sk); /* Note : set sk->sk_refcnt to 1 */
1133 sk->sk_family = PF_IRDA;
1134 sk->sk_protocol = protocol;
1135
1134 /* Register as a client with IrLMP */ 1136 /* Register as a client with IrLMP */
1135 self->ckey = irlmp_register_client(0, NULL, NULL, NULL); 1137 self->ckey = irlmp_register_client(0, NULL, NULL, NULL);
1136 self->mask.word = 0xffff; 1138 self->mask.word = 0xffff;
diff --git a/net/irda/discovery.c b/net/irda/discovery.c
index bfacef8b76f4..a6f99b5a1499 100644
--- a/net/irda/discovery.c
+++ b/net/irda/discovery.c
@@ -40,6 +40,8 @@
40 40
41#include <net/irda/discovery.h> 41#include <net/irda/discovery.h>
42 42
43#include <asm/unaligned.h>
44
43/* 45/*
44 * Function irlmp_add_discovery (cachelog, discovery) 46 * Function irlmp_add_discovery (cachelog, discovery)
45 * 47 *
@@ -87,7 +89,7 @@ void irlmp_add_discovery(hashbin_t *cachelog, discovery_t *new)
87 */ 89 */
88 hashbin_remove_this(cachelog, (irda_queue_t *) node); 90 hashbin_remove_this(cachelog, (irda_queue_t *) node);
89 /* Check if hints bits are unchanged */ 91 /* Check if hints bits are unchanged */
90 if(u16ho(node->data.hints) == u16ho(new->data.hints)) 92 if (get_unaligned((__u16 *)node->data.hints) == get_unaligned((__u16 *)new->data.hints))
91 /* Set time of first discovery for this node */ 93 /* Set time of first discovery for this node */
92 new->firststamp = node->firststamp; 94 new->firststamp = node->firststamp;
93 kfree(node); 95 kfree(node);
@@ -281,9 +283,9 @@ struct irda_device_info *irlmp_copy_discoveries(hashbin_t *log, int *pn,
281 /* Mask out the ones we don't want : 283 /* Mask out the ones we don't want :
282 * We want to match the discovery mask, and to get only 284 * We want to match the discovery mask, and to get only
283 * the most recent one (unless we want old ones) */ 285 * the most recent one (unless we want old ones) */
284 if ((u16ho(discovery->data.hints) & mask) && 286 if ((get_unaligned((__u16 *)discovery->data.hints) & mask) &&
285 ((old_entries) || 287 ((old_entries) ||
286 ((jiffies - discovery->firststamp) < j_timeout)) ) { 288 ((jiffies - discovery->firststamp) < j_timeout))) {
287 /* Create buffer as needed. 289 /* Create buffer as needed.
288 * As this function get called a lot and most time 290 * As this function get called a lot and most time
289 * we don't have anything to put in the log (we are 291 * we don't have anything to put in the log (we are
diff --git a/net/irda/iriap.c b/net/irda/iriap.c
index 9e15c82960fe..4a105dc32dcd 100644
--- a/net/irda/iriap.c
+++ b/net/irda/iriap.c
@@ -451,12 +451,14 @@ static void iriap_getvaluebyclass_confirm(struct iriap_cb *self,
451 n = 2; 451 n = 2;
452 452
453 /* Get length, MSB first */ 453 /* Get length, MSB first */
454 len = be16_to_cpu(get_unaligned((__be16 *)(fp+n))); n += 2; 454 len = get_unaligned_be16(fp + n);
455 n += 2;
455 456
456 IRDA_DEBUG(4, "%s(), len=%d\n", __func__, len); 457 IRDA_DEBUG(4, "%s(), len=%d\n", __func__, len);
457 458
458 /* Get object ID, MSB first */ 459 /* Get object ID, MSB first */
459 obj_id = be16_to_cpu(get_unaligned((__be16 *)(fp+n))); n += 2; 460 obj_id = get_unaligned_be16(fp + n);
461 n += 2;
460 462
461 type = fp[n++]; 463 type = fp[n++];
462 IRDA_DEBUG(4, "%s(), Value type = %d\n", __func__, type); 464 IRDA_DEBUG(4, "%s(), Value type = %d\n", __func__, type);
@@ -506,7 +508,7 @@ static void iriap_getvaluebyclass_confirm(struct iriap_cb *self,
506 value = irias_new_string_value(fp+n); 508 value = irias_new_string_value(fp+n);
507 break; 509 break;
508 case IAS_OCT_SEQ: 510 case IAS_OCT_SEQ:
509 value_len = be16_to_cpu(get_unaligned((__be16 *)(fp+n))); 511 value_len = get_unaligned_be16(fp + n);
510 n += 2; 512 n += 2;
511 513
512 /* Will truncate to IAS_MAX_OCTET_STRING bytes */ 514 /* Will truncate to IAS_MAX_OCTET_STRING bytes */
diff --git a/net/irda/irlmp.c b/net/irda/irlmp.c
index 1f81f8e7c61d..7bf5b913828b 100644
--- a/net/irda/irlmp.c
+++ b/net/irda/irlmp.c
@@ -1062,7 +1062,8 @@ void irlmp_discovery_expiry(discinfo_t *expiries, int number)
1062 for(i = 0; i < number; i++) { 1062 for(i = 0; i < number; i++) {
1063 /* Check if we should notify client */ 1063 /* Check if we should notify client */
1064 if ((client->expir_callback) && 1064 if ((client->expir_callback) &&
1065 (client->hint_mask.word & u16ho(expiries[i].hints) 1065 (client->hint_mask.word &
1066 get_unaligned((__u16 *)expiries[i].hints)
1066 & 0x7f7f) ) 1067 & 0x7f7f) )
1067 client->expir_callback(&(expiries[i]), 1068 client->expir_callback(&(expiries[i]),
1068 EXPIRY_TIMEOUT, 1069 EXPIRY_TIMEOUT,
@@ -1086,7 +1087,7 @@ discovery_t *irlmp_get_discovery_response(void)
1086 1087
1087 IRDA_ASSERT(irlmp != NULL, return NULL;); 1088 IRDA_ASSERT(irlmp != NULL, return NULL;);
1088 1089
1089 u16ho(irlmp->discovery_rsp.data.hints) = irlmp->hints.word; 1090 put_unaligned(irlmp->hints.word, (__u16 *)irlmp->discovery_rsp.data.hints);
1090 1091
1091 /* 1092 /*
1092 * Set character set for device name (we use ASCII), and 1093 * Set character set for device name (we use ASCII), and
diff --git a/net/irda/irnet/irnet_irda.c b/net/irda/irnet/irnet_irda.c
index 75497e55927d..cf9a4b531a98 100644
--- a/net/irda/irnet/irnet_irda.c
+++ b/net/irda/irnet/irnet_irda.c
@@ -10,6 +10,7 @@
10 10
11#include "irnet_irda.h" /* Private header */ 11#include "irnet_irda.h" /* Private header */
12#include <linux/seq_file.h> 12#include <linux/seq_file.h>
13#include <asm/unaligned.h>
13 14
14/* 15/*
15 * PPP disconnect work: we need to make sure we're in 16 * PPP disconnect work: we need to make sure we're in
@@ -1673,7 +1674,7 @@ irnet_discovery_indication(discinfo_t * discovery,
1673 /* Notify the control channel */ 1674 /* Notify the control channel */
1674 irnet_post_event(NULL, IRNET_DISCOVER, 1675 irnet_post_event(NULL, IRNET_DISCOVER,
1675 discovery->saddr, discovery->daddr, discovery->info, 1676 discovery->saddr, discovery->daddr, discovery->info,
1676 u16ho(discovery->hints)); 1677 get_unaligned((__u16 *)discovery->hints));
1677 1678
1678 DEXIT(IRDA_OCB_TRACE, "\n"); 1679 DEXIT(IRDA_OCB_TRACE, "\n");
1679} 1680}
@@ -1704,7 +1705,7 @@ irnet_expiry_indication(discinfo_t * expiry,
1704 /* Notify the control channel */ 1705 /* Notify the control channel */
1705 irnet_post_event(NULL, IRNET_EXPIRE, 1706 irnet_post_event(NULL, IRNET_EXPIRE,
1706 expiry->saddr, expiry->daddr, expiry->info, 1707 expiry->saddr, expiry->daddr, expiry->info,
1707 u16ho(expiry->hints)); 1708 get_unaligned((__u16 *)expiry->hints));
1708 1709
1709 DEXIT(IRDA_OCB_TRACE, "\n"); 1710 DEXIT(IRDA_OCB_TRACE, "\n");
1710} 1711}
diff --git a/net/irda/irnetlink.c b/net/irda/irnetlink.c
index 9e1fb82e3220..2f05ec1037ab 100644
--- a/net/irda/irnetlink.c
+++ b/net/irda/irnetlink.c
@@ -101,8 +101,8 @@ static int irda_nl_get_mode(struct sk_buff *skb, struct genl_info *info)
101 101
102 hdr = genlmsg_put(msg, info->snd_pid, info->snd_seq, 102 hdr = genlmsg_put(msg, info->snd_pid, info->snd_seq,
103 &irda_nl_family, 0, IRDA_NL_CMD_GET_MODE); 103 &irda_nl_family, 0, IRDA_NL_CMD_GET_MODE);
104 if (IS_ERR(hdr)) { 104 if (hdr == NULL) {
105 ret = PTR_ERR(hdr); 105 ret = -EMSGSIZE;
106 goto err_out; 106 goto err_out;
107 } 107 }
108 108
diff --git a/net/key/af_key.c b/net/key/af_key.c
index 9e7236ff6bcc..7470e367272b 100644
--- a/net/key/af_key.c
+++ b/net/key/af_key.c
@@ -1251,7 +1251,7 @@ static struct xfrm_state * pfkey_msg2xfrm_state(struct sadb_msg *hdr,
1251 x->sel.prefixlen_s = addr->sadb_address_prefixlen; 1251 x->sel.prefixlen_s = addr->sadb_address_prefixlen;
1252 } 1252 }
1253 1253
1254 if (x->props.mode == XFRM_MODE_TRANSPORT) 1254 if (!x->sel.family)
1255 x->sel.family = x->props.family; 1255 x->sel.family = x->props.family;
1256 1256
1257 if (ext_hdrs[SADB_X_EXT_NAT_T_TYPE-1]) { 1257 if (ext_hdrs[SADB_X_EXT_NAT_T_TYPE-1]) {
@@ -3030,6 +3030,9 @@ static int key_notify_sa_expire(struct xfrm_state *x, struct km_event *c)
3030 3030
3031static int pfkey_send_notify(struct xfrm_state *x, struct km_event *c) 3031static int pfkey_send_notify(struct xfrm_state *x, struct km_event *c)
3032{ 3032{
3033 if (atomic_read(&pfkey_socks_nr) == 0)
3034 return 0;
3035
3033 switch (c->event) { 3036 switch (c->event) {
3034 case XFRM_MSG_EXPIRE: 3037 case XFRM_MSG_EXPIRE:
3035 return key_notify_sa_expire(x, c); 3038 return key_notify_sa_expire(x, c);
diff --git a/net/llc/llc_sap.c b/net/llc/llc_sap.c
index e2ddde755019..008de1fc42ca 100644
--- a/net/llc/llc_sap.c
+++ b/net/llc/llc_sap.c
@@ -286,12 +286,14 @@ void llc_build_and_send_xid_pkt(struct llc_sap *sap, struct sk_buff *skb,
286 * 286 *
287 * Sends received pdus to the sap state machine. 287 * Sends received pdus to the sap state machine.
288 */ 288 */
289static void llc_sap_rcv(struct llc_sap *sap, struct sk_buff *skb) 289static void llc_sap_rcv(struct llc_sap *sap, struct sk_buff *skb,
290 struct sock *sk)
290{ 291{
291 struct llc_sap_state_ev *ev = llc_sap_ev(skb); 292 struct llc_sap_state_ev *ev = llc_sap_ev(skb);
292 293
293 ev->type = LLC_SAP_EV_TYPE_PDU; 294 ev->type = LLC_SAP_EV_TYPE_PDU;
294 ev->reason = 0; 295 ev->reason = 0;
296 skb->sk = sk;
295 llc_sap_state_process(sap, skb); 297 llc_sap_state_process(sap, skb);
296} 298}
297 299
@@ -360,8 +362,7 @@ static void llc_sap_mcast(struct llc_sap *sap,
360 break; 362 break;
361 363
362 sock_hold(sk); 364 sock_hold(sk);
363 skb_set_owner_r(skb1, sk); 365 llc_sap_rcv(sap, skb1, sk);
364 llc_sap_rcv(sap, skb1);
365 sock_put(sk); 366 sock_put(sk);
366 } 367 }
367 read_unlock_bh(&sap->sk_list.lock); 368 read_unlock_bh(&sap->sk_list.lock);
@@ -381,8 +382,7 @@ void llc_sap_handler(struct llc_sap *sap, struct sk_buff *skb)
381 } else { 382 } else {
382 struct sock *sk = llc_lookup_dgram(sap, &laddr); 383 struct sock *sk = llc_lookup_dgram(sap, &laddr);
383 if (sk) { 384 if (sk) {
384 skb_set_owner_r(skb, sk); 385 llc_sap_rcv(sap, skb, sk);
385 llc_sap_rcv(sap, skb);
386 sock_put(sk); 386 sock_put(sk);
387 } else 387 } else
388 kfree_skb(skb); 388 kfree_skb(skb);
diff --git a/net/mac80211/cfg.c b/net/mac80211/cfg.c
index 699d97b8de5e..a9fce4afdf21 100644
--- a/net/mac80211/cfg.c
+++ b/net/mac80211/cfg.c
@@ -672,7 +672,7 @@ static int ieee80211_add_station(struct wiphy *wiphy, struct net_device *dev,
672 if (params->vlan) { 672 if (params->vlan) {
673 sdata = IEEE80211_DEV_TO_SUB_IF(params->vlan); 673 sdata = IEEE80211_DEV_TO_SUB_IF(params->vlan);
674 674
675 if (sdata->vif.type != IEEE80211_IF_TYPE_VLAN || 675 if (sdata->vif.type != IEEE80211_IF_TYPE_VLAN &&
676 sdata->vif.type != IEEE80211_IF_TYPE_AP) 676 sdata->vif.type != IEEE80211_IF_TYPE_AP)
677 return -EINVAL; 677 return -EINVAL;
678 } else 678 } else
@@ -760,7 +760,7 @@ static int ieee80211_change_station(struct wiphy *wiphy,
760 if (params->vlan && params->vlan != sta->sdata->dev) { 760 if (params->vlan && params->vlan != sta->sdata->dev) {
761 vlansdata = IEEE80211_DEV_TO_SUB_IF(params->vlan); 761 vlansdata = IEEE80211_DEV_TO_SUB_IF(params->vlan);
762 762
763 if (vlansdata->vif.type != IEEE80211_IF_TYPE_VLAN || 763 if (vlansdata->vif.type != IEEE80211_IF_TYPE_VLAN &&
764 vlansdata->vif.type != IEEE80211_IF_TYPE_AP) { 764 vlansdata->vif.type != IEEE80211_IF_TYPE_AP) {
765 rcu_read_unlock(); 765 rcu_read_unlock();
766 return -EINVAL; 766 return -EINVAL;
diff --git a/net/mac80211/debugfs_key.c b/net/mac80211/debugfs_key.c
index 879e7210458a..19efc3a6a932 100644
--- a/net/mac80211/debugfs_key.c
+++ b/net/mac80211/debugfs_key.c
@@ -255,14 +255,23 @@ void ieee80211_debugfs_key_remove(struct ieee80211_key *key)
255void ieee80211_debugfs_key_add_default(struct ieee80211_sub_if_data *sdata) 255void ieee80211_debugfs_key_add_default(struct ieee80211_sub_if_data *sdata)
256{ 256{
257 char buf[50]; 257 char buf[50];
258 struct ieee80211_key *key;
258 259
259 if (!sdata->debugfsdir) 260 if (!sdata->debugfsdir)
260 return; 261 return;
261 262
262 sprintf(buf, "../keys/%d", sdata->default_key->debugfs.cnt); 263 /* this is running under the key lock */
263 sdata->debugfs.default_key = 264
264 debugfs_create_symlink("default_key", sdata->debugfsdir, buf); 265 key = sdata->default_key;
266 if (key) {
267 sprintf(buf, "../keys/%d", key->debugfs.cnt);
268 sdata->debugfs.default_key =
269 debugfs_create_symlink("default_key",
270 sdata->debugfsdir, buf);
271 } else
272 ieee80211_debugfs_key_remove_default(sdata);
265} 273}
274
266void ieee80211_debugfs_key_remove_default(struct ieee80211_sub_if_data *sdata) 275void ieee80211_debugfs_key_remove_default(struct ieee80211_sub_if_data *sdata)
267{ 276{
268 if (!sdata) 277 if (!sdata)
diff --git a/net/mac80211/ieee80211_i.h b/net/mac80211/ieee80211_i.h
index 8e53ce7ed444..006486b26726 100644
--- a/net/mac80211/ieee80211_i.h
+++ b/net/mac80211/ieee80211_i.h
@@ -354,7 +354,7 @@ struct ieee80211_if_sta {
354 int preq_queue_len; 354 int preq_queue_len;
355 struct mesh_stats mshstats; 355 struct mesh_stats mshstats;
356 struct mesh_config mshcfg; 356 struct mesh_config mshcfg;
357 u8 mesh_seqnum[3]; 357 u32 mesh_seqnum;
358 bool accepting_plinks; 358 bool accepting_plinks;
359#endif 359#endif
360 u16 aid; 360 u16 aid;
@@ -899,7 +899,7 @@ extern const struct iw_handler_def ieee80211_iw_handler_def;
899 899
900 900
901/* ieee80211_ioctl.c */ 901/* ieee80211_ioctl.c */
902int ieee80211_set_freq(struct ieee80211_local *local, int freq); 902int ieee80211_set_freq(struct net_device *dev, int freq);
903/* ieee80211_sta.c */ 903/* ieee80211_sta.c */
904void ieee80211_sta_timer(unsigned long data); 904void ieee80211_sta_timer(unsigned long data);
905void ieee80211_sta_work(struct work_struct *work); 905void ieee80211_sta_work(struct work_struct *work);
diff --git a/net/mac80211/iface.c b/net/mac80211/iface.c
index 80954a512185..06e88a5a036d 100644
--- a/net/mac80211/iface.c
+++ b/net/mac80211/iface.c
@@ -54,6 +54,15 @@ int ieee80211_if_add(struct net_device *dev, const char *name,
54 if (!ndev) 54 if (!ndev)
55 return -ENOMEM; 55 return -ENOMEM;
56 56
57 ndev->needed_headroom = local->tx_headroom +
58 4*6 /* four MAC addresses */
59 + 2 + 2 + 2 + 2 /* ctl, dur, seq, qos */
60 + 6 /* mesh */
61 + 8 /* rfc1042/bridge tunnel */
62 - ETH_HLEN /* ethernet hard_header_len */
63 + IEEE80211_ENCRYPT_HEADROOM;
64 ndev->needed_tailroom = IEEE80211_ENCRYPT_TAILROOM;
65
57 ret = dev_alloc_name(ndev, ndev->name); 66 ret = dev_alloc_name(ndev, ndev->name);
58 if (ret < 0) 67 if (ret < 0)
59 goto fail; 68 goto fail;
diff --git a/net/mac80211/key.c b/net/mac80211/key.c
index 150d66dbda9d..220e83be3ef4 100644
--- a/net/mac80211/key.c
+++ b/net/mac80211/key.c
@@ -380,6 +380,15 @@ void ieee80211_key_free(struct ieee80211_key *key)
380 if (!key) 380 if (!key)
381 return; 381 return;
382 382
383 if (!key->sdata) {
384 /* The key has not been linked yet, simply free it
385 * and don't Oops */
386 if (key->conf.alg == ALG_CCMP)
387 ieee80211_aes_key_free(key->u.ccmp.tfm);
388 kfree(key);
389 return;
390 }
391
383 spin_lock_irqsave(&key->sdata->local->key_lock, flags); 392 spin_lock_irqsave(&key->sdata->local->key_lock, flags);
384 __ieee80211_key_free(key); 393 __ieee80211_key_free(key);
385 spin_unlock_irqrestore(&key->sdata->local->key_lock, flags); 394 spin_unlock_irqrestore(&key->sdata->local->key_lock, flags);
diff --git a/net/mac80211/main.c b/net/mac80211/main.c
index e9a978979d38..df0836ff1a20 100644
--- a/net/mac80211/main.c
+++ b/net/mac80211/main.c
@@ -255,22 +255,8 @@ static int ieee80211_open(struct net_device *dev)
255 255
256 switch (sdata->vif.type) { 256 switch (sdata->vif.type) {
257 case IEEE80211_IF_TYPE_WDS: 257 case IEEE80211_IF_TYPE_WDS:
258 if (is_zero_ether_addr(sdata->u.wds.remote_addr)) 258 if (!is_valid_ether_addr(sdata->u.wds.remote_addr))
259 return -ENOLINK; 259 return -ENOLINK;
260
261 /* Create STA entry for the WDS peer */
262 sta = sta_info_alloc(sdata, sdata->u.wds.remote_addr,
263 GFP_KERNEL);
264 if (!sta)
265 return -ENOMEM;
266
267 sta->flags |= WLAN_STA_AUTHORIZED;
268
269 res = sta_info_insert(sta);
270 if (res) {
271 /* STA has been freed */
272 return res;
273 }
274 break; 260 break;
275 case IEEE80211_IF_TYPE_VLAN: 261 case IEEE80211_IF_TYPE_VLAN:
276 if (!sdata->u.vlan.ap) 262 if (!sdata->u.vlan.ap)
@@ -337,10 +323,8 @@ static int ieee80211_open(struct net_device *dev)
337 conf.type = sdata->vif.type; 323 conf.type = sdata->vif.type;
338 conf.mac_addr = dev->dev_addr; 324 conf.mac_addr = dev->dev_addr;
339 res = local->ops->add_interface(local_to_hw(local), &conf); 325 res = local->ops->add_interface(local_to_hw(local), &conf);
340 if (res && !local->open_count && local->ops->stop)
341 local->ops->stop(local_to_hw(local));
342 if (res) 326 if (res)
343 return res; 327 goto err_stop;
344 328
345 ieee80211_if_config(dev); 329 ieee80211_if_config(dev);
346 ieee80211_reset_erp_info(dev); 330 ieee80211_reset_erp_info(dev);
@@ -353,9 +337,29 @@ static int ieee80211_open(struct net_device *dev)
353 netif_carrier_on(dev); 337 netif_carrier_on(dev);
354 } 338 }
355 339
340 if (sdata->vif.type == IEEE80211_IF_TYPE_WDS) {
341 /* Create STA entry for the WDS peer */
342 sta = sta_info_alloc(sdata, sdata->u.wds.remote_addr,
343 GFP_KERNEL);
344 if (!sta) {
345 res = -ENOMEM;
346 goto err_del_interface;
347 }
348
349 sta->flags |= WLAN_STA_AUTHORIZED;
350
351 res = sta_info_insert(sta);
352 if (res) {
353 /* STA has been freed */
354 goto err_del_interface;
355 }
356 }
357
356 if (local->open_count == 0) { 358 if (local->open_count == 0) {
357 res = dev_open(local->mdev); 359 res = dev_open(local->mdev);
358 WARN_ON(res); 360 WARN_ON(res);
361 if (res)
362 goto err_del_interface;
359 tasklet_enable(&local->tx_pending_tasklet); 363 tasklet_enable(&local->tx_pending_tasklet);
360 tasklet_enable(&local->tasklet); 364 tasklet_enable(&local->tasklet);
361 } 365 }
@@ -390,6 +394,12 @@ static int ieee80211_open(struct net_device *dev)
390 netif_start_queue(dev); 394 netif_start_queue(dev);
391 395
392 return 0; 396 return 0;
397 err_del_interface:
398 local->ops->remove_interface(local_to_hw(local), &conf);
399 err_stop:
400 if (!local->open_count && local->ops->stop)
401 local->ops->stop(local_to_hw(local));
402 return res;
393} 403}
394 404
395static int ieee80211_stop(struct net_device *dev) 405static int ieee80211_stop(struct net_device *dev)
@@ -501,6 +511,7 @@ static int ieee80211_stop(struct net_device *dev)
501 case IEEE80211_IF_TYPE_STA: 511 case IEEE80211_IF_TYPE_STA:
502 case IEEE80211_IF_TYPE_IBSS: 512 case IEEE80211_IF_TYPE_IBSS:
503 sdata->u.sta.state = IEEE80211_DISABLED; 513 sdata->u.sta.state = IEEE80211_DISABLED;
514 memset(sdata->u.sta.bssid, 0, ETH_ALEN);
504 del_timer_sync(&sdata->u.sta.timer); 515 del_timer_sync(&sdata->u.sta.timer);
505 /* 516 /*
506 * When we get here, the interface is marked down. 517 * When we get here, the interface is marked down.
@@ -519,8 +530,6 @@ static int ieee80211_stop(struct net_device *dev)
519 local->sta_hw_scanning = 0; 530 local->sta_hw_scanning = 0;
520 } 531 }
521 532
522 flush_workqueue(local->hw.workqueue);
523
524 sdata->u.sta.flags &= ~IEEE80211_STA_PRIVACY_INVOKED; 533 sdata->u.sta.flags &= ~IEEE80211_STA_PRIVACY_INVOKED;
525 kfree(sdata->u.sta.extra_ie); 534 kfree(sdata->u.sta.extra_ie);
526 sdata->u.sta.extra_ie = NULL; 535 sdata->u.sta.extra_ie = NULL;
@@ -544,6 +553,8 @@ static int ieee80211_stop(struct net_device *dev)
544 553
545 ieee80211_led_radio(local, 0); 554 ieee80211_led_radio(local, 0);
546 555
556 flush_workqueue(local->hw.workqueue);
557
547 tasklet_disable(&local->tx_pending_tasklet); 558 tasklet_disable(&local->tx_pending_tasklet);
548 tasklet_disable(&local->tasklet); 559 tasklet_disable(&local->tasklet);
549 } 560 }
@@ -975,6 +986,7 @@ static int __ieee80211_if_config(struct net_device *dev,
975 conf.ssid_len = sdata->u.sta.ssid_len; 986 conf.ssid_len = sdata->u.sta.ssid_len;
976 } else if (ieee80211_vif_is_mesh(&sdata->vif)) { 987 } else if (ieee80211_vif_is_mesh(&sdata->vif)) {
977 conf.beacon = beacon; 988 conf.beacon = beacon;
989 conf.beacon_control = control;
978 ieee80211_start_mesh(dev); 990 ieee80211_start_mesh(dev);
979 } else if (sdata->vif.type == IEEE80211_IF_TYPE_AP) { 991 } else if (sdata->vif.type == IEEE80211_IF_TYPE_AP) {
980 conf.ssid = sdata->u.ap.ssid; 992 conf.ssid = sdata->u.ap.ssid;
@@ -1302,7 +1314,7 @@ static void ieee80211_handle_filtered_frame(struct ieee80211_local *local,
1302 /* 1314 /*
1303 * Clear the TX filter mask for this STA when sending the next 1315 * Clear the TX filter mask for this STA when sending the next
1304 * packet. If the STA went to power save mode, this will happen 1316 * packet. If the STA went to power save mode, this will happen
1305 * happen when it wakes up for the next time. 1317 * when it wakes up for the next time.
1306 */ 1318 */
1307 sta->flags |= WLAN_STA_CLEAR_PS_FILT; 1319 sta->flags |= WLAN_STA_CLEAR_PS_FILT;
1308 1320
@@ -1755,6 +1767,7 @@ fail_wep:
1755fail_rate: 1767fail_rate:
1756 ieee80211_debugfs_remove_netdev(IEEE80211_DEV_TO_SUB_IF(local->mdev)); 1768 ieee80211_debugfs_remove_netdev(IEEE80211_DEV_TO_SUB_IF(local->mdev));
1757 unregister_netdevice(local->mdev); 1769 unregister_netdevice(local->mdev);
1770 local->mdev = NULL;
1758fail_dev: 1771fail_dev:
1759 rtnl_unlock(); 1772 rtnl_unlock();
1760 sta_info_stop(local); 1773 sta_info_stop(local);
@@ -1762,8 +1775,10 @@ fail_sta_info:
1762 debugfs_hw_del(local); 1775 debugfs_hw_del(local);
1763 destroy_workqueue(local->hw.workqueue); 1776 destroy_workqueue(local->hw.workqueue);
1764fail_workqueue: 1777fail_workqueue:
1765 ieee80211_if_free(local->mdev); 1778 if (local->mdev != NULL) {
1766 local->mdev = NULL; 1779 ieee80211_if_free(local->mdev);
1780 local->mdev = NULL;
1781 }
1767fail_mdev_alloc: 1782fail_mdev_alloc:
1768 wiphy_unregister(local->hw.wiphy); 1783 wiphy_unregister(local->hw.wiphy);
1769 return result; 1784 return result;
diff --git a/net/mac80211/mesh.c b/net/mac80211/mesh.c
index 594a3356a508..697ef67f96b6 100644
--- a/net/mac80211/mesh.c
+++ b/net/mac80211/mesh.c
@@ -8,6 +8,7 @@
8 * published by the Free Software Foundation. 8 * published by the Free Software Foundation.
