aboutsummaryrefslogtreecommitdiffstats
path: root/net
diff options
context:
space:
mode:
Diffstat (limited to 'net')
-rw-r--r--net/8021q/vlan.c28
-rw-r--r--net/8021q/vlan_dev.c2
-rw-r--r--net/atm/br2684.c78
-rw-r--r--net/ax25/ax25_subr.c11
-rw-r--r--net/bluetooth/rfcomm/core.c2
-rw-r--r--net/bluetooth/rfcomm/tty.c13
-rw-r--r--net/core/dev.c52
-rw-r--r--net/core/fib_rules.c4
-rw-r--r--net/core/filter.c1
-rw-r--r--net/core/neighbour.c9
-rw-r--r--net/core/net_namespace.c3
-rw-r--r--net/core/pktgen.c4
-rw-r--r--net/core/rtnetlink.c3
-rw-r--r--net/core/skbuff.c22
-rw-r--r--net/core/user_dma.c2
-rw-r--r--net/dccp/ackvec.c29
-rw-r--r--net/dccp/ccids/ccid3.c27
-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/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/ipv4/arp.c5
-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/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/ipip.c130
-rw-r--r--net/ipv4/netfilter/nf_nat_core.c3
-rw-r--r--net/ipv4/netfilter/nf_nat_snmp_basic.c14
-rw-r--r--net/ipv4/raw.c11
-rw-r--r--net/ipv4/route.c4
-rw-r--r--net/ipv4/syncookies.c3
-rw-r--r--net/ipv4/tcp.c36
-rw-r--r--net/ipv4/tcp_input.c80
-rw-r--r--net/ipv4/tcp_ipv4.c20
-rw-r--r--net/ipv4/tcp_minisocks.c32
-rw-r--r--net/ipv4/tcp_output.c12
-rw-r--r--net/ipv4/tcp_timer.c5
-rw-r--r--net/ipv4/tunnel4.c2
-rw-r--r--net/ipv4/udp.c3
-rw-r--r--net/ipv4/xfrm4_mode_tunnel.c2
-rw-r--r--net/ipv6/addrconf.c142
-rw-r--r--net/ipv6/af_inet6.c2
-rw-r--r--net/ipv6/datagram.c50
-rw-r--r--net/ipv6/ip6_flowlabel.c2
-rw-r--r--net/ipv6/ip6_input.c9
-rw-r--r--net/ipv6/ip6mr.c2
-rw-r--r--net/ipv6/ipv6_sockglue.c44
-rw-r--r--net/ipv6/ndisc.c8
-rw-r--r--net/ipv6/netfilter/ip6table_mangle.c2
-rw-r--r--net/ipv6/netfilter/nf_conntrack_reasm.c9
-rw-r--r--net/ipv6/raw.c13
-rw-r--r--net/ipv6/reassembly.c2
-rw-r--r--net/ipv6/route.c32
-rw-r--r--net/ipv6/sit.c133
-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/key/af_key.c5
-rw-r--r--net/llc/llc_sap.c10
-rw-r--r--net/mac80211/cfg.c4
-rw-r--r--net/mac80211/ieee80211_i.h2
-rw-r--r--net/mac80211/key.c9
-rw-r--r--net/mac80211/main.c3
-rw-r--r--net/mac80211/mlme.c74
-rw-r--r--net/mac80211/rx.c4
-rw-r--r--net/mac80211/tx.c13
-rw-r--r--net/mac80211/util.c41
-rw-r--r--net/mac80211/wext.c35
-rw-r--r--net/mac80211/wme.c5
-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_proto_tcp.c13
-rw-r--r--net/netfilter/nf_log.c4
-rw-r--r--net/netfilter/xt_connlimit.c3
-rw-r--r--net/netlabel/netlabel_unlabeled.c2
-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/sched/Kconfig11
-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.c2
-rw-r--r--net/sched/sch_gred.c3
-rw-r--r--net/sched/sch_hfsc.c8
-rw-r--r--net/sched/sch_htb.c27
-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/socket.c4
-rw-r--r--net/sctp/transport.c50
-rw-r--r--net/sunrpc/auth_gss/svcauth_gss.c12
-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/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_user.c11
124 files changed, 1316 insertions, 1281 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/atm/br2684.c b/net/atm/br2684.c
index 9d52ebfc1962..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
@@ -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/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/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/core/dev.c b/net/core/dev.c
index a1607bc0cd4c..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
@@ -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);
@@ -1358,6 +1363,29 @@ void netif_device_attach(struct net_device *dev)
1358} 1363}
1359EXPORT_SYMBOL(netif_device_attach); 1364EXPORT_SYMBOL(netif_device_attach);
1360 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}
1361 1389
1362/* 1390/*
1363 * Invalidate hardware checksum when packet is to be mangled, and 1391 * Invalidate hardware checksum when packet is to be mangled, and
@@ -1636,14 +1664,8 @@ int dev_queue_xmit(struct sk_buff *skb)
1636 if (skb->ip_summed == CHECKSUM_PARTIAL) { 1664 if (skb->ip_summed == CHECKSUM_PARTIAL) {
1637 skb_set_transport_header(skb, skb->csum_start - 1665 skb_set_transport_header(skb, skb->csum_start -
1638 skb_headroom(skb)); 1666 skb_headroom(skb));
1639 1667 if (!dev_can_checksum(dev, skb) && skb_checksum_help(skb))
1640 if (!(dev->features & NETIF_F_GEN_CSUM) && 1668 goto out_kfree_skb;
1641 !((dev->features & NETIF_F_IP_CSUM) &&
1642 skb->protocol == htons(ETH_P_IP)) &&
1643 !((dev->features & NETIF_F_IPV6_CSUM) &&
1644 skb->protocol == htons(ETH_P_IPV6)))
1645 if (skb_checksum_help(skb))
1646 goto out_kfree_skb;
1647 } 1669 }
1648 1670
1649gso: 1671gso:
@@ -2055,6 +2077,10 @@ int netif_receive_skb(struct sk_buff *skb)
2055 2077
2056 rcu_read_lock(); 2078 rcu_read_lock();
2057 2079
2080 /* Don't receive packets in an exiting network namespace */
2081 if (!net_alive(dev_net(skb->dev)))
2082 goto out;
2083
2058#ifdef CONFIG_NET_CLS_ACT 2084#ifdef CONFIG_NET_CLS_ACT
2059 if (skb->tc_verd & TC_NCLS) { 2085 if (skb->tc_verd & TC_NCLS) {
2060 skb->tc_verd = CLR_TC_NCLS(skb->tc_verd); 2086 skb->tc_verd = CLR_TC_NCLS(skb->tc_verd);
@@ -2947,7 +2973,7 @@ EXPORT_SYMBOL(dev_unicast_delete);
2947/** 2973/**
2948 * dev_unicast_add - add a secondary unicast address 2974 * dev_unicast_add - add a secondary unicast address
2949 * @dev: device 2975 * @dev: device
2950 * @addr: address to delete 2976 * @addr: address to add
2951 * @alen: length of @addr 2977 * @alen: length of @addr
2952 * 2978 *
2953 * Add a secondary unicast address to the device or increase 2979 * Add a secondary unicast address to the device or increase
@@ -3137,7 +3163,7 @@ int dev_change_flags(struct net_device *dev, unsigned flags)
3137 * Load in the correct multicast list now the flags have changed. 3163 * Load in the correct multicast list now the flags have changed.