9 */ 9 */
10 10
11#include <asm/unaligned.h>
11#include "ieee80211_i.h" 12#include "ieee80211_i.h"
12#include "mesh.h" 13#include "mesh.h"
13 14
@@ -167,8 +168,8 @@ int mesh_rmc_check(u8 *sa, struct ieee80211s_hdr *mesh_hdr,
167 struct rmc_entry *p, *n; 168 struct rmc_entry *p, *n;
168 169
169 /* Don't care about endianness since only match matters */ 170 /* Don't care about endianness since only match matters */
170 memcpy(&seqnum, mesh_hdr->seqnum, sizeof(mesh_hdr->seqnum)); 171 memcpy(&seqnum, &mesh_hdr->seqnum, sizeof(mesh_hdr->seqnum));
171 idx = mesh_hdr->seqnum[0] & rmc->idx_mask; 172 idx = le32_to_cpu(mesh_hdr->seqnum) & rmc->idx_mask;
172 list_for_each_entry_safe(p, n, &rmc->bucket[idx].list, list) { 173 list_for_each_entry_safe(p, n, &rmc->bucket[idx].list, list) {
173 ++entries; 174 ++entries;
174 if (time_after(jiffies, p->exp_time) || 175 if (time_after(jiffies, p->exp_time) ||
@@ -393,18 +394,10 @@ int ieee80211_new_mesh_header(struct ieee80211s_hdr *meshhdr,
393{ 394{
394 meshhdr->flags = 0; 395 meshhdr->flags = 0;
395 meshhdr->ttl = sdata->u.sta.mshcfg.dot11MeshTTL; 396 meshhdr->ttl = sdata->u.sta.mshcfg.dot11MeshTTL;
397 put_unaligned(cpu_to_le32(sdata->u.sta.mesh_seqnum), &meshhdr->seqnum);
398 sdata->u.sta.mesh_seqnum++;
396 399
397 meshhdr->seqnum[0] = sdata->u.sta.mesh_seqnum[0]++; 400 return 6;
398 meshhdr->seqnum[1] = sdata->u.sta.mesh_seqnum[1];
399 meshhdr->seqnum[2] = sdata->u.sta.mesh_seqnum[2];
400
401 if (sdata->u.sta.mesh_seqnum[0] == 0) {
402 sdata->u.sta.mesh_seqnum[1]++;
403 if (sdata->u.sta.mesh_seqnum[1] == 0)
404 sdata->u.sta.mesh_seqnum[2]++;
405 }
406
407 return 5;
408} 401}
409 402
410void ieee80211_mesh_init_sdata(struct ieee80211_sub_if_data *sdata) 403void ieee80211_mesh_init_sdata(struct ieee80211_sub_if_data *sdata)
diff --git a/net/mac80211/mesh.h b/net/mac80211/mesh.h
index 9ee3affab346..2e161f6d8288 100644
--- a/net/mac80211/mesh.h
+++ b/net/mac80211/mesh.h
@@ -140,7 +140,7 @@ struct rmc_entry {
140 140
141struct mesh_rmc { 141struct mesh_rmc {
142 struct rmc_entry bucket[RMC_BUCKETS]; 142 struct rmc_entry bucket[RMC_BUCKETS];
143 u8 idx_mask; 143 u32 idx_mask;
144}; 144};
145 145
146 146
diff --git a/net/mac80211/mesh_hwmp.c b/net/mac80211/mesh_hwmp.c
index 3df809222d1c..af0cd1e3e213 100644
--- a/net/mac80211/mesh_hwmp.c
+++ b/net/mac80211/mesh_hwmp.c
@@ -120,7 +120,7 @@ static int mesh_path_sel_frame_tx(enum mpath_frame_type action, u8 flags,
120 *pos++ = WLAN_EID_PREP; 120 *pos++ = WLAN_EID_PREP;
121 break; 121 break;
122 default: 122 default:
123 kfree(skb); 123 kfree_skb(skb);
124 return -ENOTSUPP; 124 return -ENOTSUPP;
125 break; 125 break;
126 } 126 }
diff --git a/net/mac80211/mesh_pathtbl.c b/net/mac80211/mesh_pathtbl.c
index 5845dc21ce85..99c2d360888e 100644
--- a/net/mac80211/mesh_pathtbl.c
+++ b/net/mac80211/mesh_pathtbl.c
@@ -158,19 +158,25 @@ int mesh_path_add(u8 *dst, struct net_device *dev)
158 if (atomic_add_unless(&sdata->u.sta.mpaths, 1, MESH_MAX_MPATHS) == 0) 158 if (atomic_add_unless(&sdata->u.sta.mpaths, 1, MESH_MAX_MPATHS) == 0)
159 return -ENOSPC; 159 return -ENOSPC;
160 160
161 read_lock(&pathtbl_resize_lock);
162
163 new_mpath = kzalloc(sizeof(struct mesh_path), GFP_KERNEL); 161 new_mpath = kzalloc(sizeof(struct mesh_path), GFP_KERNEL);
164 if (!new_mpath) { 162 if (!new_mpath) {
165 atomic_dec(&sdata->u.sta.mpaths); 163 atomic_dec(&sdata->u.sta.mpaths);
166 err = -ENOMEM; 164 err = -ENOMEM;
167 goto endadd2; 165 goto endadd2;
168 } 166 }
167 new_node = kmalloc(sizeof(struct mpath_node), GFP_KERNEL);
168 if (!new_node) {
169 kfree(new_mpath);
170 atomic_dec(&sdata->u.sta.mpaths);
171 err = -ENOMEM;
172 goto endadd2;
173 }
174
175 read_lock(&pathtbl_resize_lock);
169 memcpy(new_mpath->dst, dst, ETH_ALEN); 176 memcpy(new_mpath->dst, dst, ETH_ALEN);
170 new_mpath->dev = dev; 177 new_mpath->dev = dev;
171 new_mpath->flags = 0; 178 new_mpath->flags = 0;
172 skb_queue_head_init(&new_mpath->frame_queue); 179 skb_queue_head_init(&new_mpath->frame_queue);
173 new_node = kmalloc(sizeof(struct mpath_node), GFP_KERNEL);
174 new_node->mpath = new_mpath; 180 new_node->mpath = new_mpath;
175 new_mpath->timer.data = (unsigned long) new_mpath; 181 new_mpath->timer.data = (unsigned long) new_mpath;
176 new_mpath->timer.function = mesh_path_timer; 182 new_mpath->timer.function = mesh_path_timer;
@@ -202,7 +208,6 @@ int mesh_path_add(u8 *dst, struct net_device *dev)
202 208
203endadd: 209endadd:
204 spin_unlock(&mesh_paths->hashwlock[hash_idx]); 210 spin_unlock(&mesh_paths->hashwlock[hash_idx]);
205endadd2:
206 read_unlock(&pathtbl_resize_lock); 211 read_unlock(&pathtbl_resize_lock);
207 if (!err && grow) { 212 if (!err && grow) {
208 struct mesh_table *oldtbl, *newtbl; 213 struct mesh_table *oldtbl, *newtbl;
@@ -215,10 +220,12 @@ endadd2:
215 return -ENOMEM; 220 return -ENOMEM;
216 } 221 }
217 rcu_assign_pointer(mesh_paths, newtbl); 222 rcu_assign_pointer(mesh_paths, newtbl);
223 write_unlock(&pathtbl_resize_lock);
224
218 synchronize_rcu(); 225 synchronize_rcu();
219 mesh_table_free(oldtbl, false); 226 mesh_table_free(oldtbl, false);
220 write_unlock(&pathtbl_resize_lock);
221 } 227 }
228endadd2:
222 return err; 229 return err;
223} 230}
224 231
diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
index a5e5c31c23ab..b404537c0bcd 100644
--- a/net/mac80211/mlme.c
+++ b/net/mac80211/mlme.c
@@ -44,7 +44,7 @@
44#define IEEE80211_RETRY_AUTH_INTERVAL (1 * HZ) 44#define IEEE80211_RETRY_AUTH_INTERVAL (1 * HZ)
45#define IEEE80211_SCAN_INTERVAL (2 * HZ) 45#define IEEE80211_SCAN_INTERVAL (2 * HZ)
46#define IEEE80211_SCAN_INTERVAL_SLOW (15 * HZ) 46#define IEEE80211_SCAN_INTERVAL_SLOW (15 * HZ)
47#define IEEE80211_IBSS_JOIN_TIMEOUT (20 * HZ) 47#define IEEE80211_IBSS_JOIN_TIMEOUT (7 * HZ)
48 48
49#define IEEE80211_PROBE_DELAY (HZ / 33) 49#define IEEE80211_PROBE_DELAY (HZ / 33)
50#define IEEE80211_CHANNEL_TIME (HZ / 33) 50#define IEEE80211_CHANNEL_TIME (HZ / 33)
@@ -547,15 +547,14 @@ static void ieee80211_set_associated(struct net_device *dev,
547 sdata->bss_conf.ht_bss_conf = &conf->ht_bss_conf; 547 sdata->bss_conf.ht_bss_conf = &conf->ht_bss_conf;
548 } 548 }
549 549
550 netif_carrier_on(dev);
551 ifsta->flags |= IEEE80211_STA_PREV_BSSID_SET; 550 ifsta->flags |= IEEE80211_STA_PREV_BSSID_SET;
552 memcpy(ifsta->prev_bssid, sdata->u.sta.bssid, ETH_ALEN); 551 memcpy(ifsta->prev_bssid, sdata->u.sta.bssid, ETH_ALEN);
553 memcpy(wrqu.ap_addr.sa_data, sdata->u.sta.bssid, ETH_ALEN); 552 memcpy(wrqu.ap_addr.sa_data, sdata->u.sta.bssid, ETH_ALEN);
554 ieee80211_sta_send_associnfo(dev, ifsta); 553 ieee80211_sta_send_associnfo(dev, ifsta);
555 } else { 554 } else {
555 netif_carrier_off(dev);
556 ieee80211_sta_tear_down_BA_sessions(dev, ifsta->bssid); 556 ieee80211_sta_tear_down_BA_sessions(dev, ifsta->bssid);
557 ifsta->flags &= ~IEEE80211_STA_ASSOCIATED; 557 ifsta->flags &= ~IEEE80211_STA_ASSOCIATED;
558 netif_carrier_off(dev);
559 ieee80211_reset_erp_info(dev); 558 ieee80211_reset_erp_info(dev);
560 559
561 sdata->bss_conf.assoc_ht = 0; 560 sdata->bss_conf.assoc_ht = 0;
@@ -569,6 +568,10 @@ static void ieee80211_set_associated(struct net_device *dev,
569 568
570 sdata->bss_conf.assoc = assoc; 569 sdata->bss_conf.assoc = assoc;
571 ieee80211_bss_info_change_notify(sdata, changed); 570 ieee80211_bss_info_change_notify(sdata, changed);
571
572 if (assoc)
573 netif_carrier_on(dev);
574
572 wrqu.ap_addr.sa_family = ARPHRD_ETHER; 575 wrqu.ap_addr.sa_family = ARPHRD_ETHER;
573 wireless_send_event(dev, SIOCGIWAP, &wrqu, NULL); 576 wireless_send_event(dev, SIOCGIWAP, &wrqu, NULL);
574} 577}
@@ -665,6 +668,26 @@ static void ieee80211_authenticate(struct net_device *dev,
665 mod_timer(&ifsta->timer, jiffies + IEEE80211_AUTH_TIMEOUT); 668 mod_timer(&ifsta->timer, jiffies + IEEE80211_AUTH_TIMEOUT);
666} 669}
667 670
671static int ieee80211_compatible_rates(struct ieee80211_sta_bss *bss,
672 struct ieee80211_supported_band *sband,
673 u64 *rates)
674{
675 int i, j, count;
676 *rates = 0;
677 count = 0;
678 for (i = 0; i < bss->supp_rates_len; i++) {
679 int rate = (bss->supp_rates[i] & 0x7F) * 5;
680
681 for (j = 0; j < sband->n_bitrates; j++)
682 if (sband->bitrates[j].bitrate == rate) {
683 *rates |= BIT(j);
684 count++;
685 break;
686 }
687 }
688
689 return count;
690}
668 691
669static void ieee80211_send_assoc(struct net_device *dev, 692static void ieee80211_send_assoc(struct net_device *dev,
670 struct ieee80211_if_sta *ifsta) 693 struct ieee80211_if_sta *ifsta)
@@ -673,11 +696,12 @@ static void ieee80211_send_assoc(struct net_device *dev,
673 struct sk_buff *skb; 696 struct sk_buff *skb;
674 struct ieee80211_mgmt *mgmt; 697 struct ieee80211_mgmt *mgmt;
675 u8 *pos, *ies; 698 u8 *pos, *ies;
676 int i, len; 699 int i, len, count, rates_len, supp_rates_len;
677 u16 capab; 700 u16 capab;
678 struct ieee80211_sta_bss *bss; 701 struct ieee80211_sta_bss *bss;
679 int wmm = 0; 702 int wmm = 0;
680 struct ieee80211_supported_band *sband; 703 struct ieee80211_supported_band *sband;
704 u64 rates = 0;
681 705
682 skb = dev_alloc_skb(local->hw.extra_tx_headroom + 706 skb = dev_alloc_skb(local->hw.extra_tx_headroom +
683 sizeof(*mgmt) + 200 + ifsta->extra_ie_len + 707 sizeof(*mgmt) + 200 + ifsta->extra_ie_len +
@@ -709,7 +733,17 @@ static void ieee80211_send_assoc(struct net_device *dev,
709 if (bss->wmm_ie) { 733 if (bss->wmm_ie) {
710 wmm = 1; 734 wmm = 1;
711 } 735 }
736
737 /* get all rates supported by the device and the AP as
738 * some APs don't like getting a superset of their rates
739 * in the association request (e.g. D-Link DAP 1353 in
740 * b-only mode) */
741 rates_len = ieee80211_compatible_rates(bss, sband, &rates);
742
712 ieee80211_rx_bss_put(dev, bss); 743 ieee80211_rx_bss_put(dev, bss);
744 } else {
745 rates = ~0;
746 rates_len = sband->n_bitrates;
713 } 747 }
714 748
715 mgmt = (struct ieee80211_mgmt *) skb_put(skb, 24); 749 mgmt = (struct ieee80211_mgmt *) skb_put(skb, 24);
@@ -740,24 +774,36 @@ static void ieee80211_send_assoc(struct net_device *dev,
740 *pos++ = ifsta->ssid_len; 774 *pos++ = ifsta->ssid_len;
741 memcpy(pos, ifsta->ssid, ifsta->ssid_len); 775 memcpy(pos, ifsta->ssid, ifsta->ssid_len);
742 776
777 /* add all rates which were marked to be used above */
778 supp_rates_len = rates_len;
779 if (supp_rates_len > 8)
780 supp_rates_len = 8;
781
743 len = sband->n_bitrates; 782 len = sband->n_bitrates;
744 if (len > 8) 783 pos = skb_put(skb, supp_rates_len + 2);
745 len = 8;
746 pos = skb_put(skb, len + 2);
747 *pos++ = WLAN_EID_SUPP_RATES; 784 *pos++ = WLAN_EID_SUPP_RATES;
748 *pos++ = len; 785 *pos++ = supp_rates_len;
749 for (i = 0; i < len; i++) {
750 int rate = sband->bitrates[i].bitrate;
751 *pos++ = (u8) (rate / 5);
752 }
753 786
754 if (sband->n_bitrates > len) { 787 count = 0;
755 pos = skb_put(skb, sband->n_bitrates - len + 2); 788 for (i = 0; i < sband->n_bitrates; i++) {
756 *pos++ = WLAN_EID_EXT_SUPP_RATES; 789 if (BIT(i) & rates) {
757 *pos++ = sband->n_bitrates - len;
758 for (i = len; i < sband->n_bitrates; i++) {
759 int rate = sband->bitrates[i].bitrate; 790 int rate = sband->bitrates[i].bitrate;
760 *pos++ = (u8) (rate / 5); 791 *pos++ = (u8) (rate / 5);
792 if (++count == 8)
793 break;
794 }
795 }
796
797 if (count == 8) {
798 pos = skb_put(skb, rates_len - count + 2);
799 *pos++ = WLAN_EID_EXT_SUPP_RATES;
800 *pos++ = rates_len - count;
801
802 for (i++; i < sband->n_bitrates; i++) {
803 if (BIT(i) & rates) {
804 int rate = sband->bitrates[i].bitrate;
805 *pos++ = (u8) (rate / 5);
806 }
761 } 807 }
762 } 808 }
763 809
@@ -1282,7 +1328,7 @@ static void ieee80211_sta_process_addba_request(struct net_device *dev,
1282 1328
1283 /* prepare reordering buffer */ 1329 /* prepare reordering buffer */
1284 tid_agg_rx->reorder_buf = 1330 tid_agg_rx->reorder_buf =
1285 kmalloc(buf_size * sizeof(struct sk_buf *), GFP_ATOMIC); 1331 kmalloc(buf_size * sizeof(struct sk_buff *), GFP_ATOMIC);
1286 if (!tid_agg_rx->reorder_buf) { 1332 if (!tid_agg_rx->reorder_buf) {
1287 if (net_ratelimit()) 1333 if (net_ratelimit())
1288 printk(KERN_ERR "can not allocate reordering buffer " 1334 printk(KERN_ERR "can not allocate reordering buffer "
@@ -1291,7 +1337,7 @@ static void ieee80211_sta_process_addba_request(struct net_device *dev,
1291 goto end; 1337 goto end;
1292 } 1338 }
1293 memset(tid_agg_rx->reorder_buf, 0, 1339 memset(tid_agg_rx->reorder_buf, 0,
1294 buf_size * sizeof(struct sk_buf *)); 1340 buf_size * sizeof(struct sk_buff *));
1295 1341
1296 if (local->ops->ampdu_action) 1342 if (local->ops->ampdu_action)
1297 ret = local->ops->ampdu_action(hw, IEEE80211_AMPDU_RX_START, 1343 ret = local->ops->ampdu_action(hw, IEEE80211_AMPDU_RX_START,
@@ -1571,7 +1617,7 @@ void sta_addba_resp_timer_expired(unsigned long data)
1571 * only one argument, and both sta_info and TID are needed, so init 1617 * only one argument, and both sta_info and TID are needed, so init
1572 * flow in sta_info_create gives the TID as data, while the timer_to_id 1618 * flow in sta_info_create gives the TID as data, while the timer_to_id
1573 * array gives the sta through container_of */ 1619 * array gives the sta through container_of */
1574 u16 tid = *(int *)data; 1620 u16 tid = *(u8 *)data;
1575 struct sta_info *temp_sta = container_of((void *)data, 1621 struct sta_info *temp_sta = container_of((void *)data,
1576 struct sta_info, timer_to_tid[tid]); 1622 struct sta_info, timer_to_tid[tid]);
1577 1623
@@ -1619,7 +1665,7 @@ timer_expired_exit:
1619void sta_rx_agg_session_timer_expired(unsigned long data) 1665void sta_rx_agg_session_timer_expired(unsigned long data)
1620{ 1666{
1621 /* not an elegant detour, but there is no choice as the timer passes 1667 /* not an elegant detour, but there is no choice as the timer passes
1622 * only one argument, and verious sta_info are needed here, so init 1668 * only one argument, and various sta_info are needed here, so init
1623 * flow in sta_info_create gives the TID as data, while the timer_to_id 1669 * flow in sta_info_create gives the TID as data, while the timer_to_id
1624 * array gives the sta through container_of */ 1670 * array gives the sta through container_of */
1625 u8 *ptid = (u8 *)data; 1671 u8 *ptid = (u8 *)data;
@@ -2293,6 +2339,7 @@ static int ieee80211_sta_join_ibss(struct net_device *dev,
2293 u8 *pos; 2339 u8 *pos;
2294 struct ieee80211_sub_if_data *sdata; 2340 struct ieee80211_sub_if_data *sdata;
2295 struct ieee80211_supported_band *sband; 2341 struct ieee80211_supported_band *sband;
2342 union iwreq_data wrqu;
2296 2343
2297 sband = local->hw.wiphy->bands[local->hw.conf.channel->band]; 2344 sband = local->hw.wiphy->bands[local->hw.conf.channel->band];
2298 2345
@@ -2315,13 +2362,10 @@ static int ieee80211_sta_join_ibss(struct net_device *dev,
2315 sdata->drop_unencrypted = bss->capability & 2362 sdata->drop_unencrypted = bss->capability &
2316 WLAN_CAPABILITY_PRIVACY ? 1 : 0; 2363 WLAN_CAPABILITY_PRIVACY ? 1 : 0;
2317 2364
2318 res = ieee80211_set_freq(local, bss->freq); 2365 res = ieee80211_set_freq(dev, bss->freq);
2319 2366
2320 if (local->oper_channel->flags & IEEE80211_CHAN_NO_IBSS) { 2367 if (res)
2321 printk(KERN_DEBUG "%s: IBSS not allowed on frequency " 2368 return res;
2322 "%d MHz\n", dev->name, local->oper_channel->center_freq);
2323 return -1;
2324 }
2325 2369
2326 /* Set beacon template */ 2370 /* Set beacon template */
2327 skb = dev_alloc_skb(local->hw.extra_tx_headroom + 400); 2371 skb = dev_alloc_skb(local->hw.extra_tx_headroom + 400);
@@ -2436,7 +2480,9 @@ static int ieee80211_sta_join_ibss(struct net_device *dev,
2436 ifsta->state = IEEE80211_IBSS_JOINED; 2480 ifsta->state = IEEE80211_IBSS_JOINED;
2437 mod_timer(&ifsta->timer, jiffies + IEEE80211_IBSS_MERGE_INTERVAL); 2481 mod_timer(&ifsta->timer, jiffies + IEEE80211_IBSS_MERGE_INTERVAL);
2438 2482
2439 ieee80211_rx_bss_put(dev, bss); 2483 memset(&wrqu, 0, sizeof(wrqu));
2484 memcpy(wrqu.ap_addr.sa_data, bss->bssid, ETH_ALEN);
2485 wireless_send_event(dev, SIOCGIWAP, &wrqu, NULL);
2440 2486
2441 return res; 2487 return res;
2442} 2488}
@@ -3410,21 +3456,17 @@ static int ieee80211_sta_config_auth(struct net_device *dev,
3410 struct ieee80211_sta_bss *bss, *selected = NULL; 3456 struct ieee80211_sta_bss *bss, *selected = NULL;
3411 int top_rssi = 0, freq; 3457 int top_rssi = 0, freq;
3412 3458
3413 if (!(ifsta->flags & (IEEE80211_STA_AUTO_SSID_SEL |
3414 IEEE80211_STA_AUTO_BSSID_SEL | IEEE80211_STA_AUTO_CHANNEL_SEL))) {
3415 ifsta->state = IEEE80211_AUTHENTICATE;
3416 ieee80211_sta_reset_auth(dev, ifsta);
3417 return 0;
3418 }
3419
3420 spin_lock_bh(&local->sta_bss_lock); 3459 spin_lock_bh(&local->sta_bss_lock);
3421 freq = local->oper_channel->center_freq; 3460 freq = local->oper_channel->center_freq;
3422 list_for_each_entry(bss, &local->sta_bss_list, list) { 3461 list_for_each_entry(bss, &local->sta_bss_list, list) {
3423 if (!(bss->capability & WLAN_CAPABILITY_ESS)) 3462 if (!(bss->capability & WLAN_CAPABILITY_ESS))
3424 continue; 3463 continue;
3425 3464
3426 if (!!(bss->capability & WLAN_CAPABILITY_PRIVACY) ^ 3465 if ((ifsta->flags & (IEEE80211_STA_AUTO_SSID_SEL |
3427 !!sdata->default_key) 3466 IEEE80211_STA_AUTO_BSSID_SEL |
3467 IEEE80211_STA_AUTO_CHANNEL_SEL)) &&
3468 (!!(bss->capability & WLAN_CAPABILITY_PRIVACY) ^
3469 !!sdata->default_key))
3428 continue; 3470 continue;
3429 3471
3430 if (!(ifsta->flags & IEEE80211_STA_AUTO_CHANNEL_SEL) && 3472 if (!(ifsta->flags & IEEE80211_STA_AUTO_CHANNEL_SEL) &&
@@ -3449,7 +3491,7 @@ static int ieee80211_sta_config_auth(struct net_device *dev,
3449 spin_unlock_bh(&local->sta_bss_lock); 3491 spin_unlock_bh(&local->sta_bss_lock);
3450 3492
3451 if (selected) { 3493 if (selected) {
3452 ieee80211_set_freq(local, selected->freq); 3494 ieee80211_set_freq(dev, selected->freq);
3453 if (!(ifsta->flags & IEEE80211_STA_SSID_SET)) 3495 if (!(ifsta->flags & IEEE80211_STA_SSID_SET))
3454 ieee80211_sta_set_ssid(dev, selected->ssid, 3496 ieee80211_sta_set_ssid(dev, selected->ssid,
3455 selected->ssid_len); 3497 selected->ssid_len);
@@ -3484,6 +3526,7 @@ static int ieee80211_sta_create_ibss(struct net_device *dev,
3484 struct ieee80211_supported_band *sband; 3526 struct ieee80211_supported_band *sband;
3485 u8 bssid[ETH_ALEN], *pos; 3527 u8 bssid[ETH_ALEN], *pos;
3486 int i; 3528 int i;
3529 int ret;
3487 DECLARE_MAC_BUF(mac); 3530 DECLARE_MAC_BUF(mac);
3488 3531
3489#if 0 3532#if 0
@@ -3528,7 +3571,9 @@ static int ieee80211_sta_create_ibss(struct net_device *dev,
3528 *pos++ = (u8) (rate / 5); 3571 *pos++ = (u8) (rate / 5);
3529 } 3572 }
3530 3573
3531 return ieee80211_sta_join_ibss(dev, ifsta, bss); 3574 ret = ieee80211_sta_join_ibss(dev, ifsta, bss);
3575 ieee80211_rx_bss_put(dev, bss);
3576 return ret;
3532} 3577}
3533 3578
3534 3579
@@ -3569,17 +3614,22 @@ static int ieee80211_sta_find_ibss(struct net_device *dev,
3569 spin_unlock_bh(&local->sta_bss_lock); 3614 spin_unlock_bh(&local->sta_bss_lock);
3570 3615
3571#ifdef CONFIG_MAC80211_IBSS_DEBUG 3616#ifdef CONFIG_MAC80211_IBSS_DEBUG
3572 printk(KERN_DEBUG " sta_find_ibss: selected %s current " 3617 if (found)
3573 "%s\n", print_mac(mac, bssid), print_mac(mac2, ifsta->bssid)); 3618 printk(KERN_DEBUG " sta_find_ibss: selected %s current "
3619 "%s\n", print_mac(mac, bssid),
3620 print_mac(mac2, ifsta->bssid));
3574#endif /* CONFIG_MAC80211_IBSS_DEBUG */ 3621#endif /* CONFIG_MAC80211_IBSS_DEBUG */
3575 if (found && memcmp(ifsta->bssid, bssid, ETH_ALEN) != 0 && 3622 if (found && memcmp(ifsta->bssid, bssid, ETH_ALEN) != 0 &&
3576 (bss = ieee80211_rx_bss_get(dev, bssid, 3623 (bss = ieee80211_rx_bss_get(dev, bssid,
3577 local->hw.conf.channel->center_freq, 3624 local->hw.conf.channel->center_freq,
3578 ifsta->ssid, ifsta->ssid_len))) { 3625 ifsta->ssid, ifsta->ssid_len))) {
3626 int ret;
3579 printk(KERN_DEBUG "%s: Selected IBSS BSSID %s" 3627 printk(KERN_DEBUG "%s: Selected IBSS BSSID %s"
3580 " based on configured SSID\n", 3628 " based on configured SSID\n",
3581 dev->name, print_mac(mac, bssid)); 3629 dev->name, print_mac(mac, bssid));
3582 return ieee80211_sta_join_ibss(dev, ifsta, bss); 3630 ret = ieee80211_sta_join_ibss(dev, ifsta, bss);
3631 ieee80211_rx_bss_put(dev, bss);
3632 return ret;
3583 } 3633 }
3584#ifdef CONFIG_MAC80211_IBSS_DEBUG 3634#ifdef CONFIG_MAC80211_IBSS_DEBUG
3585 printk(KERN_DEBUG " did not try to join ibss\n"); 3635 printk(KERN_DEBUG " did not try to join ibss\n");
@@ -4056,18 +4106,17 @@ ieee80211_sta_scan_result(struct net_device *dev,
4056 4106
4057 memset(&iwe, 0, sizeof(iwe)); 4107 memset(&iwe, 0, sizeof(iwe));
4058 iwe.cmd = SIOCGIWFREQ; 4108 iwe.cmd = SIOCGIWFREQ;
4059 iwe.u.freq.m = bss->freq; 4109 iwe.u.freq.m = ieee80211_frequency_to_channel(bss->freq);
4060 iwe.u.freq.e = 6; 4110 iwe.u.freq.e = 0;
4061 current_ev = iwe_stream_add_event(current_ev, end_buf, &iwe, 4111 current_ev = iwe_stream_add_event(current_ev, end_buf, &iwe,
4062 IW_EV_FREQ_LEN); 4112 IW_EV_FREQ_LEN);
4063 4113
4064 memset(&iwe, 0, sizeof(iwe)); 4114 memset(&iwe, 0, sizeof(iwe));
4065 iwe.cmd = SIOCGIWFREQ; 4115 iwe.cmd = SIOCGIWFREQ;
4066 iwe.u.freq.m = ieee80211_frequency_to_channel(bss->freq); 4116 iwe.u.freq.m = bss->freq;
4067 iwe.u.freq.e = 0; 4117 iwe.u.freq.e = 6;
4068 current_ev = iwe_stream_add_event(current_ev, end_buf, &iwe, 4118 current_ev = iwe_stream_add_event(current_ev, end_buf, &iwe,
4069 IW_EV_FREQ_LEN); 4119 IW_EV_FREQ_LEN);
4070
4071 memset(&iwe, 0, sizeof(iwe)); 4120 memset(&iwe, 0, sizeof(iwe));
4072 iwe.cmd = IWEVQUAL; 4121 iwe.cmd = IWEVQUAL;
4073 iwe.u.qual.qual = bss->signal; 4122 iwe.u.qual.qual = bss->signal;
diff --git a/net/mac80211/rc80211_pid.h b/net/mac80211/rc80211_pid.h
index 04afc13ed825..4ea7b97d1af1 100644
--- a/net/mac80211/rc80211_pid.h
+++ b/net/mac80211/rc80211_pid.h
@@ -141,7 +141,6 @@ struct rc_pid_events_file_info {
141 * rate behaviour values (lower means we should trust more what we learnt 141 * rate behaviour values (lower means we should trust more what we learnt
142 * about behaviour of rates, higher means we should trust more the natural 142 * about behaviour of rates, higher means we should trust more the natural
143 * ordering of rates) 143 * ordering of rates)
144 * @fast_start: if Y, push high rates right after initialization
145 */ 144 */
146struct rc_pid_debugfs_entries { 145struct rc_pid_debugfs_entries {
147 struct dentry *dir; 146 struct dentry *dir;
@@ -154,7 +153,6 @@ struct rc_pid_debugfs_entries {
154 struct dentry *sharpen_factor; 153 struct dentry *sharpen_factor;
155 struct dentry *sharpen_duration; 154 struct dentry *sharpen_duration;
156 struct dentry *norm_offset; 155 struct dentry *norm_offset;
157 struct dentry *fast_start;
158}; 156};
159 157
160void rate_control_pid_event_tx_status(struct rc_pid_event_buffer *buf, 158void rate_control_pid_event_tx_status(struct rc_pid_event_buffer *buf,
@@ -267,9 +265,6 @@ struct rc_pid_info {
267 /* Normalization offset. */ 265 /* Normalization offset. */
268 unsigned int norm_offset; 266 unsigned int norm_offset;
269 267
270 /* Fast starst parameter. */
271 unsigned int fast_start;
272
273 /* Rates information. */ 268 /* Rates information. */
274 struct rc_pid_rateinfo *rinfo; 269 struct rc_pid_rateinfo *rinfo;
275 270
diff --git a/net/mac80211/rc80211_pid_algo.c b/net/mac80211/rc80211_pid_algo.c
index a849b745bdb5..bcd27c1d7594 100644
--- a/net/mac80211/rc80211_pid_algo.c
+++ b/net/mac80211/rc80211_pid_algo.c
@@ -398,13 +398,25 @@ static void *rate_control_pid_alloc(struct ieee80211_local *local)
398 return NULL; 398 return NULL;
399 } 399 }
400 400
401 pinfo->target = RC_PID_TARGET_PF;
402 pinfo->sampling_period = RC_PID_INTERVAL;
403 pinfo->coeff_p = RC_PID_COEFF_P;
404 pinfo->coeff_i = RC_PID_COEFF_I;
405 pinfo->coeff_d = RC_PID_COEFF_D;
406 pinfo->smoothing_shift = RC_PID_SMOOTHING_SHIFT;
407 pinfo->sharpen_factor = RC_PID_SHARPENING_FACTOR;
408 pinfo->sharpen_duration = RC_PID_SHARPENING_DURATION;
409 pinfo->norm_offset = RC_PID_NORM_OFFSET;
410 pinfo->rinfo = rinfo;
411 pinfo->oldrate = 0;
412
401 /* Sort the rates. This is optimized for the most common case (i.e. 413 /* Sort the rates. This is optimized for the most common case (i.e.
402 * almost-sorted CCK+OFDM rates). Kind of bubble-sort with reversed 414 * almost-sorted CCK+OFDM rates). Kind of bubble-sort with reversed
403 * mapping too. */ 415 * mapping too. */
404 for (i = 0; i < sband->n_bitrates; i++) { 416 for (i = 0; i < sband->n_bitrates; i++) {
405 rinfo[i].index = i; 417 rinfo[i].index = i;
406 rinfo[i].rev_index = i; 418 rinfo[i].rev_index = i;
407 if (pinfo->fast_start) 419 if (RC_PID_FAST_START)
408 rinfo[i].diff = 0; 420 rinfo[i].diff = 0;
409 else 421 else
410 rinfo[i].diff = i * pinfo->norm_offset; 422 rinfo[i].diff = i * pinfo->norm_offset;
@@ -425,19 +437,6 @@ static void *rate_control_pid_alloc(struct ieee80211_local *local)
425 break; 437 break;
426 } 438 }
427 439
428 pinfo->target = RC_PID_TARGET_PF;
429 pinfo->sampling_period = RC_PID_INTERVAL;
430 pinfo->coeff_p = RC_PID_COEFF_P;
431 pinfo->coeff_i = RC_PID_COEFF_I;
432 pinfo->coeff_d = RC_PID_COEFF_D;
433 pinfo->smoothing_shift = RC_PID_SMOOTHING_SHIFT;
434 pinfo->sharpen_factor = RC_PID_SHARPENING_FACTOR;
435 pinfo->sharpen_duration = RC_PID_SHARPENING_DURATION;
436 pinfo->norm_offset = RC_PID_NORM_OFFSET;
437 pinfo->fast_start = RC_PID_FAST_START;
438 pinfo->rinfo = rinfo;
439 pinfo->oldrate = 0;
440
441#ifdef CONFIG_MAC80211_DEBUGFS 440#ifdef CONFIG_MAC80211_DEBUGFS
442 de = &pinfo->dentries; 441 de = &pinfo->dentries;
443 de->dir = debugfs_create_dir("rc80211_pid", 442 de->dir = debugfs_create_dir("rc80211_pid",
@@ -465,9 +464,6 @@ static void *rate_control_pid_alloc(struct ieee80211_local *local)
465 de->norm_offset = debugfs_create_u32("norm_offset", 464 de->norm_offset = debugfs_create_u32("norm_offset",
466 S_IRUSR | S_IWUSR, de->dir, 465 S_IRUSR | S_IWUSR, de->dir,
467 &pinfo->norm_offset); 466 &pinfo->norm_offset);
468 de->fast_start = debugfs_create_bool("fast_start",
469 S_IRUSR | S_IWUSR, de->dir,
470 &pinfo->fast_start);
471#endif 467#endif
472 468
473 return pinfo; 469 return pinfo;
@@ -479,7 +475,6 @@ static void rate_control_pid_free(void *priv)
479#ifdef CONFIG_MAC80211_DEBUGFS 475#ifdef CONFIG_MAC80211_DEBUGFS
480 struct rc_pid_debugfs_entries *de = &pinfo->dentries; 476 struct rc_pid_debugfs_entries *de = &pinfo->dentries;
481 477
482 debugfs_remove(de->fast_start);
483 debugfs_remove(de->norm_offset); 478 debugfs_remove(de->norm_offset);
484 debugfs_remove(de->sharpen_duration); 479 debugfs_remove(de->sharpen_duration);
485 debugfs_remove(de->sharpen_factor); 480 debugfs_remove(de->sharpen_factor);
diff --git a/net/mac80211/rc80211_pid_debugfs.c b/net/mac80211/rc80211_pid_debugfs.c
index ae75d4178739..ff5c380f3c13 100644
--- a/net/mac80211/rc80211_pid_debugfs.c
+++ b/net/mac80211/rc80211_pid_debugfs.c
@@ -85,7 +85,7 @@ static int rate_control_pid_events_open(struct inode *inode, struct file *file)
85 struct rc_pid_sta_info *sinfo = inode->i_private; 85 struct rc_pid_sta_info *sinfo = inode->i_private;
86 struct rc_pid_event_buffer *events = &sinfo->events; 86 struct rc_pid_event_buffer *events = &sinfo->events;
87 struct rc_pid_events_file_info *file_info; 87 struct rc_pid_events_file_info *file_info;
88 unsigned int status; 88 unsigned long status;
89 89
90 /* Allocate a state struct */ 90 /* Allocate a state struct */
91 file_info = kmalloc(sizeof(*file_info), GFP_KERNEL); 91 file_info = kmalloc(sizeof(*file_info), GFP_KERNEL);
@@ -135,7 +135,7 @@ static ssize_t rate_control_pid_events_read(struct file *file, char __user *buf,
135 char pb[RC_PID_PRINT_BUF_SIZE]; 135 char pb[RC_PID_PRINT_BUF_SIZE];
136 int ret; 136 int ret;
137 int p; 137 int p;
138 unsigned int status; 138 unsigned long status;
139 139
140 /* Check if there is something to read. */ 140 /* Check if there is something to read. */
141 if (events->next_entry == file_info->next_entry) { 141 if (events->next_entry == file_info->next_entry) {
diff --git a/net/mac80211/rx.c b/net/mac80211/rx.c
index 02f436a86061..0941e5d6a522 100644
--- a/net/mac80211/rx.c
+++ b/net/mac80211/rx.c
@@ -1091,7 +1091,7 @@ ieee80211_data_to_8023(struct ieee80211_rx_data *rx)
1091 u16 fc, hdrlen, ethertype; 1091 u16 fc, hdrlen, ethertype;
1092 u8 *payload; 1092 u8 *payload;
1093 u8 dst[ETH_ALEN]; 1093 u8 dst[ETH_ALEN];
1094 u8 src[ETH_ALEN]; 1094 u8 src[ETH_ALEN] __aligned(2);
1095 struct sk_buff *skb = rx->skb; 1095 struct sk_buff *skb = rx->skb;
1096 struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev); 1096 struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev);
1097 DECLARE_MAC_BUF(mac); 1097 DECLARE_MAC_BUF(mac);
@@ -1234,7 +1234,7 @@ ieee80211_data_to_8023(struct ieee80211_rx_data *rx)
1234 */ 1234 */
1235static bool ieee80211_frame_allowed(struct ieee80211_rx_data *rx) 1235static bool ieee80211_frame_allowed(struct ieee80211_rx_data *rx)
1236{ 1236{
1237 static const u8 pae_group_addr[ETH_ALEN] 1237 static const u8 pae_group_addr[ETH_ALEN] __aligned(2)
1238 = { 0x01, 0x80, 0xC2, 0x00, 0x00, 0x03 }; 1238 = { 0x01, 0x80, 0xC2, 0x00, 0x00, 0x03 };
1239 struct ethhdr *ehdr = (struct ethhdr *) rx->skb->data; 1239 struct ethhdr *ehdr = (struct ethhdr *) rx->skb->data;
1240 1240
@@ -1305,11 +1305,11 @@ ieee80211_deliver_skb(struct ieee80211_rx_data *rx)
1305 if (is_multicast_ether_addr(skb->data)) { 1305 if (is_multicast_ether_addr(skb->data)) {
1306 if (*mesh_ttl > 0) { 1306 if (*mesh_ttl > 0) {
1307 xmit_skb = skb_copy(skb, GFP_ATOMIC); 1307 xmit_skb = skb_copy(skb, GFP_ATOMIC);
1308 if (!xmit_skb && net_ratelimit()) 1308 if (xmit_skb)
1309 xmit_skb->pkt_type = PACKET_OTHERHOST;
1310 else if (net_ratelimit())
1309 printk(KERN_DEBUG "%s: failed to clone " 1311 printk(KERN_DEBUG "%s: failed to clone "
1310 "multicast frame\n", dev->name); 1312 "multicast frame\n", dev->name);
1311 else
1312 xmit_skb->pkt_type = PACKET_OTHERHOST;
1313 } else 1313 } else
1314 IEEE80211_IFSTA_MESH_CTR_INC(&sdata->u.sta, 1314 IEEE80211_IFSTA_MESH_CTR_INC(&sdata->u.sta,
1315 dropped_frames_ttl); 1315 dropped_frames_ttl);
@@ -1395,7 +1395,7 @@ ieee80211_rx_h_amsdu(struct ieee80211_rx_data *rx)
1395 padding = ((4 - subframe_len) & 0x3); 1395 padding = ((4 - subframe_len) & 0x3);
1396 /* the last MSDU has no padding */ 1396 /* the last MSDU has no padding */
1397 if (subframe_len > remaining) { 1397 if (subframe_len > remaining) {
1398 printk(KERN_DEBUG "%s: wrong buffer size", dev->name); 1398 printk(KERN_DEBUG "%s: wrong buffer size\n", dev->name);
1399 return RX_DROP_UNUSABLE; 1399 return RX_DROP_UNUSABLE;
1400 } 1400 }
1401 1401
@@ -1418,7 +1418,7 @@ ieee80211_rx_h_amsdu(struct ieee80211_rx_data *rx)
1418 eth = (struct ethhdr *) skb_pull(skb, ntohs(len) + 1418 eth = (struct ethhdr *) skb_pull(skb, ntohs(len) +
1419 padding); 1419 padding);
1420 if (!eth) { 1420 if (!eth) {
1421 printk(KERN_DEBUG "%s: wrong buffer size ", 1421 printk(KERN_DEBUG "%s: wrong buffer size\n",
1422 dev->name); 1422 dev->name);
1423 dev_kfree_skb(frame); 1423 dev_kfree_skb(frame);
1424 return RX_DROP_UNUSABLE; 1424 return RX_DROP_UNUSABLE;
@@ -1952,7 +1952,7 @@ static void __ieee80211_rx_handle_packet(struct ieee80211_hw *hw,
1952 if (!skb_new) { 1952 if (!skb_new) {
1953 if (net_ratelimit()) 1953 if (net_ratelimit())
1954 printk(KERN_DEBUG "%s: failed to copy " 1954 printk(KERN_DEBUG "%s: failed to copy "
1955 "multicast frame for %s", 1955 "multicast frame for %s\n",
1956 wiphy_name(local->hw.wiphy), 1956 wiphy_name(local->hw.wiphy),
1957 prev->dev->name); 1957 prev->dev->name);
1958 continue; 1958 continue;
diff --git a/net/mac80211/tkip.c b/net/mac80211/tkip.c
index dddbfd60f351..09093da24af6 100644
--- a/net/mac80211/tkip.c
+++ b/net/mac80211/tkip.c
@@ -230,10 +230,8 @@ void ieee80211_get_tkip_key(struct ieee80211_key_conf *keyconf,
230 230
231 iv16 = data[hdr_len] << 8; 231 iv16 = data[hdr_len] << 8;
232 iv16 += data[hdr_len + 2]; 232 iv16 += data[hdr_len + 2];
233 iv32 = data[hdr_len + 4] + 233 iv32 = data[hdr_len + 4] | (data[hdr_len + 5] << 8) |
234 (data[hdr_len + 5] >> 8) + 234 (data[hdr_len + 6] << 16) | (data[hdr_len + 7] << 24);
235 (data[hdr_len + 6] >> 16) +
236 (data[hdr_len + 7] >> 24);
237 235
238#ifdef CONFIG_TKIP_DEBUG 236#ifdef CONFIG_TKIP_DEBUG
239 printk(KERN_DEBUG "TKIP encrypt: iv16 = 0x%04x, iv32 = 0x%08x\n", 237 printk(KERN_DEBUG "TKIP encrypt: iv16 = 0x%04x, iv32 = 0x%08x\n",
diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
index f35eaea98e73..c80d5899f279 100644
--- a/net/mac80211/tx.c
+++ b/net/mac80211/tx.c
@@ -1132,7 +1132,7 @@ static int ieee80211_tx(struct net_device *dev, struct sk_buff *skb,
1132 ieee80211_tx_handler *handler; 1132 ieee80211_tx_handler *handler;
1133 struct ieee80211_tx_data tx; 1133 struct ieee80211_tx_data tx;
1134 ieee80211_tx_result res = TX_DROP, res_prepare; 1134 ieee80211_tx_result res = TX_DROP, res_prepare;
1135 int ret, i; 1135 int ret, i, retries = 0;
1136 1136
1137 WARN_ON(__ieee80211_queue_pending(local, control->queue)); 1137 WARN_ON(__ieee80211_queue_pending(local, control->queue));
1138 1138
@@ -1216,6 +1216,13 @@ retry:
1216 if (!__ieee80211_queue_stopped(local, control->queue)) { 1216 if (!__ieee80211_queue_stopped(local, control->queue)) {
1217 clear_bit(IEEE80211_LINK_STATE_PENDING, 1217 clear_bit(IEEE80211_LINK_STATE_PENDING,
1218 &local->state[control->queue]); 1218 &local->state[control->queue]);
1219 retries++;
1220 /*
1221 * Driver bug, it's rejecting packets but
1222 * not stopping queues.
1223 */
1224 if (WARN_ON_ONCE(retries > 5))
1225 goto drop;
1219 goto retry; 1226 goto retry;
1220 } 1227 }
1221 memcpy(&store->control, control, 1228 memcpy(&store->control, control,
@@ -1898,6 +1905,7 @@ struct sk_buff *ieee80211_beacon_get(struct ieee80211_hw *hw,
1898 control->flags |= IEEE80211_TXCTL_SHORT_PREAMBLE; 1905 control->flags |= IEEE80211_TXCTL_SHORT_PREAMBLE;
1899 control->antenna_sel_tx = local->hw.conf.antenna_sel_tx; 1906 control->antenna_sel_tx = local->hw.conf.antenna_sel_tx;
1900 control->flags |= IEEE80211_TXCTL_NO_ACK; 1907 control->flags |= IEEE80211_TXCTL_NO_ACK;
1908 control->flags |= IEEE80211_TXCTL_DO_NOT_ENCRYPT;
1901 control->retry_limit = 1; 1909 control->retry_limit = 1;
1902 control->flags |= IEEE80211_TXCTL_CLEAR_PS_FILT; 1910 control->flags |= IEEE80211_TXCTL_CLEAR_PS_FILT;
1903 } 1911 }
diff --git a/net/mac80211/util.c b/net/mac80211/util.c
index cc9f715c7bfc..4e97b266f907 100644
--- a/net/mac80211/util.c
+++ b/net/mac80211/util.c
@@ -34,11 +34,11 @@ void *mac80211_wiphy_privid = &mac80211_wiphy_privid;
34 34
35/* See IEEE 802.1H for LLC/SNAP encapsulation/decapsulation */ 35/* See IEEE 802.1H for LLC/SNAP encapsulation/decapsulation */
36/* Ethernet-II snap header (RFC1042 for most EtherTypes) */ 36/* Ethernet-II snap header (RFC1042 for most EtherTypes) */
37const unsigned char rfc1042_header[] = 37const unsigned char rfc1042_header[] __aligned(2) =
38 { 0xaa, 0xaa, 0x03, 0x00, 0x00, 0x00 }; 38 { 0xaa, 0xaa, 0x03, 0x00, 0x00, 0x00 };
39 39
40/* Bridge-Tunnel header (for EtherTypes ETH_P_AARP and ETH_P_IPX) */ 40/* Bridge-Tunnel header (for EtherTypes ETH_P_AARP and ETH_P_IPX) */
41const unsigned char bridge_tunnel_header[] = 41const unsigned char bridge_tunnel_header[] __aligned(2) =
42 { 0xaa, 0xaa, 0x03, 0x00, 0x00, 0xf8 }; 42 { 0xaa, 0xaa, 0x03, 0x00, 0x00, 0xf8 };
43 43
44 44
@@ -153,15 +153,15 @@ int ieee80211_get_mesh_hdrlen(struct ieee80211s_hdr *meshhdr)
153 /* 7.1.3.5a.2 */ 153 /* 7.1.3.5a.2 */
154 switch (ae) { 154 switch (ae) {
155 case 0: 155 case 0:
156 return 5; 156 return 6;
157 case 1: 157 case 1:
158 return 11; 158 return 12;
159 case 2: 159 case 2:
160 return 17; 160 return 18;
161 case 3: 161 case 3:
162 return 23; 162 return 24;
163 default: 163 default:
164 return 5; 164 return 6;
165 } 165 }
166} 166}
167 167
@@ -389,6 +389,41 @@ void ieee80211_iterate_active_interfaces(
389 struct ieee80211_local *local = hw_to_local(hw); 389 struct ieee80211_local *local = hw_to_local(hw);
390 struct ieee80211_sub_if_data *sdata; 390 struct ieee80211_sub_if_data *sdata;
391 391
392 rtnl_lock();
393
394 list_for_each_entry(sdata, &local->interfaces, list) {
395 switch (sdata->vif.type) {
396 case IEEE80211_IF_TYPE_INVALID:
397 case IEEE80211_IF_TYPE_MNTR:
398 case IEEE80211_IF_TYPE_VLAN:
399 continue;
400 case IEEE80211_IF_TYPE_AP:
401 case IEEE80211_IF_TYPE_STA:
402 case IEEE80211_IF_TYPE_IBSS:
403 case IEEE80211_IF_TYPE_WDS:
404 case IEEE80211_IF_TYPE_MESH_POINT:
405 break;
406 }
407 if (sdata->dev == local->mdev)
408 continue;
409 if (netif_running(sdata->dev))
410 iterator(data, sdata->dev->dev_addr,
411 &sdata->vif);
412 }
413
414 rtnl_unlock();
415}
416EXPORT_SYMBOL_GPL(ieee80211_iterate_active_interfaces);
417
418void ieee80211_iterate_active_interfaces_atomic(
419 struct ieee80211_hw *hw,
420 void (*iterator)(void *data, u8 *mac,
421 struct ieee80211_vif *vif),
422 void *data)
423{
424 struct ieee80211_local *local = hw_to_local(hw);
425 struct ieee80211_sub_if_data *sdata;
426
392 rcu_read_lock(); 427 rcu_read_lock();
393 428
394 list_for_each_entry_rcu(sdata, &local->interfaces, list) { 429 list_for_each_entry_rcu(sdata, &local->interfaces, list) {
@@ -413,4 +448,4 @@ void ieee80211_iterate_active_interfaces(
413 448
414 rcu_read_unlock(); 449 rcu_read_unlock();
415} 450}
416EXPORT_SYMBOL_GPL(ieee80211_iterate_active_interfaces); 451EXPORT_SYMBOL_GPL(ieee80211_iterate_active_interfaces_atomic);
diff --git a/net/mac80211/wext.c b/net/mac80211/wext.c
index 76e1de1dc735..e8404212ad57 100644
--- a/net/mac80211/wext.c
+++ b/net/mac80211/wext.c
@@ -95,6 +95,13 @@ static int ieee80211_set_encryption(struct net_device *dev, u8 *sta_addr,
95 } 95 }
96 } 96 }
97 97
98 if (alg == ALG_WEP &&
99 key_len != LEN_WEP40 && key_len != LEN_WEP104) {
100 ieee80211_key_free(key);
101 err = -EINVAL;
102 goto out_unlock;
103 }
104
98 ieee80211_key_link(key, sdata, sta); 105 ieee80211_key_link(key, sdata, sta);
99 106
100 if (set_tx_key || (!sta && !sdata->default_key && key)) 107 if (set_tx_key || (!sta && !sdata->default_key && key))
@@ -209,7 +216,6 @@ static int ieee80211_ioctl_giwrange(struct net_device *dev,
209 range->num_frequency = c; 216 range->num_frequency = c;
210 217
211 IW_EVENT_CAPA_SET_KERNEL(range->event_capa); 218 IW_EVENT_CAPA_SET_KERNEL(range->event_capa);
212 IW_EVENT_CAPA_SET(range->event_capa, SIOCGIWTHRSPY);
213 IW_EVENT_CAPA_SET(range->event_capa, SIOCGIWAP); 219 IW_EVENT_CAPA_SET(range->event_capa, SIOCGIWAP);
214 IW_EVENT_CAPA_SET(range->event_capa, SIOCGIWSCAN); 220 IW_EVENT_CAPA_SET(range->event_capa, SIOCGIWSCAN);
215 221
@@ -291,14 +297,22 @@ static int ieee80211_ioctl_giwmode(struct net_device *dev,
291 return 0; 297 return 0;
292} 298}
293 299
294int ieee80211_set_freq(struct ieee80211_local *local, int freqMHz) 300int ieee80211_set_freq(struct net_device *dev, int freqMHz)
295{ 301{
296 int ret = -EINVAL; 302 int ret = -EINVAL;
297 struct ieee80211_channel *chan; 303 struct ieee80211_channel *chan;
304 struct ieee80211_local *local = wdev_priv(dev->ieee80211_ptr);
305 struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev);
298 306
299 chan = ieee80211_get_channel(local->hw.wiphy, freqMHz); 307 chan = ieee80211_get_channel(local->hw.wiphy, freqMHz);
300 308
301 if (chan && !(chan->flags & IEEE80211_CHAN_DISABLED)) { 309 if (chan && !(chan->flags & IEEE80211_CHAN_DISABLED)) {
310 if (sdata->vif.type == IEEE80211_IF_TYPE_IBSS &&
311 chan->flags & IEEE80211_CHAN_NO_IBSS) {
312 printk(KERN_DEBUG "%s: IBSS not allowed on frequency "
313 "%d MHz\n", dev->name, chan->center_freq);
314 return ret;
315 }
302 local->oper_channel = chan; 316 local->oper_channel = chan;
303 317
304 if (local->sta_sw_scanning || local->sta_hw_scanning) 318 if (local->sta_sw_scanning || local->sta_hw_scanning)
@@ -316,7 +330,6 @@ static int ieee80211_ioctl_siwfreq(struct net_device *dev,
316 struct iw_request_info *info, 330 struct iw_request_info *info,
317 struct iw_freq *freq, char *extra) 331 struct iw_freq *freq, char *extra)
318{ 332{
319 struct ieee80211_local *local = wdev_priv(dev->ieee80211_ptr);
320 struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev); 333 struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev);
321 334
322 if (sdata->vif.type == IEEE80211_IF_TYPE_STA) 335 if (sdata->vif.type == IEEE80211_IF_TYPE_STA)
@@ -330,14 +343,14 @@ static int ieee80211_ioctl_siwfreq(struct net_device *dev,
330 IEEE80211_STA_AUTO_CHANNEL_SEL; 343 IEEE80211_STA_AUTO_CHANNEL_SEL;
331 return 0; 344 return 0;
332 } else 345 } else
333 return ieee80211_set_freq(local, 346 return ieee80211_set_freq(dev,
334 ieee80211_channel_to_frequency(freq->m)); 347 ieee80211_channel_to_frequency(freq->m));
335 } else { 348 } else {
336 int i, div = 1000000; 349 int i, div = 1000000;
337 for (i = 0; i < freq->e; i++) 350 for (i = 0; i < freq->e; i++)
338 div /= 10; 351 div /= 10;
339 if (div > 0) 352 if (div > 0)
340 return ieee80211_set_freq(local, freq->m / div); 353 return ieee80211_set_freq(dev, freq->m / div);
341 else 354 else
342 return -EINVAL; 355 return -EINVAL;
343 } 356 }
@@ -490,9 +503,15 @@ static int ieee80211_ioctl_giwap(struct net_device *dev,
490 sdata = IEEE80211_DEV_TO_SUB_IF(dev); 503 sdata = IEEE80211_DEV_TO_SUB_IF(dev);
491 if (sdata->vif.type == IEEE80211_IF_TYPE_STA || 504 if (sdata->vif.type == IEEE80211_IF_TYPE_STA ||
492 sdata->vif.type == IEEE80211_IF_TYPE_IBSS) { 505 sdata->vif.type == IEEE80211_IF_TYPE_IBSS) {
493 ap_addr->sa_family = ARPHRD_ETHER; 506 if (sdata->u.sta.state == IEEE80211_ASSOCIATED ||
494 memcpy(&ap_addr->sa_data, sdata->u.sta.bssid, ETH_ALEN); 507 sdata->u.sta.state == IEEE80211_IBSS_JOINED) {
495 return 0; 508 ap_addr->sa_family = ARPHRD_ETHER;
509 memcpy(&ap_addr->sa_data, sdata->u.sta.bssid, ETH_ALEN);
510 return 0;
511 } else {
512 memset(&ap_addr->sa_data, 0, ETH_ALEN);
513 return 0;
514 }
496 } else if (sdata->vif.type == IEEE80211_IF_TYPE_WDS) { 515 } else if (sdata->vif.type == IEEE80211_IF_TYPE_WDS) {
497 ap_addr->sa_family = ARPHRD_ETHER; 516 ap_addr->sa_family = ARPHRD_ETHER;
498 memcpy(&ap_addr->sa_data, sdata->u.wds.remote_addr, ETH_ALEN); 517 memcpy(&ap_addr->sa_data, sdata->u.wds.remote_addr, ETH_ALEN);
diff --git a/net/mac80211/wme.c b/net/mac80211/wme.c
index 64faa3dc488f..5d09e8698b57 100644
--- a/net/mac80211/wme.c
+++ b/net/mac80211/wme.c
@@ -323,8 +323,7 @@ static void wme_qdiscop_destroy(struct Qdisc* qd)
323 struct ieee80211_hw *hw = &local->hw; 323 struct ieee80211_hw *hw = &local->hw;
324 int queue; 324 int queue;
325 325
326 tcf_destroy_chain(q->filter_list); 326 tcf_destroy_chain(&q->filter_list);
327 q->filter_list = NULL;
328 327
329 for (queue=0; queue < hw->queues; queue++) { 328 for (queue=0; queue < hw->queues; queue++) {
330 skb_queue_purge(&q->requeued[queue]); 329 skb_queue_purge(&q->requeued[queue]);
@@ -394,7 +393,8 @@ static int wme_qdiscop_init(struct Qdisc *qd, struct nlattr *opt)
394 qd->handle); 393 qd->handle);
395 if (!q->queues[i]) { 394 if (!q->queues[i]) {
396 q->queues[i] = &noop_qdisc; 395 q->queues[i] = &noop_qdisc;
397 printk(KERN_ERR "%s child qdisc %i creation failed", dev->name, i); 396 printk(KERN_ERR "%s child qdisc %i creation failed\n",
397 dev->name, i);
398 } 398 }
399 } 399 }
400 400
@@ -672,7 +672,7 @@ int ieee80211_ht_agg_queue_add(struct ieee80211_local *local,
672#ifdef CONFIG_MAC80211_HT_DEBUG 672#ifdef CONFIG_MAC80211_HT_DEBUG
673 if (net_ratelimit()) 673 if (net_ratelimit())
674 printk(KERN_DEBUG "allocated aggregation queue" 674 printk(KERN_DEBUG "allocated aggregation queue"
675 " %d tid %d addr %s pool=0x%lX", 675 " %d tid %d addr %s pool=0x%lX\n",
676 i, tid, print_mac(mac, sta->addr), 676 i, tid, print_mac(mac, sta->addr),
677 q->qdisc_pool[0]); 677 q->qdisc_pool[0]);
678#endif /* CONFIG_MAC80211_HT_DEBUG */ 678#endif /* CONFIG_MAC80211_HT_DEBUG */
diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig
index c1fc0f1a641c..aa8d80c35e28 100644
--- a/net/netfilter/Kconfig
+++ b/net/netfilter/Kconfig
@@ -90,6 +90,7 @@ config NF_CT_PROTO_DCCP
90 tristate 'DCCP protocol connection tracking support (EXPERIMENTAL)' 90 tristate 'DCCP protocol connection tracking support (EXPERIMENTAL)'
91 depends on EXPERIMENTAL && NF_CONNTRACK 91 depends on EXPERIMENTAL && NF_CONNTRACK
92 depends on NETFILTER_ADVANCED 92 depends on NETFILTER_ADVANCED
93 default IP_DCCP
93 help 94 help
94 With this option enabled, the layer 3 independent connection 95 With this option enabled, the layer 3 independent connection
95 tracking code will be able to do state tracking on DCCP connections. 96 tracking code will be able to do state tracking on DCCP connections.
@@ -104,6 +105,7 @@ config NF_CT_PROTO_SCTP
104 tristate 'SCTP protocol connection tracking support (EXPERIMENTAL)' 105 tristate 'SCTP protocol connection tracking support (EXPERIMENTAL)'
105 depends on EXPERIMENTAL && NF_CONNTRACK 106 depends on EXPERIMENTAL && NF_CONNTRACK
106 depends on NETFILTER_ADVANCED 107 depends on NETFILTER_ADVANCED
108 default IP_SCTP
107 help 109 help
108 With this option enabled, the layer 3 independent connection 110 With this option enabled, the layer 3 independent connection
109 tracking code will be able to do state tracking on SCTP connections. 111 tracking code will be able to do state tracking on SCTP connections.
@@ -532,6 +534,7 @@ config NETFILTER_XT_MATCH_DCCP
532 tristate '"dccp" protocol match support' 534 tristate '"dccp" protocol match support'
533 depends on NETFILTER_XTABLES 535 depends on NETFILTER_XTABLES
534 depends on NETFILTER_ADVANCED 536 depends on NETFILTER_ADVANCED
537 default IP_DCCP
535 help 538 help
536 With this option enabled, you will be able to use the iptables 539 With this option enabled, you will be able to use the iptables
537 `dccp' match in order to match on DCCP source/destination ports 540 `dccp' match in order to match on DCCP source/destination ports
@@ -725,6 +728,7 @@ config NETFILTER_XT_MATCH_SCTP
725 tristate '"sctp" protocol match support (EXPERIMENTAL)' 728 tristate '"sctp" protocol match support (EXPERIMENTAL)'
726 depends on NETFILTER_XTABLES && EXPERIMENTAL 729 depends on NETFILTER_XTABLES && EXPERIMENTAL
727 depends on NETFILTER_ADVANCED 730 depends on NETFILTER_ADVANCED
731 default IP_SCTP
728 help 732 help
729 With this option enabled, you will be able to use the 733 With this option enabled, you will be able to use the
730 `sctp' match in order to match on SCTP source/destination ports 734 `sctp' match in order to match on SCTP source/destination ports
diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
index c4b1799da5d7..662c1ccfee26 100644
--- a/net/netfilter/nf_conntrack_core.c
+++ b/net/netfilter/nf_conntrack_core.c
@@ -196,8 +196,6 @@ destroy_conntrack(struct nf_conntrack *nfct)
196 if (l4proto && l4proto->destroy) 196 if (l4proto && l4proto->destroy)
197 l4proto->destroy(ct); 197 l4proto->destroy(ct);
198 198
199 nf_ct_ext_destroy(ct);
200
201 rcu_read_unlock(); 199 rcu_read_unlock();
202 200
203 spin_lock_bh(&nf_conntrack_lock); 201 spin_lock_bh(&nf_conntrack_lock);
@@ -520,6 +518,7 @@ static void nf_conntrack_free_rcu(struct rcu_head *head)
520 518
521void nf_conntrack_free(struct nf_conn *ct) 519void nf_conntrack_free(struct nf_conn *ct)
522{ 520{
521 nf_ct_ext_destroy(ct);
523 call_rcu(&ct->rcu, nf_conntrack_free_rcu); 522 call_rcu(&ct->rcu, nf_conntrack_free_rcu);
524} 523}
525EXPORT_SYMBOL_GPL(nf_conntrack_free); 524EXPORT_SYMBOL_GPL(nf_conntrack_free);
diff --git a/net/netfilter/nf_conntrack_expect.c b/net/netfilter/nf_conntrack_expect.c
index e31beeb33b2b..e8f0dead267f 100644
--- a/net/netfilter/nf_conntrack_expect.c
+++ b/net/netfilter/nf_conntrack_expect.c
@@ -587,10 +587,10 @@ int __init nf_conntrack_expect_init(void)
587 return 0; 587 return 0;
588 588
589err3: 589err3:
590 kmem_cache_destroy(nf_ct_expect_cachep);
591err2:
590 nf_ct_free_hashtable(nf_ct_expect_hash, nf_ct_expect_vmalloc, 592 nf_ct_free_hashtable(nf_ct_expect_hash, nf_ct_expect_vmalloc,
591 nf_ct_expect_hsize); 593 nf_ct_expect_hsize);
592err2:
593 kmem_cache_destroy(nf_ct_expect_cachep);
594err1: 594err1:
595 return err; 595 return err;
596} 596}
diff --git a/net/netfilter/nf_conntrack_extend.c b/net/netfilter/nf_conntrack_extend.c
index bcc19fa4ed1e..8a3f8b34e466 100644
--- a/net/netfilter/nf_conntrack_extend.c
+++ b/net/netfilter/nf_conntrack_extend.c
@@ -59,12 +59,19 @@ nf_ct_ext_create(struct nf_ct_ext **ext, enum nf_ct_ext_id id, gfp_t gfp)
59 if (!*ext) 59 if (!*ext)
60 return NULL; 60 return NULL;
61 61
62 INIT_RCU_HEAD(&(*ext)->rcu);
62 (*ext)->offset[id] = off; 63 (*ext)->offset[id] = off;
63 (*ext)->len = len; 64 (*ext)->len = len;
64 65
65 return (void *)(*ext) + off; 66 return (void *)(*ext) + off;
66} 67}
67 68
69static void __nf_ct_ext_free_rcu(struct rcu_head *head)
70{
71 struct nf_ct_ext *ext = container_of(head, struct nf_ct_ext, rcu);
72 kfree(ext);
73}
74
68void *__nf_ct_ext_add(struct nf_conn *ct, enum nf_ct_ext_id id, gfp_t gfp) 75void *__nf_ct_ext_add(struct nf_conn *ct, enum nf_ct_ext_id id, gfp_t gfp)
69{ 76{
70 struct nf_ct_ext *new; 77 struct nf_ct_ext *new;
@@ -106,7 +113,7 @@ void *__nf_ct_ext_add(struct nf_conn *ct, enum nf_ct_ext_id id, gfp_t gfp)
106 (void *)ct->ext + ct->ext->offset[i]); 113 (void *)ct->ext + ct->ext->offset[i]);
107 rcu_read_unlock(); 114 rcu_read_unlock();
108 } 115 }
109 kfree(ct->ext); 116 call_rcu(&ct->ext->rcu, __nf_ct_ext_free_rcu);
110 ct->ext = new; 117 ct->ext = new;
111 } 118 }
112 119
diff --git a/net/netfilter/nf_conntrack_h323_main.c b/net/netfilter/nf_conntrack_h323_main.c
index 95da1a24aab7..2f83c158934d 100644
--- a/net/netfilter/nf_conntrack_h323_main.c
+++ b/net/netfilter/nf_conntrack_h323_main.c
@@ -619,6 +619,7 @@ static const struct nf_conntrack_expect_policy h245_exp_policy = {
619static struct nf_conntrack_helper nf_conntrack_helper_h245 __read_mostly = { 619static struct nf_conntrack_helper nf_conntrack_helper_h245 __read_mostly = {
620 .name = "H.245", 620 .name = "H.245",
621 .me = THIS_MODULE, 621 .me = THIS_MODULE,
622 .tuple.src.l3num = AF_UNSPEC,
622 .tuple.dst.protonum = IPPROTO_UDP, 623 .tuple.dst.protonum = IPPROTO_UDP,
623 .help = h245_help, 624 .help = h245_help,
624 .expect_policy = &h245_exp_policy, 625 .expect_policy = &h245_exp_policy,
@@ -1765,6 +1766,7 @@ static void __exit nf_conntrack_h323_fini(void)
1765 nf_conntrack_helper_unregister(&nf_conntrack_helper_ras[0]); 1766 nf_conntrack_helper_unregister(&nf_conntrack_helper_ras[0]);
1766 nf_conntrack_helper_unregister(&nf_conntrack_helper_q931[1]); 1767 nf_conntrack_helper_unregister(&nf_conntrack_helper_q931[1]);
1767 nf_conntrack_helper_unregister(&nf_conntrack_helper_q931[0]); 1768 nf_conntrack_helper_unregister(&nf_conntrack_helper_q931[0]);
1769 nf_conntrack_helper_unregister(&nf_conntrack_helper_h245);
1768 kfree(h323_buffer); 1770 kfree(h323_buffer);
1769 pr_debug("nf_ct_h323: fini\n"); 1771 pr_debug("nf_ct_h323: fini\n");
1770} 1772}
@@ -1777,28 +1779,34 @@ static int __init nf_conntrack_h323_init(void)
1777 h323_buffer = kmalloc(65536, GFP_KERNEL); 1779 h323_buffer = kmalloc(65536, GFP_KERNEL);
1778 if (!h323_buffer) 1780 if (!h323_buffer)
1779 return -ENOMEM; 1781 return -ENOMEM;
1780 ret = nf_conntrack_helper_register(&nf_conntrack_helper_q931[0]); 1782 ret = nf_conntrack_helper_register(&nf_conntrack_helper_h245);
1781 if (ret < 0) 1783 if (ret < 0)
1782 goto err1; 1784 goto err1;
1783 ret = nf_conntrack_helper_register(&nf_conntrack_helper_q931[1]); 1785 ret = nf_conntrack_helper_register(&nf_conntrack_helper_q931[0]);
1784 if (ret < 0) 1786 if (ret < 0)
1785 goto err2; 1787 goto err2;
1786 ret = nf_conntrack_helper_register(&nf_conntrack_helper_ras[0]); 1788 ret = nf_conntrack_helper_register(&nf_conntrack_helper_q931[1]);
1787 if (ret < 0) 1789 if (ret < 0)
1788 goto err3; 1790 goto err3;
1789 ret = nf_conntrack_helper_register(&nf_conntrack_helper_ras[1]); 1791 ret = nf_conntrack_helper_register(&nf_conntrack_helper_ras[0]);
1790 if (ret < 0) 1792 if (ret < 0)
1791 goto err4; 1793 goto err4;
1794 ret = nf_conntrack_helper_register(&nf_conntrack_helper_ras[1]);
1795 if (ret < 0)
1796 goto err5;
1792 pr_debug("nf_ct_h323: init success\n"); 1797 pr_debug("nf_ct_h323: init success\n");
1793 return 0; 1798 return 0;
1794 1799
1795err4: 1800err5:
1796 nf_conntrack_helper_unregister(&nf_conntrack_helper_ras[0]); 1801 nf_conntrack_helper_unregister(&nf_conntrack_helper_ras[0]);
1797err3: 1802err4:
1798 nf_conntrack_helper_unregister(&nf_conntrack_helper_q931[1]); 1803 nf_conntrack_helper_unregister(&nf_conntrack_helper_q931[1]);
1799err2: 1804err3:
1800 nf_conntrack_helper_unregister(&nf_conntrack_helper_q931[0]); 1805 nf_conntrack_helper_unregister(&nf_conntrack_helper_q931[0]);
1806err2:
1807 nf_conntrack_helper_unregister(&nf_conntrack_helper_h245);
1801err1: 1808err1:
1809 kfree(h323_buffer);
1802 return ret; 1810 return ret;
1803} 1811}
1804 1812
diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c
index 16774ecd1c4e..0edefcfc5949 100644
--- a/net/netfilter/nf_conntrack_netlink.c
+++ b/net/netfilter/nf_conntrack_netlink.c
@@ -472,6 +472,9 @@ static int ctnetlink_conntrack_event(struct notifier_block *this,
472 goto nla_put_failure; 472 goto nla_put_failure;
473 nla_nest_end(skb, nest_parms); 473 nla_nest_end(skb, nest_parms);
474 474
475 if (ctnetlink_dump_id(skb, ct) < 0)
476 goto nla_put_failure;
477
475 if (events & IPCT_DESTROY) { 478 if (events & IPCT_DESTROY) {
476 if (ctnetlink_dump_counters(skb, ct, IP_CT_DIR_ORIGINAL) < 0 || 479 if (ctnetlink_dump_counters(skb, ct, IP_CT_DIR_ORIGINAL) < 0 ||
477 ctnetlink_dump_counters(skb, ct, IP_CT_DIR_REPLY) < 0) 480 ctnetlink_dump_counters(skb, ct, IP_CT_DIR_REPLY) < 0)
diff --git a/net/netfilter/nf_conntrack_proto_tcp.c b/net/netfilter/nf_conntrack_proto_tcp.c
index ba94004fe323..dd28fb239a60 100644
--- a/net/netfilter/nf_conntrack_proto_tcp.c
+++ b/net/netfilter/nf_conntrack_proto_tcp.c
@@ -331,12 +331,13 @@ static unsigned int get_conntrack_index(const struct tcphdr *tcph)
331 331
332 I. Upper bound for valid data: seq <= sender.td_maxend 332 I. Upper bound for valid data: seq <= sender.td_maxend
333 II. Lower bound for valid data: seq + len >= sender.td_end - receiver.td_maxwin 333 II. Lower bound for valid data: seq + len >= sender.td_end - receiver.td_maxwin
334 III. Upper bound for valid ack: sack <= receiver.td_end 334 III. Upper bound for valid (s)ack: sack <= receiver.td_end
335 IV. Lower bound for valid ack: ack >= receiver.td_end - MAXACKWINDOW 335 IV. Lower bound for valid (s)ack: sack >= receiver.td_end - MAXACKWINDOW
336 336
337 where sack is the highest right edge of sack block found in the packet. 337 where sack is the highest right edge of sack block found in the packet
338 or ack in the case of packet without SACK option.
338 339
339 The upper bound limit for a valid ack is not ignored - 340 The upper bound limit for a valid (s)ack is not ignored -
340 we doesn't have to deal with fragments. 341 we doesn't have to deal with fragments.
341*/ 342*/
342 343
@@ -606,12 +607,12 @@ static bool tcp_in_window(const struct nf_conn *ct,
606 before(seq, sender->td_maxend + 1), 607 before(seq, sender->td_maxend + 1),
607 after(end, sender->td_end - receiver->td_maxwin - 1), 608 after(end, sender->td_end - receiver->td_maxwin - 1),
608 before(sack, receiver->td_end + 1), 609 before(sack, receiver->td_end + 1),
609 after(ack, receiver->td_end - MAXACKWINDOW(sender))); 610 after(sack, receiver->td_end - MAXACKWINDOW(sender) - 1));
610 611
611 if (before(seq, sender->td_maxend + 1) && 612 if (before(seq, sender->td_maxend + 1) &&
612 after(end, sender->td_end - receiver->td_maxwin - 1) && 613 after(end, sender->td_end - receiver->td_maxwin - 1) &&
613 before(sack, receiver->td_end + 1) && 614 before(sack, receiver->td_end + 1) &&
614 after(ack, receiver->td_end - MAXACKWINDOW(sender))) { 615 after(sack, receiver->td_end - MAXACKWINDOW(sender) - 1)) {
615 /* 616 /*
616 * Take into account window scaling (RFC 1323). 617 * Take into account window scaling (RFC 1323).
617 */ 618 */
@@ -843,9 +844,15 @@ static int tcp_packet(struct nf_conn *ct,
843 /* Attempt to reopen a closed/aborted connection. 844 /* Attempt to reopen a closed/aborted connection.
844 * Delete this connection and look up again. */ 845 * Delete this connection and look up again. */
845 write_unlock_bh(&tcp_lock); 846 write_unlock_bh(&tcp_lock);
846 if (del_timer(&ct->timeout)) 847 /* Only repeat if we can actually remove the timer.
848 * Destruction may already be in progress in process
849 * context and we must give it a chance to terminate.