3138 */ 3164 */
3139 3165
3140 if (dev->change_rx_flags && (dev->flags ^ flags) & IFF_MULTICAST) 3166 if (dev->change_rx_flags && (old_flags ^ flags) & IFF_MULTICAST)
3141 dev->change_rx_flags(dev, IFF_MULTICAST); 3167 dev->change_rx_flags(dev, IFF_MULTICAST);
3142 3168
3143 dev_set_rx_mode(dev); 3169 dev_set_rx_mode(dev);
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 4f8369729a4e..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
diff --git a/net/core/neighbour.c b/net/core/neighbour.c
index 5d9d7130bd6e..65f01f71b3f3 100644
--- a/net/core/neighbour.c
+++ b/net/core/neighbour.c
@@ -1714,7 +1714,8 @@ static int neightbl_fill_parms(struct sk_buff *skb, struct neigh_parms *parms)
1714 return nla_nest_end(skb, nest); 1714 return nla_nest_end(skb, nest);
1715 1715
1716nla_put_failure: 1716nla_put_failure:
1717 return nla_nest_cancel(skb, nest); 1717 nla_nest_cancel(skb, nest);
1718 return -EMSGSIZE;
1718} 1719}
1719 1720
1720static 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,
@@ -2057,9 +2058,9 @@ static int neigh_fill_info(struct sk_buff *skb, struct neighbour *neigh,
2057 goto nla_put_failure; 2058 goto nla_put_failure;
2058 } 2059 }
2059 2060
2060 ci.ndm_used = now - neigh->used; 2061 ci.ndm_used = jiffies_to_clock_t(now - neigh->used);
2061 ci.ndm_confirmed = now - neigh->confirmed; 2062 ci.ndm_confirmed = jiffies_to_clock_t(now - neigh->confirmed);
2062 ci.ndm_updated = now - neigh->updated; 2063 ci.ndm_updated = jiffies_to_clock_t(now - neigh->updated);
2063 ci.ndm_refcnt = atomic_read(&neigh->refcnt) - 1; 2064 ci.ndm_refcnt = atomic_read(&neigh->refcnt) - 1;
2064 read_unlock_bh(&neigh->lock); 2065 read_unlock_bh(&neigh->lock);
2065 2066
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/pktgen.c b/net/core/pktgen.c
index 8dca21110493..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
@@ -3615,6 +3617,7 @@ static int __init pktgen_create_thread(int cpu)
3615 INIT_LIST_HEAD(&t->if_list); 3617 INIT_LIST_HEAD(&t->if_list);
3616 3618
3617 list_add_tail(&t->th_list, &pktgen_threads); 3619 list_add_tail(&t->th_list, &pktgen_threads);
3620 init_completion(&t->start_done);
3618 3621
3619 p = kthread_create(pktgen_thread_worker, t, "kpktgend_%d", cpu); 3622 p = kthread_create(pktgen_thread_worker, t, "kpktgend_%d", cpu);
3620 if (IS_ERR(p)) { 3623 if (IS_ERR(p)) {
@@ -3639,6 +3642,7 @@ static int __init pktgen_create_thread(int cpu)
3639 } 3642 }
3640 3643
3641 wake_up_process(p); 3644 wake_up_process(p);
3645 wait_for_completion(&t->start_done);
3642 3646
3643 return 0; 3647 return 0;
3644} 3648}
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 5c459f2b7985..366621610e76 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -1292,12 +1292,14 @@ static int __skb_splice_bits(struct sk_buff *skb, unsigned int *offset,
1292{ 1292{
1293 unsigned int nr_pages = spd->nr_pages; 1293 unsigned int nr_pages = spd->nr_pages;
1294 unsigned int poff, plen, len, toff, tlen; 1294 unsigned int poff, plen, len, toff, tlen;
1295 int headlen, seg; 1295 int headlen, seg, error = 0;
1296 1296
1297 toff = *offset; 1297 toff = *offset;
1298 tlen = *total_len; 1298 tlen = *total_len;
1299 if (!tlen) 1299 if (!tlen) {
1300 error = 1;
1300 goto err; 1301 goto err;
1302 }
1301 1303
1302 /* 1304 /*
1303 * 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
@@ -1339,7 +1341,8 @@ static int __skb_splice_bits(struct sk_buff *skb, unsigned int *offset,
1339 * just jump directly to update and return, no point 1341 * just jump directly to update and return, no point
1340 * in going over fragments when the output is full. 1342 * in going over fragments when the output is full.
1341 */ 1343 */
1342 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)
1343 goto done; 1346 goto done;
1344 1347
1345 tlen -= plen; 1348 tlen -= plen;
@@ -1369,7 +1372,8 @@ map_frag:
1369 if (!plen) 1372 if (!plen)
1370 break; 1373 break;
1371 1374
1372 if (spd_fill_page(spd, f->page, plen, poff, skb)) 1375 error = spd_fill_page(spd, f->page, plen, poff, skb);
1376 if (error)
1373 break; 1377 break;
1374 1378
1375 tlen -= plen; 1379 tlen -= plen;
@@ -1382,7 +1386,10 @@ done:
1382 return 0; 1386 return 0;
1383 } 1387 }
1384err: 1388err:
1385 return 1; 1389 /* update the offset to reflect the linear part skip, if any */
1390 if (!error)
1391 *offset = toff;
1392 return error;
1386} 1393}
1387 1394
1388/* 1395/*
@@ -1445,6 +1452,7 @@ done:
1445 1452
1446 if (spd.nr_pages) { 1453 if (spd.nr_pages) {
1447 int ret; 1454 int ret;
1455 struct sock *sk = __skb->sk;
1448 1456
1449 /* 1457 /*
1450 * Drop the socket lock, otherwise we have reverse 1458 * Drop the socket lock, otherwise we have reverse
@@ -1455,9 +1463,9 @@ done:
1455 * we call into ->sendpage() with the i_mutex lock held 1463 * we call into ->sendpage() with the i_mutex lock held
1456 * and networking will grab the socket lock. 1464 * and networking will grab the socket lock.