850 */
851 if (del_timer(&ct->timeout)) {
847 ct->timeout.function((unsigned long)ct); 852 ct->timeout.function((unsigned long)ct);
848 return -NF_REPEAT; 853 return -NF_REPEAT;
854 }
855 return -NF_DROP;
849 } 856 }
850 /* Fall through */ 857 /* Fall through */
851 case TCP_CONNTRACK_IGNORE: 858 case TCP_CONNTRACK_IGNORE:
diff --git a/net/netfilter/nf_conntrack_sip.c b/net/netfilter/nf_conntrack_sip.c
index 9f4900069561..2f9bbc058b48 100644
--- a/net/netfilter/nf_conntrack_sip.c
+++ b/net/netfilter/nf_conntrack_sip.c
@@ -870,6 +870,7 @@ static int process_sdp(struct sk_buff *skb,
870{ 870{
871 enum ip_conntrack_info ctinfo; 871 enum ip_conntrack_info ctinfo;
872 struct nf_conn *ct = nf_ct_get(skb, &ctinfo); 872 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
873 struct nf_conn_help *help = nfct_help(ct);
873 unsigned int matchoff, matchlen; 874 unsigned int matchoff, matchlen;
874 unsigned int mediaoff, medialen; 875 unsigned int mediaoff, medialen;
875 unsigned int sdpoff; 876 unsigned int sdpoff;
@@ -959,6 +960,9 @@ static int process_sdp(struct sk_buff *skb,
959 if (nf_nat_sdp_session && ct->status & IPS_NAT_MASK) 960 if (nf_nat_sdp_session && ct->status & IPS_NAT_MASK)
960 ret = nf_nat_sdp_session(skb, dptr, sdpoff, datalen, &rtp_addr); 961 ret = nf_nat_sdp_session(skb, dptr, sdpoff, datalen, &rtp_addr);
961 962
963 if (ret == NF_ACCEPT && i > 0)
964 help->help.ct_sip_info.invite_cseq = cseq;
965
962 return ret; 966 return ret;
963} 967}
964static int process_invite_response(struct sk_buff *skb, 968static int process_invite_response(struct sk_buff *skb,
@@ -967,14 +971,14 @@ static int process_invite_response(struct sk_buff *skb,
967{ 971{
968 enum ip_conntrack_info ctinfo; 972 enum ip_conntrack_info ctinfo;
969 struct nf_conn *ct = nf_ct_get(skb, &ctinfo); 973 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
974 struct nf_conn_help *help = nfct_help(ct);
970 975
971 if ((code >= 100 && code <= 199) || 976 if ((code >= 100 && code <= 199) ||
972 (code >= 200 && code <= 299)) 977 (code >= 200 && code <= 299))
973 return process_sdp(skb, dptr, datalen, cseq); 978 return process_sdp(skb, dptr, datalen, cseq);
974 else { 979 else if (help->help.ct_sip_info.invite_cseq == cseq)
975 flush_expectations(ct, true); 980 flush_expectations(ct, true);
976 return NF_ACCEPT; 981 return NF_ACCEPT;
977 }
978} 982}
979 983
980static int process_update_response(struct sk_buff *skb, 984static int process_update_response(struct sk_buff *skb,
@@ -983,14 +987,14 @@ static int process_update_response(struct sk_buff *skb,
983{ 987{
984 enum ip_conntrack_info ctinfo; 988 enum ip_conntrack_info ctinfo;
985 struct nf_conn *ct = nf_ct_get(skb, &ctinfo); 989 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
990 struct nf_conn_help *help = nfct_help(ct);
986 991
987 if ((code >= 100 && code <= 199) || 992 if ((code >= 100 && code <= 199) ||
988 (code >= 200 && code <= 299)) 993 (code >= 200 && code <= 299))
989 return process_sdp(skb, dptr, datalen, cseq); 994 return process_sdp(skb, dptr, datalen, cseq);
990 else { 995 else if (help->help.ct_sip_info.invite_cseq == cseq)
991 flush_expectations(ct, true); 996 flush_expectations(ct, true);
992 return NF_ACCEPT; 997 return NF_ACCEPT;
993 }
994} 998}
995 999
996static int process_prack_response(struct sk_buff *skb, 1000static int process_prack_response(struct sk_buff *skb,
@@ -999,14 +1003,14 @@ static int process_prack_response(struct sk_buff *skb,
999{ 1003{
1000 enum ip_conntrack_info ctinfo; 1004 enum ip_conntrack_info ctinfo;
1001 struct nf_conn *ct = nf_ct_get(skb, &ctinfo); 1005 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1006 struct nf_conn_help *help = nfct_help(ct);
1002 1007
1003 if ((code >= 100 && code <= 199) || 1008 if ((code >= 100 && code <= 199) ||
1004 (code >= 200 && code <= 299)) 1009 (code >= 200 && code <= 299))
1005 return process_sdp(skb, dptr, datalen, cseq); 1010 return process_sdp(skb, dptr, datalen, cseq);
1006 else { 1011 else if (help->help.ct_sip_info.invite_cseq == cseq)
1007 flush_expectations(ct, true); 1012 flush_expectations(ct, true);
1008 return NF_ACCEPT; 1013 return NF_ACCEPT;
1009 }
1010} 1014}
1011 1015
1012static int process_bye_request(struct sk_buff *skb, 1016static int process_bye_request(struct sk_buff *skb,
diff --git a/net/netfilter/nf_conntrack_standalone.c b/net/netfilter/nf_conntrack_standalone.c
index b59871f6bdda..46ea542d0df9 100644
--- a/net/netfilter/nf_conntrack_standalone.c
+++ b/net/netfilter/nf_conntrack_standalone.c
@@ -296,11 +296,11 @@ static int nf_conntrack_standalone_init_proc(void)
296 pde = proc_net_fops_create(&init_net, "nf_conntrack", 0440, &ct_file_ops); 296 pde = proc_net_fops_create(&init_net, "nf_conntrack", 0440, &ct_file_ops);
297 if (!pde) 297 if (!pde)
298 goto out_nf_conntrack; 298 goto out_nf_conntrack;
299 pde = create_proc_entry("nf_conntrack", S_IRUGO, init_net.proc_net_stat); 299
300 pde = proc_create("nf_conntrack", S_IRUGO, init_net.proc_net_stat,
301 &ct_cpu_seq_fops);
300 if (!pde) 302 if (!pde)
301 goto out_stat_nf_conntrack; 303 goto out_stat_nf_conntrack;
302 pde->proc_fops = &ct_cpu_seq_fops;
303 pde->owner = THIS_MODULE;
304 return 0; 304 return 0;
305 305
306out_stat_nf_conntrack: 306out_stat_nf_conntrack:
diff --git a/net/netfilter/nf_log.c b/net/netfilter/nf_log.c
index bc11d7092032..9fda6ee95a31 100644
--- a/net/netfilter/nf_log.c
+++ b/net/netfilter/nf_log.c
@@ -92,10 +92,6 @@ void nf_log_packet(int pf,
92 vsnprintf(prefix, sizeof(prefix), fmt, args); 92 vsnprintf(prefix, sizeof(prefix), fmt, args);
93 va_end(args); 93 va_end(args);
94 logger->logfn(pf, hooknum, skb, in, out, loginfo, prefix); 94 logger->logfn(pf, hooknum, skb, in, out, loginfo, prefix);
95 } else if (net_ratelimit()) {
96 printk(KERN_WARNING "nf_log_packet: can\'t log since "
97 "no backend logging module loaded in! Please either "
98 "load one, or disable logging explicitly\n");
99 } 95 }
100 rcu_read_unlock(); 96 rcu_read_unlock();
101} 97}
diff --git a/net/netfilter/x_tables.c b/net/netfilter/x_tables.c
index 11b22abc2b70..5d75cd86ebb3 100644
--- a/net/netfilter/x_tables.c
+++ b/net/netfilter/x_tables.c
@@ -936,25 +936,24 @@ int xt_proto_init(struct net *net, int af)
936#ifdef CONFIG_PROC_FS 936#ifdef CONFIG_PROC_FS
937 strlcpy(buf, xt_prefix[af], sizeof(buf)); 937 strlcpy(buf, xt_prefix[af], sizeof(buf));
938 strlcat(buf, FORMAT_TABLES, sizeof(buf)); 938 strlcat(buf, FORMAT_TABLES, sizeof(buf));
939 proc = proc_net_fops_create(net, buf, 0440, &xt_table_ops); 939 proc = proc_create_data(buf, 0440, net->proc_net, &xt_table_ops,
940 (void *)(unsigned long)af);
940 if (!proc) 941 if (!proc)
941 goto out; 942 goto out;
942 proc->data = (void *)(unsigned long)af;
943
944 943
945 strlcpy(buf, xt_prefix[af], sizeof(buf)); 944 strlcpy(buf, xt_prefix[af], sizeof(buf));
946 strlcat(buf, FORMAT_MATCHES, sizeof(buf)); 945 strlcat(buf, FORMAT_MATCHES, sizeof(buf));
947 proc = proc_net_fops_create(net, buf, 0440, &xt_match_ops); 946 proc = proc_create_data(buf, 0440, net->proc_net, &xt_match_ops,
947 (void *)(unsigned long)af);
948 if (!proc) 948 if (!proc)
949 goto out_remove_tables; 949 goto out_remove_tables;
950 proc->data = (void *)(unsigned long)af;
951 950
952 strlcpy(buf, xt_prefix[af], sizeof(buf)); 951 strlcpy(buf, xt_prefix[af], sizeof(buf));
953 strlcat(buf, FORMAT_TARGETS, sizeof(buf)); 952 strlcat(buf, FORMAT_TARGETS, sizeof(buf));
954 proc = proc_net_fops_create(net, buf, 0440, &xt_target_ops); 953 proc = proc_create_data(buf, 0440, net->proc_net, &xt_target_ops,
954 (void *)(unsigned long)af);
955 if (!proc) 955 if (!proc)
956 goto out_remove_matches; 956 goto out_remove_matches;
957 proc->data = (void *)(unsigned long)af;
958#endif 957#endif
959 958
960 return 0; 959 return 0;
diff --git a/net/netfilter/xt_connlimit.c b/net/netfilter/xt_connlimit.c
index 2e89a00df92c..70907f6baac3 100644
--- a/net/netfilter/xt_connlimit.c
+++ b/net/netfilter/xt_connlimit.c
@@ -73,7 +73,8 @@ connlimit_iphash6(const union nf_inet_addr *addr,
73static inline bool already_closed(const struct nf_conn *conn) 73static inline bool already_closed(const struct nf_conn *conn)
74{ 74{
75 if (nf_ct_protonum(conn) == IPPROTO_TCP) 75 if (nf_ct_protonum(conn) == IPPROTO_TCP)
76 return conn->proto.tcp.state == TCP_CONNTRACK_TIME_WAIT; 76 return conn->proto.tcp.state == TCP_CONNTRACK_TIME_WAIT ||
77 conn->proto.tcp.state == TCP_CONNTRACK_CLOSE;
77 else 78 else
78 return 0; 79 return 0;
79} 80}
diff --git a/net/netfilter/xt_hashlimit.c b/net/netfilter/xt_hashlimit.c
index 40d344b21453..6809af542a2c 100644
--- a/net/netfilter/xt_hashlimit.c
+++ b/net/netfilter/xt_hashlimit.c
@@ -237,15 +237,15 @@ static int htable_create_v0(struct xt_hashlimit_info *minfo, int family)
237 hinfo->family = family; 237 hinfo->family = family;
238 hinfo->rnd_initialized = 0; 238 hinfo->rnd_initialized = 0;
239 spin_lock_init(&hinfo->lock); 239 spin_lock_init(&hinfo->lock);
240 hinfo->pde = proc_create(minfo->name, 0, 240 hinfo->pde =
241 proc_create_data(minfo->name, 0,
241 family == AF_INET ? hashlimit_procdir4 : 242 family == AF_INET ? hashlimit_procdir4 :
242 hashlimit_procdir6, 243 hashlimit_procdir6,
243 &dl_file_ops); 244 &dl_file_ops, hinfo);
244 if (!hinfo->pde) { 245 if (!hinfo->pde) {
245 vfree(hinfo); 246 vfree(hinfo);
246 return -1; 247 return -1;
247 } 248 }
248 hinfo->pde->data = hinfo;
249 249
250 setup_timer(&hinfo->timer, htable_gc, (unsigned long )hinfo); 250 setup_timer(&hinfo->timer, htable_gc, (unsigned long )hinfo);
251 hinfo->timer.expires = jiffies + msecs_to_jiffies(hinfo->cfg.gc_interval); 251 hinfo->timer.expires = jiffies + msecs_to_jiffies(hinfo->cfg.gc_interval);
@@ -301,15 +301,15 @@ static int htable_create(struct xt_hashlimit_mtinfo1 *minfo,
301 hinfo->rnd_initialized = 0; 301 hinfo->rnd_initialized = 0;
302 spin_lock_init(&hinfo->lock); 302 spin_lock_init(&hinfo->lock);
303 303
304 hinfo->pde = proc_create(minfo->name, 0, 304 hinfo->pde =
305 proc_create_data(minfo->name, 0,
305 family == AF_INET ? hashlimit_procdir4 : 306 family == AF_INET ? hashlimit_procdir4 :
306 hashlimit_procdir6, 307 hashlimit_procdir6,
307 &dl_file_ops); 308 &dl_file_ops, hinfo);
308 if (hinfo->pde == NULL) { 309 if (hinfo->pde == NULL) {
309 vfree(hinfo); 310 vfree(hinfo);
310 return -1; 311 return -1;
311 } 312 }
312 hinfo->pde->data = hinfo;
313 313
314 setup_timer(&hinfo->timer, htable_gc, (unsigned long)hinfo); 314 setup_timer(&hinfo->timer, htable_gc, (unsigned long)hinfo);
315 hinfo->timer.expires = jiffies + msecs_to_jiffies(hinfo->cfg.gc_interval); 315 hinfo->timer.expires = jiffies + msecs_to_jiffies(hinfo->cfg.gc_interval);
diff --git a/net/netfilter/xt_iprange.c b/net/netfilter/xt_iprange.c
index 500528d60cd7..c63e9333c755 100644
--- a/net/netfilter/xt_iprange.c
+++ b/net/netfilter/xt_iprange.c
@@ -179,3 +179,5 @@ module_exit(iprange_mt_exit);
179MODULE_LICENSE("GPL"); 179MODULE_LICENSE("GPL");
180MODULE_AUTHOR("Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>, Jan Engelhardt <jengelh@computergmbh.de>"); 180MODULE_AUTHOR("Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>, Jan Engelhardt <jengelh@computergmbh.de>");
181MODULE_DESCRIPTION("Xtables: arbitrary IPv4 range matching"); 181MODULE_DESCRIPTION("Xtables: arbitrary IPv4 range matching");
182MODULE_ALIAS("ipt_iprange");
183MODULE_ALIAS("ip6t_iprange");
diff --git a/net/netlabel/netlabel_cipso_v4.c b/net/netlabel/netlabel_cipso_v4.c
index fdc14a0d21af..9080c61b71a5 100644
--- a/net/netlabel/netlabel_cipso_v4.c
+++ b/net/netlabel/netlabel_cipso_v4.c
@@ -584,12 +584,7 @@ list_start:
584 rcu_read_unlock(); 584 rcu_read_unlock();
585 585
586 genlmsg_end(ans_skb, data); 586 genlmsg_end(ans_skb, data);
587 587 return genlmsg_reply(ans_skb, info);
588 ret_val = genlmsg_reply(ans_skb, info);
589 if (ret_val != 0)
590 goto list_failure;
591
592 return 0;
593 588
594list_retry: 589list_retry:
595 /* XXX - this limit is a guesstimate */ 590 /* XXX - this limit is a guesstimate */
diff --git a/net/netlabel/netlabel_mgmt.c b/net/netlabel/netlabel_mgmt.c
index 22c191267808..44be5d5261f4 100644
--- a/net/netlabel/netlabel_mgmt.c
+++ b/net/netlabel/netlabel_mgmt.c
@@ -386,11 +386,7 @@ static int netlbl_mgmt_listdef(struct sk_buff *skb, struct genl_info *info)
386 rcu_read_unlock(); 386 rcu_read_unlock();
387 387
388 genlmsg_end(ans_skb, data); 388 genlmsg_end(ans_skb, data);
389 389 return genlmsg_reply(ans_skb, info);
390 ret_val = genlmsg_reply(ans_skb, info);
391 if (ret_val != 0)
392 goto listdef_failure;
393 return 0;
394 390
395listdef_failure_lock: 391listdef_failure_lock:
396 rcu_read_unlock(); 392 rcu_read_unlock();
@@ -501,11 +497,7 @@ static int netlbl_mgmt_version(struct sk_buff *skb, struct genl_info *info)
501 goto version_failure; 497 goto version_failure;
502 498
503 genlmsg_end(ans_skb, data); 499 genlmsg_end(ans_skb, data);
504 500 return genlmsg_reply(ans_skb, info);
505 ret_val = genlmsg_reply(ans_skb, info);
506 if (ret_val != 0)
507 goto version_failure;
508 return 0;
509 501
510version_failure: 502version_failure:
511 kfree_skb(ans_skb); 503 kfree_skb(ans_skb);
diff --git a/net/netlabel/netlabel_unlabeled.c b/net/netlabel/netlabel_unlabeled.c
index 0099da5b2591..56f80872924e 100644
--- a/net/netlabel/netlabel_unlabeled.c
+++ b/net/netlabel/netlabel_unlabeled.c
@@ -1107,11 +1107,7 @@ static int netlbl_unlabel_list(struct sk_buff *skb, struct genl_info *info)
1107 goto list_failure; 1107 goto list_failure;
1108 1108
1109 genlmsg_end(ans_skb, data); 1109 genlmsg_end(ans_skb, data);
1110 1110 return genlmsg_reply(ans_skb, info);
1111 ret_val = genlmsg_reply(ans_skb, info);
1112 if (ret_val != 0)
1113 goto list_failure;
1114 return 0;
1115 1111
1116list_failure: 1112list_failure:
1117 kfree_skb(ans_skb); 1113 kfree_skb(ans_skb);
@@ -1534,7 +1530,7 @@ static int netlbl_unlabel_staticlistdef(struct sk_buff *skb,
1534 } 1530 }
1535 } 1531 }
1536 list_for_each_entry_rcu(addr6, &iface->addr6_list, list) { 1532 list_for_each_entry_rcu(addr6, &iface->addr6_list, list) {
1537 if (addr6->valid || iter_addr6++ < skip_addr6) 1533 if (!addr6->valid || iter_addr6++ < skip_addr6)
1538 continue; 1534 continue;
1539 if (netlbl_unlabel_staticlist_gen(NLBL_UNLABEL_C_STATICLISTDEF, 1535 if (netlbl_unlabel_staticlist_gen(NLBL_UNLABEL_C_STATICLISTDEF,
1540 iface, 1536 iface,
diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index 9b97f8006c9c..349aba189558 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c
@@ -886,7 +886,7 @@ retry:
886 return netlink_unicast_kernel(sk, skb); 886 return netlink_unicast_kernel(sk, skb);
887 887
888 if (sk_filter(sk, skb)) { 888 if (sk_filter(sk, skb)) {
889 int err = skb->len; 889 err = skb->len;
890 kfree_skb(skb); 890 kfree_skb(skb);
891 sock_put(sk); 891 sock_put(sk);
892 return err; 892 return err;
diff --git a/net/netlink/attr.c b/net/netlink/attr.c
index feb326f4a752..2d106cfe1d27 100644
--- a/net/netlink/attr.c
+++ b/net/netlink/attr.c
@@ -132,6 +132,7 @@ errout:
132 * @maxtype: maximum attribute type to be expected 132 * @maxtype: maximum attribute type to be expected
133 * @head: head of attribute stream 133 * @head: head of attribute stream
134 * @len: length of attribute stream 134 * @len: length of attribute stream
135 * @policy: validation policy
135 * 136 *
136 * Parses a stream of attributes and stores a pointer to each attribute in 137 * Parses a stream of attributes and stores a pointer to each attribute in
137 * the tb array accessable via the attribute type. Attributes with a type 138 * the tb array accessable via the attribute type. Attributes with a type
@@ -194,7 +195,7 @@ struct nlattr *nla_find(struct nlattr *head, int len, int attrtype)
194/** 195/**
195 * nla_strlcpy - Copy string attribute payload into a sized buffer 196 * nla_strlcpy - Copy string attribute payload into a sized buffer
196 * @dst: where to copy the string to 197 * @dst: where to copy the string to
197 * @src: attribute to copy the string from 198 * @nla: attribute to copy the string from
198 * @dstsize: size of destination buffer 199 * @dstsize: size of destination buffer
199 * 200 *
200 * Copies at most dstsize - 1 bytes into the destination buffer. 201 * Copies at most dstsize - 1 bytes into the destination buffer.
@@ -340,9 +341,9 @@ struct nlattr *nla_reserve(struct sk_buff *skb, int attrtype, int attrlen)
340} 341}
341 342
342/** 343/**
343 * nla_reserve - reserve room for attribute without header 344 * nla_reserve_nohdr - reserve room for attribute without header
344 * @skb: socket buffer to reserve room on 345 * @skb: socket buffer to reserve room on
345 * @len: length of attribute payload 346 * @attrlen: length of attribute payload
346 * 347 *
347 * Reserves room for attribute payload without a header. 348 * Reserves room for attribute payload without a header.
348 * 349 *
@@ -400,13 +401,13 @@ void __nla_put_nohdr(struct sk_buff *skb, int attrlen, const void *data)
400 * @attrlen: length of attribute payload 401 * @attrlen: length of attribute payload
401 * @data: head of attribute payload 402 * @data: head of attribute payload
402 * 403 *
403 * Returns -1 if the tailroom of the skb is insufficient to store 404 * Returns -EMSGSIZE if the tailroom of the skb is insufficient to store
404 * the attribute header and payload. 405 * the attribute header and payload.
405 */ 406 */
406int nla_put(struct sk_buff *skb, int attrtype, int attrlen, const void *data) 407int nla_put(struct sk_buff *skb, int attrtype, int attrlen, const void *data)
407{ 408{
408 if (unlikely(skb_tailroom(skb) < nla_total_size(attrlen))) 409 if (unlikely(skb_tailroom(skb) < nla_total_size(attrlen)))
409 return -1; 410 return -EMSGSIZE;
410 411
411 __nla_put(skb, attrtype, attrlen, data); 412 __nla_put(skb, attrtype, attrlen, data);
412 return 0; 413 return 0;
@@ -418,13 +419,13 @@ int nla_put(struct sk_buff *skb, int attrtype, int attrlen, const void *data)
418 * @attrlen: length of attribute payload 419 * @attrlen: length of attribute payload
419 * @data: head of attribute payload 420 * @data: head of attribute payload
420 * 421 *
421 * Returns -1 if the tailroom of the skb is insufficient to store 422 * Returns -EMSGSIZE if the tailroom of the skb is insufficient to store
422 * the attribute payload. 423 * the attribute payload.
423 */ 424 */
424int nla_put_nohdr(struct sk_buff *skb, int attrlen, const void *data) 425int nla_put_nohdr(struct sk_buff *skb, int attrlen, const void *data)
425{ 426{
426 if (unlikely(skb_tailroom(skb) < NLA_ALIGN(attrlen))) 427 if (unlikely(skb_tailroom(skb) < NLA_ALIGN(attrlen)))
427 return -1; 428 return -EMSGSIZE;
428 429
429 __nla_put_nohdr(skb, attrlen, data); 430 __nla_put_nohdr(skb, attrlen, data);
430 return 0; 431 return 0;
@@ -436,13 +437,13 @@ int nla_put_nohdr(struct sk_buff *skb, int attrlen, const void *data)
436 * @attrlen: length of attribute payload 437 * @attrlen: length of attribute payload
437 * @data: head of attribute payload 438 * @data: head of attribute payload
438 * 439 *
439 * Returns -1 if the tailroom of the skb is insufficient to store 440 * Returns -EMSGSIZE if the tailroom of the skb is insufficient to store
440 * the attribute payload. 441 * the attribute payload.
441 */ 442 */
442int nla_append(struct sk_buff *skb, int attrlen, const void *data) 443int nla_append(struct sk_buff *skb, int attrlen, const void *data)
443{ 444{
444 if (unlikely(skb_tailroom(skb) < NLA_ALIGN(attrlen))) 445 if (unlikely(skb_tailroom(skb) < NLA_ALIGN(attrlen)))
445 return -1; 446 return -EMSGSIZE;
446 447
447 memcpy(skb_put(skb, attrlen), data, attrlen); 448 memcpy(skb_put(skb, attrlen), data, attrlen);
448 return 0; 449 return 0;
diff --git a/net/netlink/genetlink.c b/net/netlink/genetlink.c
index d16929c9b4bc..3e1191cecaf0 100644
--- a/net/netlink/genetlink.c
+++ b/net/netlink/genetlink.c
@@ -444,8 +444,11 @@ static int genl_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh)
444 if (ops->dumpit == NULL) 444 if (ops->dumpit == NULL)
445 return -EOPNOTSUPP; 445 return -EOPNOTSUPP;
446 446
447 return netlink_dump_start(genl_sock, skb, nlh, 447 genl_unlock();
448 ops->dumpit, ops->done); 448 err = netlink_dump_start(genl_sock, skb, nlh,
449 ops->dumpit, ops->done);
450 genl_lock();
451 return err;
449 } 452 }
450 453
451 if (ops->doit == NULL) 454 if (ops->doit == NULL)
@@ -554,7 +557,8 @@ static int ctrl_fill_info(struct genl_family *family, u32 pid, u32 seq,
554 return genlmsg_end(skb, hdr); 557 return genlmsg_end(skb, hdr);
555 558
556nla_put_failure: 559nla_put_failure:
557 return genlmsg_cancel(skb, hdr); 560 genlmsg_cancel(skb, hdr);
561 return -EMSGSIZE;
558} 562}
559 563
560static int ctrl_fill_mcgrp_info(struct genl_multicast_group *grp, u32 pid, 564static int ctrl_fill_mcgrp_info(struct genl_multicast_group *grp, u32 pid,
@@ -590,7 +594,8 @@ static int ctrl_fill_mcgrp_info(struct genl_multicast_group *grp, u32 pid,
590 return genlmsg_end(skb, hdr); 594 return genlmsg_end(skb, hdr);
591 595
592nla_put_failure: 596nla_put_failure:
593 return genlmsg_cancel(skb, hdr); 597 genlmsg_cancel(skb, hdr);
598 return -EMSGSIZE;
594} 599}
595 600
596static int ctrl_dumpfamily(struct sk_buff *skb, struct netlink_callback *cb) 601static int ctrl_dumpfamily(struct sk_buff *skb, struct netlink_callback *cb)
@@ -601,9 +606,6 @@ static int ctrl_dumpfamily(struct sk_buff *skb, struct netlink_callback *cb)
601 int chains_to_skip = cb->args[0]; 606 int chains_to_skip = cb->args[0];
602 int fams_to_skip = cb->args[1]; 607 int fams_to_skip = cb->args[1];
603 608
604 if (chains_to_skip != 0)
605 genl_lock();
606
607 for (i = 0; i < GENL_FAM_TAB_SIZE; i++) { 609 for (i = 0; i < GENL_FAM_TAB_SIZE; i++) {
608 if (i < chains_to_skip) 610 if (i < chains_to_skip)
609 continue; 611 continue;
@@ -621,9 +623,6 @@ static int ctrl_dumpfamily(struct sk_buff *skb, struct netlink_callback *cb)
621 } 623 }
622 624
623errout: 625errout:
624 if (chains_to_skip != 0)
625 genl_unlock();
626
627 cb->args[0] = i; 626 cb->args[0] = i;
628 cb->args[1] = n; 627 cb->args[1] = n;
629 628
@@ -768,7 +767,7 @@ static int __init genl_init(void)
768 767
769 /* we'll bump the group number right afterwards */ 768 /* we'll bump the group number right afterwards */
770 genl_sock = netlink_kernel_create(&init_net, NETLINK_GENERIC, 0, 769 genl_sock = netlink_kernel_create(&init_net, NETLINK_GENERIC, 0,
771 genl_rcv, NULL, THIS_MODULE); 770 genl_rcv, &genl_mutex, THIS_MODULE);
772 if (genl_sock == NULL) 771 if (genl_sock == NULL)
773 panic("GENL: Cannot initialize generic netlink\n"); 772 panic("GENL: Cannot initialize generic netlink\n");
774 773
diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 25070240d4ae..2cee87da4441 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -743,7 +743,7 @@ static int packet_sendmsg(struct kiocb *iocb, struct socket *sock,
743 if (len > dev->mtu+reserve) 743 if (len > dev->mtu+reserve)
744 goto out_unlock; 744 goto out_unlock;
745 745
746 skb = sock_alloc_send_skb(sk, len + LL_RESERVED_SPACE(dev), 746 skb = sock_alloc_send_skb(sk, len + LL_ALLOCATED_SPACE(dev),
747 msg->msg_flags & MSG_DONTWAIT, &err); 747 msg->msg_flags & MSG_DONTWAIT, &err);
748 if (skb==NULL) 748 if (skb==NULL)
749 goto out_unlock; 749 goto out_unlock;
diff --git a/net/rose/rose_route.c b/net/rose/rose_route.c
index 5053a53ba24f..bd593871c81e 100644
--- a/net/rose/rose_route.c
+++ b/net/rose/rose_route.c
@@ -1066,12 +1066,12 @@ out:
1066#ifdef CONFIG_PROC_FS 1066#ifdef CONFIG_PROC_FS
1067 1067
1068static void *rose_node_start(struct seq_file *seq, loff_t *pos) 1068static void *rose_node_start(struct seq_file *seq, loff_t *pos)
1069 __acquires(rose_neigh_list_lock) 1069 __acquires(rose_node_list_lock)
1070{ 1070{
1071 struct rose_node *rose_node; 1071 struct rose_node *rose_node;
1072 int i = 1; 1072 int i = 1;
1073 1073
1074 spin_lock_bh(&rose_neigh_list_lock); 1074 spin_lock_bh(&rose_node_list_lock);
1075 if (*pos == 0) 1075 if (*pos == 0)
1076 return SEQ_START_TOKEN; 1076 return SEQ_START_TOKEN;
1077 1077
@@ -1090,9 +1090,9 @@ static void *rose_node_next(struct seq_file *seq, void *v, loff_t *pos)
1090} 1090}
1091 1091
1092static void rose_node_stop(struct seq_file *seq, void *v) 1092static void rose_node_stop(struct seq_file *seq, void *v)
1093 __releases(rose_neigh_list_lock) 1093 __releases(rose_node_list_lock)
1094{ 1094{
1095 spin_unlock_bh(&rose_neigh_list_lock); 1095 spin_unlock_bh(&rose_node_list_lock);
1096} 1096}
1097 1097
1098static int rose_node_show(struct seq_file *seq, void *v) 1098static int rose_node_show(struct seq_file *seq, void *v)
diff --git a/net/rxrpc/ar-transport.c b/net/rxrpc/ar-transport.c
index bb282a6a19f0..64069c8769a5 100644
--- a/net/rxrpc/ar-transport.c
+++ b/net/rxrpc/ar-transport.c
@@ -184,12 +184,13 @@ void rxrpc_put_transport(struct rxrpc_transport *trans)
184 ASSERTCMP(atomic_read(&trans->usage), >, 0); 184 ASSERTCMP(atomic_read(&trans->usage), >, 0);
185 185
186 trans->put_time = get_seconds(); 186 trans->put_time = get_seconds();
187 if (unlikely(atomic_dec_and_test(&trans->usage))) 187 if (unlikely(atomic_dec_and_test(&trans->usage))) {
188 _debug("zombie"); 188 _debug("zombie");
189 /* let the reaper determine the timeout to avoid a race with 189 /* let the reaper determine the timeout to avoid a race with
190 * overextending the timeout if the reaper is running at the 190 * overextending the timeout if the reaper is running at the
191 * same time */ 191 * same time */
192 rxrpc_queue_delayed_work(&rxrpc_transport_reap, 0); 192 rxrpc_queue_delayed_work(&rxrpc_transport_reap, 0);
193 }
193 _leave(""); 194 _leave("");
194} 195}
195 196
diff --git a/net/sched/Kconfig b/net/sched/Kconfig
index 82adfe6447d7..9437b27ff84d 100644
--- a/net/sched/Kconfig
+++ b/net/sched/Kconfig
@@ -106,17 +106,6 @@ config NET_SCH_PRIO
106 To compile this code as a module, choose M here: the 106 To compile this code as a module, choose M here: the
107 module will be called sch_prio. 107 module will be called sch_prio.
108 108
109config NET_SCH_RR
110 tristate "Multi Band Round Robin Queuing (RR)"
111 select NET_SCH_PRIO
112 ---help---
113 Say Y here if you want to use an n-band round robin packet
114 scheduler.
115
116 The module uses sch_prio for its framework and is aliased as
117 sch_rr, so it will load sch_prio, although it is referred
118 to using sch_rr.
119
120config NET_SCH_RED 109config NET_SCH_RED
121 tristate "Random Early Detection (RED)" 110 tristate "Random Early Detection (RED)"
122 ---help--- 111 ---help---
diff --git a/net/sched/act_simple.c b/net/sched/act_simple.c
index 64b2d136c78e..1d421d059caf 100644
--- a/net/sched/act_simple.c
+++ b/net/sched/act_simple.c
@@ -6,7 +6,7 @@
6 * as published by the Free Software Foundation; either version 6 * as published by the Free Software Foundation; either version
7 * 2 of the License, or (at your option) any later version. 7 * 2 of the License, or (at your option) any later version.
8 * 8 *
9 * Authors: Jamal Hadi Salim (2005) 9 * Authors: Jamal Hadi Salim (2005-8)
10 * 10 *
11 */ 11 */
12 12
@@ -34,6 +34,7 @@ static struct tcf_hashinfo simp_hash_info = {
34 .lock = &simp_lock, 34 .lock = &simp_lock,
35}; 35};
36 36
37#define SIMP_MAX_DATA 32
37static int tcf_simp(struct sk_buff *skb, struct tc_action *a, struct tcf_result *res) 38static int tcf_simp(struct sk_buff *skb, struct tc_action *a, struct tcf_result *res)
38{ 39{
39 struct tcf_defact *d = a->priv; 40 struct tcf_defact *d = a->priv;
@@ -69,23 +70,28 @@ static int tcf_simp_release(struct tcf_defact *d, int bind)
69 return ret; 70 return ret;
70} 71}
71 72
72static int alloc_defdata(struct tcf_defact *d, u32 datalen, void *defdata) 73static int alloc_defdata(struct tcf_defact *d, char *defdata)
73{ 74{
74 d->tcfd_defdata = kmemdup(defdata, datalen, GFP_KERNEL); 75 d->tcfd_defdata = kstrndup(defdata, SIMP_MAX_DATA, GFP_KERNEL);
75 if (unlikely(!d->tcfd_defdata)) 76 if (unlikely(!d->tcfd_defdata))
76 return -ENOMEM; 77 return -ENOMEM;
77 d->tcfd_datalen = datalen; 78
78 return 0; 79 return 0;
79} 80}
80 81
81static int realloc_defdata(struct tcf_defact *d, u32 datalen, void *defdata) 82static void reset_policy(struct tcf_defact *d, char *defdata,
83 struct tc_defact *p)
82{ 84{
83 kfree(d->tcfd_defdata); 85 spin_lock_bh(&d->tcf_lock);
84 return alloc_defdata(d, datalen, defdata); 86 d->tcf_action = p->action;
87 memset(d->tcfd_defdata, 0, SIMP_MAX_DATA);
88 strlcpy(d->tcfd_defdata, defdata, SIMP_MAX_DATA);
89 spin_unlock_bh(&d->tcf_lock);
85} 90}
86 91
87static const struct nla_policy simple_policy[TCA_DEF_MAX + 1] = { 92static const struct nla_policy simple_policy[TCA_DEF_MAX + 1] = {
88 [TCA_DEF_PARMS] = { .len = sizeof(struct tc_defact) }, 93 [TCA_DEF_PARMS] = { .len = sizeof(struct tc_defact) },
94 [TCA_DEF_DATA] = { .type = NLA_STRING, .len = SIMP_MAX_DATA },
89}; 95};
90 96
91static int tcf_simp_init(struct nlattr *nla, struct nlattr *est, 97static int tcf_simp_init(struct nlattr *nla, struct nlattr *est,
@@ -95,28 +101,24 @@ static int tcf_simp_init(struct nlattr *nla, struct nlattr *est,
95 struct tc_defact *parm; 101 struct tc_defact *parm;
96 struct tcf_defact *d; 102 struct tcf_defact *d;
97 struct tcf_common *pc; 103 struct tcf_common *pc;
98 void *defdata; 104 char *defdata;
99 u32 datalen = 0;
100 int ret = 0, err; 105 int ret = 0, err;
101 106
102 if (nla == NULL) 107 if (nla == NULL)
103 return -EINVAL; 108 return -EINVAL;
104 109
105 err = nla_parse_nested(tb, TCA_DEF_MAX, nla, NULL); 110 err = nla_parse_nested(tb, TCA_DEF_MAX, nla, simple_policy);
106 if (err < 0) 111 if (err < 0)
107 return err; 112 return err;
108 113
109 if (tb[TCA_DEF_PARMS] == NULL) 114 if (tb[TCA_DEF_PARMS] == NULL)
110 return -EINVAL; 115 return -EINVAL;
111 116
112 parm = nla_data(tb[TCA_DEF_PARMS]); 117 if (tb[TCA_DEF_DATA] == NULL)
113 defdata = nla_data(tb[TCA_DEF_DATA]);
114 if (defdata == NULL)
115 return -EINVAL; 118 return -EINVAL;
116 119
117 datalen = nla_len(tb[TCA_DEF_DATA]); 120 parm = nla_data(tb[TCA_DEF_PARMS]);
118 if (datalen == 0) 121 defdata = nla_data(tb[TCA_DEF_DATA]);
119 return -EINVAL;
120 122
121 pc = tcf_hash_check(parm->index, a, bind, &simp_hash_info); 123 pc = tcf_hash_check(parm->index, a, bind, &simp_hash_info);
122 if (!pc) { 124 if (!pc) {
@@ -126,11 +128,12 @@ static int tcf_simp_init(struct nlattr *nla, struct nlattr *est,
126 return -ENOMEM; 128 return -ENOMEM;
127 129
128 d = to_defact(pc); 130 d = to_defact(pc);
129 ret = alloc_defdata(d, datalen, defdata); 131 ret = alloc_defdata(d, defdata);
130 if (ret < 0) { 132 if (ret < 0) {
131 kfree(pc); 133 kfree(pc);
132 return ret; 134 return ret;
133 } 135 }
136 d->tcf_action = parm->action;
134 ret = ACT_P_CREATED; 137 ret = ACT_P_CREATED;
135 } else { 138 } else {
136 d = to_defact(pc); 139 d = to_defact(pc);
@@ -138,13 +141,9 @@ static int tcf_simp_init(struct nlattr *nla, struct nlattr *est,
138 tcf_simp_release(d, bind); 141 tcf_simp_release(d, bind);
139 return -EEXIST; 142 return -EEXIST;
140 } 143 }
141 realloc_defdata(d, datalen, defdata); 144 reset_policy(d, defdata, parm);
142 } 145 }
143 146
144 spin_lock_bh(&d->tcf_lock);
145 d->tcf_action = parm->action;
146 spin_unlock_bh(&d->tcf_lock);
147
148 if (ret == ACT_P_CREATED) 147 if (ret == ACT_P_CREATED)
149 tcf_hash_insert(pc, &simp_hash_info); 148 tcf_hash_insert(pc, &simp_hash_info);
150 return ret; 149 return ret;
@@ -172,7 +171,7 @@ static inline int tcf_simp_dump(struct sk_buff *skb, struct tc_action *a,
172 opt.bindcnt = d->tcf_bindcnt - bind; 171 opt.bindcnt = d->tcf_bindcnt - bind;
173 opt.action = d->tcf_action; 172 opt.action = d->tcf_action;
174 NLA_PUT(skb, TCA_DEF_PARMS, sizeof(opt), &opt); 173 NLA_PUT(skb, TCA_DEF_PARMS, sizeof(opt), &opt);
175 NLA_PUT(skb, TCA_DEF_DATA, d->tcfd_datalen, d->tcfd_defdata); 174 NLA_PUT_STRING(skb, TCA_DEF_DATA, d->tcfd_defdata);
176 t.install = jiffies_to_clock_t(jiffies - d->tcf_tm.install); 175 t.install = jiffies_to_clock_t(jiffies - d->tcf_tm.install);
177 t.lastuse = jiffies_to_clock_t(jiffies - d->tcf_tm.lastuse); 176 t.lastuse = jiffies_to_clock_t(jiffies - d->tcf_tm.lastuse);
178 t.expires = jiffies_to_clock_t(d->tcf_tm.expires); 177 t.expires = jiffies_to_clock_t(d->tcf_tm.expires);
diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index 1086df7478bc..9360fc81e8c7 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c
@@ -220,7 +220,7 @@ replay:
220 tp = kzalloc(sizeof(*tp), GFP_KERNEL); 220 tp = kzalloc(sizeof(*tp), GFP_KERNEL);
221 if (tp == NULL) 221 if (tp == NULL)
222 goto errout; 222 goto errout;
223 err = -EINVAL; 223 err = -ENOENT;
224 tp_ops = tcf_proto_lookup_ops(tca[TCA_KIND]); 224 tp_ops = tcf_proto_lookup_ops(tca[TCA_KIND]);
225 if (tp_ops == NULL) { 225 if (tp_ops == NULL) {
226#ifdef CONFIG_KMOD 226#ifdef CONFIG_KMOD
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index c40773cdbe45..10f01ad04380 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -1252,12 +1252,12 @@ void tcf_destroy(struct tcf_proto *tp)
1252 kfree(tp); 1252 kfree(tp);
1253} 1253}
1254 1254
1255void tcf_destroy_chain(struct tcf_proto *fl) 1255void tcf_destroy_chain(struct tcf_proto **fl)
1256{ 1256{
1257 struct tcf_proto *tp; 1257 struct tcf_proto *tp;
1258 1258
1259 while ((tp = fl) != NULL) { 1259 while ((tp = *fl) != NULL) {
1260 fl = tp->next; 1260 *fl = tp->next;
1261 tcf_destroy(tp); 1261 tcf_destroy(tp);
1262 } 1262 }
1263} 1263}
diff --git a/net/sched/sch_atm.c b/net/sched/sch_atm.c
index 335273416384..db0e23ae85f8 100644
--- a/net/sched/sch_atm.c
+++ b/net/sched/sch_atm.c
@@ -160,7 +160,7 @@ static void atm_tc_put(struct Qdisc *sch, unsigned long cl)
160 *prev = flow->next; 160 *prev = flow->next;
161 pr_debug("atm_tc_put: qdisc %p\n", flow->q); 161 pr_debug("atm_tc_put: qdisc %p\n", flow->q);
162 qdisc_destroy(flow->q); 162 qdisc_destroy(flow->q);
163 tcf_destroy_chain(flow->filter_list); 163 tcf_destroy_chain(&flow->filter_list);
164 if (flow->sock) { 164 if (flow->sock) {
165 pr_debug("atm_tc_put: f_count %d\n", 165 pr_debug("atm_tc_put: f_count %d\n",
166 file_count(flow->sock->file)); 166 file_count(flow->sock->file));
@@ -586,10 +586,11 @@ static void atm_tc_destroy(struct Qdisc *sch)
586 struct atm_flow_data *flow; 586 struct atm_flow_data *flow;
587 587
588 pr_debug("atm_tc_destroy(sch %p,[qdisc %p])\n", sch, p); 588 pr_debug("atm_tc_destroy(sch %p,[qdisc %p])\n", sch, p);
589 for (flow = p->flows; flow; flow = flow->next)
590 tcf_destroy_chain(&flow->filter_list);
591
589 /* races ? */ 592 /* races ? */
590 while ((flow = p->flows)) { 593 while ((flow = p->flows)) {
591 tcf_destroy_chain(flow->filter_list);
592 flow->filter_list = NULL;
593 if (flow->ref > 1) 594 if (flow->ref > 1)
594 printk(KERN_ERR "atm_destroy: %p->ref = %d\n", flow, 595 printk(KERN_ERR "atm_destroy: %p->ref = %d\n", flow,
595 flow->ref); 596 flow->ref);
diff --git a/net/sched/sch_cbq.c b/net/sched/sch_cbq.c
index 09969c1fbc08..2a3c97f7dc63 100644
--- a/net/sched/sch_cbq.c
+++ b/net/sched/sch_cbq.c
@@ -1704,7 +1704,7 @@ static void cbq_destroy_class(struct Qdisc *sch, struct cbq_class *cl)
1704 1704
1705 BUG_TRAP(!cl->filters); 1705 BUG_TRAP(!cl->filters);
1706 1706
1707 tcf_destroy_chain(cl->filter_list); 1707 tcf_destroy_chain(&cl->filter_list);
1708 qdisc_destroy(cl->q); 1708 qdisc_destroy(cl->q);
1709 qdisc_put_rtab(cl->R_tab); 1709 qdisc_put_rtab(cl->R_tab);
1710 gen_kill_estimator(&cl->bstats, &cl->rate_est); 1710 gen_kill_estimator(&cl->bstats, &cl->rate_est);
@@ -1728,10 +1728,8 @@ cbq_destroy(struct Qdisc* sch)
1728 * be bound to classes which have been destroyed already. --TGR '04 1728 * be bound to classes which have been destroyed already. --TGR '04
1729 */ 1729 */
1730 for (h = 0; h < 16; h++) { 1730 for (h = 0; h < 16; h++) {
1731 for (cl = q->classes[h]; cl; cl = cl->next) { 1731 for (cl = q->classes[h]; cl; cl = cl->next)
1732 tcf_destroy_chain(cl->filter_list); 1732 tcf_destroy_chain(&cl->filter_list);
1733 cl->filter_list = NULL;
1734 }
1735 } 1733 }
1736 for (h = 0; h < 16; h++) { 1734 for (h = 0; h < 16; h++) {
1737 struct cbq_class *next; 1735 struct cbq_class *next;
diff --git a/net/sched/sch_dsmark.c b/net/sched/sch_dsmark.c
index 0df911fd67b1..c4c1317cd47d 100644
--- a/net/sched/sch_dsmark.c
+++ b/net/sched/sch_dsmark.c
@@ -416,7 +416,7 @@ static void dsmark_destroy(struct Qdisc *sch)
416 416
417 pr_debug("dsmark_destroy(sch %p,[qdisc %p])\n", sch, p); 417 pr_debug("dsmark_destroy(sch %p,[qdisc %p])\n", sch, p);
418 418
419 tcf_destroy_chain(p->filter_list); 419 tcf_destroy_chain(&p->filter_list);
420 qdisc_destroy(p->q); 420 qdisc_destroy(p->q);
421 kfree(p->mask); 421 kfree(p->mask);
422} 422}
@@ -444,7 +444,8 @@ static int dsmark_dump_class(struct Qdisc *sch, unsigned long cl,
444 return nla_nest_end(skb, opts); 444 return nla_nest_end(skb, opts);
445 445
446nla_put_failure: 446nla_put_failure:
447 return nla_nest_cancel(skb, opts); 447 nla_nest_cancel(skb, opts);
448 return -EMSGSIZE;
448} 449}
449 450
450static int dsmark_dump(struct Qdisc *sch, struct sk_buff *skb) 451static int dsmark_dump(struct Qdisc *sch, struct sk_buff *skb)
@@ -466,7 +467,8 @@ static int dsmark_dump(struct Qdisc *sch, struct sk_buff *skb)
466 return nla_nest_end(skb, opts); 467 return nla_nest_end(skb, opts);
467 468
468nla_put_failure: 469nla_put_failure:
469 return nla_nest_cancel(skb, opts); 470 nla_nest_cancel(skb, opts);
471 return -EMSGSIZE;
470} 472}
471 473
472static const struct Qdisc_class_ops dsmark_class_ops = { 474static const struct Qdisc_class_ops dsmark_class_ops = {
diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index b741618e4d54..13afa7214392 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -219,6 +219,7 @@ static void dev_watchdog(unsigned long arg)
219 printk(KERN_INFO "NETDEV WATCHDOG: %s: transmit timed out\n", 219 printk(KERN_INFO "NETDEV WATCHDOG: %s: transmit timed out\n",
220 dev->name); 220 dev->name);
221 dev->tx_timeout(dev); 221 dev->tx_timeout(dev);
222 WARN_ON_ONCE(1);
222 } 223 }
223 if (!mod_timer(&dev->watchdog_timer, round_jiffies(jiffies + dev->watchdog_timeo))) 224 if (!mod_timer(&dev->watchdog_timer, round_jiffies(jiffies + dev->watchdog_timeo)))
224 dev_hold(dev); 225 dev_hold(dev);
@@ -467,7 +468,7 @@ struct Qdisc *qdisc_alloc(struct net_device *dev, struct Qdisc_ops *ops)
467 468
468 return sch; 469 return sch;
469errout: 470errout:
470 return ERR_PTR(-err); 471 return ERR_PTR(err);
471} 472}
472 473
473struct Qdisc * qdisc_create_dflt(struct net_device *dev, struct Qdisc_ops *ops, 474struct Qdisc * qdisc_create_dflt(struct net_device *dev, struct Qdisc_ops *ops,
diff --git a/net/sched/sch_gred.c b/net/sched/sch_gred.c
index 3a9d226ff1e4..c89fba56db56 100644
--- a/net/sched/sch_gred.c
+++ b/net/sched/sch_gred.c
@@ -582,7 +582,8 @@ append_opt:
582 return nla_nest_end(skb, opts); 582 return nla_nest_end(skb, opts);
583 583
584nla_put_failure: 584nla_put_failure:
585 return nla_nest_cancel(skb, opts); 585 nla_nest_cancel(skb, opts);
586 return -EMSGSIZE;
586} 587}
587 588
588static void gred_destroy(struct Qdisc *sch) 589static void gred_destroy(struct Qdisc *sch)
diff --git a/net/sched/sch_hfsc.c b/net/sched/sch_hfsc.c
index 87293d0db1d7..e817aa00441d 100644
--- a/net/sched/sch_hfsc.c
+++ b/net/sched/sch_hfsc.c
@@ -1123,7 +1123,7 @@ hfsc_destroy_class(struct Qdisc *sch, struct hfsc_class *cl)
1123{ 1123{
1124 struct hfsc_sched *q = qdisc_priv(sch); 1124 struct hfsc_sched *q = qdisc_priv(sch);
1125 1125
1126 tcf_destroy_chain(cl->filter_list); 1126 tcf_destroy_chain(&cl->filter_list);
1127 qdisc_destroy(cl->qdisc); 1127 qdisc_destroy(cl->qdisc);
1128 gen_kill_estimator(&cl->bstats, &cl->rate_est); 1128 gen_kill_estimator(&cl->bstats, &cl->rate_est);
1129 if (cl != &q->root) 1129 if (cl != &q->root)
@@ -1360,7 +1360,7 @@ hfsc_dump_class(struct Qdisc *sch, unsigned long arg, struct sk_buff *skb,
1360 1360
1361 nla_put_failure: 1361 nla_put_failure:
1362 nla_nest_cancel(skb, nest); 1362 nla_nest_cancel(skb, nest);
1363 return -1; 1363 return -EMSGSIZE;
1364} 1364}
1365 1365
1366static int 1366static int
@@ -1541,6 +1541,10 @@ hfsc_destroy_qdisc(struct Qdisc *sch)
1541 unsigned int i; 1541 unsigned int i;
1542 1542
1543 for (i = 0; i < HFSC_HSIZE; i++) { 1543 for (i = 0; i < HFSC_HSIZE; i++) {
1544 list_for_each_entry(cl, &q->clhash[i], hlist)
1545 tcf_destroy_chain(&cl->filter_list);
1546 }
1547 for (i = 0; i < HFSC_HSIZE; i++) {
1544 list_for_each_entry_safe(cl, next, &q->clhash[i], hlist) 1548 list_for_each_entry_safe(cl, next, &q->clhash[i], hlist)
1545 hfsc_destroy_class(sch, cl); 1549 hfsc_destroy_class(sch, cl);
1546 } 1550 }
diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c
index 66148cc4759e..3fb58f428f72 100644
--- a/net/sched/sch_htb.c
+++ b/net/sched/sch_htb.c
@@ -28,6 +28,7 @@
28 * $Id: sch_htb.c,v 1.25 2003/12/07 11:08:25 devik Exp devik $ 28 * $Id: sch_htb.c,v 1.25 2003/12/07 11:08:25 devik Exp devik $
29 */ 29 */
30#include <linux/module.h> 30#include <linux/module.h>
31#include <linux/moduleparam.h>
31#include <linux/types.h> 32#include <linux/types.h>
32#include <linux/kernel.h> 33#include <linux/kernel.h>
33#include <linux/string.h> 34#include <linux/string.h>
@@ -53,13 +54,17 @@
53*/ 54*/
54 55
55#define HTB_HSIZE 16 /* classid hash size */ 56#define HTB_HSIZE 16 /* classid hash size */
56#define HTB_HYSTERESIS 1 /* whether to use mode hysteresis for speedup */ 57static int htb_hysteresis __read_mostly = 0; /* whether to use mode hysteresis for speedup */
57#define HTB_VER 0x30011 /* major must be matched with number suplied by TC as version */ 58#define HTB_VER 0x30011 /* major must be matched with number suplied by TC as version */
58 59
59#if HTB_VER >> 16 != TC_HTB_PROTOVER 60#if HTB_VER >> 16 != TC_HTB_PROTOVER
60#error "Mismatched sch_htb.c and pkt_sch.h" 61#error "Mismatched sch_htb.c and pkt_sch.h"
61#endif 62#endif
62 63
64/* Module parameter and sysfs export */
65module_param (htb_hysteresis, int, 0640);
66MODULE_PARM_DESC(htb_hysteresis, "Hysteresis mode, less CPU load, less accurate");
67
63/* used internaly to keep status of single class */ 68/* used internaly to keep status of single class */
64enum htb_cmode { 69enum htb_cmode {
65 HTB_CANT_SEND, /* class can't send and can't borrow */ 70 HTB_CANT_SEND, /* class can't send and can't borrow */
@@ -462,19 +467,21 @@ static void htb_deactivate_prios(struct htb_sched *q, struct htb_class *cl)
462 htb_remove_class_from_row(q, cl, mask); 467 htb_remove_class_from_row(q, cl, mask);
463} 468}
464 469
465#if HTB_HYSTERESIS
466static inline long htb_lowater(const struct htb_class *cl) 470static inline long htb_lowater(const struct htb_class *cl)
467{ 471{
468 return cl->cmode != HTB_CANT_SEND ? -cl->cbuffer : 0; 472 if (htb_hysteresis)
473 return cl->cmode != HTB_CANT_SEND ? -cl->cbuffer : 0;
474 else
475 return 0;
469} 476}
470static inline long htb_hiwater(const struct htb_class *cl) 477static inline long htb_hiwater(const struct htb_class *cl)
471{ 478{
472 return cl->cmode == HTB_CAN_SEND ? -cl->buffer : 0; 479 if (htb_hysteresis)
480 return cl->cmode == HTB_CAN_SEND ? -cl->buffer : 0;
481 else
482 return 0;
473} 483}
474#else 484
475#define htb_lowater(cl) (0)
476#define htb_hiwater(cl) (0)
477#endif
478 485
479/** 486/**
480 * htb_class_mode - computes and returns current class mode 487 * htb_class_mode - computes and returns current class mode
@@ -1197,12 +1204,16 @@ static inline int htb_parent_last_child(struct htb_class *cl)
1197 return 1; 1204 return 1;
1198} 1205}
1199 1206
1200static void htb_parent_to_leaf(struct htb_class *cl, struct Qdisc *new_q) 1207static void htb_parent_to_leaf(struct htb_sched *q, struct htb_class *cl,
1208 struct Qdisc *new_q)
1201{ 1209{
1202 struct htb_class *parent = cl->parent; 1210 struct htb_class *parent = cl->parent;
1203 1211
1204 BUG_TRAP(!cl->level && cl->un.leaf.q && !cl->prio_activity); 1212 BUG_TRAP(!cl->level && cl->un.leaf.q && !cl->prio_activity);
1205 1213
1214 if (parent->cmode != HTB_CAN_SEND)
1215 htb_safe_rb_erase(&parent->pq_node, q->wait_pq + parent->level);
1216
1206 parent->level = 0; 1217 parent->level = 0;
1207 memset(&parent->un.inner, 0, sizeof(parent->un.inner)); 1218 memset(&parent->un.inner, 0, sizeof(parent->un.inner));
1208 INIT_LIST_HEAD(&parent->un.leaf.drop_list); 1219 INIT_LIST_HEAD(&parent->un.leaf.drop_list);
@@ -1227,7 +1238,7 @@ static void htb_destroy_class(struct Qdisc *sch, struct htb_class *cl)
1227 qdisc_put_rtab(cl->rate); 1238 qdisc_put_rtab(cl->rate);
1228 qdisc_put_rtab(cl->ceil); 1239 qdisc_put_rtab(cl->ceil);
1229 1240
1230 tcf_destroy_chain(cl->filter_list); 1241 tcf_destroy_chain(&cl->filter_list);
1231 1242
1232 while (!list_empty(&cl->children)) 1243 while (!list_empty(&cl->children))
1233 htb_destroy_class(sch, list_entry(cl->children.next, 1244 htb_destroy_class(sch, list_entry(cl->children.next,
@@ -1256,7 +1267,7 @@ static void htb_destroy(struct Qdisc *sch)
1256 and surprisingly it worked in 2.4. But it must precede it 1267 and surprisingly it worked in 2.4. But it must precede it
1257 because filter need its target class alive to be able to call 1268 because filter need its target class alive to be able to call
1258 unbind_filter on it (without Oops). */ 1269 unbind_filter on it (without Oops). */
1259 tcf_destroy_chain(q->filter_list); 1270 tcf_destroy_chain(&q->filter_list);
1260 1271
1261 while (!list_empty(&q->root)) 1272 while (!list_empty(&q->root))
1262 htb_destroy_class(sch, list_entry(q->root.next, 1273 htb_destroy_class(sch, list_entry(q->root.next,
@@ -1300,7 +1311,7 @@ static int htb_delete(struct Qdisc *sch, unsigned long arg)
1300 htb_deactivate(q, cl); 1311 htb_deactivate(q, cl);
1301 1312
1302 if (last_child) 1313 if (last_child)
1303 htb_parent_to_leaf(cl, new_q); 1314 htb_parent_to_leaf(q, cl, new_q);
1304 1315
1305 if (--cl->refcnt == 0) 1316 if (--cl->refcnt == 0)
1306 htb_destroy_class(sch, cl); 1317 htb_destroy_class(sch, cl);
diff --git a/net/sched/sch_ingress.c b/net/sched/sch_ingress.c
index 274b1ddb160c..956c80ad5965 100644
--- a/net/sched/sch_ingress.c
+++ b/net/sched/sch_ingress.c
@@ -104,7 +104,7 @@ static void ingress_destroy(struct Qdisc *sch)
104{ 104{
105 struct ingress_qdisc_data *p = qdisc_priv(sch); 105 struct ingress_qdisc_data *p = qdisc_priv(sch);
106 106
107 tcf_destroy_chain(p->filter_list); 107 tcf_destroy_chain(&p->filter_list);
108} 108}
109 109
110static int ingress_dump(struct Qdisc *sch, struct sk_buff *skb) 110static int ingress_dump(struct Qdisc *sch, struct sk_buff *skb)
diff --git a/net/sched/sch_prio.c b/net/sched/sch_prio.c
index 4aa2b45dad0a..5532f1031ab5 100644
--- a/net/sched/sch_prio.c
+++ b/net/sched/sch_prio.c
@@ -219,7 +219,7 @@ prio_destroy(struct Qdisc* sch)
219 int prio; 219 int prio;
220 struct prio_sched_data *q = qdisc_priv(sch); 220 struct prio_sched_data *q = qdisc_priv(sch);
221 221
222 tcf_destroy_chain(q->filter_list); 222 tcf_destroy_chain(&q->filter_list);
223 for (prio=0; prio<q->bands; prio++) 223 for (prio=0; prio<q->bands; prio++)
224 qdisc_destroy(q->queues[prio]); 224 qdisc_destroy(q->queues[prio]);
225} 225}
diff --git a/net/sched/sch_red.c b/net/sched/sch_red.c
index 3dcd493f4f4a..5c569853b9c0 100644
--- a/net/sched/sch_red.c
+++ b/net/sched/sch_red.c
@@ -281,7 +281,8 @@ static int red_dump(struct Qdisc *sch, struct sk_buff *skb)
281 return nla_nest_end(skb, opts); 281 return nla_nest_end(skb, opts);
282 282
283nla_put_failure: 283nla_put_failure:
284 return nla_nest_cancel(skb, opts); 284 nla_nest_cancel(skb, opts);
285 return -EMSGSIZE;
285} 286}
286 287
287static int red_dump_stats(struct Qdisc *sch, struct gnet_dump *d) 288static int red_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index f0463d757a98..6a97afbfb952 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -520,7 +520,7 @@ static void sfq_destroy(struct Qdisc *sch)
520{ 520{
521 struct sfq_sched_data *q = qdisc_priv(sch); 521 struct sfq_sched_data *q = qdisc_priv(sch);
522 522
523 tcf_destroy_chain(q->filter_list); 523 tcf_destroy_chain(&q->filter_list);
524 q->perturb_period = 0; 524 q->perturb_period = 0;
525 del_timer_sync(&q->perturb_timer); 525 del_timer_sync(&q->perturb_timer);
526} 526}
diff --git a/net/sctp/associola.c b/net/sctp/associola.c
index b4cd2b71953f..024c3ebd9661 100644
--- a/net/sctp/associola.c
+++ b/net/sctp/associola.c
@@ -474,6 +474,15 @@ static void sctp_association_destroy(struct sctp_association *asoc)
474void sctp_assoc_set_primary(struct sctp_association *asoc, 474void sctp_assoc_set_primary(struct sctp_association *asoc,
475 struct sctp_transport *transport) 475 struct sctp_transport *transport)
476{ 476{
477 int changeover = 0;
478
479 /* it's a changeover only if we already have a primary path
480 * that we are changing
481 */
482 if (asoc->peer.primary_path != NULL &&
483 asoc->peer.primary_path != transport)
484 changeover = 1 ;
485
477 asoc->peer.primary_path = transport; 486 asoc->peer.primary_path = transport;
478 487
479 /* Set a default msg_name for events. */ 488 /* Set a default msg_name for events. */
@@ -499,12 +508,12 @@ void sctp_assoc_set_primary(struct sctp_association *asoc,
499 * double switch to the same destination address. 508 * double switch to the same destination address.
500 */ 509 */
501 if (transport->cacc.changeover_active) 510 if (transport->cacc.changeover_active)
502 transport->cacc.cycling_changeover = 1; 511 transport->cacc.cycling_changeover = changeover;
503 512
504 /* 2) The sender MUST set CHANGEOVER_ACTIVE to indicate that 513 /* 2) The sender MUST set CHANGEOVER_ACTIVE to indicate that
505 * a changeover has occurred. 514 * a changeover has occurred.
506 */ 515 */
507 transport->cacc.changeover_active = 1; 516 transport->cacc.changeover_active = changeover;
508 517
509 /* 3) The sender MUST store the next TSN to be sent in 518 /* 3) The sender MUST store the next TSN to be sent in
510 * next_tsn_at_change. 519 * next_tsn_at_change.
@@ -1203,6 +1212,9 @@ void sctp_assoc_update_retran_path(struct sctp_association *asoc)
1203 struct list_head *head = &asoc->peer.transport_addr_list; 1212 struct list_head *head = &asoc->peer.transport_addr_list;
1204 struct list_head *pos; 1213 struct list_head *pos;
1205 1214
1215 if (asoc->peer.transport_count == 1)
1216 return;
1217
1206 /* Find the next transport in a round-robin fashion. */ 1218 /* Find the next transport in a round-robin fashion. */
1207 t = asoc->peer.retran_path; 1219 t = asoc->peer.retran_path;
1208 pos = &t->transports; 1220 pos = &t->transports;
@@ -1217,6 +1229,15 @@ void sctp_assoc_update_retran_path(struct sctp_association *asoc)
1217 1229
1218 t = list_entry(pos, struct sctp_transport, transports); 1230 t = list_entry(pos, struct sctp_transport, transports);
1219 1231
1232 /* We have exhausted the list, but didn't find any
1233 * other active transports. If so, use the next
1234 * transport.
1235 */
1236 if (t == asoc->peer.retran_path) {
1237 t = next;
1238 break;
1239 }
1240
1220 /* Try to find an active transport. */ 1241 /* Try to find an active transport. */
1221 1242
1222 if ((t->state == SCTP_ACTIVE) || 1243 if ((t->state == SCTP_ACTIVE) ||
@@ -1229,15 +1250,6 @@ void sctp_assoc_update_retran_path(struct sctp_association *asoc)
1229 if (!next) 1250 if (!next)
1230 next = t; 1251 next = t;
1231 } 1252 }
1232
1233 /* We have exhausted the list, but didn't find any
1234 * other active transports. If so, use the next
1235 * transport.
1236 */
1237 if (t == asoc->peer.retran_path) {
1238 t = next;
1239 break;
1240 }
1241 } 1253 }
1242 1254
1243 asoc->peer.retran_path = t; 1255 asoc->peer.retran_path = t;
diff --git a/net/sctp/ipv6.c b/net/sctp/ipv6.c
index e45e44c60635..a2f4d4d51593 100644
--- a/net/sctp/ipv6.c
+++ b/net/sctp/ipv6.c
@@ -299,7 +299,8 @@ static inline int sctp_v6_addr_match_len(union sctp_addr *s1,
299/* Fills in the source address(saddr) based on the destination address(daddr) 299/* Fills in the source address(saddr) based on the destination address(daddr)
300 * and asoc's bind address list. 300 * and asoc's bind address list.
301 */ 301 */
302static void sctp_v6_get_saddr(struct sctp_association *asoc, 302static void sctp_v6_get_saddr(struct sctp_sock *sk,
303 struct sctp_association *asoc,
303 struct dst_entry *dst, 304 struct dst_entry *dst,
304 union sctp_addr *daddr, 305 union sctp_addr *daddr,
305 union sctp_addr *saddr) 306 union sctp_addr *saddr)
@@ -318,7 +319,7 @@ static void sctp_v6_get_saddr(struct sctp_association *asoc,
318 if (!asoc) { 319 if (!asoc) {
319 ipv6_dev_get_saddr(dst ? ip6_dst_idev(dst)->dev : NULL, 320 ipv6_dev_get_saddr(dst ? ip6_dst_idev(dst)->dev : NULL,
320 &daddr->v6.sin6_addr, 321 &daddr->v6.sin6_addr,
321 inet6_sk(asoc->base.sk)->srcprefs, 322 inet6_sk(&sk->inet.sk)->srcprefs,
322 &saddr->v6.sin6_addr); 323 &saddr->v6.sin6_addr);
323 SCTP_DEBUG_PRINTK("saddr from ipv6_get_saddr: " NIP6_FMT "\n", 324 SCTP_DEBUG_PRINTK("saddr from ipv6_get_saddr: " NIP6_FMT "\n",
324 NIP6(saddr->v6.sin6_addr)); 325 NIP6(saddr->v6.sin6_addr));
@@ -726,6 +727,11 @@ static void sctp_v6_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr)
726 seq_printf(seq, NIP6_FMT " ", NIP6(addr->v6.sin6_addr)); 727 seq_printf(seq, NIP6_FMT " ", NIP6(addr->v6.sin6_addr));
727} 728}
728 729
730static void sctp_v6_ecn_capable(struct sock *sk)
731{
732 inet6_sk(sk)->tclass |= INET_ECN_ECT_0;
733}
734
729/* Initialize a PF_INET6 socket msg_name. */ 735/* Initialize a PF_INET6 socket msg_name. */
730static void sctp_inet6_msgname(char *msgname, int *addr_len) 736static void sctp_inet6_msgname(char *msgname, int *addr_len)
731{ 737{
@@ -996,6 +1002,7 @@ static struct sctp_af sctp_af_inet6 = {
996 .skb_iif = sctp_v6_skb_iif, 1002 .skb_iif = sctp_v6_skb_iif,
997 .is_ce = sctp_v6_is_ce, 1003 .is_ce = sctp_v6_is_ce,
998 .seq_dump_addr = sctp_v6_seq_dump_addr, 1004 .seq_dump_addr = sctp_v6_seq_dump_addr,
1005 .ecn_capable = sctp_v6_ecn_capable,
999 .net_header_len = sizeof(struct ipv6hdr), 1006 .net_header_len = sizeof(struct ipv6hdr),
1000 .sockaddr_len = sizeof(struct sockaddr_in6), 1007 .sockaddr_len = sizeof(struct sockaddr_in6),
1001#ifdef CONFIG_COMPAT 1008#ifdef CONFIG_COMPAT
diff --git a/net/sctp/output.c b/net/sctp/output.c
index cf4f9fb6819d..6d45bae93b46 100644
--- a/net/sctp/output.c
+++ b/net/sctp/output.c
@@ -548,7 +548,7 @@ int sctp_packet_transmit(struct sctp_packet *packet)
548 * Note: The works for IPv6 layer checks this bit too later 548 * Note: The works for IPv6 layer checks this bit too later
549 * in transmission. See IP6_ECN_flow_xmit(). 549 * in transmission. See IP6_ECN_flow_xmit().
550 */ 550 */
551 INET_ECN_xmit(nskb->sk); 551 (*tp->af_specific->ecn_capable)(nskb->sk);
552 552
553 /* Set up the IP options. */ 553 /* Set up the IP options. */
554 /* BUG: not implemented 554 /* BUG: not implemented
diff --git a/net/sctp/outqueue.c b/net/sctp/outqueue.c
index 59edfd25a19c..ace6770e9048 100644
--- a/net/sctp/outqueue.c
+++ b/net/sctp/outqueue.c
@@ -208,6 +208,7 @@ void sctp_outq_init(struct sctp_association *asoc, struct sctp_outq *q)
208 INIT_LIST_HEAD(&q->sacked); 208 INIT_LIST_HEAD(&q->sacked);
209 INIT_LIST_HEAD(&q->abandoned); 209 INIT_LIST_HEAD(&q->abandoned);
210 210
211 q->fast_rtx = 0;
211 q->outstanding_bytes = 0; 212 q->outstanding_bytes = 0;
212 q->empty = 1; 213 q->empty = 1;
213 q->cork = 0; 214 q->cork = 0;
@@ -500,6 +501,7 @@ void sctp_retransmit(struct sctp_outq *q, struct sctp_transport *transport,
500 case SCTP_RTXR_FAST_RTX: 501 case SCTP_RTXR_FAST_RTX:
501 SCTP_INC_STATS(SCTP_MIB_FAST_RETRANSMITS); 502 SCTP_INC_STATS(SCTP_MIB_FAST_RETRANSMITS);
502 sctp_transport_lower_cwnd(transport, SCTP_LOWER_CWND_FAST_RTX); 503 sctp_transport_lower_cwnd(transport, SCTP_LOWER_CWND_FAST_RTX);
504 q->fast_rtx = 1;
503 break; 505 break;
504 case SCTP_RTXR_PMTUD: 506 case SCTP_RTXR_PMTUD:
505 SCTP_INC_STATS(SCTP_MIB_PMTUD_RETRANSMITS); 507 SCTP_INC_STATS(SCTP_MIB_PMTUD_RETRANSMITS);
@@ -518,9 +520,15 @@ void sctp_retransmit(struct sctp_outq *q, struct sctp_transport *transport,
518 * the sender SHOULD try to advance the "Advanced.Peer.Ack.Point" by 520 * the sender SHOULD try to advance the "Advanced.Peer.Ack.Point" by
519 * following the procedures outlined in C1 - C5. 521 * following the procedures outlined in C1 - C5.
520 */ 522 */
521 sctp_generate_fwdtsn(q, q->asoc->ctsn_ack_point); 523 if (reason == SCTP_RTXR_T3_RTX)
524 sctp_generate_fwdtsn(q, q->asoc->ctsn_ack_point);
522 525
523 error = sctp_outq_flush(q, /* rtx_timeout */ 1); 526 /* Flush the queues only on timeout, since fast_rtx is only
527 * triggered during sack processing and the queue
528 * will be flushed at the end.
529 */
530 if (reason != SCTP_RTXR_FAST_RTX)
531 error = sctp_outq_flush(q, /* rtx_timeout */ 1);
524 532
525 if (error) 533 if (error)
526 q->asoc->base.sk->sk_err = -error; 534 q->asoc->base.sk->sk_err = -error;
@@ -538,17 +546,23 @@ static int sctp_outq_flush_rtx(struct sctp_outq *q, struct sctp_packet *pkt,
538 int rtx_timeout, int *start_timer) 546 int rtx_timeout, int *start_timer)
539{ 547{
540 struct list_head *lqueue; 548 struct list_head *lqueue;
541 struct list_head *lchunk;
542 struct sctp_transport *transport = pkt->transport; 549 struct sctp_transport *transport = pkt->transport;
543 sctp_xmit_t status; 550 sctp_xmit_t status;
544 struct sctp_chunk *chunk, *chunk1; 551 struct sctp_chunk *chunk, *chunk1;
545 struct sctp_association *asoc; 552 struct sctp_association *asoc;
553 int fast_rtx;
546 int error = 0; 554 int error = 0;
555 int timer = 0;
556 int done = 0;
547 557
548 asoc = q->asoc; 558 asoc = q->asoc;
549 lqueue = &q->retransmit; 559 lqueue = &q->retransmit;
560 fast_rtx = q->fast_rtx;
550 561
551 /* RFC 2960 6.3.3 Handle T3-rtx Expiration 562 /* This loop handles time-out retransmissions, fast retransmissions,
563 * and retransmissions due to opening of whindow.
564 *
565 * RFC 2960 6.3.3 Handle T3-rtx Expiration
552 * 566 *
553 * E3) Determine how many of the earliest (i.e., lowest TSN) 567 * E3) Determine how many of the earliest (i.e., lowest TSN)
554 * outstanding DATA chunks for the address for which the 568 * outstanding DATA chunks for the address for which the
@@ -563,12 +577,12 @@ static int sctp_outq_flush_rtx(struct sctp_outq *q, struct sctp_packet *pkt,
563 * [Just to be painfully clear, if we are retransmitting 577 * [Just to be painfully clear, if we are retransmitting
564 * because a timeout just happened, we should send only ONE 578 * because a timeout just happened, we should send only ONE
565 * packet of retransmitted data.] 579 * packet of retransmitted data.]
580 *
581 * For fast retransmissions we also send only ONE packet. However,
582 * if we are just flushing the queue due to open window, we'll
583 * try to send as much as possible.
566 */ 584 */
567 lchunk = sctp_list_dequeue(lqueue); 585 list_for_each_entry_safe(chunk, chunk1, lqueue, transmitted_list) {
568
569 while (lchunk) {
570 chunk = list_entry(lchunk, struct sctp_chunk,
571 transmitted_list);
572 586
573 /* Make sure that Gap Acked TSNs are not retransmitted. A 587 /* Make sure that Gap Acked TSNs are not retransmitted. A
574 * simple approach is just to move such TSNs out of the 588 * simple approach is just to move such TSNs out of the
@@ -576,58 +590,60 @@ static int sctp_outq_flush_rtx(struct sctp_outq *q, struct sctp_packet *pkt,
576 * next chunk. 590 * next chunk.
577 */ 591 */
578 if (chunk->tsn_gap_acked) { 592 if (chunk->tsn_gap_acked) {
579 list_add_tail(lchunk, &transport->transmitted); 593 list_del(&chunk->transmitted_list);
580 lchunk = sctp_list_dequeue(lqueue); 594 list_add_tail(&chunk->transmitted_list,
595 &transport->transmitted);
581 continue; 596 continue;
582 } 597 }
583 598
599 /* If we are doing fast retransmit, ignore non-fast_rtransmit
600 * chunks
601 */
602 if (fast_rtx && !chunk->fast_retransmit)
603 continue;
604
584 /* Attempt to append this chunk to the packet. */ 605 /* Attempt to append this chunk to the packet. */
585 status = sctp_packet_append_chunk(pkt, chunk); 606 status = sctp_packet_append_chunk(pkt, chunk);
586 607
587 switch (status) { 608 switch (status) {
588 case SCTP_XMIT_PMTU_FULL: 609 case SCTP_XMIT_PMTU_FULL:
589 /* Send this packet. */ 610 /* Send this packet. */
590 if ((error = sctp_packet_transmit(pkt)) == 0) 611 error = sctp_packet_transmit(pkt);
591 *start_timer = 1;
592 612
593 /* If we are retransmitting, we should only 613 /* If we are retransmitting, we should only
594 * send a single packet. 614 * send a single packet.
595 */ 615 */
596 if (rtx_timeout) { 616 if (rtx_timeout || fast_rtx)
597 list_add(lchunk, lqueue); 617 done = 1;
598 lchunk = NULL;
599 }
600 618
601 /* Bundle lchunk in the next round. */ 619 /* Bundle next chunk in the next round. */
602 break; 620 break;
603 621
604 case SCTP_XMIT_RWND_FULL: 622 case SCTP_XMIT_RWND_FULL:
605 /* Send this packet. */ 623 /* Send this packet. */
606 if ((error = sctp_packet_transmit(pkt)) == 0) 624 error = sctp_packet_transmit(pkt);
607 *start_timer = 1;
608 625
609 /* Stop sending DATA as there is no more room 626 /* Stop sending DATA as there is no more room
610 * at the receiver. 627 * at the receiver.