1457 */ 1465 */
1458 release_sock(__skb->sk); 1466 release_sock(sk);
1459 ret = splice_to_pipe(pipe, &spd); 1467 ret = splice_to_pipe(pipe, &spd);
1460 lock_sock(__skb->sk); 1468 lock_sock(sk);
1461 return ret; 1469 return ret;
1462 } 1470 }
1463 1471
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/ccid3.c b/net/dccp/ccids/ccid3.c
index cd61dea2eea1..a1929f33d703 100644
--- a/net/dccp/ccids/ccid3.c
+++ b/net/dccp/ccids/ccid3.c
@@ -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/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/ipv4/arp.c b/net/ipv4/arp.c
index 418862f1bf22..9b539fa9fe18 100644
--- a/net/ipv4/arp.c
+++ b/net/ipv4/arp.c
@@ -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/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/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/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/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..7750c97fde7b 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
@@ -434,6 +443,11 @@ static unsigned char asn1_oid_decode(struct asn1_ctx *ctx,
434 unsigned long *optr; 443 unsigned long *optr;
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 fead049daf43..37a1ecd9d600 100644
--- a/net/ipv4/raw.c
+++ b/net/ipv4/raw.c
@@ -608,6 +608,14 @@ static void raw_close(struct sock *sk, long timeout)
608 sk_common_release(sk); 608 sk_common_release(sk);
609} 609}
610 610
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
611/* This gets rid of all the nasties in af_inet. -DaveM */ 619/* This gets rid of all the nasties in af_inet. -DaveM */
612static 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)
613{ 621{
@@ -820,6 +828,7 @@ struct proto raw_prot = {
820 .name = "RAW", 828 .name = "RAW",
821 .owner = THIS_MODULE, 829 .owner = THIS_MODULE,
822 .close = raw_close, 830 .close = raw_close,
831 .destroy = raw_destroy,
823 .connect = ip4_datagram_connect, 832 .connect = ip4_datagram_connect,
824 .disconnect = udp_disconnect, 833 .disconnect = udp_disconnect,
825 .ioctl = raw_ioctl, 834 .ioctl = raw_ioctl,
@@ -925,7 +934,7 @@ static void raw_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
925 srcp = inet->num; 934 srcp = inet->num;
926 935
927 seq_printf(seq, "%4d: %08X:%04X %08X:%04X" 936 seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
928 " %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",
929 i, src, srcp, dest, destp, sp->sk_state, 938 i, src, srcp, dest, destp, sp->sk_state,
930 atomic_read(&sp->sk_wmem_alloc), 939 atomic_read(&sp->sk_wmem_alloc),
931 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 92f90ae46f4a..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};
@@ -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..850825dc86e6 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -260,6 +260,8 @@
260#include <linux/socket.h> 260#include <linux/socket.h>
261#include <linux/random.h> 261#include <linux/random.h>
262#include <linux/bootmem.h> 262#include <linux/bootmem.h>
263#include <linux/highmem.h>
264#include <linux/swap.h>
263#include <linux/cache.h> 265#include <linux/cache.h>
264#include <linux/err.h> 266#include <linux/err.h>
265#include <linux/crypto.h> 267#include <linux/crypto.h>
@@ -1227,7 +1229,14 @@ int tcp_read_sock(struct sock *sk, read_descriptor_t *desc,
1227 copied += used; 1229 copied += used;
1228 offset += used; 1230 offset += used;
1229 } 1231 }
1230 if (offset != skb->len) 1232 /*
1233 * If recv_actor drops the lock (e.g. TCP splice
1234 * receive) the skb pointer might be invalid when
1235 * getting here: tcp_collapse might have deleted it
1236 * while aggregating skbs from the socket queue.
1237 */
1238 skb = tcp_recv_skb(sk, seq-1, &offset);
1239 if (!skb || (offset+1 != skb->len))
1231 break; 1240 break;
1232 } 1241 }
1233 if (tcp_hdr(skb)->fin) { 1242 if (tcp_hdr(skb)->fin) {
@@ -2105,12 +2114,15 @@ static int do_tcp_setsockopt(struct sock *sk, int level,
2105 break; 2114 break;
2106 2115
2107 case TCP_DEFER_ACCEPT: 2116 case TCP_DEFER_ACCEPT:
2108 if (val < 0) { 2117 icsk->icsk_accept_queue.rskq_defer_accept = 0;
2109 err = -EINVAL; 2118 if (val > 0) {
2110 } else { 2119 /* Translate value in seconds to number of
2111 if (val > MAX_TCP_ACCEPT_DEFERRED) 2120 * retransmits */
2112 val = MAX_TCP_ACCEPT_DEFERRED; 2121 while (icsk->icsk_accept_queue.rskq_defer_accept < 32 &&
2113 icsk->icsk_accept_queue.rskq_defer_accept = val; 2122 val > ((TCP_TIMEOUT_INIT / HZ) <<
2123 icsk->icsk_accept_queue.rskq_defer_accept))
2124 icsk->icsk_accept_queue.rskq_defer_accept++;
2125 icsk->icsk_accept_queue.rskq_defer_accept++;
2114 } 2126 }
2115 break; 2127 break;
2116 2128
@@ -2292,7 +2304,8 @@ static int do_tcp_getsockopt(struct sock *sk, int level,
2292 val = (val ? : sysctl_tcp_fin_timeout) / HZ; 2304 val = (val ? : sysctl_tcp_fin_timeout) / HZ;
2293 break; 2305 break;
2294 case TCP_DEFER_ACCEPT: 2306 case TCP_DEFER_ACCEPT:
2295 val = icsk->icsk_accept_queue.rskq_defer_accept; 2307 val = !icsk->icsk_accept_queue.rskq_defer_accept ? 0 :
2308 ((TCP_TIMEOUT_INIT / HZ) << (icsk->icsk_accept_queue.rskq_defer_accept - 1));
2296 break; 2309 break;
2297 case TCP_WINDOW_CLAMP: 2310 case TCP_WINDOW_CLAMP:
2298 val = tp->window_clamp; 2311 val = tp->window_clamp;
@@ -2609,7 +2622,7 @@ __setup("thash_entries=", set_thash_entries);
2609void __init tcp_init(void) 2622void __init tcp_init(void)
2610{ 2623{
2611 struct sk_buff *skb = NULL; 2624 struct sk_buff *skb = NULL;
2612 unsigned long limit; 2625 unsigned long nr_pages, limit;
2613 int order, i, max_share; 2626 int order, i, max_share;
2614 2627
2615 BUILD_BUG_ON(sizeof(struct tcp_skb_cb) > sizeof(skb->cb)); 2628 BUILD_BUG_ON(sizeof(struct tcp_skb_cb) > sizeof(skb->cb));
@@ -2678,8 +2691,9 @@ void __init tcp_init(void)
2678 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of 2691 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of
2679 * memory, with a floor of 128 pages. 2692 * memory, with a floor of 128 pages.