611 */ 628 */
612 list_add(lchunk, lqueue); 629 done = 1;
613 lchunk = NULL;
614 break; 630 break;
615 631
616 case SCTP_XMIT_NAGLE_DELAY: 632 case SCTP_XMIT_NAGLE_DELAY:
617 /* Send this packet. */ 633 /* Send this packet. */
618 if ((error = sctp_packet_transmit(pkt)) == 0) 634 error = sctp_packet_transmit(pkt);
619 *start_timer = 1;
620 635
621 /* Stop sending DATA because of nagle delay. */ 636 /* Stop sending DATA because of nagle delay. */
622 list_add(lchunk, lqueue); 637 done = 1;
623 lchunk = NULL;
624 break; 638 break;
625 639
626 default: 640 default:
627 /* The append was successful, so add this chunk to 641 /* The append was successful, so add this chunk to
628 * the transmitted list. 642 * the transmitted list.
629 */ 643 */
630 list_add_tail(lchunk, &transport->transmitted); 644 list_del(&chunk->transmitted_list);
645 list_add_tail(&chunk->transmitted_list,
646 &transport->transmitted);
631 647
632 /* Mark the chunk as ineligible for fast retransmit 648 /* Mark the chunk as ineligible for fast retransmit
633 * after it is retransmitted. 649 * after it is retransmitted.
@@ -635,27 +651,44 @@ static int sctp_outq_flush_rtx(struct sctp_outq *q, struct sctp_packet *pkt,
635 if (chunk->fast_retransmit > 0) 651 if (chunk->fast_retransmit > 0)
636 chunk->fast_retransmit = -1; 652 chunk->fast_retransmit = -1;
637 653
638 *start_timer = 1; 654 /* Force start T3-rtx timer when fast retransmitting
639 q->empty = 0; 655 * the earliest outstanding TSN
656 */
657 if (!timer && fast_rtx &&
658 ntohl(chunk->subh.data_hdr->tsn) ==
659 asoc->ctsn_ack_point + 1)
660 timer = 2;
640 661
641 /* Retrieve a new chunk to bundle. */ 662 q->empty = 0;
642 lchunk = sctp_list_dequeue(lqueue);
643 break; 663 break;
644 } 664 }
645 665
646 /* If we are here due to a retransmit timeout or a fast 666 /* Set the timer if there were no errors */
647 * retransmit and if there are any chunks left in the retransmit 667 if (!error && !timer)
648 * queue that could not fit in the PMTU sized packet, they need 668 timer = 1;
649 * to be marked as ineligible for a subsequent fast retransmit. 669
650 */ 670 if (done)
651 if (rtx_timeout && !lchunk) { 671 break;
652 list_for_each_entry(chunk1, lqueue, transmitted_list) { 672 }
653 if (chunk1->fast_retransmit > 0) 673
654 chunk1->fast_retransmit = -1; 674 /* If we are here due to a retransmit timeout or a fast
655 } 675 * retransmit and if there are any chunks left in the retransmit
676 * queue that could not fit in the PMTU sized packet, they need
677 * to be marked as ineligible for a subsequent fast retransmit.
678 */
679 if (rtx_timeout || fast_rtx) {
680 list_for_each_entry(chunk1, lqueue, transmitted_list) {
681 if (chunk1->fast_retransmit > 0)
682 chunk1->fast_retransmit = -1;
656 } 683 }
657 } 684 }
658 685
686 *start_timer = timer;
687
688 /* Clear fast retransmit hint */
689 if (fast_rtx)
690 q->fast_rtx = 0;
691
659 return error; 692 return error;
660} 693}
661 694
@@ -862,7 +895,8 @@ int sctp_outq_flush(struct sctp_outq *q, int rtx_timeout)
862 rtx_timeout, &start_timer); 895 rtx_timeout, &start_timer);
863 896
864 if (start_timer) 897 if (start_timer)
865 sctp_transport_reset_timers(transport); 898 sctp_transport_reset_timers(transport,
899 start_timer-1);
866 900
867 /* This can happen on COOKIE-ECHO resend. Only 901 /* This can happen on COOKIE-ECHO resend. Only
868 * one chunk can get bundled with a COOKIE-ECHO. 902 * one chunk can get bundled with a COOKIE-ECHO.
@@ -977,7 +1011,7 @@ int sctp_outq_flush(struct sctp_outq *q, int rtx_timeout)
977 list_add_tail(&chunk->transmitted_list, 1011 list_add_tail(&chunk->transmitted_list,
978 &transport->transmitted); 1012 &transport->transmitted);
979 1013
980 sctp_transport_reset_timers(transport); 1014 sctp_transport_reset_timers(transport, start_timer-1);
981 1015
982 q->empty = 0; 1016 q->empty = 0;
983 1017
diff --git a/net/sctp/protocol.c b/net/sctp/protocol.c
index 0ec234b762c2..9258dfe784ae 100644
--- a/net/sctp/protocol.c
+++ b/net/sctp/protocol.c
@@ -108,14 +108,23 @@ static __init int sctp_proc_init(void)
108 } 108 }
109 109
110 if (sctp_snmp_proc_init()) 110 if (sctp_snmp_proc_init())
111 goto out_nomem; 111 goto out_snmp_proc_init;
112 if (sctp_eps_proc_init()) 112 if (sctp_eps_proc_init())
113 goto out_nomem; 113 goto out_eps_proc_init;
114 if (sctp_assocs_proc_init()) 114 if (sctp_assocs_proc_init())
115 goto out_nomem; 115 goto out_assocs_proc_init;
116 116
117 return 0; 117 return 0;
118 118
119out_assocs_proc_init:
120 sctp_eps_proc_exit();
121out_eps_proc_init:
122 sctp_snmp_proc_exit();
123out_snmp_proc_init:
124 if (proc_net_sctp) {
125 proc_net_sctp = NULL;
126 remove_proc_entry("sctp", init_net.proc_net);
127 }
119out_nomem: 128out_nomem:
120 return -ENOMEM; 129 return -ENOMEM;
121} 130}
@@ -470,11 +479,11 @@ static struct dst_entry *sctp_v4_get_dst(struct sctp_association *asoc,
470 /* Walk through the bind address list and look for a bind 479 /* Walk through the bind address list and look for a bind
471 * address that matches the source address of the returned dst. 480 * address that matches the source address of the returned dst.
472 */ 481 */
482 sctp_v4_dst_saddr(&dst_saddr, dst, htons(bp->port));
473 rcu_read_lock(); 483 rcu_read_lock();
474 list_for_each_entry_rcu(laddr, &bp->address_list, list) { 484 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
475 if (!laddr->valid || (laddr->state != SCTP_ADDR_SRC)) 485 if (!laddr->valid || (laddr->state != SCTP_ADDR_SRC))
476 continue; 486 continue;
477 sctp_v4_dst_saddr(&dst_saddr, dst, htons(bp->port));
478 if (sctp_v4_cmp_addr(&dst_saddr, &laddr->a)) 487 if (sctp_v4_cmp_addr(&dst_saddr, &laddr->a))
479 goto out_unlock; 488 goto out_unlock;
480 } 489 }
@@ -519,7 +528,8 @@ out:
519/* For v4, the source address is cached in the route entry(dst). So no need 528/* For v4, the source address is cached in the route entry(dst). So no need
520 * to cache it separately and hence this is an empty routine. 529 * to cache it separately and hence this is an empty routine.
521 */ 530 */
522static void sctp_v4_get_saddr(struct sctp_association *asoc, 531static void sctp_v4_get_saddr(struct sctp_sock *sk,
532 struct sctp_association *asoc,
523 struct dst_entry *dst, 533 struct dst_entry *dst,
524 union sctp_addr *daddr, 534 union sctp_addr *daddr,
525 union sctp_addr *saddr) 535 union sctp_addr *saddr)
@@ -616,6 +626,11 @@ static void sctp_v4_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr)
616 seq_printf(seq, "%d.%d.%d.%d ", NIPQUAD(addr->v4.sin_addr)); 626 seq_printf(seq, "%d.%d.%d.%d ", NIPQUAD(addr->v4.sin_addr));
617} 627}
618 628
629static void sctp_v4_ecn_capable(struct sock *sk)
630{
631 INET_ECN_xmit(sk);
632}
633
619/* Event handler for inet address addition/deletion events. 634/* Event handler for inet address addition/deletion events.
620 * The sctp_local_addr_list needs to be protocted by a spin lock since 635 * The sctp_local_addr_list needs to be protocted by a spin lock since
621 * multiple notifiers (say IPv4 and IPv6) may be running at the same 636 * multiple notifiers (say IPv4 and IPv6) may be running at the same
@@ -934,6 +949,7 @@ static struct sctp_af sctp_af_inet = {
934 .skb_iif = sctp_v4_skb_iif, 949 .skb_iif = sctp_v4_skb_iif,
935 .is_ce = sctp_v4_is_ce, 950 .is_ce = sctp_v4_is_ce,
936 .seq_dump_addr = sctp_v4_seq_dump_addr, 951 .seq_dump_addr = sctp_v4_seq_dump_addr,
952 .ecn_capable = sctp_v4_ecn_capable,
937 .net_header_len = sizeof(struct iphdr), 953 .net_header_len = sizeof(struct iphdr),
938 .sockaddr_len = sizeof(struct sockaddr_in), 954 .sockaddr_len = sizeof(struct sockaddr_in),
939#ifdef CONFIG_COMPAT 955#ifdef CONFIG_COMPAT
diff --git a/net/sctp/sm_make_chunk.c b/net/sctp/sm_make_chunk.c
index 81b606424e12..bbc7107c86cf 100644
--- a/net/sctp/sm_make_chunk.c
+++ b/net/sctp/sm_make_chunk.c
@@ -2418,7 +2418,8 @@ static int sctp_process_param(struct sctp_association *asoc,
2418 break; 2418 break;
2419 2419
2420 case SCTP_PARAM_IPV6_ADDRESS: 2420 case SCTP_PARAM_IPV6_ADDRESS:
2421 asoc->peer.ipv6_address = 1; 2421 if (PF_INET6 == asoc->base.sk->sk_family)
2422 asoc->peer.ipv6_address = 1;
2422 break; 2423 break;
2423 2424
2424 case SCTP_PARAM_HOST_NAME_ADDRESS: 2425 case SCTP_PARAM_HOST_NAME_ADDRESS:
@@ -2829,6 +2830,19 @@ static __be16 sctp_process_asconf_param(struct sctp_association *asoc,
2829 addr_param = (union sctp_addr_param *) 2830 addr_param = (union sctp_addr_param *)
2830 ((void *)asconf_param + sizeof(sctp_addip_param_t)); 2831 ((void *)asconf_param + sizeof(sctp_addip_param_t));
2831 2832
2833 switch (addr_param->v4.param_hdr.type) {
2834 case SCTP_PARAM_IPV6_ADDRESS:
2835 if (!asoc->peer.ipv6_address)
2836 return SCTP_ERROR_INV_PARAM;
2837 break;
2838 case SCTP_PARAM_IPV4_ADDRESS:
2839 if (!asoc->peer.ipv4_address)
2840 return SCTP_ERROR_INV_PARAM;
2841 break;
2842 default:
2843 return SCTP_ERROR_INV_PARAM;
2844 }
2845
2832 af = sctp_get_af_specific(param_type2af(addr_param->v4.param_hdr.type)); 2846 af = sctp_get_af_specific(param_type2af(addr_param->v4.param_hdr.type));
2833 if (unlikely(!af)) 2847 if (unlikely(!af))
2834 return SCTP_ERROR_INV_PARAM; 2848 return SCTP_ERROR_INV_PARAM;
diff --git a/net/sctp/sm_statefuns.c b/net/sctp/sm_statefuns.c
index 0c9d5a6950fe..fcdb45d1071b 100644
--- a/net/sctp/sm_statefuns.c
+++ b/net/sctp/sm_statefuns.c
@@ -5899,12 +5899,6 @@ static int sctp_eat_data(const struct sctp_association *asoc,
5899 return SCTP_IERROR_NO_DATA; 5899 return SCTP_IERROR_NO_DATA;
5900 } 5900 }
5901 5901
5902 /* If definately accepting the DATA chunk, record its TSN, otherwise
5903 * wait for renege processing.
5904 */
5905 if (SCTP_CMD_CHUNK_ULP == deliver)
5906 sctp_add_cmd_sf(commands, SCTP_CMD_REPORT_TSN, SCTP_U32(tsn));
5907
5908 chunk->data_accepted = 1; 5902 chunk->data_accepted = 1;
5909 5903
5910 /* Note: Some chunks may get overcounted (if we drop) or overcounted 5904 /* Note: Some chunks may get overcounted (if we drop) or overcounted
@@ -5924,6 +5918,9 @@ static int sctp_eat_data(const struct sctp_association *asoc,
5924 * and discard the DATA chunk. 5918 * and discard the DATA chunk.
5925 */ 5919 */
5926 if (ntohs(data_hdr->stream) >= asoc->c.sinit_max_instreams) { 5920 if (ntohs(data_hdr->stream) >= asoc->c.sinit_max_instreams) {
5921 /* Mark tsn as received even though we drop it */
5922 sctp_add_cmd_sf(commands, SCTP_CMD_REPORT_TSN, SCTP_U32(tsn));
5923
5927 err = sctp_make_op_error(asoc, chunk, SCTP_ERROR_INV_STRM, 5924 err = sctp_make_op_error(asoc, chunk, SCTP_ERROR_INV_STRM,
5928 &data_hdr->stream, 5925 &data_hdr->stream,
5929 sizeof(data_hdr->stream)); 5926 sizeof(data_hdr->stream));
diff --git a/net/sctp/socket.c b/net/sctp/socket.c
index e7e3baf7009e..0dbcde6758ea 100644
--- a/net/sctp/socket.c
+++ b/net/sctp/socket.c
@@ -4401,7 +4401,9 @@ static int sctp_getsockopt_local_addrs_old(struct sock *sk, int len,
4401 if (copy_from_user(&getaddrs, optval, len)) 4401 if (copy_from_user(&getaddrs, optval, len))
4402 return -EFAULT; 4402 return -EFAULT;
4403 4403
4404 if (getaddrs.addr_num <= 0) return -EINVAL; 4404 if (getaddrs.addr_num <= 0 ||
4405 getaddrs.addr_num >= (INT_MAX / sizeof(union sctp_addr)))
4406 return -EINVAL;
4405 /* 4407 /*
4406 * For UDP-style sockets, id specifies the association to query. 4408 * For UDP-style sockets, id specifies the association to query.
4407 * If the id field is set to the value '0' then the locally bound 4409 * If the id field is set to the value '0' then the locally bound
diff --git a/net/sctp/transport.c b/net/sctp/transport.c
index f4938f6c5abe..3f34f61221ec 100644
--- a/net/sctp/transport.c
+++ b/net/sctp/transport.c
@@ -79,6 +79,7 @@ static struct sctp_transport *sctp_transport_init(struct sctp_transport *peer,
79 peer->rttvar = 0; 79 peer->rttvar = 0;
80 peer->srtt = 0; 80 peer->srtt = 0;
81 peer->rto_pending = 0; 81 peer->rto_pending = 0;
82 peer->fast_recovery = 0;
82 83
83 peer->last_time_heard = jiffies; 84 peer->last_time_heard = jiffies;
84 peer->last_time_used = jiffies; 85 peer->last_time_used = jiffies;
@@ -190,7 +191,7 @@ static void sctp_transport_destroy(struct sctp_transport *transport)
190/* Start T3_rtx timer if it is not already running and update the heartbeat 191/* Start T3_rtx timer if it is not already running and update the heartbeat
191 * timer. This routine is called every time a DATA chunk is sent. 192 * timer. This routine is called every time a DATA chunk is sent.
192 */ 193 */
193void sctp_transport_reset_timers(struct sctp_transport *transport) 194void sctp_transport_reset_timers(struct sctp_transport *transport, int force)
194{ 195{
195 /* RFC 2960 6.3.2 Retransmission Timer Rules 196 /* RFC 2960 6.3.2 Retransmission Timer Rules
196 * 197 *
@@ -200,7 +201,7 @@ void sctp_transport_reset_timers(struct sctp_transport *transport)
200 * address. 201 * address.
201 */ 202 */
202 203
203 if (!timer_pending(&transport->T3_rtx_timer)) 204 if (force || !timer_pending(&transport->T3_rtx_timer))
204 if (!mod_timer(&transport->T3_rtx_timer, 205 if (!mod_timer(&transport->T3_rtx_timer,
205 jiffies + transport->rto)) 206 jiffies + transport->rto))
206 sctp_transport_hold(transport); 207 sctp_transport_hold(transport);
@@ -291,7 +292,7 @@ void sctp_transport_route(struct sctp_transport *transport,
291 if (saddr) 292 if (saddr)
292 memcpy(&transport->saddr, saddr, sizeof(union sctp_addr)); 293 memcpy(&transport->saddr, saddr, sizeof(union sctp_addr));
293 else 294 else
294 af->get_saddr(asoc, dst, daddr, &transport->saddr); 295 af->get_saddr(opt, asoc, dst, daddr, &transport->saddr);
295 296
296 transport->dst = dst; 297 transport->dst = dst;
297 if ((transport->param_flags & SPP_PMTUD_DISABLE) && transport->pathmtu) { 298 if ((transport->param_flags & SPP_PMTUD_DISABLE) && transport->pathmtu) {
@@ -403,11 +404,16 @@ void sctp_transport_raise_cwnd(struct sctp_transport *transport,
403 cwnd = transport->cwnd; 404 cwnd = transport->cwnd;
404 flight_size = transport->flight_size; 405 flight_size = transport->flight_size;
405 406
407 /* See if we need to exit Fast Recovery first */
408 if (transport->fast_recovery &&
409 TSN_lte(transport->fast_recovery_exit, sack_ctsn))
410 transport->fast_recovery = 0;
411
406 /* The appropriate cwnd increase algorithm is performed if, and only 412 /* The appropriate cwnd increase algorithm is performed if, and only
407 * if the cumulative TSN has advanced and the congestion window is 413 * if the cumulative TSN whould advanced and the congestion window is
408 * being fully utilized. 414 * being fully utilized.
409 */ 415 */
410 if ((transport->asoc->ctsn_ack_point >= sack_ctsn) || 416 if (TSN_lte(sack_ctsn, transport->asoc->ctsn_ack_point) ||
411 (flight_size < cwnd)) 417 (flight_size < cwnd))
412 return; 418 return;
413 419
@@ -416,17 +422,23 @@ void sctp_transport_raise_cwnd(struct sctp_transport *transport,
416 pmtu = transport->asoc->pathmtu; 422 pmtu = transport->asoc->pathmtu;
417 423
418 if (cwnd <= ssthresh) { 424 if (cwnd <= ssthresh) {
419 /* RFC 2960 7.2.1, sctpimpguide-05 2.14.2 When cwnd is less 425 /* RFC 4960 7.2.1
420 * than or equal to ssthresh an SCTP endpoint MUST use the 426 * o When cwnd is less than or equal to ssthresh, an SCTP
421 * slow start algorithm to increase cwnd only if the current 427 * endpoint MUST use the slow-start algorithm to increase
422 * congestion window is being fully utilized and an incoming 428 * cwnd only if the current congestion window is being fully
423 * SACK advances the Cumulative TSN Ack Point. Only when these 429 * utilized, an incoming SACK advances the Cumulative TSN
424 * two conditions are met can the cwnd be increased otherwise 430 * Ack Point, and the data sender is not in Fast Recovery.
425 * the cwnd MUST not be increased. If these conditions are met 431 * Only when these three conditions are met can the cwnd be
426 * then cwnd MUST be increased by at most the lesser of 432 * increased; otherwise, the cwnd MUST not be increased.
427 * 1) the total size of the previously outstanding DATA 433 * If these conditions are met, then cwnd MUST be increased
428 * chunk(s) acknowledged, and 2) the destination's path MTU. 434 * by, at most, the lesser of 1) the total size of the
435 * previously outstanding DATA chunk(s) acknowledged, and
436 * 2) the destination's path MTU. This upper bound protects
437 * against the ACK-Splitting attack outlined in [SAVAGE99].
429 */ 438 */
439 if (transport->fast_recovery)
440 return;
441
430 if (bytes_acked > pmtu) 442 if (bytes_acked > pmtu)
431 cwnd += pmtu; 443 cwnd += pmtu;
432 else 444 else
@@ -502,6 +514,13 @@ void sctp_transport_lower_cwnd(struct sctp_transport *transport,
502 * cwnd = ssthresh 514 * cwnd = ssthresh
503 * partial_bytes_acked = 0 515 * partial_bytes_acked = 0
504 */ 516 */
517 if (transport->fast_recovery)
518 return;
519
520 /* Mark Fast recovery */
521 transport->fast_recovery = 1;
522 transport->fast_recovery_exit = transport->asoc->next_tsn - 1;
523
505 transport->ssthresh = max(transport->cwnd/2, 524 transport->ssthresh = max(transport->cwnd/2,
506 4*transport->asoc->pathmtu); 525 4*transport->asoc->pathmtu);
507 transport->cwnd = transport->ssthresh; 526 transport->cwnd = transport->ssthresh;
@@ -586,6 +605,7 @@ void sctp_transport_reset(struct sctp_transport *t)
586 t->flight_size = 0; 605 t->flight_size = 0;
587 t->error_count = 0; 606 t->error_count = 0;
588 t->rto_pending = 0; 607 t->rto_pending = 0;
608 t->fast_recovery = 0;
589 609
590 /* Initialize the state information for SFR-CACC */ 610 /* Initialize the state information for SFR-CACC */
591 t->cacc.changeover_active = 0; 611 t->cacc.changeover_active = 0;
diff --git a/net/sctp/ulpevent.c b/net/sctp/ulpevent.c
index ce6cda6b6994..a1f654aea268 100644
--- a/net/sctp/ulpevent.c
+++ b/net/sctp/ulpevent.c
@@ -710,6 +710,11 @@ struct sctp_ulpevent *sctp_ulpevent_make_rcvmsg(struct sctp_association *asoc,
710 if (!skb) 710 if (!skb)
711 goto fail; 711 goto fail;
712 712
713 /* Now that all memory allocations for this chunk succeeded, we
714 * can mark it as received so the tsn_map is updated correctly.
715 */
716 sctp_tsnmap_mark(&asoc->peer.tsn_map, ntohl(chunk->subh.data_hdr->tsn));
717
713 /* First calculate the padding, so we don't inadvertently 718 /* First calculate the padding, so we don't inadvertently
714 * pass up the wrong length to the user. 719 * pass up the wrong length to the user.
715 * 720 *
diff --git a/net/sunrpc/auth_generic.c b/net/sunrpc/auth_generic.c
index d927d9f57412..744b79fdcb19 100644
--- a/net/sunrpc/auth_generic.c
+++ b/net/sunrpc/auth_generic.c
@@ -17,8 +17,8 @@
17# define RPCDBG_FACILITY RPCDBG_AUTH 17# define RPCDBG_FACILITY RPCDBG_AUTH
18#endif 18#endif
19 19
20#define RPC_ANONYMOUS_USERID ((uid_t)-2) 20#define RPC_MACHINE_CRED_USERID ((uid_t)0)
21#define RPC_ANONYMOUS_GROUPID ((gid_t)-2) 21#define RPC_MACHINE_CRED_GROUPID ((gid_t)0)
22 22
23struct generic_cred { 23struct generic_cred {
24 struct rpc_cred gc_base; 24 struct rpc_cred gc_base;
@@ -44,8 +44,8 @@ EXPORT_SYMBOL_GPL(rpc_lookup_cred);
44struct rpc_cred *rpc_lookup_machine_cred(void) 44struct rpc_cred *rpc_lookup_machine_cred(void)
45{ 45{
46 struct auth_cred acred = { 46 struct auth_cred acred = {
47 .uid = RPC_ANONYMOUS_USERID, 47 .uid = RPC_MACHINE_CRED_USERID,
48 .gid = RPC_ANONYMOUS_GROUPID, 48 .gid = RPC_MACHINE_CRED_GROUPID,
49 .machine_cred = 1, 49 .machine_cred = 1,
50 }; 50 };
51 51
diff --git a/net/sunrpc/auth_gss/svcauth_gss.c b/net/sunrpc/auth_gss/svcauth_gss.c
index 5905d56737d6..81ae3d62a0cc 100644
--- a/net/sunrpc/auth_gss/svcauth_gss.c
+++ b/net/sunrpc/auth_gss/svcauth_gss.c
@@ -1144,20 +1144,20 @@ svcauth_gss_accept(struct svc_rqst *rqstp, __be32 *authp)
1144 case RPC_GSS_SVC_NONE: 1144 case RPC_GSS_SVC_NONE:
1145 break; 1145 break;
1146 case RPC_GSS_SVC_INTEGRITY: 1146 case RPC_GSS_SVC_INTEGRITY:
1147 /* placeholders for length and seq. number: */
1148 svc_putnl(resv, 0);
1149 svc_putnl(resv, 0);
1147 if (unwrap_integ_data(&rqstp->rq_arg, 1150 if (unwrap_integ_data(&rqstp->rq_arg,
1148 gc->gc_seq, rsci->mechctx)) 1151 gc->gc_seq, rsci->mechctx))
1149 goto garbage_args; 1152 goto garbage_args;
1153 break;
1154 case RPC_GSS_SVC_PRIVACY:
1150 /* placeholders for length and seq. number: */ 1155 /* placeholders for length and seq. number: */
1151 svc_putnl(resv, 0); 1156 svc_putnl(resv, 0);
1152 svc_putnl(resv, 0); 1157 svc_putnl(resv, 0);
1153 break;
1154 case RPC_GSS_SVC_PRIVACY:
1155 if (unwrap_priv_data(rqstp, &rqstp->rq_arg, 1158 if (unwrap_priv_data(rqstp, &rqstp->rq_arg,
1156 gc->gc_seq, rsci->mechctx)) 1159 gc->gc_seq, rsci->mechctx))
1157 goto garbage_args; 1160 goto garbage_args;
1158 /* placeholders for length and seq. number: */
1159 svc_putnl(resv, 0);
1160 svc_putnl(resv, 0);
1161 break; 1161 break;
1162 default: 1162 default:
1163 goto auth_err; 1163 goto auth_err;
@@ -1170,8 +1170,6 @@ svcauth_gss_accept(struct svc_rqst *rqstp, __be32 *authp)
1170 goto out; 1170 goto out;
1171 } 1171 }
1172garbage_args: 1172garbage_args:
1173 /* Restore write pointer to its original value: */
1174 xdr_ressize_check(rqstp, reject_stat);
1175 ret = SVC_GARBAGE; 1173 ret = SVC_GARBAGE;
1176 goto out; 1174 goto out;
1177auth_err: 1175auth_err:
diff --git a/net/sunrpc/cache.c b/net/sunrpc/cache.c
index d75530ff2a6d..c9966713282a 100644
--- a/net/sunrpc/cache.c
+++ b/net/sunrpc/cache.c
@@ -316,31 +316,28 @@ static int create_cache_proc_entries(struct cache_detail *cd)
316 cd->proc_ent->owner = cd->owner; 316 cd->proc_ent->owner = cd->owner;
317 cd->channel_ent = cd->content_ent = NULL; 317 cd->channel_ent = cd->content_ent = NULL;
318 318
319 p = proc_create("flush", S_IFREG|S_IRUSR|S_IWUSR, 319 p = proc_create_data("flush", S_IFREG|S_IRUSR|S_IWUSR,
320 cd->proc_ent, &cache_flush_operations); 320 cd->proc_ent, &cache_flush_operations, cd);
321 cd->flush_ent = p; 321 cd->flush_ent = p;
322 if (p == NULL) 322 if (p == NULL)
323 goto out_nomem; 323 goto out_nomem;
324 p->owner = cd->owner; 324 p->owner = cd->owner;
325 p->data = cd;
326 325
327 if (cd->cache_request || cd->cache_parse) { 326 if (cd->cache_request || cd->cache_parse) {
328 p = proc_create("channel", S_IFREG|S_IRUSR|S_IWUSR, 327 p = proc_create_data("channel", S_IFREG|S_IRUSR|S_IWUSR,
329 cd->proc_ent, &cache_file_operations); 328 cd->proc_ent, &cache_file_operations, cd);
330 cd->channel_ent = p; 329 cd->channel_ent = p;
331 if (p == NULL) 330 if (p == NULL)
332 goto out_nomem; 331 goto out_nomem;
333 p->owner = cd->owner; 332 p->owner = cd->owner;
334 p->data = cd;
335 } 333 }
336 if (cd->cache_show) { 334 if (cd->cache_show) {
337 p = proc_create("content", S_IFREG|S_IRUSR|S_IWUSR, 335 p = proc_create_data("content", S_IFREG|S_IRUSR|S_IWUSR,
338 cd->proc_ent, &content_file_operations); 336 cd->proc_ent, &content_file_operations, cd);
339 cd->content_ent = p; 337 cd->content_ent = p;
340 if (p == NULL) 338 if (p == NULL)
341 goto out_nomem; 339 goto out_nomem;
342 p->owner = cd->owner; 340 p->owner = cd->owner;
343 p->data = cd;
344 } 341 }
345 return 0; 342 return 0;
346out_nomem: 343out_nomem:
diff --git a/net/sunrpc/rpcb_clnt.c b/net/sunrpc/rpcb_clnt.c
index 0517967a68bf..e6fb21b19b86 100644
--- a/net/sunrpc/rpcb_clnt.c
+++ b/net/sunrpc/rpcb_clnt.c
@@ -243,10 +243,10 @@ int rpcb_getport_sync(struct sockaddr_in *sin, u32 prog, u32 vers, int prot)
243} 243}
244EXPORT_SYMBOL_GPL(rpcb_getport_sync); 244EXPORT_SYMBOL_GPL(rpcb_getport_sync);
245 245
246static struct rpc_task *rpcb_call_async(struct rpc_clnt *rpcb_clnt, struct rpcbind_args *map, int version) 246static struct rpc_task *rpcb_call_async(struct rpc_clnt *rpcb_clnt, struct rpcbind_args *map, struct rpc_procinfo *proc)
247{ 247{
248 struct rpc_message msg = { 248 struct rpc_message msg = {
249 .rpc_proc = rpcb_next_version[version].rpc_proc, 249 .rpc_proc = proc,
250 .rpc_argp = map, 250 .rpc_argp = map,
251 .rpc_resp = &map->r_port, 251 .rpc_resp = &map->r_port,
252 }; 252 };
@@ -271,6 +271,7 @@ static struct rpc_task *rpcb_call_async(struct rpc_clnt *rpcb_clnt, struct rpcbi
271void rpcb_getport_async(struct rpc_task *task) 271void rpcb_getport_async(struct rpc_task *task)
272{ 272{
273 struct rpc_clnt *clnt = task->tk_client; 273 struct rpc_clnt *clnt = task->tk_client;
274 struct rpc_procinfo *proc;
274 u32 bind_version; 275 u32 bind_version;
275 struct rpc_xprt *xprt = task->tk_xprt; 276 struct rpc_xprt *xprt = task->tk_xprt;
276 struct rpc_clnt *rpcb_clnt; 277 struct rpc_clnt *rpcb_clnt;
@@ -280,7 +281,6 @@ void rpcb_getport_async(struct rpc_task *task)
280 struct sockaddr *sap = (struct sockaddr *)&addr; 281 struct sockaddr *sap = (struct sockaddr *)&addr;
281 size_t salen; 282 size_t salen;
282 int status; 283 int status;
283 struct rpcb_info *info;
284 284
285 dprintk("RPC: %5u %s(%s, %u, %u, %d)\n", 285 dprintk("RPC: %5u %s(%s, %u, %u, %d)\n",
286 task->tk_pid, __func__, 286 task->tk_pid, __func__,
@@ -313,10 +313,12 @@ void rpcb_getport_async(struct rpc_task *task)
313 /* Don't ever use rpcbind v2 for AF_INET6 requests */ 313 /* Don't ever use rpcbind v2 for AF_INET6 requests */
314 switch (sap->sa_family) { 314 switch (sap->sa_family) {
315 case AF_INET: 315 case AF_INET:
316 info = rpcb_next_version; 316 proc = rpcb_next_version[xprt->bind_index].rpc_proc;
317 bind_version = rpcb_next_version[xprt->bind_index].rpc_vers;
317 break; 318 break;
318 case AF_INET6: 319 case AF_INET6:
319 info = rpcb_next_version6; 320 proc = rpcb_next_version6[xprt->bind_index].rpc_proc;
321 bind_version = rpcb_next_version6[xprt->bind_index].rpc_vers;
320 break; 322 break;
321 default: 323 default:
322 status = -EAFNOSUPPORT; 324 status = -EAFNOSUPPORT;
@@ -324,14 +326,13 @@ void rpcb_getport_async(struct rpc_task *task)
324 task->tk_pid, __func__); 326 task->tk_pid, __func__);
325 goto bailout_nofree; 327 goto bailout_nofree;
326 } 328 }
327 if (info[xprt->bind_index].rpc_proc == NULL) { 329 if (proc == NULL) {
328 xprt->bind_index = 0; 330 xprt->bind_index = 0;
329 status = -EPFNOSUPPORT; 331 status = -EPFNOSUPPORT;
330 dprintk("RPC: %5u %s: no more getport versions available\n", 332 dprintk("RPC: %5u %s: no more getport versions available\n",
331 task->tk_pid, __func__); 333 task->tk_pid, __func__);
332 goto bailout_nofree; 334 goto bailout_nofree;
333 } 335 }
334 bind_version = info[xprt->bind_index].rpc_vers;
335 336
336 dprintk("RPC: %5u %s: trying rpcbind version %u\n", 337 dprintk("RPC: %5u %s: trying rpcbind version %u\n",
337 task->tk_pid, __func__, bind_version); 338 task->tk_pid, __func__, bind_version);
@@ -361,22 +362,20 @@ void rpcb_getport_async(struct rpc_task *task)
361 map->r_addr = rpc_peeraddr2str(rpcb_clnt, RPC_DISPLAY_UNIVERSAL_ADDR); 362 map->r_addr = rpc_peeraddr2str(rpcb_clnt, RPC_DISPLAY_UNIVERSAL_ADDR);
362 map->r_owner = RPCB_OWNER_STRING; /* ignored for GETADDR */ 363 map->r_owner = RPCB_OWNER_STRING; /* ignored for GETADDR */
363 364
364 child = rpcb_call_async(rpcb_clnt, map, xprt->bind_index); 365 child = rpcb_call_async(rpcb_clnt, map, proc);
365 rpc_release_client(rpcb_clnt); 366 rpc_release_client(rpcb_clnt);
366 if (IS_ERR(child)) { 367 if (IS_ERR(child)) {
367 status = -EIO; 368 status = -EIO;
369 /* rpcb_map_release() has freed the arguments */
368 dprintk("RPC: %5u %s: rpc_run_task failed\n", 370 dprintk("RPC: %5u %s: rpc_run_task failed\n",
369 task->tk_pid, __func__); 371 task->tk_pid, __func__);
370 goto bailout; 372 goto bailout_nofree;
371 } 373 }
372 rpc_put_task(child); 374 rpc_put_task(child);
373 375
374 task->tk_xprt->stat.bind_count++; 376 task->tk_xprt->stat.bind_count++;
375 return; 377 return;
376 378
377bailout:
378 kfree(map);
379 xprt_put(xprt);
380bailout_nofree: 379bailout_nofree:
381 rpcb_wake_rpcbind_waiters(xprt, status); 380 rpcb_wake_rpcbind_waiters(xprt, status);
382bailout_nowake: 381bailout_nowake:
diff --git a/net/sunrpc/stats.c b/net/sunrpc/stats.c
index c6061a4346c8..50b049c6598a 100644
--- a/net/sunrpc/stats.c
+++ b/net/sunrpc/stats.c
@@ -224,16 +224,10 @@ EXPORT_SYMBOL_GPL(rpc_print_iostats);
224static inline struct proc_dir_entry * 224static inline struct proc_dir_entry *
225do_register(const char *name, void *data, const struct file_operations *fops) 225do_register(const char *name, void *data, const struct file_operations *fops)
226{ 226{
227 struct proc_dir_entry *ent;
228
229 rpc_proc_init(); 227 rpc_proc_init();
230 dprintk("RPC: registering /proc/net/rpc/%s\n", name); 228 dprintk("RPC: registering /proc/net/rpc/%s\n", name);
231 229
232 ent = proc_create(name, 0, proc_net_rpc, fops); 230 return proc_create_data(name, 0, proc_net_rpc, fops, data);
233 if (ent) {
234 ent->data = data;
235 }
236 return ent;
237} 231}
238 232
239struct proc_dir_entry * 233struct proc_dir_entry *
diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
index d74c2d269539..01c7e311b904 100644
--- a/net/sunrpc/svc.c
+++ b/net/sunrpc/svc.c
@@ -18,7 +18,6 @@
18#include <linux/mm.h> 18#include <linux/mm.h>
19#include <linux/interrupt.h> 19#include <linux/interrupt.h>
20#include <linux/module.h> 20#include <linux/module.h>
21#include <linux/sched.h>
22 21
23#include <linux/sunrpc/types.h> 22#include <linux/sunrpc/types.h>
24#include <linux/sunrpc/xdr.h> 23#include <linux/sunrpc/xdr.h>
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index d8e8d79a8451..e46c825f4954 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -6,30 +6,9 @@
6 6
7#include <linux/sched.h> 7#include <linux/sched.h>
8#include <linux/errno.h> 8#include <linux/errno.h>
9#include <linux/fcntl.h>
10#include <linux/net.h>
11#include <linux/in.h>
12#include <linux/inet.h>
13#include <linux/udp.h>
14#include <linux/tcp.h>
15#include <linux/unistd.h>
16#include <linux/slab.h>
17#include <linux/netdevice.h>
18#include <linux/skbuff.h>
19#include <linux/file.h>
20#include <linux/freezer.h> 9#include <linux/freezer.h>
21#include <linux/kthread.h> 10#include <linux/kthread.h>
22#include <net/sock.h> 11#include <net/sock.h>
23#include <net/checksum.h>
24#include <net/ip.h>
25#include <net/ipv6.h>
26#include <net/tcp_states.h>
27#include <linux/uaccess.h>
28#include <asm/ioctls.h>
29
30#include <linux/sunrpc/types.h>
31#include <linux/sunrpc/clnt.h>
32#include <linux/sunrpc/xdr.h>
33#include <linux/sunrpc/stats.h> 12#include <linux/sunrpc/stats.h>
34#include <linux/sunrpc/svc_xprt.h> 13#include <linux/sunrpc/svc_xprt.h>
35 14
@@ -296,8 +275,6 @@ void svc_xprt_enqueue(struct svc_xprt *xprt)
296 if (!(xprt->xpt_flags & 275 if (!(xprt->xpt_flags &
297 ((1<<XPT_CONN)|(1<<XPT_DATA)|(1<<XPT_CLOSE)|(1<<XPT_DEFERRED)))) 276 ((1<<XPT_CONN)|(1<<XPT_DATA)|(1<<XPT_CLOSE)|(1<<XPT_DEFERRED))))
298 return; 277 return;
299 if (test_bit(XPT_DEAD, &xprt->xpt_flags))
300 return;
301 278
302 cpu = get_cpu(); 279 cpu = get_cpu();
303 pool = svc_pool_for_cpu(xprt->xpt_server, cpu); 280 pool = svc_pool_for_cpu(xprt->xpt_server, cpu);
diff --git a/net/sunrpc/svcauth_unix.c b/net/sunrpc/svcauth_unix.c
index 3f30ee6006ae..f24800f2c098 100644
--- a/net/sunrpc/svcauth_unix.c
+++ b/net/sunrpc/svcauth_unix.c
@@ -278,7 +278,7 @@ static int ip_map_show(struct seq_file *m,
278 dom = im->m_client->h.name; 278 dom = im->m_client->h.name;
279 279
280 if (ipv6_addr_v4mapped(&addr)) { 280 if (ipv6_addr_v4mapped(&addr)) {
281 seq_printf(m, "%s" NIPQUAD_FMT "%s\n", 281 seq_printf(m, "%s " NIPQUAD_FMT " %s\n",
282 im->m_class, 282 im->m_class,
283 ntohl(addr.s6_addr32[3]) >> 24 & 0xff, 283 ntohl(addr.s6_addr32[3]) >> 24 & 0xff,
284 ntohl(addr.s6_addr32[3]) >> 16 & 0xff, 284 ntohl(addr.s6_addr32[3]) >> 16 & 0xff,
@@ -286,7 +286,7 @@ static int ip_map_show(struct seq_file *m,
286 ntohl(addr.s6_addr32[3]) >> 0 & 0xff, 286 ntohl(addr.s6_addr32[3]) >> 0 & 0xff,
287 dom); 287 dom);
288 } else { 288 } else {
289 seq_printf(m, "%s" NIP6_FMT "%s\n", 289 seq_printf(m, "%s " NIP6_FMT " %s\n",
290 im->m_class, NIP6(addr), dom); 290 im->m_class, NIP6(addr), dom);
291 } 291 }
292 return 0; 292 return 0;
diff --git a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
index c22d6b6f2db4..06ab4841537b 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
@@ -260,11 +260,16 @@ static int rdma_read_max_sge(struct svcxprt_rdma *xprt, int sge_count)
260 * On our side, we need to read into a pagelist. The first page immediately 260 * On our side, we need to read into a pagelist. The first page immediately
261 * follows the RPC header. 261 * follows the RPC header.
262 * 262 *
263 * This function returns 1 to indicate success. The data is not yet in 263 * This function returns:
264 * 0 - No error and no read-list found.
265 *
266 * 1 - Successful read-list processing. The data is not yet in
264 * the pagelist and therefore the RPC request must be deferred. The 267 * the pagelist and therefore the RPC request must be deferred. The
265 * I/O completion will enqueue the transport again and 268 * I/O completion will enqueue the transport again and
266 * svc_rdma_recvfrom will complete the request. 269 * svc_rdma_recvfrom will complete the request.
267 * 270 *
271 * <0 - Error processing/posting read-list.
272 *
268 * NOTE: The ctxt must not be touched after the last WR has been posted 273 * NOTE: The ctxt must not be touched after the last WR has been posted
269 * because the I/O completion processing may occur on another 274 * because the I/O completion processing may occur on another
270 * processor and free / modify the context. Ne touche pas! 275 * processor and free / modify the context. Ne touche pas!
@@ -284,7 +289,6 @@ static int rdma_read_xdr(struct svcxprt_rdma *xprt,
284 u64 sgl_offset; 289 u64 sgl_offset;
285 struct rpcrdma_read_chunk *ch; 290 struct rpcrdma_read_chunk *ch;
286 struct svc_rdma_op_ctxt *ctxt = NULL; 291 struct svc_rdma_op_ctxt *ctxt = NULL;
287 struct svc_rdma_op_ctxt *head;
288 struct svc_rdma_op_ctxt *tmp_sge_ctxt; 292 struct svc_rdma_op_ctxt *tmp_sge_ctxt;
289 struct svc_rdma_op_ctxt *tmp_ch_ctxt; 293 struct svc_rdma_op_ctxt *tmp_ch_ctxt;
290 struct chunk_sge *ch_sge_ary; 294 struct chunk_sge *ch_sge_ary;
@@ -302,25 +306,19 @@ static int rdma_read_xdr(struct svcxprt_rdma *xprt,
302 ch_sge_ary = (struct chunk_sge *)tmp_ch_ctxt->sge; 306 ch_sge_ary = (struct chunk_sge *)tmp_ch_ctxt->sge;
303 307
304 svc_rdma_rcl_chunk_counts(ch, &ch_count, &byte_count); 308 svc_rdma_rcl_chunk_counts(ch, &ch_count, &byte_count);
309 if (ch_count > RPCSVC_MAXPAGES)
310 return -EINVAL;
305 sge_count = rdma_rcl_to_sge(xprt, rqstp, hdr_ctxt, rmsgp, 311 sge_count = rdma_rcl_to_sge(xprt, rqstp, hdr_ctxt, rmsgp,
306 sge, ch_sge_ary, 312 sge, ch_sge_ary,
307 ch_count, byte_count); 313 ch_count, byte_count);
308 head = svc_rdma_get_context(xprt);
309 sgl_offset = 0; 314 sgl_offset = 0;
310 ch_no = 0; 315 ch_no = 0;
311 316
312 for (ch = (struct rpcrdma_read_chunk *)&rmsgp->rm_body.rm_chunks[0]; 317 for (ch = (struct rpcrdma_read_chunk *)&rmsgp->rm_body.rm_chunks[0];
313 ch->rc_discrim != 0; ch++, ch_no++) { 318 ch->rc_discrim != 0; ch++, ch_no++) {
314next_sge: 319next_sge:
315 if (!ctxt) 320 ctxt = svc_rdma_get_context(xprt);
316 ctxt = head;
317 else {
318 ctxt->next = svc_rdma_get_context(xprt);
319 ctxt = ctxt->next;
320 }
321 ctxt->next = NULL;
322 ctxt->direction = DMA_FROM_DEVICE; 321 ctxt->direction = DMA_FROM_DEVICE;
323 clear_bit(RDMACTXT_F_READ_DONE, &ctxt->flags);
324 clear_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags); 322 clear_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags);
325 323
326 /* Prepare READ WR */ 324 /* Prepare READ WR */
@@ -347,20 +345,15 @@ next_sge:
347 * the client and the RPC needs to be enqueued. 345 * the client and the RPC needs to be enqueued.
348 */ 346 */
349 set_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags); 347 set_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags);
350 ctxt->next = hdr_ctxt; 348 ctxt->read_hdr = hdr_ctxt;
351 hdr_ctxt->next = head;
352 } 349 }
353 /* Post the read */ 350 /* Post the read */
354 err = svc_rdma_send(xprt, &read_wr); 351 err = svc_rdma_send(xprt, &read_wr);
355 if (err) { 352 if (err) {
356 printk(KERN_ERR "svcrdma: Error posting send = %d\n", 353 printk(KERN_ERR "svcrdma: Error %d posting RDMA_READ\n",
357 err); 354 err);
358 /* 355 set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags);
359 * Break the circular list so free knows when 356 svc_rdma_put_context(ctxt, 0);
360 * to stop if the error happened to occur on
361 * the last read
362 */
363 ctxt->next = NULL;
364 goto out; 357 goto out;
365 } 358 }
366 atomic_inc(&rdma_stat_read); 359 atomic_inc(&rdma_stat_read);
@@ -371,7 +364,7 @@ next_sge:
371 goto next_sge; 364 goto next_sge;
372 } 365 }
373 sgl_offset = 0; 366 sgl_offset = 0;
374 err = 0; 367 err = 1;
375 } 368 }
376 369
377 out: 370 out:
@@ -389,25 +382,12 @@ next_sge:
389 while (rqstp->rq_resused) 382 while (rqstp->rq_resused)
390 rqstp->rq_respages[--rqstp->rq_resused] = NULL; 383 rqstp->rq_respages[--rqstp->rq_resused] = NULL;
391 384
392 if (err) { 385 return err;
393 printk(KERN_ERR "svcrdma : RDMA_READ error = %d\n", err);
394 set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags);
395 /* Free the linked list of read contexts */
396 while (head != NULL) {
397 ctxt = head->next;
398 svc_rdma_put_context(head, 1);
399 head = ctxt;
400 }
401 return 0;
402 }
403
404 return 1;
405} 386}
406 387
407static int rdma_read_complete(struct svc_rqst *rqstp, 388static int rdma_read_complete(struct svc_rqst *rqstp,
408 struct svc_rdma_op_ctxt *data) 389 struct svc_rdma_op_ctxt *head)
409{ 390{
410 struct svc_rdma_op_ctxt *head = data->next;
411 int page_no; 391 int page_no;
412 int ret; 392 int ret;
413 393
@@ -433,21 +413,12 @@ static int rdma_read_complete(struct svc_rqst *rqstp,
433 rqstp->rq_arg.len = head->arg.len; 413 rqstp->rq_arg.len = head->arg.len;
434 rqstp->rq_arg.buflen = head->arg.buflen; 414 rqstp->rq_arg.buflen = head->arg.buflen;
435 415
416 /* Free the context */
417 svc_rdma_put_context(head, 0);
418
436 /* XXX: What should this be? */ 419 /* XXX: What should this be? */
437 rqstp->rq_prot = IPPROTO_MAX; 420 rqstp->rq_prot = IPPROTO_MAX;
438 421 svc_xprt_copy_addrs(rqstp, rqstp->rq_xprt);
439 /*
440 * Free the contexts we used to build the RDMA_READ. We have
441 * to be careful here because the context list uses the same
442 * next pointer used to chain the contexts associated with the
443 * RDMA_READ
444 */
445 data->next = NULL; /* terminate circular list */
446 do {
447 data = head->next;
448 svc_rdma_put_context(head, 0);
449 head = data;
450 } while (head != NULL);
451 422
452 ret = rqstp->rq_arg.head[0].iov_len 423 ret = rqstp->rq_arg.head[0].iov_len
453 + rqstp->rq_arg.page_len 424 + rqstp->rq_arg.page_len
@@ -457,8 +428,6 @@ static int rdma_read_complete(struct svc_rqst *rqstp,
457 ret, rqstp->rq_arg.len, rqstp->rq_arg.head[0].iov_base, 428 ret, rqstp->rq_arg.len, rqstp->rq_arg.head[0].iov_base,
458 rqstp->rq_arg.head[0].iov_len); 429 rqstp->rq_arg.head[0].iov_len);
459 430
460 /* Indicate that we've consumed an RQ credit */
461 rqstp->rq_xprt_ctxt = rqstp->rq_xprt;
462 svc_xprt_received(rqstp->rq_xprt); 431 svc_xprt_received(rqstp->rq_xprt);
463 return ret; 432 return ret;
464} 433}
@@ -480,13 +449,6 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp)
480 449
481 dprintk("svcrdma: rqstp=%p\n", rqstp); 450 dprintk("svcrdma: rqstp=%p\n", rqstp);
482 451
483 /*
484 * The rq_xprt_ctxt indicates if we've consumed an RQ credit
485 * or not. It is used in the rdma xpo_release_rqst function to
486 * determine whether or not to return an RQ WQE to the RQ.
487 */
488 rqstp->rq_xprt_ctxt = NULL;
489
490 spin_lock_bh(&rdma_xprt->sc_read_complete_lock); 452 spin_lock_bh(&rdma_xprt->sc_read_complete_lock);
491 if (!list_empty(&rdma_xprt->sc_read_complete_q)) { 453 if (!list_empty(&rdma_xprt->sc_read_complete_q)) {
492 ctxt = list_entry(rdma_xprt->sc_read_complete_q.next, 454 ctxt = list_entry(rdma_xprt->sc_read_complete_q.next,
@@ -537,21 +499,22 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp)
537 /* If the request is invalid, reply with an error */ 499 /* If the request is invalid, reply with an error */
538 if (len < 0) { 500 if (len < 0) {
539 if (len == -ENOSYS) 501 if (len == -ENOSYS)
540 (void)svc_rdma_send_error(rdma_xprt, rmsgp, ERR_VERS); 502 svc_rdma_send_error(rdma_xprt, rmsgp, ERR_VERS);
541 goto close_out; 503 goto close_out;
542 } 504 }
543 505
544 /* Read read-list data. If we would need to wait, defer 506 /* Read read-list data. */
545 * it. Not that in this case, we don't return the RQ credit 507 ret = rdma_read_xdr(rdma_xprt, rmsgp, rqstp, ctxt);
546 * until after the read completes. 508 if (ret > 0) {
547 */ 509 /* read-list posted, defer until data received from client. */
548 if (rdma_read_xdr(rdma_xprt, rmsgp, rqstp, ctxt)) {
549 svc_xprt_received(xprt); 510 svc_xprt_received(xprt);
550 return 0; 511 return 0;
551 } 512 }
552 513 if (ret < 0) {
553 /* Indicate we've consumed an RQ credit */ 514 /* Post of read-list failed, free context. */
554 rqstp->rq_xprt_ctxt = rqstp->rq_xprt; 515 svc_rdma_put_context(ctxt, 1);
516 return 0;
517 }
555 518
556 ret = rqstp->rq_arg.head[0].iov_len 519 ret = rqstp->rq_arg.head[0].iov_len
557 + rqstp->rq_arg.page_len 520 + rqstp->rq_arg.page_len
@@ -569,11 +532,8 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp)
569 return ret; 532 return ret;
570 533
571 close_out: 534 close_out:
572 if (ctxt) { 535 if (ctxt)
573 svc_rdma_put_context(ctxt, 1); 536 svc_rdma_put_context(ctxt, 1);
574 /* Indicate we've consumed an RQ credit */
575 rqstp->rq_xprt_ctxt = rqstp->rq_xprt;
576 }
577 dprintk("svcrdma: transport %p is closing\n", xprt); 537 dprintk("svcrdma: transport %p is closing\n", xprt);
578 /* 538 /*
579 * Set the close bit and enqueue it. svc_recv will see the 539 * Set the close bit and enqueue it. svc_recv will see the
diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
index 981f190c1b39..fb82b1b683f8 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
@@ -389,6 +389,17 @@ static int send_reply(struct svcxprt_rdma *rdma,
389 int page_no; 389 int page_no;
390 int ret; 390 int ret;
391 391
392 /* Post a recv buffer to handle another request. */
393 ret = svc_rdma_post_recv(rdma);
394 if (ret) {
395 printk(KERN_INFO
396 "svcrdma: could not post a receive buffer, err=%d."
397 "Closing transport %p.\n", ret, rdma);
398 set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags);
399 svc_rdma_put_context(ctxt, 0);
400 return -ENOTCONN;
401 }
402
392 /* Prepare the context */ 403 /* Prepare the context */
393 ctxt->pages[0] = page; 404 ctxt->pages[0] = page;
394 ctxt->count = 1; 405 ctxt->count = 1;
diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c
index af408fc12634..e132509d1db0 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_transport.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c
@@ -103,8 +103,8 @@ static int rdma_bump_context_cache(struct svcxprt_rdma *xprt)
103 spin_lock_bh(&xprt->sc_ctxt_lock); 103 spin_lock_bh(&xprt->sc_ctxt_lock);
104 if (ctxt) { 104 if (ctxt) {
105 at_least_one = 1; 105 at_least_one = 1;
106 ctxt->next = xprt->sc_ctxt_head; 106 INIT_LIST_HEAD(&ctxt->free_list);
107 xprt->sc_ctxt_head = ctxt; 107 list_add(&ctxt->free_list, &xprt->sc_ctxt_free);
108 } else { 108 } else {
109 /* kmalloc failed...give up for now */ 109 /* kmalloc failed...give up for now */
110 xprt->sc_ctxt_cnt--; 110 xprt->sc_ctxt_cnt--;
@@ -123,7 +123,7 @@ struct svc_rdma_op_ctxt *svc_rdma_get_context(struct svcxprt_rdma *xprt)
123 123
124 while (1) { 124 while (1) {
125 spin_lock_bh(&xprt->sc_ctxt_lock); 125 spin_lock_bh(&xprt->sc_ctxt_lock);
126 if (unlikely(xprt->sc_ctxt_head == NULL)) { 126 if (unlikely(list_empty(&xprt->sc_ctxt_free))) {
127 /* Try to bump my cache. */ 127 /* Try to bump my cache. */
128 spin_unlock_bh(&xprt->sc_ctxt_lock); 128 spin_unlock_bh(&xprt->sc_ctxt_lock);
129 129
@@ -136,12 +136,15 @@ struct svc_rdma_op_ctxt *svc_rdma_get_context(struct svcxprt_rdma *xprt)
136 schedule_timeout_uninterruptible(msecs_to_jiffies(500)); 136 schedule_timeout_uninterruptible(msecs_to_jiffies(500));
137 continue; 137 continue;
138 } 138 }
139 ctxt = xprt->sc_ctxt_head; 139 ctxt = list_entry(xprt->sc_ctxt_free.next,
140 xprt->sc_ctxt_head = ctxt->next; 140 struct svc_rdma_op_ctxt,
141 free_list);
142 list_del_init(&ctxt->free_list);
141 spin_unlock_bh(&xprt->sc_ctxt_lock); 143 spin_unlock_bh(&xprt->sc_ctxt_lock);
142 ctxt->xprt = xprt; 144 ctxt->xprt = xprt;
143 INIT_LIST_HEAD(&ctxt->dto_q); 145 INIT_LIST_HEAD(&ctxt->dto_q);
144 ctxt->count = 0; 146 ctxt->count = 0;
147 atomic_inc(&xprt->sc_ctxt_used);
145 break; 148 break;
146 } 149 }
147 return ctxt; 150 return ctxt;
@@ -159,14 +162,15 @@ void svc_rdma_put_context(struct svc_rdma_op_ctxt *ctxt, int free_pages)
159 put_page(ctxt->pages[i]); 162 put_page(ctxt->pages[i]);
160 163
161 for (i = 0; i < ctxt->count; i++) 164 for (i = 0; i < ctxt->count; i++)
162 dma_unmap_single(xprt->sc_cm_id->device->dma_device, 165 ib_dma_unmap_single(xprt->sc_cm_id->device,
163 ctxt->sge[i].addr, 166 ctxt->sge[i].addr,
164 ctxt->sge[i].length, 167 ctxt->sge[i].length,
165 ctxt->direction); 168 ctxt->direction);
169
166 spin_lock_bh(&xprt->sc_ctxt_lock); 170 spin_lock_bh(&xprt->sc_ctxt_lock);
167 ctxt->next = xprt->sc_ctxt_head; 171 list_add(&ctxt->free_list, &xprt->sc_ctxt_free);
168 xprt->sc_ctxt_head = ctxt;
169 spin_unlock_bh(&xprt->sc_ctxt_lock); 172 spin_unlock_bh(&xprt->sc_ctxt_lock);
173 atomic_dec(&xprt->sc_ctxt_used);
170} 174}
171 175
172/* ib_cq event handler */ 176/* ib_cq event handler */
@@ -228,23 +232,8 @@ static void dto_tasklet_func(unsigned long data)
228 list_del_init(&xprt->sc_dto_q); 232 list_del_init(&xprt->sc_dto_q);
229 spin_unlock_irqrestore(&dto_lock, flags); 233 spin_unlock_irqrestore(&dto_lock, flags);
230 234
231 if (test_and_clear_bit(RDMAXPRT_RQ_PENDING, &xprt->sc_flags)) { 235 rq_cq_reap(xprt);
232 ib_req_notify_cq(xprt->sc_rq_cq, IB_CQ_NEXT_COMP); 236 sq_cq_reap(xprt);
233 rq_cq_reap(xprt);
234 set_bit(XPT_DATA, &xprt->sc_xprt.xpt_flags);
235 /*
236 * If data arrived before established event,
237 * don't enqueue. This defers RPC I/O until the
238 * RDMA connection is complete.
239 */
240 if (!test_bit(RDMAXPRT_CONN_PENDING, &xprt->sc_flags))
241 svc_xprt_enqueue(&xprt->sc_xprt);
242 }
243
244 if (test_and_clear_bit(RDMAXPRT_SQ_PENDING, &xprt->sc_flags)) {
245 ib_req_notify_cq(xprt->sc_sq_cq, IB_CQ_NEXT_COMP);
246 sq_cq_reap(xprt);
247 }
248 237
249 svc_xprt_put(&xprt->sc_xprt); 238 svc_xprt_put(&xprt->sc_xprt);
250 spin_lock_irqsave(&dto_lock, flags); 239 spin_lock_irqsave(&dto_lock, flags);
@@ -263,11 +252,15 @@ static void rq_comp_handler(struct ib_cq *cq, void *cq_context)
263 struct svcxprt_rdma *xprt = cq_context; 252 struct svcxprt_rdma *xprt = cq_context;
264 unsigned long flags; 253 unsigned long flags;
265 254
255 /* Guard against unconditional flush call for destroyed QP */
256 if (atomic_read(&xprt->sc_xprt.xpt_ref.refcount)==0)
257 return;
258
266 /* 259 /*
267 * Set the bit regardless of whether or not it's on the list 260 * Set the bit regardless of whether or not it's on the list
268 * because it may be on the list already due to an SQ 261 * because it may be on the list already due to an SQ
269 * completion. 262 * completion.
270 */ 263 */
271 set_bit(RDMAXPRT_RQ_PENDING, &xprt->sc_flags); 264 set_bit(RDMAXPRT_RQ_PENDING, &xprt->sc_flags);
272 265
273 /* 266 /*
@@ -290,6 +283,8 @@ static void rq_comp_handler(struct ib_cq *cq, void *cq_context)
290 * 283 *
291 * Take all completing WC off the CQE and enqueue the associated DTO 284 * Take all completing WC off the CQE and enqueue the associated DTO
292 * context on the dto_q for the transport. 285 * context on the dto_q for the transport.
286 *
287 * Note that caller must hold a transport reference.
293 */ 288 */
294static void rq_cq_reap(struct svcxprt_rdma *xprt) 289static void rq_cq_reap(struct svcxprt_rdma *xprt)
295{ 290{
@@ -297,29 +292,47 @@ static void rq_cq_reap(struct svcxprt_rdma *xprt)
297 struct ib_wc wc; 292 struct ib_wc wc;
298 struct svc_rdma_op_ctxt *ctxt = NULL; 293 struct svc_rdma_op_ctxt *ctxt = NULL;
299 294
295 if (!test_and_clear_bit(RDMAXPRT_RQ_PENDING, &xprt->sc_flags))
296 return;
297
298 ib_req_notify_cq(xprt->sc_rq_cq, IB_CQ_NEXT_COMP);
300 atomic_inc(&rdma_stat_rq_poll); 299 atomic_inc(&rdma_stat_rq_poll);
301 300
302 spin_lock_bh(&xprt->sc_rq_dto_lock);
303 while ((ret = ib_poll_cq(xprt->sc_rq_cq, 1, &wc)) > 0) { 301 while ((ret = ib_poll_cq(xprt->sc_rq_cq, 1, &wc)) > 0) {
304 ctxt = (struct svc_rdma_op_ctxt *)(unsigned long)wc.wr_id; 302 ctxt = (struct svc_rdma_op_ctxt *)(unsigned long)wc.wr_id;
305 ctxt->wc_status = wc.status; 303 ctxt->wc_status = wc.status;
306 ctxt->byte_len = wc.byte_len; 304 ctxt->byte_len = wc.byte_len;
307 if (wc.status != IB_WC_SUCCESS) { 305 if (wc.status != IB_WC_SUCCESS) {
308 /* Close the transport */ 306 /* Close the transport */
307 dprintk("svcrdma: transport closing putting ctxt %p\n", ctxt);
309 set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); 308 set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags);
310 svc_rdma_put_context(ctxt, 1); 309 svc_rdma_put_context(ctxt, 1);
310 svc_xprt_put(&xprt->sc_xprt);
311 continue; 311 continue;
312 } 312 }
313 spin_lock_bh(&xprt->sc_rq_dto_lock);
313 list_add_tail(&ctxt->dto_q, &xprt->sc_rq_dto_q); 314 list_add_tail(&ctxt->dto_q, &xprt->sc_rq_dto_q);
315 spin_unlock_bh(&xprt->sc_rq_dto_lock);
316 svc_xprt_put(&xprt->sc_xprt);
314 } 317 }
315 spin_unlock_bh(&xprt->sc_rq_dto_lock);
316 318
317 if (ctxt) 319 if (ctxt)
318 atomic_inc(&rdma_stat_rq_prod); 320 atomic_inc(&rdma_stat_rq_prod);
321
322 set_bit(XPT_DATA, &xprt->sc_xprt.xpt_flags);
323 /*
324 * If data arrived before established event,
325 * don't enqueue. This defers RPC I/O until the
326 * RDMA connection is complete.
327 */
328 if (!test_bit(RDMAXPRT_CONN_PENDING, &xprt->sc_flags))
329 svc_xprt_enqueue(&xprt->sc_xprt);
319} 330}
320 331
321/* 332/*
322 * Send Queue Completion Handler - potentially called on interrupt context. 333 * Send Queue Completion Handler - potentially called on interrupt context.
334 *
335 * Note that caller must hold a transport reference.
323 */ 336 */
324static void sq_cq_reap(struct svcxprt_rdma *xprt) 337static void sq_cq_reap(struct svcxprt_rdma *xprt)
325{ 338{
@@ -328,6 +341,11 @@ static void sq_cq_reap(struct svcxprt_rdma *xprt)
328 struct ib_cq *cq = xprt->sc_sq_cq; 341 struct ib_cq *cq = xprt->sc_sq_cq;
329 int ret; 342 int ret;
330 343
344
345 if (!test_and_clear_bit(RDMAXPRT_SQ_PENDING, &xprt->sc_flags))
346 return;
347
348 ib_req_notify_cq(xprt->sc_sq_cq, IB_CQ_NEXT_COMP);
331 atomic_inc(&rdma_stat_sq_poll); 349 atomic_inc(&rdma_stat_sq_poll);
332 while ((ret = ib_poll_cq(cq, 1, &wc)) > 0) { 350 while ((ret = ib_poll_cq(cq, 1, &wc)) > 0) {
333 ctxt = (struct svc_rdma_op_ctxt *)(unsigned long)wc.wr_id; 351 ctxt = (struct svc_rdma_op_ctxt *)(unsigned long)wc.wr_id;
@@ -349,14 +367,16 @@ static void sq_cq_reap(struct svcxprt_rdma *xprt)
349 367
350 case IB_WR_RDMA_READ: 368 case IB_WR_RDMA_READ:
351 if (test_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags)) { 369 if (test_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags)) {
370 struct svc_rdma_op_ctxt *read_hdr = ctxt->read_hdr;
371 BUG_ON(!read_hdr);
352 set_bit(XPT_DATA, &xprt->sc_xprt.xpt_flags); 372 set_bit(XPT_DATA, &xprt->sc_xprt.xpt_flags);
353 set_bit(RDMACTXT_F_READ_DONE, &ctxt->flags);
354 spin_lock_bh(&xprt->sc_read_complete_lock); 373 spin_lock_bh(&xprt->sc_read_complete_lock);
355 list_add_tail(&ctxt->dto_q, 374 list_add_tail(&read_hdr->dto_q,
356 &xprt->sc_read_complete_q); 375 &xprt->sc_read_complete_q);
357 spin_unlock_bh(&xprt->sc_read_complete_lock); 376 spin_unlock_bh(&xprt->sc_read_complete_lock);
358 svc_xprt_enqueue(&xprt->sc_xprt); 377 svc_xprt_enqueue(&xprt->sc_xprt);
359 } 378 }
379 svc_rdma_put_context(ctxt, 0);
360 break; 380 break;
361 381
362 default: 382 default:
@@ -365,6 +385,7 @@ static void sq_cq_reap(struct svcxprt_rdma *xprt)
365 wc.opcode, wc.status); 385 wc.opcode, wc.status);
366 break; 386 break;
367 } 387 }
388 svc_xprt_put(&xprt->sc_xprt);
368 } 389 }
369 390
370 if (ctxt) 391 if (ctxt)
@@ -376,11 +397,15 @@ static void sq_comp_handler(struct ib_cq *cq, void *cq_context)
376 struct svcxprt_rdma *xprt = cq_context; 397 struct svcxprt_rdma *xprt = cq_context;
377 unsigned long flags; 398 unsigned long flags;
378 399
400 /* Guard against unconditional flush call for destroyed QP */
401 if (atomic_read(&xprt->sc_xprt.xpt_ref.refcount)==0)
402 return;
403
379 /* 404 /*
380 * Set the bit regardless of whether or not it's on the list 405 * Set the bit regardless of whether or not it's on the list
381 * because it may be on the list already due to an RQ 406 * because it may be on the list already due to an RQ
382 * completion. 407 * completion.