2680 */ 2693 */
2681 limit = min(nr_all_pages, 1UL<<(28-PAGE_SHIFT)) >> (20-PAGE_SHIFT); 2694 nr_pages = totalram_pages - totalhigh_pages;
2682 limit = (limit * (nr_all_pages >> (20-PAGE_SHIFT))) >> (PAGE_SHIFT-11); 2695 limit = min(nr_pages, 1UL<<(28-PAGE_SHIFT)) >> (20-PAGE_SHIFT);
2696 limit = (limit * (nr_pages >> (20-PAGE_SHIFT))) >> (PAGE_SHIFT-11);
2683 limit = max(limit, 128UL); 2697 limit = max(limit, 128UL);
2684 sysctl_tcp_mem[0] = limit / 4 * 3; 2698 sysctl_tcp_mem[0] = limit / 4 * 3;
2685 sysctl_tcp_mem[1] = limit; 2699 sysctl_tcp_mem[1] = limit;
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index b54d9d37b636..cad73b7dfef0 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -1392,9 +1392,9 @@ static struct sk_buff *tcp_maybe_skipping_dsack(struct sk_buff *skb,
1392 1392
1393 if (before(next_dup->start_seq, skip_to_seq)) { 1393 if (before(next_dup->start_seq, skip_to_seq)) {
1394 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);
1395 tcp_sacktag_walk(skb, sk, NULL, 1395 skb = tcp_sacktag_walk(skb, sk, NULL,
1396 next_dup->start_seq, next_dup->end_seq, 1396 next_dup->start_seq, next_dup->end_seq,
1397 1, fack_count, reord, flag); 1397 1, fack_count, reord, flag);
1398 } 1398 }
1399 1399
1400 return skb; 1400 return skb;
@@ -2483,6 +2483,20 @@ static inline void tcp_complete_cwr(struct sock *sk)
2483 tcp_ca_event(sk, CA_EVENT_COMPLETE_CWR); 2483 tcp_ca_event(sk, CA_EVENT_COMPLETE_CWR);
2484} 2484}
2485 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
2486static void tcp_try_to_open(struct sock *sk, int flag) 2500static void tcp_try_to_open(struct sock *sk, int flag)
2487{ 2501{
2488 struct tcp_sock *tp = tcp_sk(sk); 2502 struct tcp_sock *tp = tcp_sk(sk);
@@ -2496,15 +2510,7 @@ static void tcp_try_to_open(struct sock *sk, int flag)
2496 tcp_enter_cwr(sk, 1); 2510 tcp_enter_cwr(sk, 1);
2497 2511
2498 if (inet_csk(sk)->icsk_ca_state != TCP_CA_CWR) { 2512 if (inet_csk(sk)->icsk_ca_state != TCP_CA_CWR) {
2499 int state = TCP_CA_Open; 2513 tcp_try_keep_open(sk);
2500
2501 if (tcp_left_out(tp) || tp->retrans_out || tp->undo_marker)
2502 state = TCP_CA_Disorder;
2503
2504 if (inet_csk(sk)->icsk_ca_state != state) {
2505 tcp_set_ca_state(sk, state);
2506 tp->high_seq = tp->snd_nxt;
2507 }
2508 tcp_moderate_cwnd(tp); 2514 tcp_moderate_cwnd(tp);
2509 } else { 2515 } else {
2510 tcp_cwnd_down(sk, flag); 2516 tcp_cwnd_down(sk, flag);
@@ -3310,8 +3316,11 @@ no_queue:
3310 return 1; 3316 return 1;
3311 3317
3312old_ack: 3318old_ack:
3313 if (TCP_SKB_CB(skb)->sacked) 3319 if (TCP_SKB_CB(skb)->sacked) {
3314 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 }
3315 3324
3316uninteresting_ack: 3325uninteresting_ack:
3317 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);
@@ -4532,49 +4541,6 @@ static void tcp_urg(struct sock *sk, struct sk_buff *skb, struct tcphdr *th)
4532 } 4541 }
4533} 4542}
4534 4543
4535static int tcp_defer_accept_check(struct sock *sk)
4536{
4537 struct tcp_sock *tp = tcp_sk(sk);
4538
4539 if (tp->defer_tcp_accept.request) {
4540 int queued_data = tp->rcv_nxt - tp->copied_seq;
4541 int hasfin = !skb_queue_empty(&sk->sk_receive_queue) ?
4542 tcp_hdr((struct sk_buff *)
4543 sk->sk_receive_queue.prev)->fin : 0;
4544
4545 if (queued_data && hasfin)
4546 queued_data--;
4547
4548 if (queued_data &&
4549 tp->defer_tcp_accept.listen_sk->sk_state == TCP_LISTEN) {
4550 if (sock_flag(sk, SOCK_KEEPOPEN)) {
4551 inet_csk_reset_keepalive_timer(sk,
4552 keepalive_time_when(tp));
4553 } else {
4554 inet_csk_delete_keepalive_timer(sk);
4555 }
4556
4557 inet_csk_reqsk_queue_add(
4558 tp->defer_tcp_accept.listen_sk,
4559 tp->defer_tcp_accept.request,
4560 sk);
4561
4562 tp->defer_tcp_accept.listen_sk->sk_data_ready(
4563 tp->defer_tcp_accept.listen_sk, 0);
4564
4565 sock_put(tp->defer_tcp_accept.listen_sk);
4566 sock_put(sk);
4567 tp->defer_tcp_accept.listen_sk = NULL;
4568 tp->defer_tcp_accept.request = NULL;
4569 } else if (hasfin ||
4570 tp->defer_tcp_accept.listen_sk->sk_state != TCP_LISTEN) {
4571 tcp_reset(sk);
4572 return -1;
4573 }
4574 }
4575 return 0;
4576}
4577
4578static 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)
4579{ 4545{
4580 struct tcp_sock *tp = tcp_sk(sk); 4546 struct tcp_sock *tp = tcp_sk(sk);
@@ -4935,8 +4901,6 @@ step5:
4935 4901
4936 tcp_data_snd_check(sk); 4902 tcp_data_snd_check(sk);
4937 tcp_ack_snd_check(sk); 4903 tcp_ack_snd_check(sk);
4938
4939 tcp_defer_accept_check(sk);
4940 return 0; 4904 return 0;
4941 4905
4942csum_error: 4906csum_error:
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index cd601a866c2f..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;
@@ -2303,7 +2291,7 @@ static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i, int *len)
2303 } 2291 }
2304 2292
2305 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 "
2306 "%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",
2307 i, src, srcp, dest, destp, sk->sk_state, 2295 i, src, srcp, dest, destp, sk->sk_state,
2308 tp->write_seq - tp->snd_una, 2296 tp->write_seq - tp->snd_una,
2309 sk->sk_state == TCP_LISTEN ? sk->sk_ack_backlog : 2297 sk->sk_state == TCP_LISTEN ? sk->sk_ack_backlog :
@@ -2315,8 +2303,8 @@ static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i, int *len)
2315 icsk->icsk_probes_out, 2303 icsk->icsk_probes_out,
2316 sock_i_ino(sk), 2304 sock_i_ino(sk),
2317 atomic_read(&sk->sk_refcnt), sk, 2305 atomic_read(&sk->sk_refcnt), sk,
2318 icsk->icsk_rto, 2306 jiffies_to_clock_t(icsk->icsk_rto),
2319 icsk->icsk_ack.ato, 2307 jiffies_to_clock_t(icsk->icsk_ack.ato),
2320 (icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong, 2308 (icsk->icsk_ack.quick << 1) | icsk->icsk_ack.pingpong,
2321 tp->snd_cwnd, 2309 tp->snd_cwnd,
2322 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_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/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 db1cb7c96d63..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
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..147588f4c7c0 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;
@@ -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/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/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/ndisc.c b/net/ipv6/ndisc.c
index a55fc05b8125..282fdb31f8ed 100644
--- a/net/ipv6/ndisc.c
+++ b/net/ipv6/ndisc.c
@@ -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/raw.c b/net/ipv6/raw.c
index 232e0dc45bf5..3aee12310d94 100644
--- a/net/ipv6/raw.c
+++ b/net/ipv6/raw.c
@@ -813,7 +813,7 @@ static int rawv6_sendmsg(struct kiocb *iocb, struct sock *sk,
813 memset(opt, 0, sizeof(struct ipv6_txoptions)); 813 memset(opt, 0, sizeof(struct ipv6_txoptions));
814 opt->tot_len = sizeof(struct ipv6_txoptions); 814 opt->tot_len = sizeof(struct ipv6_txoptions);
815 815
816 err = datagram_send_ctl(msg, &fl, opt, &hlimit, &tclass); 816 err = datagram_send_ctl(sock_net(sk), msg, &fl, opt, &hlimit, &tclass);
817 if (err < 0) { 817 if (err < 0) {
818 fl6_sock_release(flowlabel); 818 fl6_sock_release(flowlabel);
819 return err; 819 return err;
@@ -1164,6 +1164,15 @@ static void rawv6_close(struct sock *sk, long timeout)
1164 sk_common_release(sk); 1164 sk_common_release(sk);
1165} 1165}
1166 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
1167static int rawv6_init_sk(struct sock *sk) 1176static int rawv6_init_sk(struct sock *sk)
1168{ 1177{
1169 struct raw6_sock *rp = raw6_sk(sk); 1178 struct raw6_sock *rp = raw6_sk(sk);
@@ -1187,11 +1196,11 @@ struct proto rawv6_prot = {
1187 .name = "RAWv6", 1196 .name = "RAWv6",
1188 .owner = THIS_MODULE, 1197 .owner = THIS_MODULE,
1189 .close = rawv6_close, 1198 .close = rawv6_close,
1199 .destroy = raw6_destroy,
1190 .connect = ip6_datagram_connect, 1200 .connect = ip6_datagram_connect,
1191 .disconnect = udp_disconnect, 1201 .disconnect = udp_disconnect,
1192 .ioctl = rawv6_ioctl, 1202 .ioctl = rawv6_ioctl,
1193 .init = rawv6_init_sk, 1203 .init = rawv6_init_sk,
1194 .destroy = inet6_destroy_sock,
1195 .setsockopt = rawv6_setsockopt, 1204 .setsockopt = rawv6_setsockopt,
1196 .getsockopt = rawv6_getsockopt, 1205 .getsockopt = rawv6_getsockopt,
1197 .sendmsg = rawv6_sendmsg, 1206 .sendmsg = rawv6_sendmsg,
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 798cabc7535b..a60d7d129713 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -247,6 +247,8 @@ fq_find(struct net *net, __be32 id, struct in6_addr *src, struct in6_addr *dst,
247 arg.id = id; 247 arg.id = id;
248 arg.src = src; 248 arg.src = src;
249 arg.dst = dst; 249 arg.dst = dst;
250
251 read_lock(&ip6_frags.lock);
250 hash = ip6qhashfn(id, src, dst); 252 hash = ip6qhashfn(id, src, dst);
251 253
252 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 12bba0880345..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;
@@ -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 5a6fab95569f..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)
@@ -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/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/ieee80211_i.h b/net/mac80211/ieee80211_i.h
index c7314bf4bec2..006486b26726 100644
--- a/net/mac80211/ieee80211_i.h
+++ b/net/mac80211/ieee80211_i.h
@@ -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/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 915afadb0602..98c0b5e56ecc 100644
--- a/net/mac80211/main.c
+++ b/net/mac80211/main.c
@@ -511,6 +511,7 @@ static int ieee80211_stop(struct net_device *dev)
511 case IEEE80211_IF_TYPE_STA: 511 case IEEE80211_IF_TYPE_STA:
512 case IEEE80211_IF_TYPE_IBSS: 512 case IEEE80211_IF_TYPE_IBSS:
513 sdata->u.sta.state = IEEE80211_DISABLED; 513 sdata->u.sta.state = IEEE80211_DISABLED;
514 memset(sdata->u.sta.bssid, 0, ETH_ALEN);
514 del_timer_sync(&sdata->u.sta.timer); 515 del_timer_sync(&sdata->u.sta.timer);
515 /* 516 /*
516 * When we get here, the interface is marked down. 517 * When we get here, the interface is marked down.
@@ -1313,7 +1314,7 @@ static void ieee80211_handle_filtered_frame(struct ieee80211_local *local,
1313 /* 1314 /*
1314 * 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
1315 * 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
1316 * happen when it wakes up for the next time. 1317 * when it wakes up for the next time.