383 */ 408 */
384 set_bit(RDMAXPRT_SQ_PENDING, &xprt->sc_flags); 409 set_bit(RDMAXPRT_SQ_PENDING, &xprt->sc_flags);
385 410
386 /* 411 /*
@@ -407,28 +432,29 @@ static void create_context_cache(struct svcxprt_rdma *xprt,
407 xprt->sc_ctxt_max = ctxt_max; 432 xprt->sc_ctxt_max = ctxt_max;
408 xprt->sc_ctxt_bump = ctxt_bump; 433 xprt->sc_ctxt_bump = ctxt_bump;
409 xprt->sc_ctxt_cnt = 0; 434 xprt->sc_ctxt_cnt = 0;
410 xprt->sc_ctxt_head = NULL; 435 atomic_set(&xprt->sc_ctxt_used, 0);
436
437 INIT_LIST_HEAD(&xprt->sc_ctxt_free);
411 for (i = 0; i < ctxt_count; i++) { 438 for (i = 0; i < ctxt_count; i++) {
412 ctxt = kmalloc(sizeof(*ctxt), GFP_KERNEL); 439 ctxt = kmalloc(sizeof(*ctxt), GFP_KERNEL);
413 if (ctxt) { 440 if (ctxt) {
414 ctxt->next = xprt->sc_ctxt_head; 441 INIT_LIST_HEAD(&ctxt->free_list);
415 xprt->sc_ctxt_head = ctxt; 442 list_add(&ctxt->free_list, &xprt->sc_ctxt_free);
416 xprt->sc_ctxt_cnt++; 443 xprt->sc_ctxt_cnt++;
417 } 444 }
418 } 445 }
419} 446}
420 447
421static void destroy_context_cache(struct svc_rdma_op_ctxt *ctxt) 448static void destroy_context_cache(struct svcxprt_rdma *xprt)
422{ 449{
423 struct svc_rdma_op_ctxt *next; 450 while (!list_empty(&xprt->sc_ctxt_free)) {
424 if (!ctxt) 451 struct svc_rdma_op_ctxt *ctxt;
425 return; 452 ctxt = list_entry(xprt->sc_ctxt_free.next,
426 453 struct svc_rdma_op_ctxt,
427 do { 454 free_list);
428 next = ctxt->next; 455 list_del_init(&ctxt->free_list);
429 kfree(ctxt); 456 kfree(ctxt);
430 ctxt = next; 457 }
431 } while (next);
432} 458}
433 459
434static struct svcxprt_rdma *rdma_create_xprt(struct svc_serv *serv, 460static struct svcxprt_rdma *rdma_create_xprt(struct svc_serv *serv,
@@ -465,7 +491,7 @@ static struct svcxprt_rdma *rdma_create_xprt(struct svc_serv *serv,
465 reqs + 491 reqs +
466 cma_xprt->sc_sq_depth + 492 cma_xprt->sc_sq_depth +
467 RPCRDMA_MAX_THREADS + 1); /* max */ 493 RPCRDMA_MAX_THREADS + 1); /* max */
468 if (!cma_xprt->sc_ctxt_head) { 494 if (list_empty(&cma_xprt->sc_ctxt_free)) {
469 kfree(cma_xprt); 495 kfree(cma_xprt);
470 return NULL; 496 return NULL;
471 } 497 }
@@ -520,7 +546,12 @@ int svc_rdma_post_recv(struct svcxprt_rdma *xprt)
520 recv_wr.num_sge = ctxt->count; 546 recv_wr.num_sge = ctxt->count;
521 recv_wr.wr_id = (u64)(unsigned long)ctxt; 547 recv_wr.wr_id = (u64)(unsigned long)ctxt;
522 548
549 svc_xprt_get(&xprt->sc_xprt);
523 ret = ib_post_recv(xprt->sc_qp, &recv_wr, &bad_recv_wr); 550 ret = ib_post_recv(xprt->sc_qp, &recv_wr, &bad_recv_wr);
551 if (ret) {
552 svc_xprt_put(&xprt->sc_xprt);
553 svc_rdma_put_context(ctxt, 1);
554 }
524 return ret; 555 return ret;
525} 556}
526 557
@@ -539,6 +570,7 @@ static void handle_connect_req(struct rdma_cm_id *new_cma_id)
539{ 570{
540 struct svcxprt_rdma *listen_xprt = new_cma_id->context; 571 struct svcxprt_rdma *listen_xprt = new_cma_id->context;
541 struct svcxprt_rdma *newxprt; 572 struct svcxprt_rdma *newxprt;
573 struct sockaddr *sa;
542 574
543 /* Create a new transport */ 575 /* Create a new transport */
544 newxprt = rdma_create_xprt(listen_xprt->sc_xprt.xpt_server, 0); 576 newxprt = rdma_create_xprt(listen_xprt->sc_xprt.xpt_server, 0);
@@ -551,6 +583,12 @@ static void handle_connect_req(struct rdma_cm_id *new_cma_id)
551 dprintk("svcrdma: Creating newxprt=%p, cm_id=%p, listenxprt=%p\n", 583 dprintk("svcrdma: Creating newxprt=%p, cm_id=%p, listenxprt=%p\n",
552 newxprt, newxprt->sc_cm_id, listen_xprt); 584 newxprt, newxprt->sc_cm_id, listen_xprt);
553 585
586 /* Set the local and remote addresses in the transport */
587 sa = (struct sockaddr *)&newxprt->sc_cm_id->route.addr.dst_addr;
588 svc_xprt_set_remote(&newxprt->sc_xprt, sa, svc_addr_len(sa));
589 sa = (struct sockaddr *)&newxprt->sc_cm_id->route.addr.src_addr;
590 svc_xprt_set_local(&newxprt->sc_xprt, sa, svc_addr_len(sa));
591
554 /* 592 /*
555 * Enqueue the new transport on the accept queue of the listening 593 * Enqueue the new transport on the accept queue of the listening
556 * transport 594 * transport
@@ -627,6 +665,7 @@ static int rdma_cma_handler(struct rdma_cm_id *cma_id,
627 if (xprt) { 665 if (xprt) {
628 set_bit(XPT_CLOSE, &xprt->xpt_flags); 666 set_bit(XPT_CLOSE, &xprt->xpt_flags);
629 svc_xprt_enqueue(xprt); 667 svc_xprt_enqueue(xprt);
668 svc_xprt_put(xprt);
630 } 669 }
631 break; 670 break;
632 case RDMA_CM_EVENT_DEVICE_REMOVAL: 671 case RDMA_CM_EVENT_DEVICE_REMOVAL:
@@ -661,31 +700,27 @@ static struct svc_xprt *svc_rdma_create(struct svc_serv *serv,
661 700
662 cma_xprt = rdma_create_xprt(serv, 1); 701 cma_xprt = rdma_create_xprt(serv, 1);
663 if (!cma_xprt) 702 if (!cma_xprt)
664 return ERR_PTR(ENOMEM); 703 return ERR_PTR(-ENOMEM);
665 xprt = &cma_xprt->sc_xprt; 704 xprt = &cma_xprt->sc_xprt;
666 705
667 listen_id = rdma_create_id(rdma_listen_handler, cma_xprt, RDMA_PS_TCP); 706 listen_id = rdma_create_id(rdma_listen_handler, cma_xprt, RDMA_PS_TCP);
668 if (IS_ERR(listen_id)) { 707 if (IS_ERR(listen_id)) {
669 svc_xprt_put(&cma_xprt->sc_xprt); 708 ret = PTR_ERR(listen_id);
670 dprintk("svcrdma: rdma_create_id failed = %ld\n", 709 dprintk("svcrdma: rdma_create_id failed = %d\n", ret);
671 PTR_ERR(listen_id)); 710 goto err0;
672 return (void *)listen_id;
673 } 711 }
712
674 ret = rdma_bind_addr(listen_id, sa); 713 ret = rdma_bind_addr(listen_id, sa);
675 if (ret) { 714 if (ret) {
676 rdma_destroy_id(listen_id);
677 svc_xprt_put(&cma_xprt->sc_xprt);
678 dprintk("svcrdma: rdma_bind_addr failed = %d\n", ret); 715 dprintk("svcrdma: rdma_bind_addr failed = %d\n", ret);
679 return ERR_PTR(ret); 716 goto err1;
680 } 717 }
681 cma_xprt->sc_cm_id = listen_id; 718 cma_xprt->sc_cm_id = listen_id;
682 719
683 ret = rdma_listen(listen_id, RPCRDMA_LISTEN_BACKLOG); 720 ret = rdma_listen(listen_id, RPCRDMA_LISTEN_BACKLOG);
684 if (ret) { 721 if (ret) {
685 rdma_destroy_id(listen_id);
686 svc_xprt_put(&cma_xprt->sc_xprt);
687 dprintk("svcrdma: rdma_listen failed = %d\n", ret); 722 dprintk("svcrdma: rdma_listen failed = %d\n", ret);
688 return ERR_PTR(ret); 723 goto err1;
689 } 724 }
690 725
691 /* 726 /*
@@ -696,6 +731,12 @@ static struct svc_xprt *svc_rdma_create(struct svc_serv *serv,
696 svc_xprt_set_local(&cma_xprt->sc_xprt, sa, salen); 731 svc_xprt_set_local(&cma_xprt->sc_xprt, sa, salen);
697 732
698 return &cma_xprt->sc_xprt; 733 return &cma_xprt->sc_xprt;
734
735 err1:
736 rdma_destroy_id(listen_id);
737 err0:
738 kfree(cma_xprt);
739 return ERR_PTR(ret);
699} 740}
700 741
701/* 742/*
@@ -716,7 +757,6 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
716 struct rdma_conn_param conn_param; 757 struct rdma_conn_param conn_param;
717 struct ib_qp_init_attr qp_attr; 758 struct ib_qp_init_attr qp_attr;
718 struct ib_device_attr devattr; 759 struct ib_device_attr devattr;
719 struct sockaddr *sa;
720 int ret; 760 int ret;
721 int i; 761 int i;
722 762
@@ -826,7 +866,6 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
826 newxprt->sc_sq_depth = qp_attr.cap.max_send_wr; 866 newxprt->sc_sq_depth = qp_attr.cap.max_send_wr;
827 newxprt->sc_max_requests = qp_attr.cap.max_recv_wr; 867 newxprt->sc_max_requests = qp_attr.cap.max_recv_wr;
828 } 868 }
829 svc_xprt_get(&newxprt->sc_xprt);
830 newxprt->sc_qp = newxprt->sc_cm_id->qp; 869 newxprt->sc_qp = newxprt->sc_cm_id->qp;
831 870
832 /* Register all of physical memory */ 871 /* Register all of physical memory */
@@ -850,6 +889,13 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
850 /* Swap out the handler */ 889 /* Swap out the handler */
851 newxprt->sc_cm_id->event_handler = rdma_cma_handler; 890 newxprt->sc_cm_id->event_handler = rdma_cma_handler;
852 891
892 /*
893 * Arm the CQs for the SQ and RQ before accepting so we can't
894 * miss the first message
895 */
896 ib_req_notify_cq(newxprt->sc_sq_cq, IB_CQ_NEXT_COMP);
897 ib_req_notify_cq(newxprt->sc_rq_cq, IB_CQ_NEXT_COMP);
898
853 /* Accept Connection */ 899 /* Accept Connection */
854 set_bit(RDMAXPRT_CONN_PENDING, &newxprt->sc_flags); 900 set_bit(RDMAXPRT_CONN_PENDING, &newxprt->sc_flags);
855 memset(&conn_param, 0, sizeof conn_param); 901 memset(&conn_param, 0, sizeof conn_param);
@@ -886,58 +932,26 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt)
886 newxprt->sc_max_requests, 932 newxprt->sc_max_requests,
887 newxprt->sc_ord); 933 newxprt->sc_ord);
888 934
889 /* Set the local and remote addresses in the transport */
890 sa = (struct sockaddr *)&newxprt->sc_cm_id->route.addr.dst_addr;
891 svc_xprt_set_remote(&newxprt->sc_xprt, sa, svc_addr_len(sa));
892 sa = (struct sockaddr *)&newxprt->sc_cm_id->route.addr.src_addr;
893 svc_xprt_set_local(&newxprt->sc_xprt, sa, svc_addr_len(sa));
894
895 ib_req_notify_cq(newxprt->sc_sq_cq, IB_CQ_NEXT_COMP);
896 ib_req_notify_cq(newxprt->sc_rq_cq, IB_CQ_NEXT_COMP);
897 return &newxprt->sc_xprt; 935 return &newxprt->sc_xprt;
898 936
899 errout: 937 errout:
900 dprintk("svcrdma: failure accepting new connection rc=%d.\n", ret); 938 dprintk("svcrdma: failure accepting new connection rc=%d.\n", ret);
901 /* Take a reference in case the DTO handler runs */ 939 /* Take a reference in case the DTO handler runs */
902 svc_xprt_get(&newxprt->sc_xprt); 940 svc_xprt_get(&newxprt->sc_xprt);
903 if (newxprt->sc_qp && !IS_ERR(newxprt->sc_qp)) { 941 if (newxprt->sc_qp && !IS_ERR(newxprt->sc_qp))
904 ib_destroy_qp(newxprt->sc_qp); 942 ib_destroy_qp(newxprt->sc_qp);
905 svc_xprt_put(&newxprt->sc_xprt);
906 }
907 rdma_destroy_id(newxprt->sc_cm_id); 943 rdma_destroy_id(newxprt->sc_cm_id);
908 /* This call to put will destroy the transport */ 944 /* This call to put will destroy the transport */
909 svc_xprt_put(&newxprt->sc_xprt); 945 svc_xprt_put(&newxprt->sc_xprt);
910 return NULL; 946 return NULL;
911} 947}
912 948
913/*
914 * Post an RQ WQE to the RQ when the rqst is being released. This
915 * effectively returns an RQ credit to the client. The rq_xprt_ctxt
916 * will be null if the request is deferred due to an RDMA_READ or the
917 * transport had no data ready (EAGAIN). Note that an RPC deferred in
918 * svc_process will still return the credit, this is because the data
919 * is copied and no longer consume a WQE/WC.
920 */
921static void svc_rdma_release_rqst(struct svc_rqst *rqstp) 949static void svc_rdma_release_rqst(struct svc_rqst *rqstp)
922{ 950{
923 int err;
924 struct svcxprt_rdma *rdma =
925 container_of(rqstp->rq_xprt, struct svcxprt_rdma, sc_xprt);
926 if (rqstp->rq_xprt_ctxt) {
927 BUG_ON(rqstp->rq_xprt_ctxt != rdma);
928 err = svc_rdma_post_recv(rdma);
929 if (err)
930 dprintk("svcrdma: failed to post an RQ WQE error=%d\n",
931 err);
932 }
933 rqstp->rq_xprt_ctxt = NULL;
934} 951}
935 952
936/* 953/*
937 * When connected, an svc_xprt has at least three references: 954 * When connected, an svc_xprt has at least two references:
938 *
939 * - A reference held by the QP. We still hold that here because this
940 * code deletes the QP and puts the reference.
941 * 955 *
942 * - A reference held by the cm_id between the ESTABLISHED and 956 * - A reference held by the cm_id between the ESTABLISHED and
943 * DISCONNECTED events. If the remote peer disconnected first, this 957 * DISCONNECTED events. If the remote peer disconnected first, this
@@ -946,7 +960,7 @@ static void svc_rdma_release_rqst(struct svc_rqst *rqstp)
946 * - A reference held by the svc_recv code that called this function 960 * - A reference held by the svc_recv code that called this function
947 * as part of close processing. 961 * as part of close processing.
948 * 962 *
949 * At a minimum two references should still be held. 963 * At a minimum one references should still be held.
950 */ 964 */
951static void svc_rdma_detach(struct svc_xprt *xprt) 965static void svc_rdma_detach(struct svc_xprt *xprt)
952{ 966{
@@ -956,23 +970,53 @@ static void svc_rdma_detach(struct svc_xprt *xprt)
956 970
957 /* Disconnect and flush posted WQE */ 971 /* Disconnect and flush posted WQE */
958 rdma_disconnect(rdma->sc_cm_id); 972 rdma_disconnect(rdma->sc_cm_id);
959
960 /* Destroy the QP if present (not a listener) */
961 if (rdma->sc_qp && !IS_ERR(rdma->sc_qp)) {
962 ib_destroy_qp(rdma->sc_qp);
963 svc_xprt_put(xprt);
964 }
965
966 /* Destroy the CM ID */
967 rdma_destroy_id(rdma->sc_cm_id);
968} 973}
969 974
970static void svc_rdma_free(struct svc_xprt *xprt) 975static void __svc_rdma_free(struct work_struct *work)
971{ 976{
972 struct svcxprt_rdma *rdma = (struct svcxprt_rdma *)xprt; 977 struct svcxprt_rdma *rdma =
978 container_of(work, struct svcxprt_rdma, sc_work);
973 dprintk("svcrdma: svc_rdma_free(%p)\n", rdma); 979 dprintk("svcrdma: svc_rdma_free(%p)\n", rdma);
980
974 /* We should only be called from kref_put */ 981 /* We should only be called from kref_put */
975 BUG_ON(atomic_read(&xprt->xpt_ref.refcount) != 0); 982 BUG_ON(atomic_read(&rdma->sc_xprt.xpt_ref.refcount) != 0);
983
984 /*
985 * Destroy queued, but not processed read completions. Note
986 * that this cleanup has to be done before destroying the
987 * cm_id because the device ptr is needed to unmap the dma in
988 * svc_rdma_put_context.
989 */
990 spin_lock_bh(&rdma->sc_read_complete_lock);
991 while (!list_empty(&rdma->sc_read_complete_q)) {
992 struct svc_rdma_op_ctxt *ctxt;
993 ctxt = list_entry(rdma->sc_read_complete_q.next,
994 struct svc_rdma_op_ctxt,
995 dto_q);
996 list_del_init(&ctxt->dto_q);
997 svc_rdma_put_context(ctxt, 1);
998 }
999 spin_unlock_bh(&rdma->sc_read_complete_lock);
1000
1001 /* Destroy queued, but not processed recv completions */
1002 spin_lock_bh(&rdma->sc_rq_dto_lock);
1003 while (!list_empty(&rdma->sc_rq_dto_q)) {
1004 struct svc_rdma_op_ctxt *ctxt;
1005 ctxt = list_entry(rdma->sc_rq_dto_q.next,
1006 struct svc_rdma_op_ctxt,
1007 dto_q);
1008 list_del_init(&ctxt->dto_q);
1009 svc_rdma_put_context(ctxt, 1);
1010 }
1011 spin_unlock_bh(&rdma->sc_rq_dto_lock);
1012
1013 /* Warn if we leaked a resource or under-referenced */
1014 WARN_ON(atomic_read(&rdma->sc_ctxt_used) != 0);
1015
1016 /* Destroy the QP if present (not a listener) */
1017 if (rdma->sc_qp && !IS_ERR(rdma->sc_qp))
1018 ib_destroy_qp(rdma->sc_qp);
1019
976 if (rdma->sc_sq_cq && !IS_ERR(rdma->sc_sq_cq)) 1020 if (rdma->sc_sq_cq && !IS_ERR(rdma->sc_sq_cq))
977 ib_destroy_cq(rdma->sc_sq_cq); 1021 ib_destroy_cq(rdma->sc_sq_cq);
978 1022
@@ -985,10 +1029,21 @@ static void svc_rdma_free(struct svc_xprt *xprt)
985 if (rdma->sc_pd && !IS_ERR(rdma->sc_pd)) 1029 if (rdma->sc_pd && !IS_ERR(rdma->sc_pd))
986 ib_dealloc_pd(rdma->sc_pd); 1030 ib_dealloc_pd(rdma->sc_pd);
987 1031
988 destroy_context_cache(rdma->sc_ctxt_head); 1032 /* Destroy the CM ID */
1033 rdma_destroy_id(rdma->sc_cm_id);
1034
1035 destroy_context_cache(rdma);
989 kfree(rdma); 1036 kfree(rdma);
990} 1037}
991 1038
1039static void svc_rdma_free(struct svc_xprt *xprt)
1040{
1041 struct svcxprt_rdma *rdma =
1042 container_of(xprt, struct svcxprt_rdma, sc_xprt);
1043 INIT_WORK(&rdma->sc_work, __svc_rdma_free);
1044 schedule_work(&rdma->sc_work);
1045}
1046
992static int svc_rdma_has_wspace(struct svc_xprt *xprt) 1047static int svc_rdma_has_wspace(struct svc_xprt *xprt)
993{ 1048{
994 struct svcxprt_rdma *rdma = 1049 struct svcxprt_rdma *rdma =
@@ -1018,7 +1073,7 @@ int svc_rdma_send(struct svcxprt_rdma *xprt, struct ib_send_wr *wr)
1018 int ret; 1073 int ret;
1019 1074
1020 if (test_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags)) 1075 if (test_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags))
1021 return 0; 1076 return -ENOTCONN;
1022 1077
1023 BUG_ON(wr->send_flags != IB_SEND_SIGNALED); 1078 BUG_ON(wr->send_flags != IB_SEND_SIGNALED);
1024 BUG_ON(((struct svc_rdma_op_ctxt *)(unsigned long)wr->wr_id)->wr_op != 1079 BUG_ON(((struct svc_rdma_op_ctxt *)(unsigned long)wr->wr_id)->wr_op !=
@@ -1029,7 +1084,8 @@ int svc_rdma_send(struct svcxprt_rdma *xprt, struct ib_send_wr *wr)
1029 if (xprt->sc_sq_depth == atomic_read(&xprt->sc_sq_count)) { 1084 if (xprt->sc_sq_depth == atomic_read(&xprt->sc_sq_count)) {
1030 spin_unlock_bh(&xprt->sc_lock); 1085 spin_unlock_bh(&xprt->sc_lock);
1031 atomic_inc(&rdma_stat_sq_starve); 1086 atomic_inc(&rdma_stat_sq_starve);
1032 /* See if we can reap some SQ WR */ 1087
1088 /* See if we can opportunistically reap SQ WR to make room */
1033 sq_cq_reap(xprt); 1089 sq_cq_reap(xprt);
1034 1090
1035 /* Wait until SQ WR available if SQ still full */ 1091 /* Wait until SQ WR available if SQ still full */
@@ -1041,22 +1097,25 @@ int svc_rdma_send(struct svcxprt_rdma *xprt, struct ib_send_wr *wr)
1041 continue; 1097 continue;
1042 } 1098 }
1043 /* Bumped used SQ WR count and post */ 1099 /* Bumped used SQ WR count and post */
1100 svc_xprt_get(&xprt->sc_xprt);
1044 ret = ib_post_send(xprt->sc_qp, wr, &bad_wr); 1101 ret = ib_post_send(xprt->sc_qp, wr, &bad_wr);
1045 if (!ret) 1102 if (!ret)
1046 atomic_inc(&xprt->sc_sq_count); 1103 atomic_inc(&xprt->sc_sq_count);
1047 else 1104 else {
1105 svc_xprt_put(&xprt->sc_xprt);
1048 dprintk("svcrdma: failed to post SQ WR rc=%d, " 1106 dprintk("svcrdma: failed to post SQ WR rc=%d, "
1049 "sc_sq_count=%d, sc_sq_depth=%d\n", 1107 "sc_sq_count=%d, sc_sq_depth=%d\n",
1050 ret, atomic_read(&xprt->sc_sq_count), 1108 ret, atomic_read(&xprt->sc_sq_count),
1051 xprt->sc_sq_depth); 1109 xprt->sc_sq_depth);
1110 }
1052 spin_unlock_bh(&xprt->sc_lock); 1111 spin_unlock_bh(&xprt->sc_lock);
1053 break; 1112 break;
1054 } 1113 }
1055 return ret; 1114 return ret;
1056} 1115}
1057 1116
1058int svc_rdma_send_error(struct svcxprt_rdma *xprt, struct rpcrdma_msg *rmsgp, 1117void svc_rdma_send_error(struct svcxprt_rdma *xprt, struct rpcrdma_msg *rmsgp,
1059 enum rpcrdma_errcode err) 1118 enum rpcrdma_errcode err)
1060{ 1119{
1061 struct ib_send_wr err_wr; 1120 struct ib_send_wr err_wr;
1062 struct ib_sge sge; 1121 struct ib_sge sge;
@@ -1094,9 +1153,8 @@ int svc_rdma_send_error(struct svcxprt_rdma *xprt, struct rpcrdma_msg *rmsgp,
1094 /* Post It */ 1153 /* Post It */
1095 ret = svc_rdma_send(xprt, &err_wr); 1154 ret = svc_rdma_send(xprt, &err_wr);
1096 if (ret) { 1155 if (ret) {
1097 dprintk("svcrdma: Error posting send = %d\n", ret); 1156 dprintk("svcrdma: Error %d posting send for protocol error\n",
1157 ret);
1098 svc_rdma_put_context(ctxt, 1); 1158 svc_rdma_put_context(ctxt, 1);
1099 } 1159 }
1100
1101 return ret;
1102} 1160}
diff --git a/net/sysctl_net.c b/net/sysctl_net.c
index 665e856675a4..b4f0525f91af 100644
--- a/net/sysctl_net.c
+++ b/net/sysctl_net.c
@@ -82,6 +82,6 @@ EXPORT_SYMBOL_GPL(register_net_sysctl_table);
82 82
83void unregister_net_sysctl_table(struct ctl_table_header *header) 83void unregister_net_sysctl_table(struct ctl_table_header *header)
84{ 84{
85 return unregister_sysctl_table(header); 85 unregister_sysctl_table(header);
86} 86}
87EXPORT_SYMBOL_GPL(unregister_net_sysctl_table); 87EXPORT_SYMBOL_GPL(unregister_net_sysctl_table);
diff --git a/net/tipc/core.h b/net/tipc/core.h
index 325404fd4eb5..5a0e4878d3b7 100644
--- a/net/tipc/core.h
+++ b/net/tipc/core.h
@@ -279,15 +279,14 @@ static inline void k_term_timer(struct timer_list *timer)
279/* 279/*
280 * TIPC message buffer code 280 * TIPC message buffer code
281 * 281 *
282 * TIPC message buffer headroom reserves space for a link-level header 282 * TIPC message buffer headroom reserves space for the worst-case
283 * (in case the message is sent off-node), 283 * link-level device header (in case the message is sent off-node).
284 * while ensuring TIPC header is word aligned for quicker access
285 * 284 *
286 * The largest header currently supported is 18 bytes, which is used when 285 * Note: Headroom should be a multiple of 4 to ensure the TIPC header fields
287 * the standard 14 byte Ethernet header has 4 added bytes for VLAN info 286 * are word aligned for quicker access
288 */ 287 */
289 288
290#define BUF_HEADROOM 20u 289#define BUF_HEADROOM LL_MAX_HEADER
291 290
292struct tipc_skb_cb { 291struct tipc_skb_cb {
293 void *handle; 292 void *handle;
diff --git a/net/unix/af_unix.c b/net/unix/af_unix.c
index e18cd3628db4..783317dacd30 100644
--- a/net/unix/af_unix.c
+++ b/net/unix/af_unix.c
@@ -169,6 +169,11 @@ static inline int unix_may_send(struct sock *sk, struct sock *osk)
169 return (unix_peer(osk) == NULL || unix_our_peer(sk, osk)); 169 return (unix_peer(osk) == NULL || unix_our_peer(sk, osk));
170} 170}
171 171
172static inline int unix_recvq_full(struct sock const *sk)
173{
174 return skb_queue_len(&sk->sk_receive_queue) > sk->sk_max_ack_backlog;
175}
176
172static struct sock *unix_peer_get(struct sock *s) 177static struct sock *unix_peer_get(struct sock *s)
173{ 178{
174 struct sock *peer; 179 struct sock *peer;
@@ -482,6 +487,8 @@ static int unix_socketpair(struct socket *, struct socket *);
482static int unix_accept(struct socket *, struct socket *, int); 487static int unix_accept(struct socket *, struct socket *, int);
483static int unix_getname(struct socket *, struct sockaddr *, int *, int); 488static int unix_getname(struct socket *, struct sockaddr *, int *, int);
484static unsigned int unix_poll(struct file *, struct socket *, poll_table *); 489static unsigned int unix_poll(struct file *, struct socket *, poll_table *);
490static unsigned int unix_dgram_poll(struct file *, struct socket *,
491 poll_table *);
485static int unix_ioctl(struct socket *, unsigned int, unsigned long); 492static int unix_ioctl(struct socket *, unsigned int, unsigned long);
486static int unix_shutdown(struct socket *, int); 493static int unix_shutdown(struct socket *, int);
487static int unix_stream_sendmsg(struct kiocb *, struct socket *, 494static int unix_stream_sendmsg(struct kiocb *, struct socket *,
@@ -527,7 +534,7 @@ static const struct proto_ops unix_dgram_ops = {
527 .socketpair = unix_socketpair, 534 .socketpair = unix_socketpair,
528 .accept = sock_no_accept, 535 .accept = sock_no_accept,
529 .getname = unix_getname, 536 .getname = unix_getname,
530 .poll = datagram_poll, 537 .poll = unix_dgram_poll,
531 .ioctl = unix_ioctl, 538 .ioctl = unix_ioctl,
532 .listen = sock_no_listen, 539 .listen = sock_no_listen,
533 .shutdown = unix_shutdown, 540 .shutdown = unix_shutdown,
@@ -548,7 +555,7 @@ static const struct proto_ops unix_seqpacket_ops = {
548 .socketpair = unix_socketpair, 555 .socketpair = unix_socketpair,
549 .accept = unix_accept, 556 .accept = unix_accept,
550 .getname = unix_getname, 557 .getname = unix_getname,
551 .poll = datagram_poll, 558 .poll = unix_dgram_poll,
552 .ioctl = unix_ioctl, 559 .ioctl = unix_ioctl,
553 .listen = unix_listen, 560 .listen = unix_listen,
554 .shutdown = unix_shutdown, 561 .shutdown = unix_shutdown,
@@ -983,8 +990,7 @@ static long unix_wait_for_peer(struct sock *other, long timeo)
983 990
984 sched = !sock_flag(other, SOCK_DEAD) && 991 sched = !sock_flag(other, SOCK_DEAD) &&
985 !(other->sk_shutdown & RCV_SHUTDOWN) && 992 !(other->sk_shutdown & RCV_SHUTDOWN) &&
986 (skb_queue_len(&other->sk_receive_queue) > 993 unix_recvq_full(other);
987 other->sk_max_ack_backlog);
988 994
989 unix_state_unlock(other); 995 unix_state_unlock(other);
990 996
@@ -1058,8 +1064,7 @@ restart:
1058 if (other->sk_state != TCP_LISTEN) 1064 if (other->sk_state != TCP_LISTEN)
1059 goto out_unlock; 1065 goto out_unlock;
1060 1066
1061 if (skb_queue_len(&other->sk_receive_queue) > 1067 if (unix_recvq_full(other)) {
1062 other->sk_max_ack_backlog) {
1063 err = -EAGAIN; 1068 err = -EAGAIN;
1064 if (!timeo) 1069 if (!timeo)
1065 goto out_unlock; 1070 goto out_unlock;
@@ -1428,9 +1433,7 @@ restart:
1428 goto out_unlock; 1433 goto out_unlock;
1429 } 1434 }
1430 1435
1431 if (unix_peer(other) != sk && 1436 if (unix_peer(other) != sk && unix_recvq_full(other)) {
1432 (skb_queue_len(&other->sk_receive_queue) >
1433 other->sk_max_ack_backlog)) {
1434 if (!timeo) { 1437 if (!timeo) {
1435 err = -EAGAIN; 1438 err = -EAGAIN;
1436 goto out_unlock; 1439 goto out_unlock;
@@ -1991,6 +1994,60 @@ static unsigned int unix_poll(struct file * file, struct socket *sock, poll_tabl
1991 return mask; 1994 return mask;
1992} 1995}
1993 1996
1997static unsigned int unix_dgram_poll(struct file *file, struct socket *sock,
1998 poll_table *wait)
1999{
2000 struct sock *sk = sock->sk, *other;
2001 unsigned int mask, writable;
2002
2003 poll_wait(file, sk->sk_sleep, wait);
2004 mask = 0;
2005
2006 /* exceptional events? */
2007 if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue))
2008 mask |= POLLERR;
2009 if (sk->sk_shutdown & RCV_SHUTDOWN)
2010 mask |= POLLRDHUP;
2011 if (sk->sk_shutdown == SHUTDOWN_MASK)
2012 mask |= POLLHUP;
2013
2014 /* readable? */
2015 if (!skb_queue_empty(&sk->sk_receive_queue) ||
2016 (sk->sk_shutdown & RCV_SHUTDOWN))
2017 mask |= POLLIN | POLLRDNORM;
2018
2019 /* Connection-based need to check for termination and startup */
2020 if (sk->sk_type == SOCK_SEQPACKET) {
2021 if (sk->sk_state == TCP_CLOSE)
2022 mask |= POLLHUP;
2023 /* connection hasn't started yet? */
2024 if (sk->sk_state == TCP_SYN_SENT)
2025 return mask;
2026 }
2027
2028 /* writable? */
2029 writable = unix_writable(sk);
2030 if (writable) {
2031 other = unix_peer_get(sk);
2032 if (other) {
2033 if (unix_peer(other) != sk) {
2034 poll_wait(file, &unix_sk(other)->peer_wait,
2035 wait);
2036 if (unix_recvq_full(other))
2037 writable = 0;
2038 }
2039
2040 sock_put(other);
2041 }
2042 }
2043
2044 if (writable)
2045 mask |= POLLOUT | POLLWRNORM | POLLWRBAND;
2046 else
2047 set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags);
2048
2049 return mask;
2050}
1994 2051
1995#ifdef CONFIG_PROC_FS 2052#ifdef CONFIG_PROC_FS
1996static struct sock *first_unix_socket(int *i) 2053static struct sock *first_unix_socket(int *i)
diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c
index 2bdd4dddc0e1..fb75f265b39c 100644
--- a/net/wireless/nl80211.c
+++ b/net/wireless/nl80211.c
@@ -187,7 +187,8 @@ static int nl80211_send_wiphy(struct sk_buff *msg, u32 pid, u32 seq, int flags,
187 return genlmsg_end(msg, hdr); 187 return genlmsg_end(msg, hdr);
188 188
189 nla_put_failure: 189 nla_put_failure:
190 return genlmsg_cancel(msg, hdr); 190 genlmsg_cancel(msg, hdr);
191 return -EMSGSIZE;
191} 192}
192 193
193static int nl80211_dump_wiphy(struct sk_buff *skb, struct netlink_callback *cb) 194static int nl80211_dump_wiphy(struct sk_buff *skb, struct netlink_callback *cb)
@@ -273,7 +274,8 @@ static int nl80211_send_iface(struct sk_buff *msg, u32 pid, u32 seq, int flags,
273 return genlmsg_end(msg, hdr); 274 return genlmsg_end(msg, hdr);
274 275
275 nla_put_failure: 276 nla_put_failure:
276 return genlmsg_cancel(msg, hdr); 277 genlmsg_cancel(msg, hdr);
278 return -EMSGSIZE;
277} 279}
278 280
279static int nl80211_dump_interface(struct sk_buff *skb, struct netlink_callback *cb) 281static int nl80211_dump_interface(struct sk_buff *skb, struct netlink_callback *cb)
@@ -928,7 +930,8 @@ static int nl80211_send_station(struct sk_buff *msg, u32 pid, u32 seq,
928 return genlmsg_end(msg, hdr); 930 return genlmsg_end(msg, hdr);
929 931
930 nla_put_failure: 932 nla_put_failure:
931 return genlmsg_cancel(msg, hdr); 933 genlmsg_cancel(msg, hdr);
934 return -EMSGSIZE;
932} 935}
933 936
934static int nl80211_dump_station(struct sk_buff *skb, 937static int nl80211_dump_station(struct sk_buff *skb,
@@ -1267,7 +1270,8 @@ static int nl80211_send_mpath(struct sk_buff *msg, u32 pid, u32 seq,
1267 return genlmsg_end(msg, hdr); 1270 return genlmsg_end(msg, hdr);
1268 1271
1269 nla_put_failure: 1272 nla_put_failure:
1270 return genlmsg_cancel(msg, hdr); 1273 genlmsg_cancel(msg, hdr);
1274 return -EMSGSIZE;
1271} 1275}
1272 1276
1273static int nl80211_dump_mpath(struct sk_buff *skb, 1277static int nl80211_dump_mpath(struct sk_buff *skb,
diff --git a/net/wireless/reg.c b/net/wireless/reg.c
index 185488da2466..855bff4b3250 100644
--- a/net/wireless/reg.c
+++ b/net/wireless/reg.c
@@ -80,6 +80,23 @@ static const struct ieee80211_channel_range ieee80211_JP_channels[] = {
80 IEEE80211_CHAN_RADAR), 80 IEEE80211_CHAN_RADAR),
81}; 81};
82 82
83static const struct ieee80211_channel_range ieee80211_EU_channels[] = {
84 /* IEEE 802.11b/g, channels 1..13 */
85 RANGE_PWR(2412, 2472, 20, 6, 0),
86 /* IEEE 802.11a, channel 36*/
87 RANGE_PWR(5180, 5180, 23, 6, IEEE80211_CHAN_PASSIVE_SCAN),
88 /* IEEE 802.11a, channel 40*/
89 RANGE_PWR(5200, 5200, 23, 6, IEEE80211_CHAN_PASSIVE_SCAN),
90 /* IEEE 802.11a, channel 44*/
91 RANGE_PWR(5220, 5220, 23, 6, IEEE80211_CHAN_PASSIVE_SCAN),
92 /* IEEE 802.11a, channels 48..64 */
93 RANGE_PWR(5240, 5320, 23, 6, IEEE80211_CHAN_NO_IBSS |
94 IEEE80211_CHAN_RADAR),
95 /* IEEE 802.11a, channels 100..140 */
96 RANGE_PWR(5500, 5700, 30, 6, IEEE80211_CHAN_NO_IBSS |
97 IEEE80211_CHAN_RADAR),
98};
99
83#define REGDOM(_code) \ 100#define REGDOM(_code) \
84 { \ 101 { \
85 .code = __stringify(_code), \ 102 .code = __stringify(_code), \
@@ -90,6 +107,7 @@ static const struct ieee80211_channel_range ieee80211_JP_channels[] = {
90static const struct ieee80211_regdomain ieee80211_regdoms[] = { 107static const struct ieee80211_regdomain ieee80211_regdoms[] = {
91 REGDOM(US), 108 REGDOM(US),
92 REGDOM(JP), 109 REGDOM(JP),
110 REGDOM(EU),
93}; 111};
94 112
95 113
diff --git a/net/xfrm/xfrm_algo.c b/net/xfrm/xfrm_algo.c
index ac765dd9c7f5..23a2cc04b8cd 100644
--- a/net/xfrm/xfrm_algo.c
+++ b/net/xfrm/xfrm_algo.c
@@ -200,8 +200,8 @@ static struct xfrm_algo_desc aalg_list[] = {
200 } 200 }
201}, 201},
202{ 202{
203 .name = "hmac(ripemd160)", 203 .name = "hmac(rmd160)",
204 .compat = "ripemd160", 204 .compat = "rmd160",
205 205
206 .uinfo = { 206 .uinfo = {
207 .auth = { 207 .auth = {
diff --git a/net/xfrm/xfrm_output.c b/net/xfrm/xfrm_output.c
index 09cd9c0c2d80..3f964db908a7 100644
--- a/net/xfrm/xfrm_output.c
+++ b/net/xfrm/xfrm_output.c
@@ -25,11 +25,11 @@ static int xfrm_state_check_space(struct xfrm_state *x, struct sk_buff *skb)
25 struct dst_entry *dst = skb->dst; 25 struct dst_entry *dst = skb->dst;
26 int nhead = dst->header_len + LL_RESERVED_SPACE(dst->dev) 26 int nhead = dst->header_len + LL_RESERVED_SPACE(dst->dev)
27 - skb_headroom(skb); 27 - skb_headroom(skb);
28 int ntail = dst->dev->needed_tailroom - skb_tailroom(skb);
28 29
29 if (nhead > 0) 30 if (nhead > 0 || ntail > 0)
30 return pskb_expand_head(skb, nhead, 0, GFP_ATOMIC); 31 return pskb_expand_head(skb, nhead, ntail, GFP_ATOMIC);
31 32
32 /* Check tail too... */
33 return 0; 33 return 0;
34} 34}
35 35
diff --git a/net/xfrm/xfrm_user.c b/net/xfrm/xfrm_user.c
index a1b0fbe3ea35..04c41504f84c 100644
--- a/net/xfrm/xfrm_user.c
+++ b/net/xfrm/xfrm_user.c
@@ -50,19 +50,8 @@ static int verify_one_alg(struct nlattr **attrs, enum xfrm_attr_type_t type)
50 50
51 switch (type) { 51 switch (type) {
52 case XFRMA_ALG_AUTH: 52 case XFRMA_ALG_AUTH:
53 if (!algp->alg_key_len &&
54 strcmp(algp->alg_name, "digest_null") != 0)
55 return -EINVAL;
56 break;
57
58 case XFRMA_ALG_CRYPT: 53 case XFRMA_ALG_CRYPT:
59 if (!algp->alg_key_len &&
60 strcmp(algp->alg_name, "cipher_null") != 0)
61 return -EINVAL;
62 break;
63
64 case XFRMA_ALG_COMP: 54 case XFRMA_ALG_COMP:
65 /* Zero length keys are legal. */
66 break; 55 break;
67 56
68 default: 57 default:
@@ -288,9 +277,8 @@ static void copy_from_user_state(struct xfrm_state *x, struct xfrm_usersa_info *
288 memcpy(&x->props.saddr, &p->saddr, sizeof(x->props.saddr)); 277 memcpy(&x->props.saddr, &p->saddr, sizeof(x->props.saddr));
289 x->props.flags = p->flags; 278 x->props.flags = p->flags;
290 279
291 if (!x->sel.family) 280 if (!x->sel.family && !(p->flags & XFRM_STATE_AF_UNSPEC))
292 x->sel.family = p->family; 281 x->sel.family = p->family;
293
294} 282}
295 283
296/* 284/*