1317 */ 1318 */
1318 sta->flags |= WLAN_STA_CLEAR_PS_FILT; 1319 sta->flags |= WLAN_STA_CLEAR_PS_FILT;
1319 1320
diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
index 4adba09e80ca..4d2b582dd055 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)
@@ -730,7 +730,17 @@ static void ieee80211_send_assoc(struct net_device *dev,
730 if (bss->wmm_ie) { 730 if (bss->wmm_ie) {
731 wmm = 1; 731 wmm = 1;
732 } 732 }
733
734 /* get all rates supported by the device and the AP as
735 * some APs don't like getting a superset of their rates
736 * in the association request (e.g. D-Link DAP 1353 in
737 * b-only mode) */
738 rates_len = ieee80211_compatible_rates(bss, sband, &rates);
739
733 ieee80211_rx_bss_put(dev, bss); 740 ieee80211_rx_bss_put(dev, bss);
741 } else {
742 rates = ~0;
743 rates_len = sband->n_bitrates;
734 } 744 }
735 745
736 mgmt = (struct ieee80211_mgmt *) skb_put(skb, 24); 746 mgmt = (struct ieee80211_mgmt *) skb_put(skb, 24);
@@ -761,10 +771,7 @@ static void ieee80211_send_assoc(struct net_device *dev,
761 *pos++ = ifsta->ssid_len; 771 *pos++ = ifsta->ssid_len;
762 memcpy(pos, ifsta->ssid, ifsta->ssid_len); 772 memcpy(pos, ifsta->ssid, ifsta->ssid_len);
763 773
764 /* all supported rates should be added here but some APs 774 /* add all rates which were marked to be used above */
765 * (e.g. D-Link DAP 1353 in b-only mode) don't like that
766 * Therefore only add rates the AP supports */
767 rates_len = ieee80211_compatible_rates(bss, sband, &rates);
768 supp_rates_len = rates_len; 775 supp_rates_len = rates_len;
769 if (supp_rates_len > 8) 776 if (supp_rates_len > 8)
770 supp_rates_len = 8; 777 supp_rates_len = 8;
@@ -1318,7 +1325,7 @@ static void ieee80211_sta_process_addba_request(struct net_device *dev,
1318 1325
1319 /* prepare reordering buffer */ 1326 /* prepare reordering buffer */
1320 tid_agg_rx->reorder_buf = 1327 tid_agg_rx->reorder_buf =
1321 kmalloc(buf_size * sizeof(struct sk_buf *), GFP_ATOMIC); 1328 kmalloc(buf_size * sizeof(struct sk_buff *), GFP_ATOMIC);
1322 if (!tid_agg_rx->reorder_buf) { 1329 if (!tid_agg_rx->reorder_buf) {
1323 if (net_ratelimit()) 1330 if (net_ratelimit())
1324 printk(KERN_ERR "can not allocate reordering buffer " 1331 printk(KERN_ERR "can not allocate reordering buffer "
@@ -1327,7 +1334,7 @@ static void ieee80211_sta_process_addba_request(struct net_device *dev,
1327 goto end; 1334 goto end;
1328 } 1335 }
1329 memset(tid_agg_rx->reorder_buf, 0, 1336 memset(tid_agg_rx->reorder_buf, 0,
1330 buf_size * sizeof(struct sk_buf *)); 1337 buf_size * sizeof(struct sk_buff *));
1331 1338
1332 if (local->ops->ampdu_action) 1339 if (local->ops->ampdu_action)
1333 ret = local->ops->ampdu_action(hw, IEEE80211_AMPDU_RX_START, 1340 ret = local->ops->ampdu_action(hw, IEEE80211_AMPDU_RX_START,
@@ -1607,7 +1614,7 @@ void sta_addba_resp_timer_expired(unsigned long data)
1607 * only one argument, and both sta_info and TID are needed, so init 1614 * only one argument, and both sta_info and TID are needed, so init
1608 * flow in sta_info_create gives the TID as data, while the timer_to_id 1615 * flow in sta_info_create gives the TID as data, while the timer_to_id
1609 * array gives the sta through container_of */ 1616 * array gives the sta through container_of */
1610 u16 tid = *(int *)data; 1617 u16 tid = *(u8 *)data;
1611 struct sta_info *temp_sta = container_of((void *)data, 1618 struct sta_info *temp_sta = container_of((void *)data,
1612 struct sta_info, timer_to_tid[tid]); 1619 struct sta_info, timer_to_tid[tid]);
1613 1620
@@ -1655,7 +1662,7 @@ timer_expired_exit:
1655void sta_rx_agg_session_timer_expired(unsigned long data) 1662void sta_rx_agg_session_timer_expired(unsigned long data)
1656{ 1663{
1657 /* not an elegant detour, but there is no choice as the timer passes 1664 /* not an elegant detour, but there is no choice as the timer passes
1658 * only one argument, and verious sta_info are needed here, so init 1665 * only one argument, and various sta_info are needed here, so init
1659 * flow in sta_info_create gives the TID as data, while the timer_to_id 1666 * flow in sta_info_create gives the TID as data, while the timer_to_id
1660 * array gives the sta through container_of */ 1667 * array gives the sta through container_of */
1661 u8 *ptid = (u8 *)data; 1668 u8 *ptid = (u8 *)data;
@@ -2329,6 +2336,7 @@ static int ieee80211_sta_join_ibss(struct net_device *dev,
2329 u8 *pos; 2336 u8 *pos;
2330 struct ieee80211_sub_if_data *sdata; 2337 struct ieee80211_sub_if_data *sdata;
2331 struct ieee80211_supported_band *sband; 2338 struct ieee80211_supported_band *sband;
2339 union iwreq_data wrqu;
2332 2340
2333 sband = local->hw.wiphy->bands[local->hw.conf.channel->band]; 2341 sband = local->hw.wiphy->bands[local->hw.conf.channel->band];
2334 2342
@@ -2351,13 +2359,10 @@ static int ieee80211_sta_join_ibss(struct net_device *dev,
2351 sdata->drop_unencrypted = bss->capability & 2359 sdata->drop_unencrypted = bss->capability &
2352 WLAN_CAPABILITY_PRIVACY ? 1 : 0; 2360 WLAN_CAPABILITY_PRIVACY ? 1 : 0;
2353 2361
2354 res = ieee80211_set_freq(local, bss->freq); 2362 res = ieee80211_set_freq(dev, bss->freq);
2355 2363
2356 if (local->oper_channel->flags & IEEE80211_CHAN_NO_IBSS) { 2364 if (res)
2357 printk(KERN_DEBUG "%s: IBSS not allowed on frequency " 2365 return res;
2358 "%d MHz\n", dev->name, local->oper_channel->center_freq);
2359 return -1;
2360 }
2361 2366
2362 /* Set beacon template */ 2367 /* Set beacon template */
2363 skb = dev_alloc_skb(local->hw.extra_tx_headroom + 400); 2368 skb = dev_alloc_skb(local->hw.extra_tx_headroom + 400);
@@ -2472,7 +2477,9 @@ static int ieee80211_sta_join_ibss(struct net_device *dev,
2472 ifsta->state = IEEE80211_IBSS_JOINED; 2477 ifsta->state = IEEE80211_IBSS_JOINED;
2473 mod_timer(&ifsta->timer, jiffies + IEEE80211_IBSS_MERGE_INTERVAL); 2478 mod_timer(&ifsta->timer, jiffies + IEEE80211_IBSS_MERGE_INTERVAL);
2474 2479
2475 ieee80211_rx_bss_put(dev, bss); 2480 memset(&wrqu, 0, sizeof(wrqu));
2481 memcpy(wrqu.ap_addr.sa_data, bss->bssid, ETH_ALEN);
2482 wireless_send_event(dev, SIOCGIWAP, &wrqu, NULL);
2476 2483
2477 return res; 2484 return res;
2478} 2485}
@@ -3446,21 +3453,17 @@ static int ieee80211_sta_config_auth(struct net_device *dev,
3446 struct ieee80211_sta_bss *bss, *selected = NULL; 3453 struct ieee80211_sta_bss *bss, *selected = NULL;
3447 int top_rssi = 0, freq; 3454 int top_rssi = 0, freq;
3448 3455
3449 if (!(ifsta->flags & (IEEE80211_STA_AUTO_SSID_SEL |
3450 IEEE80211_STA_AUTO_BSSID_SEL | IEEE80211_STA_AUTO_CHANNEL_SEL))) {
3451 ifsta->state = IEEE80211_AUTHENTICATE;
3452 ieee80211_sta_reset_auth(dev, ifsta);
3453 return 0;
3454 }
3455
3456 spin_lock_bh(&local->sta_bss_lock); 3456 spin_lock_bh(&local->sta_bss_lock);
3457 freq = local->oper_channel->center_freq; 3457 freq = local->oper_channel->center_freq;
3458 list_for_each_entry(bss, &local->sta_bss_list, list) { 3458 list_for_each_entry(bss, &local->sta_bss_list, list) {
3459 if (!(bss->capability & WLAN_CAPABILITY_ESS)) 3459 if (!(bss->capability & WLAN_CAPABILITY_ESS))
3460 continue; 3460 continue;
3461 3461
3462 if (!!(bss->capability & WLAN_CAPABILITY_PRIVACY) ^ 3462 if ((ifsta->flags & (IEEE80211_STA_AUTO_SSID_SEL |
3463 !!sdata->default_key) 3463 IEEE80211_STA_AUTO_BSSID_SEL |
3464 IEEE80211_STA_AUTO_CHANNEL_SEL)) &&
3465 (!!(bss->capability & WLAN_CAPABILITY_PRIVACY) ^
3466 !!sdata->default_key))
3464 continue; 3467 continue;
3465 3468
3466 if (!(ifsta->flags & IEEE80211_STA_AUTO_CHANNEL_SEL) && 3469 if (!(ifsta->flags & IEEE80211_STA_AUTO_CHANNEL_SEL) &&
@@ -3485,7 +3488,7 @@ static int ieee80211_sta_config_auth(struct net_device *dev,
3485 spin_unlock_bh(&local->sta_bss_lock); 3488 spin_unlock_bh(&local->sta_bss_lock);
3486 3489
3487 if (selected) { 3490 if (selected) {
3488 ieee80211_set_freq(local, selected->freq); 3491 ieee80211_set_freq(dev, selected->freq);
3489 if (!(ifsta->flags & IEEE80211_STA_SSID_SET)) 3492 if (!(ifsta->flags & IEEE80211_STA_SSID_SET))
3490 ieee80211_sta_set_ssid(dev, selected->ssid, 3493 ieee80211_sta_set_ssid(dev, selected->ssid,
3491 selected->ssid_len); 3494 selected->ssid_len);
@@ -3520,6 +3523,7 @@ static int ieee80211_sta_create_ibss(struct net_device *dev,
3520 struct ieee80211_supported_band *sband; 3523 struct ieee80211_supported_band *sband;
3521 u8 bssid[ETH_ALEN], *pos; 3524 u8 bssid[ETH_ALEN], *pos;
3522 int i; 3525 int i;
3526 int ret;
3523 DECLARE_MAC_BUF(mac); 3527 DECLARE_MAC_BUF(mac);
3524 3528
3525#if 0 3529#if 0
@@ -3564,7 +3568,9 @@ static int ieee80211_sta_create_ibss(struct net_device *dev,
3564 *pos++ = (u8) (rate / 5); 3568 *pos++ = (u8) (rate / 5);
3565 } 3569 }
3566 3570
3567 return ieee80211_sta_join_ibss(dev, ifsta, bss); 3571 ret = ieee80211_sta_join_ibss(dev, ifsta, bss);
3572 ieee80211_rx_bss_put(dev, bss);
3573 return ret;
3568} 3574}
3569 3575
3570 3576
@@ -3612,10 +3618,13 @@ static int ieee80211_sta_find_ibss(struct net_device *dev,
3612 (bss = ieee80211_rx_bss_get(dev, bssid, 3618 (bss = ieee80211_rx_bss_get(dev, bssid,
3613 local->hw.conf.channel->center_freq, 3619 local->hw.conf.channel->center_freq,
3614 ifsta->ssid, ifsta->ssid_len))) { 3620 ifsta->ssid, ifsta->ssid_len))) {
3621 int ret;
3615 printk(KERN_DEBUG "%s: Selected IBSS BSSID %s" 3622 printk(KERN_DEBUG "%s: Selected IBSS BSSID %s"
3616 " based on configured SSID\n", 3623 " based on configured SSID\n",
3617 dev->name, print_mac(mac, bssid)); 3624 dev->name, print_mac(mac, bssid));
3618 return ieee80211_sta_join_ibss(dev, ifsta, bss); 3625 ret = ieee80211_sta_join_ibss(dev, ifsta, bss);
3626 ieee80211_rx_bss_put(dev, bss);
3627 return ret;
3619 } 3628 }
3620#ifdef CONFIG_MAC80211_IBSS_DEBUG 3629#ifdef CONFIG_MAC80211_IBSS_DEBUG
3621 printk(KERN_DEBUG " did not try to join ibss\n"); 3630 printk(KERN_DEBUG " did not try to join ibss\n");
@@ -4092,18 +4101,17 @@ ieee80211_sta_scan_result(struct net_device *dev,
4092 4101
4093 memset(&iwe, 0, sizeof(iwe)); 4102 memset(&iwe, 0, sizeof(iwe));
4094 iwe.cmd = SIOCGIWFREQ; 4103 iwe.cmd = SIOCGIWFREQ;
4095 iwe.u.freq.m = bss->freq; 4104 iwe.u.freq.m = ieee80211_frequency_to_channel(bss->freq);
4096 iwe.u.freq.e = 6; 4105 iwe.u.freq.e = 0;
4097 current_ev = iwe_stream_add_event(current_ev, end_buf, &iwe, 4106 current_ev = iwe_stream_add_event(current_ev, end_buf, &iwe,
4098 IW_EV_FREQ_LEN); 4107 IW_EV_FREQ_LEN);
4099 4108
4100 memset(&iwe, 0, sizeof(iwe)); 4109 memset(&iwe, 0, sizeof(iwe));
4101 iwe.cmd = SIOCGIWFREQ; 4110 iwe.cmd = SIOCGIWFREQ;
4102 iwe.u.freq.m = ieee80211_frequency_to_channel(bss->freq); 4111 iwe.u.freq.m = bss->freq;
4103 iwe.u.freq.e = 0; 4112 iwe.u.freq.e = 6;
4104 current_ev = iwe_stream_add_event(current_ev, end_buf, &iwe, 4113 current_ev = iwe_stream_add_event(current_ev, end_buf, &iwe,
4105 IW_EV_FREQ_LEN); 4114 IW_EV_FREQ_LEN);
4106
4107 memset(&iwe, 0, sizeof(iwe)); 4115 memset(&iwe, 0, sizeof(iwe));
4108 iwe.cmd = IWEVQUAL; 4116 iwe.cmd = IWEVQUAL;
4109 iwe.u.qual.qual = bss->signal; 4117 iwe.u.qual.qual = bss->signal;
diff --git a/net/mac80211/rx.c b/net/mac80211/rx.c
index 1958bfb361c6..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
diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
index 1d7dd54aacef..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,
@@ -1562,13 +1569,13 @@ int ieee80211_subif_start_xmit(struct sk_buff *skb,
1562 * be cloned. This could happen, e.g., with Linux bridge code passing 1569 * be cloned. This could happen, e.g., with Linux bridge code passing
1563 * us broadcast frames. */ 1570 * us broadcast frames. */
1564 1571
1565 if (head_need > 0 || skb_header_cloned(skb)) { 1572 if (head_need > 0 || skb_cloned(skb)) {
1566#if 0 1573#if 0
1567 printk(KERN_DEBUG "%s: need to reallocate buffer for %d bytes " 1574 printk(KERN_DEBUG "%s: need to reallocate buffer for %d bytes "
1568 "of headroom\n", dev->name, head_need); 1575 "of headroom\n", dev->name, head_need);
1569#endif 1576#endif
1570 1577
1571 if (skb_header_cloned(skb)) 1578 if (skb_cloned(skb))
1572 I802_DEBUG_INC(local->tx_expand_skb_head_cloned); 1579 I802_DEBUG_INC(local->tx_expand_skb_head_cloned);
1573 else 1580 else
1574 I802_DEBUG_INC(local->tx_expand_skb_head); 1581 I802_DEBUG_INC(local->tx_expand_skb_head);
diff --git a/net/mac80211/util.c b/net/mac80211/util.c
index 24a465c4df09..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
@@ -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 dc1598b86004..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]);
@@ -673,7 +672,7 @@ int ieee80211_ht_agg_queue_add(struct ieee80211_local *local,
673#ifdef CONFIG_MAC80211_HT_DEBUG 672#ifdef CONFIG_MAC80211_HT_DEBUG
674 if (net_ratelimit()) 673 if (net_ratelimit())
675 printk(KERN_DEBUG "allocated aggregation queue" 674 printk(KERN_DEBUG "allocated aggregation queue"
676 " %d tid %d addr %s pool=0x%lX", 675 " %d tid %d addr %s pool=0x%lX\n",
677 i, tid, print_mac(mac, sta->addr), 676 i, tid, print_mac(mac, sta->addr),
678 q->qdisc_pool[0]); 677 q->qdisc_pool[0]);
679#endif /* CONFIG_MAC80211_HT_DEBUG */ 678#endif /* CONFIG_MAC80211_HT_DEBUG */
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_proto_tcp.c b/net/netfilter/nf_conntrack_proto_tcp.c
index ba94004fe323..271cd01d57ae 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 */
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/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/netlabel/netlabel_unlabeled.c b/net/netlabel/netlabel_unlabeled.c
index 0099da5b2591..52b2611a6eb6 100644
--- a/net/netlabel/netlabel_unlabeled.c
+++ b/net/netlabel/netlabel_unlabeled.c
@@ -1534,7 +1534,7 @@ static int netlbl_unlabel_staticlistdef(struct sk_buff *skb,
1534 } 1534 }
1535 } 1535 }
1536 list_for_each_entry_rcu(addr6, &iface->addr6_list, list) { 1536 list_for_each_entry_rcu(addr6, &iface->addr6_list, list) {
1537 if (addr6->valid || iter_addr6++ < skip_addr6) 1537 if (!addr6->valid || iter_addr6++ < skip_addr6)
1538 continue; 1538 continue;
1539 if (netlbl_unlabel_staticlist_gen(NLBL_UNLABEL_C_STATICLISTDEF, 1539 if (netlbl_unlabel_staticlist_gen(NLBL_UNLABEL_C_STATICLISTDEF,
1540 iface, 1540 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/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/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 d355e5e47fe3..13afa7214392 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -468,7 +468,7 @@ struct Qdisc *qdisc_alloc(struct net_device *dev, struct Qdisc_ops *ops)
468 468
469 return sch; 469 return sch;
470errout: 470errout:
471 return ERR_PTR(-err); 471 return ERR_PTR(err);
472} 472}
473 473
474struct 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 5bc1ed490180..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
@@ -1231,7 +1238,7 @@ static void htb_destroy_class(struct Qdisc *sch, struct htb_class *cl)
1231 qdisc_put_rtab(cl->rate); 1238 qdisc_put_rtab(cl->rate);
1232 qdisc_put_rtab(cl->ceil); 1239 qdisc_put_rtab(cl->ceil);
1233 1240
1234 tcf_destroy_chain(cl->filter_list); 1241 tcf_destroy_chain(&cl->filter_list);
1235 1242
1236 while (!list_empty(&cl->children)) 1243 while (!list_empty(&cl->children))
1237 htb_destroy_class(sch, list_entry(cl->children.next, 1244 htb_destroy_class(sch, list_entry(cl->children.next,
@@ -1260,7 +1267,7 @@ static void htb_destroy(struct Qdisc *sch)
1260 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
1261 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
1262 unbind_filter on it (without Oops). */ 1269 unbind_filter on it (without Oops). */
1263 tcf_destroy_chain(q->filter_list); 1270 tcf_destroy_chain(&q->filter_list);
1264 1271
1265 while (!list_empty(&q->root)) 1272 while (!list_empty(&q->root))
1266 htb_destroy_class(sch, list_entry(q->root.next, 1273 htb_destroy_class(sch, list_entry(q->root.next,
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/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/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/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/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_user.c b/net/xfrm/xfrm_user.c
index a1b0fbe3ea35..b976d9ed10e4 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: